The Evolution of Senescence in the Tree of Life 9781107078505, 1107078504

Citation preview

1

Introduction: Wilting Leaves and Rotting Branches Reconciling Evolutionary Perspectives on Senescence Richard P. Shefferson, Owen R. Jones and Roberto Salguero-Gómez

It don’t make no difference ‘cos I ain’t gonna be, easy, easy, the only time I’m gonna be easy’s when I’m Killed by death – Motörhead, Killed By Death (1984)

Short Summary We humans have long wondered about the seemingly inevitable physiological decline that happens after our maturity. This phenomenon, known as ‘senescence’, is recognised as the physiological deterioration that results in increasing age-specific mortality or decreasing age-specific fertility at or beyond some age in maturity. But is this phenomenon universal, and is it always the result of the same processes? Although evolutionary theories of ageing exist that suggest that senescence should be universal, empirical data increasingly suggest that senescence may not necessarily be a ubiquitous feature among multicellular organisms, and where it does occur, it is not clear that it is always the result of the same ultimate or proximate mechanisms. In this contributed book, some of the leading scientists in ageing research offer an in-depth, updated understanding of the mechanisms behind senescence, using cutting-edge approaches and species representing a wide evolutionary diversity of life.

Introduction We are all aware of the march of time in our lives. We are born, we grow, we (may) reproduce, we advance in age and we die. So enveloping is our awareness of this phenomenon that, at least in the English language, we do not differentiate between ageing as the progress of time (i.e. chronological ageing) and senescence, or the decline in physiological well-being that begins at some point at or beyond our age at reproductive maturity. In this sense, it is often assumed that senescence is such an unshakeable Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:08:00, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.001

2

Introduction

part of ageing that it is universal – no organism can escape from it, except by dying prematurely from an exogenous factor (e.g. a wild fire, a ferocious predator). But is senescence truly universal? While ageing is certainly universal in the sense that we cannot escape the march of time, senescence itself is probably not. The test for the universality of senescence is fairly straightforward: there must not exist a single species in which physiological decline with age cannot be documented. A species that escapes from senescence would exhibit the lack of such decline, referred to as ‘negligible senescence’ (Vaupel et al. 2004), or perhaps even physiological improvement with age (i.e. ‘negative senescence’, per Vaupel et al. 2004). Recent analyses suggest that such species exist (Baudisch et al. 2013; Garcia et al. 2011; Jones et al. 2014). However, over much of its history, the study of senescence has been strongly taxonomically biased, focusing heavily on humans, other mammals and birds, and a small number of model organisms including Drosophila spp. and Caenorhabditis elegans, among others. This book represents an overdue attempt to examine the evolutionary consequences and mechanisms of senescence in organisms across the Tree of Life and to assess whether senescence is truly universal and why. We paint an evolutionarily broad picture in order to relate what we know to the world’s biodiversity. This is a long-overdue attempt because, since the first evolutionary theory of senescence was advanced by August Weismann over a century ago (Weismann 1892, 1893), we have only recently managed to accumulate sufficient genetic, physiological and demographic data on a diverse enough group of organisms to begin to address this topic rigorously. Biologists studying senescence in different groups of organisms have tended to use disparate methods, met at sub-discipline-specific scientific meetings, and have established their own distinct traditions and terminology. We have attempted to bring these traditions together to look for common trends and answers. For this purpose, this book is divided into a first part focused on the general evolutionary theory of ageing, followed by parts focused on animals (naturally, including humans), plants and microbes. We also keep to some common definitions throughout the book. Most importantly, we and all contributing authors strictly define ‘senescence’ as the process of physiological or biological decay leading to increasing mortality rates and/or decreasing fertility rates with age; we distinguish this from ‘ageing’, which here is viewed simply as the march of time, with no physiological decay implied.

A Short History of the Senescing Universe The modern evolutionary theory of senescence is rooted in the late nineteenth century. August Weismann postulated that an organism’s cell lines can be decomposed into a germ line and a soma (Weismann 1892, 1893). The germ line constitutes the cells responsible for the organism’s reproduction. The soma constitutes cell lineages created to keep the germ line alive and reproducing. While all cells are ‘born’ and die, the line of germ cells is potentially immortal, and somatic lineages always die when the organism itself dies – but we do note that the picture gets rather quickly complicated in organisms with clonal abilities (see Chapters 11, 13, and 15 to 17). The corollary of Weismann’s Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:08:00, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.001

A Short History of the Senescing Universe

3

work is that the soma exists to maintain the germ line and that the soma is essentially disposable (Kirkwood 1977). In essence, Weismann suggested that the physiological performance of the soma declines with age by interacting with the environment and buffering the germ line from it. Remarkably, he even briefly supposed that death itself was adaptive because, via death, senescing individuals would alleviate competition for younger individuals with greater reproductive potential, although he quickly abandoned this idea, recognising its inherent flaws. Weismann’s germ-soma theory was profoundly influential and continues to inspire researchers to this day. However, his theory predicts only that senescence will occur in all organisms with a strict germ/soma separation and so cannot account for senescence observed in unicellular life, plants, fungi, some animals such as corals, and many microbes. Comfort (1964) even criticised the theory as being somewhat circular for assuming what it is supposed to explain and therefore not really explaining why senescence may evolve in the first place. In addition, Comfort argued that it does not truly relate any particular genetic, physiological or demographic processes to fitness. Although more mechanical, physiological views of senescence most certainly exist (Comfort 1964), some have suggested that the ways in which senescence is thought to evolve preclude any common mechanism of decline (Silvertown 2013). The mathematical foundations for a synthetic evolutionary theory of senescence were laid by Ronald Fisher. He argued that senescence was likely a result of the accumulation of deleterious age-specific traits that could not be effectively removed by natural selection. The strength of natural selection was related to the reproductive value of a particular age, the remaining number of offspring an individual could expect to produce before death (Fisher 1930). John B. S. Haldane had a similar notion about a decade later (Haldane 1941), but it was Peter Medawar (1952) who succinctly emphasised the consequences for senescence in his essay ‘An Unsolved Problem of Biology’, where he remarked that, after sexual maturity, ‘[t]he force of natural selection weakens with increasing age – even in a theoretically immortal population, provided only that it is exposed to real hazards of mortality. If a genetic disaster . . . happens late enough in individual life, its consequences may be completely unimportant.’ George Williams extended this argument by pointing out that in addition to the accumulation of deleterious mutations late in life, senescence could also be caused by pleiotropic genetic effects that are positive in early life but negative in late life (Medawar 1952; Williams 1957; see review in Ljubuncic & Reznick 2009). These mechanisms for the evolution of senescence – mutation accumulation and antagonistic pleiotropy – are also potentially universal, but at the time that they were developed, once again, little to no data actually existed on senescence outside humans. Medawar and others even assumed that animals could not live long enough in the wild to experience senescence (Medawar 1952), an assumption that we know now to be false (Jones et al. 2008;Nussey et al. 2013; see also Chapter 7). Until relatively recently, senescence was viewed as a universal phenomenon, and for senescence to be truly universal, an evolutionary mechanism for its universality must explain genetic, physiological and demographic patterns throughout the life span in all Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:08:00, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.001

4

Introduction

lineages across the tree of life. In 1966, the Journal of Theoretical Biology published a paper in which William Hamilton laid out the mathematics behind a theory, drawing on the ideas of Williams and Medawar, leading to the outcome that ‘senescence is an inevitable outcome of evolution’ (Hamilton 1966). Though others had certainly had a large impact on the development of the field, this was the first seemingly testable theory to explain the evolution of ageing as a universal. Fortunately, it was developed at a time when technological advances (Holliday 1990; Medina 2005; Rabinow 1997) facilitated studies of the underlying genetics (e.g. Hoffman et al. 2014; Moorad & Promislow 2011) and physiology of senescence (e.g. Kumar et al. 2012). It was also done at a time when genetic perspectives began to revolutionise evolutionary biology (Dawkins 1978). Therefore, beginning around the start of the 1970s, senescence research truly started broadening its evolutionary scope. Since Hamilton’s work, the most prominent theory for the evolution of senescence has been Thomas Kirkwood’s ‘disposable soma theory’ (Kirkwood 1977; see also Chapter 2). According to this theory, an organism’s cells accumulate deleterious mutations with age, and repairing this damage is costly. Since the soma functions merely as a vessel for ensuring the replication/continuation of the immortal germ line, mutations in soma cells are less important to repair than those in the germ line, which are carefully protected to ensure accurate replication from generation to generation. Therefore, his theory suggests that the evolutionary stable strategy (ESS) of the trade-off between resource allocation to processes of somatic maintenance and maintenance of the germ line inescapably favours the germ line, resulting in senescence. It is noteworthy that these evolutionary theories were developed seemingly with mostly mammals and birds in mind. Other organisms have presented more perplexing mysteries for the development of a universal evolutionary theory of senescence, particularly long-lived plants such as the redwood but also fungi and other microbes. Indeed, although Hamilton considered mortality patterns in extremely long-lived plants (Hamilton 1966), it is doubtful that he accessed the data available on some longerlived tree species that contemporary evolutionary plant ecologists were working with (Harper 1967; Harper & White 1974). Although some of the most important work in evolutionary biology historically has been done on plants (Clausen et al. 1947), plant life histories themselves were not the subject of truly rigorous study until John Harper invigorated the subject with his classic book that inspired a generation of plant ecologists (Harper 1977). From then on, we have witnessed the gradual development of the field of plant senescence biology (e.g. Roach 1993; Silvertown et al. 2001). Most recently the subject has even covered a special issue of a major journal focused on the ecological consequences and evolutionary origin of or escape from senescence in plants (Salguero-Gómez et al. 2013). Other groups of organisms, particularly microbes, have only recently become major targets of research, mostly due to the much more recent development of methods to enable experimentation on them and to basic challenges in defining what the individual is in many organisms (e.g. hyphae of some fungal species may contain many different nuclei (Roper et al. 2011); some forests of quaking aspen (Populus tremuloides) are in fact a huge clone of a single individual (DeWoody et al. 2008)). Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:08:00, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.001

The State of the Science and Further Challenges

5

The State of the Science and Further Challenges Alex Comfort’s classic book on senescence, now over fifty years old, summarised the state of the evolutionary study of senescence as lacking the scientific rigor seen in other disciplines (Comfort 1964). He criticised the field for yielding a number of evolutionary theories but failing to generate sufficient data to test those theories. Implicit in this criticism was the acknowledgement that biologists used a range of incompatible definitions of senescence and that the majority of research was too narrowly focused on humans to evaluate the universality of senescence. We have come some way since Comfort (1964). We have documented demographic patterns across the life course in a growing number of organisms and have finally produced some comparative research suggesting that senescence may not be universal after all (Jones et al. 2014). However, even Comfort himself with his survey of the limited data available in the 1960s suggested that senescence was unlikely to be universal and that the mechanisms for its evolution may differ across species (Comfort 1964). Yet, he also felt that senescence was a relatively rare phenomenon among species, particularly restricted to some, but not all, metazoa, and that life in the wild was harsh enough to prevent most species from reaching senescent ages. In many cases, we know now that a lack of senescence is likely a real phenomenon in some groups rather than simply a case of inadequate or insufficient data. It was in the 1990s that Caleb Finch (1990, 1998) gave serious consideration to organisms that exhibit ‘negligible senescence’ and experience no, or only very small, increases in mortality rate with age. Finch noted that high-quality demographic data were lacking for most species at the time, but his contenders with supporting evidence included sexually reproducing species known to reach advanced age, such as the trees bristlecone pine (Pinus longaeva) and yew (Taxus baccata), lobsters (e.g. Homarus spp.), bivalves such as the quahog (Arctica islandica), marine fish including rockfish (Sebastes spp.) and halibut (Hippoglossus spp.) and the Testudinidae (tortoises) (Finch 1990, 1998). Although we still lack high-quality data on most of these groups, this mode of senescence has recently been confirmed in Hydra (Schaible et al. 2015; see also Chapter 12). As far as we have come in the field of senescence biology, we find nonetheless that a number of important challenges and opportunities exist for further research. We would like to outline five particularly important and promising venues of research, both theoretical and experimental, to develop a unifying theory of the evolution of and escape from senescence.

1.

Meaning and Mechanisms First and, far from sounding trivial, most importantly, we find a great variety of interpretation in the meaning of and assumed mechanisms behind senescence. For some, senescence is unrepaired wear and tear similar to the eventual breakdown of a mechanical system (Comfort 1964). This definition implies that the environment drives senescence and that senescence is caused by unavoidable physical and chemical

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:08:00, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.001

6

Introduction

processes. In contrast, others view senescence as the natural consequence of the accumulation of genes expressed later in life that are detrimental to health. This view implies that senescence is not fundamentally driven by physical wear and tear but rather by intrinsic processes regulated by the genome and patterns in gene expression (Lundberg et al. 2000). Regardless, senescence is most often interpreted scientifically as actuarial senescence, in which age-specific mortality is measured at the population level, while others argue that other forms of senescence, such as reproductive and physiological senescence, may be just as important to consider and perhaps even easier to measure (see Chapter 7). Ultimately, many of us are interested in senescence because of our own awareness of mortality and our hope that understanding senescence can lead to measures to counter it. However, a less self-centred perspective deserves attention here too: ecology and evolution attempt to understand variation, and why some organisms senesce and others seem to escape from senescence is a fundamental question for how the world works, aside from our own hopes and aspirations to live forever. The wear-and-tear view of its nature has led to research on physiochemical processes that lead to decay, such as oxidative stress (Møller 2007), while the genetic view has led to research on the genes driving physiological decline and increased mortality risk, such as cancer genes (Itahana et al. 2004). These are radically different views of senescence, and they need to be reconciled for a comprehensive evolutionary theory to explain the phenomenon.

2.

The Confounding Impact of Life History Some taxa have proven to be particularly challenging to study demographically, mostly because (1) they are long-lived, (2) their age-related demographic patterns are highly variable or (3) the individual is difficult to define and identify (Abrahamson 1980; see also Chapters 17 and 19). While it may seem strange that life span should be a confounding variable, the life histories of the longest-lived organisms are also among the most difficult to study. Long life spans typically require study commitments that fall outside the horizons of modern-day PhD dissertations and funding agencies and can even be challenged by the life span of the experimenter (e.g. the herbaceous perennial plant Borderea pyrenaica, which can live 300 years) (Garcia et al. 2011). Individuals with long life spans also have a tendency to be either fairly large or to have requirements for growth that prevent experimentation. Both of these challenges would be evident for scientists working with some of the world’s long-lived mammals, such as whales and elephants, which are logistically challenging to study in wild conditions and in terms of following a pedigree over several generations, as would be required for a thorough experimental study on the micro-evolutionary context of senescence (though there are some notable exceptions, e.g. Foote 2008; Foster et al. 2012; Hayward et al. 2014). The latter problem is also evident in many long-lived herbaceous perennials, which have stringent germination requirements, long juvenile periods often measuring decades and cryptic life history stages (Gremer et al. 2010; Rasmussen et al. 2015; Shefferson 2009). A corollary of these problems is that they usually prevent the inclusion of the large cohort and sample sizes necessary to ensure sufficient statistical power

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:08:00, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.001

The State of the Science and Further Challenges

7

to precisely measure traits at advanced (perhaps senescent) ages. For example, in what will no doubt be a classic study on plant senescence, Roach et al. (2009) planted over 30,000 seedlings of the short-lived perennial Plantago lanceolata in order to keep statistical power high for those living to old age, finding that over 90 per cent had died by age ten (Shefferson and Roach 2012). An important limitation in the study of senescence in herbs is also that size and age are often decoupled (Salguero-Gómez & Casper 2010) and that anatomic legacies of age (e.g. growth rings) are often difficult to assess (but see Schweingruber & Poschlod 2005). Luckily, some new analytical developments may remedy these limitations (Colchero & Schaible 2014). Clonal plants, clonal animals and hyphal fungi have proven particularly challenging because of their modularity and continual growth, which create strongly size-based patterns in mortality and fecundity that override age-based patterns and create inherent difficulties in delineating individuals (Bierzychudek 1982; Salguero-Gómez et al. 2013). Modularity in plants and clonal animals creates a role for genetic mosaicism in promoting individual fitness, where seemingly ‘somatic’ mutations are nonetheless propagated to future generations because of the lack of a clear germ-soma distinction in clonally growing structures (Gill et al. 1995; see also Chapter 11). This phenomenon is taken to a whole new level in many glomeromycotan, ascomycetous, and basidiomycetous hyphal fungi, which often exchange and keep multiple different nuclei within their physiologically ‘individual’ but genetically diverse mycelia (Roper et al. 2011). These challenges are compounded with empirical difficulties, such as unobservable life stages including vegetative dormancy (Shefferson 2009; Tuomi et al. 2013), or migration (Barthold et al. 2016) and the difficulty in detecting microscopic individuals of fungi in the environment (see Chapter 17). It is only recently that we have seen the growth of research programmes on senescence in these clonal organisms (Jones et al. 2014; Salguero-Gómez et al. 2013). Yet, pioneering theoretical work suggests that clonal indeterminate growers are precisely the organisms in which we need to understand senescence most, because they are the most likely to have escaped from it altogether (Bidder 1932; Vaupel et al. 2004). Amazingly, not only do we still need to better resolve the difficulties in analysing the demography of clonal plants, clonal animals, and fungi, but we also need to agree on a definition of fitness that can allow further study (although some theory currently exists) (see Orive 1995). In the latter vein, evolutionary biology’s heavy reliance on short-term fitness metrics may confound comparisons between short-lived organisms and long-lived organisms with overlapping generations, such as most – if not all – perennial plant species.

3.

Controlled Conditions versus the Wild A further challenge lies in the collection of empirical data on senescence. Analyses of observational, wild-collected data rarely seem to match predictions originating from classical evolutionary theory, except perhaps in large animals (Bronikowski & Promislow 2005; Jones et al. 2014; Reznick et al. 2004;). In long-lived plants, for example, observational approaches have certainly yielded many interesting patterns

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:08:00, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.001

8

Introduction

that yield far more questions than answers (Garcia et al. 2011; Jones et al. 2014; Martı́nez 1998). In this book, some authors argue that the wild is not the place to look for senescence; rather, a controlled and safe environment in which organisms actually have the potential to age even perhaps beyond reproduction is necessary to observe predicted patterns (see Chapter 3). Others here argue that wild-collected data often show evidence of senescence (see Chapter 7), and the real challenge is not to test the theory under controlled conditions but instead to build a theory that incorporates ecological context better (see Chapters 6, 15, 17 and Chapter 19). A further viewpoint notes that many animal species tend to show senescence even in the wild (Nussey et al. 2013; Reznick et al. 2006) but that experimentation in controlled conditions may yield more consistent results (see Chapter 9). We argue that both controlled experiments and in situ observational approaches need to be pursued simply because there is so little agreement on how senescence actually evolves.

4.

The Role of Environment We further argue that the role of the environment in senescence may be more complicated than is currently acknowledged. Plant and wildlife demographers, who are used to having access to long-term, high-resolution data with pedigree and causes of mortality, generally view mortality risk as being derived from either intrinsic (e.g. disease genes) or extrinsic (e.g. climatic catastrophes) sources (e.g. Abrams 1993; Ricklefs 2000). In reality, we argue, the separation of intrinsic versus extrinsic causes of mortality is more complicated than this dichotomy, and these are more likely to take place as interactions rather than as simple additive effects. For example, a tree that has been attacked by a fungus is more likely to succumb to an ‘internal’ cause of mortality, whatever that may be, than an un-attacked tree is likely to succumb to that same internal cause of mortality. Evolutionary demographers have been strongly influenced by W. Hamilton, who viewed senescence as primarily an intrinsically driven phenomenon caused by the cumulative impact of deleterious genes (Hamilton 1966), though with the ability to evolve depending on patterns in age-specific extrinsic mortality (Caswell 2007; see also Chapters 4 and 9). Intrinsic factors can make individuals more susceptible to extrinsic sources of mortality. However, evolutionary biologists have long noted that many genes exhibit differential expression patterns under different environmental contexts, a phenomenon known as ‘phenotypic plasticity’ (West-Eberhard 2003). Because trait expression depends on cellular sensing of differentially expressed or concentrated gene products in the immediate environment, many biologists now argue that most, if not all, traits are actually plastic (Pigliucci 2005; West-Eberhard 2003). In evolutionary demography, a small subset of researchers has similarly asked if patterns of senescence might be strongly driven by both intrinsic and extrinsic factors and has suggested useful frameworks for investigating the dependence of senescence-related patterns in mortality, fertility and physiological traits as a function of environmental variation (Hammers et al. 2012; Koons et al. 2014; Ricklefs 2000; Shefferson & Roach 2013).

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:08:00, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.001

The State of the Science and Further Challenges

9

Along similar lines, some authors have noted that senescence is strongly influenced by ecological context. For example, predation can shape senescence by imposing selection against the senescent, provided that the organisms are still able to reproduce and that predation is higher upon older-aged individuals (see Chapter 9). Predation and other causes of mortality also continually weed out the weakest individuals from the population, yielding a ‘disappearing fraction’, or heterogeneity effect (Vaupel & Yashin 1985), that removes senescent individuals and may result in the measurement of senescencerelated traits in only the most physiologically fit individuals (Bennington & McGraw 1995; Nussey et al. 2011; see also Chapter 8). Among the most important consequences of the relatively poor consideration of the complexity of sources of mortality and environmental influences in general is the somewhat simplistic view of trade-offs in much of the evolutionary literature on senescence. Trade-offs are often thought of as dichotomous splits in resource allocation between two traits, but the truth is far more complicated. Though trade-offs are always assumed to operate and so prevent the evolution of Darwinian demons (i.e. organisms with infinite fitness, per Law 1979), environmental influences on the expression of related traits can lead to trade-offs disappearing altogether or becoming positive relationships (Reznick et al. 2000; Spitze 1991). Trade-offs are further complicated by the order of resource allocation among traits (see Chapter 5), which can lead to many traits becoming positively related (de Jong and van Noordwijk 1992); by relationships among traits via commonly used gene expression pathways (Stearns and Magwene 2003); and by the influence of past trade-offs (Stearns 1992). Such complexities are not formally included in any model of evolutionary senescence, and indeed the most commonly used models posit dichotomous relationships between fitness components such as survival and reproduction that may or may not hold in much of the Tree of Life. With the need for a better understanding of the influence of ecological context on senescence, we argue that the distinction between intrinsic and extrinsic mortality in senescence is not helpful and indeed that many factors typically considered to be one or the other actually have elements of both (see Chapter 6). The field may grow most productively through its incorporation of viewpoints and advances in the fields of epigenetics and developmental biology (Hoffman et al. 2014; Zhu et al. 2015), as well as the inclusion of ecological context (see also Chapters 6 and 9), in a rigorous theoretical framework that can yield strong, testable predictions. It would also grow through the inclusion of further ‘forensic’ data into demographic analysis, for example, on the actual causes of mortality. The latter is something that has allowed studies on human senescence to make important contributions and that studies of the rest of the animal and plant kingdoms could do with.

5.

A Robust Phylogenetic Comparative Framework Although even some of the very earliest work on senescence had a comparative component, a lack of rate-of-ageing data (as opposed to maximum-life-span data) for many groups and the nascent state of phylogenetic comparative methods itself hampered analyses until around fifty years ago. The macro-evolutionary tradition in senescence

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:08:00, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.001

10

Introduction

biology goes back longer, being influenced by the development of comparative demography primarily in the United Kingdom and the foundation of formal phylogenetic analysis in the 1950s and 1960s with Willi Hennig’s seminal contributions (Austad & Fischer 1991; Botkin & Miller 1974; Franco & Silvertown 1997; Grime 1974; Hennig 1966; Lack 1954; Pianka 1970). The first analyses made use of available data, which included sparse coverage of the Tree of Life. Over time, these analyses have generally found support for the idea that senescence varies across organisms, with the most sophisticated recent analyses suggesting that existing theory does not account for the diversity of mortality and fecundity patterns in the world’s organisms (e.g. Jones et al. 2014). Now that we know that senescence varies across the Tree of Life, we need to know how (what are its predictors?) and why (what are its mechanisms for and against it?). We have already made the case for expanding existing theory to account for ecological context, which likely causes severe deviations from expectations in at least some cases (see also Chapters 5, 6 and 8). Evolutionary history may also be relevant to understanding patterns in senescence, particularly since the physiology involved in senescence, or its escape, is probably based on genetic predispositions that are themselves shared by closely related lineages. Do sister species senesce (or not) in similar ways? Phylogenetic hypotheses have not been explored in any systematic way within the field of senescence biology. However, some clear hypotheses are suggested, even by the dominant theories of senescence evolution. For example, if senescence is the result of a declining force of natural selection with age (Hamilton 1966; Medawar 1952), then age-related mortality and fecundity patterns, and the physiological traits that lead to senescence, may evolve slowly over macro-evolutionary time in ways similar to Brownian motion models of evolution. If so, senescence may exhibit phylogenetic signal, which would lead to a very specific phylogenetic pattern: more closely related species should exhibit more similar demographic patterns than more distantly related ones (Blomberg et al. 2003). Indeed, some groups may be particularly prone to a particular physiological path in senescence, just as some orchid lineages are genetically predisposed to mutate into non-photosynthetic plants, even within otherwise photosynthetic populations (Julou et al. 2005; Motomura et al. 2010). Further analyses can also lead to a better understanding of the evolution of specific demographic patterns and senescence-related traits via phylogenetic contrasts (Felsenstein 1985). For example, although some have suggested that clonal plants may escape senescence because clonality may yield particularly strong benefits to fitness under some circumstances (Vaupel et al. 2004), as far as we know, a systematic macro-evolutionary comparison between sister plant species differing in this trait has never been performed, even though large data sets exist that may allow such analyses (Klimešová & De Bello 2009; Salguero-Gómez et al. 2015). Phylogenetic analyses are under-utilised in senescence biology and can advance our understanding of senescence immeasurably. However, they require a perspective that can fit the different paradigms of evolutionary demography and phylogenetics together meaningfully. Importantly, as phylogenetics has developed into a strongly quantitative and computationally intensive field, it has developed rigorous tests of assumptions Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:08:00, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.001

Conventions in This Book and Chapter Introductions

11

frequently made in ecologically oriented analyses with a phylogenetic component. First and foremost, is the assumption that phylogenetic analyses can be conducted with a small sampling of taxa. While micro-evolutionary biologists rely on the methods of classical statistics, which allow unbiased estimation of parameters with even less than 10 per cent of a population fully sampled, macro-evolutionary analyses have repeatedly found that a large coverage of operational taxonomic units (OTUs) and a large character matrix are both required for accurate phylogenetic analysis (Beaulieu et al. 2012; Delsuc et al. 2005; Hillis 1998). Poor taxon coverage can lead to important artefactual patterns such as long-branch attraction in phylogeny construction and support for adaptive patterns where none exist in ancestral character reconstructions (Felsenstein 1985). A further assumption often made is that taxonomy can serve as an adequate model for evolutionary history. This approach was useful in the early days of comparative demography, when evolutionary histories for even the most commonly worked with organisms were missing. However, genetic data for ever-increasing numbers of taxa keep accumulating and, with them, our understanding of evolutionary relationships and their uncertainties throughout the Tree of Life (Bininda-Emonds et al., 2002; Bruns 2006; Bruns & Shefferson 2004; Delsuc et al. 2005; Givnish et al. 2010; Keeling et al. 2005; Lutzoni et al. 2004). While there has been an effort to make biological nomenclature reflect ‘natural’ evolutionary relationships, the fact remains that taxon levels above species lack common definitions for common comparisons and ignore the vast majority of phylogenetic information. The approach also focuses on topology while omitting information on branch lengths and can yield erroneous analytical outcomes (DíazUriarte & Garland 1998). We argue that the future of comparative demography lies in the proper integration of the ecological data (e.g. Jones et al. 2014; Salguero-Gómez et al. 2015, 2016) with known phylogenetic patterns (e.g. James et al. 2006; Shefferson 2009) while acknowledging the uncertainties in both. Another important assumption is that species-level physiological and/or demographic patterns can be inferred from a single site/population. While a perfectly understandable assumption from the logistical point of view, nonetheless, physiological and demographic patterns are known to exhibit a strong degree of plasticity and geographic variation (Shefferson & Tali 2007). Plasticity, natural selection and population history can drive strong differences in character evolution among populations (Fordyce 2006). Although methods do exist for incorporating within-species variation in phylogenetic comparative analysis (Felsenstein 2008; Ives et al. 2007; Martins & Hansen 1997), they have not been used widely. The wider use of such methods would no doubt advance macro-evolutionary demography.

Conventions in This Book and Chapter Introductions This book is divided into five major parts. Part I introduces the evolutionary theories of senescence. Chapter 2 describes the disposable soma theory and notes how this theory both compares and aligns with other theories attempting to explain senescence across the entire Tree of Life. Chapter 3 presents the strengths and challenges inherent in Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:08:00, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.001

12

Introduction

a Hamiltonian approach to the study of senescence – a programme that relies on Hamilton’s well-defined theory explaining the decline in strength of natural selection acting on mortality and fecundity with age. Chapter 4 shows how mortality affects the selection gradients involved in senescence and notes that any change to mortality can only affect senescence if it targets a particular age or stage of life rather than indiscriminately affecting the whole life span. Chapter 5 examines the weight of evidence in favour or against the ‘classical’ models of the evolution of senescence, notably mutation accumulation, antagonistic pleiotropy and disposable soma theory, and comes to the conclusion that none of these theories adequately explains the diversity of patterns of age-specific mortality and physiological ageing observed across the Tree of Life. This chapter also attempts to reconcile contemporary complex systems models of senescence, which are inherently physiological. Part II explores senescence in animals. Chapter 6 addresses age-related mortality in humans and notes that human populations often exhibit senescence patterns that differ quantitatively from classical evolutionary theory in ways that may require more general theory incorporating ecological context. Chapter 7 explores senescence in mammals, noting that although wild mammals were historically assumed not to senesce because of high extrinsic mortality in the wild, the wealth of data accumulated via mark-recapture studies suggests otherwise. Chapter 8 showcases senescence patterns in birds, noting similar historical trends as in Chapter 7 (including qualitative support for classical evolutionary models), but also noting the importance of confounding factors such as the ‘invisible fraction’ (also known as the ‘disappearing fraction’) and age-related behavioural response to trapping. Chapter 9 assesses the qualitative support for classical evolutionary models of senescence that have been performed in wild animals and issues a call for research in the 1960s evolution paradigm – longitudinal studies in the wild combined with stringent, controlled laboratory breeding studies to assess the mechanisms causing traits to evolve in senescence-like patterns. Chapter 10 explores the fascinating differences in life span and senescence in eusocial insects, where queens often live far longer than other colony members, and suggests a role for social structure in determining age-related mortality patterns across the animal kingdom. Chapter 11 showcases physiological and demographic patterns in modular, indeterminately growing animals, which are often given as examples of possible escape from senescence in the literature, offering perspectives both in support of and against the classical evolutionary theories of senescence. Finally, Chapter 12 explores senescence in Hydra, a genus of cnidarians famous for having demographic patterns suggestive of escape from senescence, particularly noting the roles of clonality and stem cells in producing seemingly constant age-specific mortality. Part III focuses on senescence in plants. Chapter 13 overviews plant physiological and biochemical processes related to senescence and offers perspectives on whole-plant senescence as it relates to organ senescence, hormones and growth. Like other chapters, particularly within Part III, it offers advice and insights on the development of experimental approaches to studying senescence in plants. Chapter 14 focuses on the influences of seasonal and general environmental variation on senescence in annual plants, noting the importance of classical phenotypic plasticity (i.e. phenological response to Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:08:00, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.001

Conclusions

13

predictable environmental variation, per Roff 2002) and also relates these perspectives to perennial herbs. Chapter 15 reviews demographic senescence in herbaceous plants, particularly focusing on this phenomenon in perennials and showcasing a number of plant traits that may influence the tendency of many plants to seemingly escape from senescence: clonality, the development of multiple meristems and growth and density effects, among others. Finally, Chapter 16 examines senescence in extremely long-lived herbaceous plants and in particular offers perspectives on how senescence may relate to the phenomenon of vegetative, or prolonged, dormancy, in which the plant lives underground as a rootstock without photosynthesising (Shefferson 2009). Part IV studies senescence in microbes and microbial symbioses. Chapter 17 explores senescence in hyphal fungi and showcases both the cellular/physiological processes and ecological conditions that are involved. This chapter also notes important differences in fungal biology that prevent the application to fungi of biological concepts used to understand senescence in clonal plants. Chapter 18 explores senescence in yeasts, which are essentially non-hyphal, single-celled fungi, and in particular notes the genetic and biochemical context under which yeasts experience both physiological and demographic decline. Chapter 19 explores a previously overlooked phenomenon from the senescence perspective: the role of symbioses. This chapter posits that symbioses likely influence the evolution of senescence in major and predictable ways, showcases the mycorrhiza as a useful case study for this research endeavour and further postulates two means of exploring these influences. We conclude the book with Chapter 20, which is the sole chapter in Part V, the part devoted to the broadest macro-evolutionary perspective on senescence. This chapter examines how correlations in trade-offs of various life history traits may be used as robust predictors of senescence rates across hundreds of fungal, animal and plant species drawing from rich demographic data.

Conclusions Senescence is an old topic (pun intended!) that was documented in Aristotle’s scripts on youth, where he suggested that dietary restriction (or long walks under a strict diet) could be the key to a long, prosperous life (Aristotle 1984). One of many hotly debated controversies in the field of senescence research is whether most theories, theorems and laws are universal (e.g. Kirkwood 2005; Zhao et al. 2014). The lack of predictive power of the current theories of senescence across taxa is ever more startling when we consider that the topic of senescence cuts across all biological, sociological and cultural disciplines. The retirement age in human societies, plans for zoo exchanges for fertility treatments and even the harvest of crops are all tightly linked to the age-specific performance of individuals along their life trajectories. The study of senescence is challenging but inherently rewarding. We hope that this book will serve as an inspiration to readers to study the biology of ageing, how unusual traits such as senescence evolve and to ask how their own lives relate to the broader patterns that evolution has woven in nature. Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:08:00, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.001

14

Introduction

Acknowledgements The editors wish to thank many researchers who contributed directly or indirectly to the completion of this book. RPS thanks Junco Nagata, Taro Shefferson-Nagata and Hazuki Shefferson-Nagata for their support and encouragement, D. Roach for helpful discussion, and the University of Tokyo’s Departments of General Systems Studies and Biology for financial and logistical support. ORJ and RS-G thank interactions with the members of the Evolutionary Demography Laboratory of the Max Planck Institute for Demographic Research and the Max Planck Center on the Biodemography of Aging at the University of Southern Denmark, led by J. Vaupel.

References Abrahamson, W. G. (1980). Demography and vegetative reproduction. In O. T. Solbrig (ed.), Demography and Evolution in Plant Populations (Berkeley: University of California Press). Abrams, P. A. (1993). Does increased mortality favour the evolution of more rapid senescence? Evolution, 47, 877–87. Aristotle (1984). On longevity and shortness of life. In The Complete Works of Aristotle, J. Barnes & G. R. T. Ross (Princeton, NJ: Princeton University Press). Austad, S. N., & Fischer, K. E. (1991). Mammalian aging, metabolism, and ecology: evidence from the bats and marsupials. Journal of Gerontology, 46, B47–53. Barthold, J., Loveridge, A., MacDonald, D., et al. (2016). Bayesian estimates of male and female African lion mortality for future use in population management. Journal of Applied Ecology, 53, 295–304. Baudisch, A., Salguero-Gómez, R., Jones, O. R., et al. (2013). The pace and shape of senescence in angiosperms. Journal of Ecology, 101, 596–606. Beaulieu, J. M., Ree, R. H., Cavender-Bares, J., et al. (2012). Synthesizing phylogenetic knowledge for ecological research. Ecology, 93, S4–S13. Bennington, C. C. & McGraw, J. B. (1995). Phenotypic selection in an artificial population of Impatiens pallida: the importance of the invisible fraction. Evolution, 49, 317–24. Bidder, G. P. (1932). Senescence. British Medical Journal, 2, 583–5. Bierzychudek, P. (1982). Life histories and demography of shade-tolerant temperate forest herbs: a review. New Phytologist, 90, 757–76. Bininda-Emonds, O. R. P., Gittleman, J. L. & Steel, M.A. (2002). The (super) tree of life: procedures, problems, and prospects. Annual Review of Ecology and Systematics, 33, 265–89. Blomberg, S. P., Garland, T. & Ives, A.R. (2003). Testing for phylogenetic signal in comparative data: behavioral traits are more labile. Evolution, 57, 717–45. Botkin, D. B. & Miller, R. S. (1974). Mortality rates and survival of birds. American Naturalist, 108, 181–92. Bronikowski, A. M. & Promislow, D. E. L. (2005). Testing evolutionary theories of aging in wild populations. Trends in Ecology and Evolution, 20, 271. Bruns, T. D. (2006). A kingdom revised. Nature, 443, 758–60.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:08:00, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.001

References

15

Bruns, T. D. & Shefferson, R. P. (2004). Evolutionary studies of ectomycorrhizal fungi: recent advances and future directions. Canadian Journal of Botany, 82, 1122–32. Caswell, H. (2007). Extrinsic mortality and the evolution of senescence. Trends in Ecology & Evolution, 22, 173–4. Clausen, J., Keck, D. D. & Hiesey, W. M. (1947). Heredity of geographically and ecologically isolated races. American Naturalist, 81, 114–33. Colchero, F. & Schaible, R. (2014). Mortality as a bivariate function of age and size in indeterminate growers. Ecosphere, 5, art. 161. Comfort, A. (1964). Ageing: The Biology of Senescence (London: Routledge Kegan Paul). Dawkins, R. (1978). The Selfish Gene (Oxford University Press). De Jong, G. & Van Noordwijk, A. J. (1992). Acquisition and allocation of resources: genetic (co)variances, selection, and life histories. American Naturalist, 139, 749–70. Delsuc, F., Brinkmann, H. & Philippe, H. (2005). Phylogenomics and the reconstruction of the tree of life. Nature Reviews Genetics, 6, 361–75. Dewoody, J., Rowe, C. A., Hipkins, V. D. & Mock, K.E. (2008). ‘Pando’ lives: molecular genetic evidence of a giant aspen clone in central Utah. Western North American Naturalist, 68, 493–7. Díaz-Uriarte, R. & Garland, T. J. (1998). Effects of branch length errors on the performance of phylogenetically independent contrasts. Systematic Biology, 47, 654–72. Felsenstein, J. (1985). Phylogenies and the comparative method. American Naturalist, 125, 1–15. Felsenstein, J. (2008). Comparative methods with sampling error and within-species variation: contrasts revisited and revised. American Naturalist, 171, 713–25. Finch, C. E. (1990). Longevity, Senescence, and the Genome (University of Chicago Press). Finch, C. E. (1998). Variations in senescence and longevity include the possibility of negligible senescence. Journals of Gerontology Series A: Biological Sciences and Medical Sciences, 53A, B235–39. Fisher, R. A. (1930). The Genetical Theory of Natural Selection (New York: Clarendon Press). Foote, A. D. (2008). Mortality rate acceleration and post-reproductive lifespan in matrilineal whale species. Biology Letters, 4, 189–91. Fordyce, J. A. (2006). The evolutionary consequences of ecological interactions mediated through phenotypic plasticity. Journal of Experimental Biology, 209, 2377–83. Foster, E. A., Franks, D. W., Mazzi, S., et al. (2012). Adaptive prolonged postreproductive life span in killer whales. Science, 337, 1313. Franco, M. & Silvertown, J. (1997). Life history variation in plants: an exploration of the fast-slow continuum hypothesis. In Plant Life Histories: Ecology, Phylogeny and Evolution, ed. J. Silvertown, M. Franco & J. L. Harper (Cambridge University Press). Garcia, M. B., Dahlgren, J. P. & Ehrlén, J. (2011). No evidence of senescence in a 300-year-old mountain herb. Journal of Ecology, 99, 1424–30. Gill, D. E., Chao, L., Perkins, S. L. & Wolf, J. B. (1995). Genetic mosaicism in plants and clonal animals. Annual Review of Ecology and Systematics, 26, 423–44.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:08:00, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.001

16

Introduction

Givnish, T. J., Ames, M., McNeal, J. R., et al. (2010). Assembling the trees of the monocotyledons: plastome sequence phylogeny and evolution of Poales 1. Annals of the Missouri Botanical Garden, 97, 584–616. Gremer, J. R., Sala, A. & Crone, E. E. (2010). Disappearing plants: why they hide and how they return. Ecology, 91, 3407–13. Grime, J. P. (1974). Vegetation classification by reference to strategies. Nature, 250, 26–31. Haldane, J. B. S. (1941). New Paths in Genetics (London: Allen & Unwin). Hamilton, W. D. (1966). The moulding of senescence by natural selection. Journal of Theoretical Biology, 12, 12–45. Hammers, M., Richardson, D. S., Burke, T. & Komdeur, J. (2012). Age-dependent terminal declines in reproductive output in a wild bird. PLoS ONE, 7, e40413. Harper, J. L. (1967). A Darwinian approach to plant ecology. Journal of Ecology, 55, 247–70. Harper, J. L. (1977). Population Biology of Plants (New York: Academic Press). Harper, J. L. & White, J. (1974). The demography of plants. Annual Review of Ecology and Systematics, 5, 419–63. Hayward, A. D., Mar, K. U., Lahdenperä, M. & Lummaa, V. (2014). Early reproductive investment, senescence and lifetime reproductive success in female Asian elephants. Journal of Evolutionary Biology, 27, 772–83. Hennig, W. (1966). Phylogenetic Systematics (Urbana: University of Illinois Press). Hillis, D. M. (1998). Taxonomic sampling, phylogenetic accuracy, and investigator bias. Systematic Biology, 47, 3–8. Hoffman, J. M., Soltow, Q. A., Li, S., et al. (2014). Effects of age, sex, and genotype on high-sensitivity metabolomic profiles in the fruit fly, Drosophila melanogaster. Aging Cell, 13, 596–604. Holliday, R. (1990). DNA methylation and epigenetic inheritance. Philosophical Transactions of the Royal Society of London Series B: Biological Science, 326(1235), 329–38. Itahana, K., Campisi, J. & Dimri, G. (2004). Mechanisms of cellular senescence in human and mouse cells. Biogerontology, 5, 1–10. Ives, A. R., Midford, P. E. & Garland, T. (2007). Within-species variation and measurement error in phylogenetic comparative methods. Systematic Biology, 56, 252–70. James, T. Y., Kauff, F., Schoch, C. L., et al. (2006). Reconstructing the early evolution of Fungi using a six-gene phylogeny. Nature, 443, 818–22. Jones, O. R., Gaillard, J.-M., Tuljapurkar, S., et al. (2008). Senescence rates are determined by ranking on the fast-slow life-history continuum. Ecology Letters, 11, 664–73. Jones, O. R., Scheuerlein, A., Salguero-Gómez, R., et al. (2014). Diversity of ageing across the tree of life. Nature, 505, 169–73. Julou, T., Burghardt, B., Gebauer, G., et al. (2005). Mixotrophy in orchids: insights from a comparative study of green individuals and nonphotosynthetic individuals of Cephalanthera damasonium. New Phytologist, 166, 639–53. Keeling, P. J., Burger, G., Durnford, D. G., et al. (2005). The tree of eukaryotes. Trends in Ecology & Evolution, 20, 670–6. Kirkwood, T. B. L. (1977). Evolution of ageing. Nature, 270, 301–4. Kirkwood, T. B. L. (2005). Understanding the odd science of aging. Cell, 120, 437–47.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:08:00, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.001

References

17

Klimešová, J. & De Bello, F. (2009). CLO-PLA: the database of clonal and bud bank traits of Central European flora. Journal of Vegetation Science, 20, 511–16. Koons, D. N., Gamelon, M., Gaillard, J.-M., et al. (2014). Methods for studying cause-specific senescence in the wild. Methods in Ecology and Evolution, 5, 924–33. Kumar, A., Rani, A., Tchigranova, O., et al. (2012). Influence of late-life exposure to environmental enrichment or exercise on hippocampal function and CA1 senescent physiology. Neurobiology of Aging, 33, 828.e1–17. Lack, D. L. (1954). The Natural Regulation of Animal Numbers (Oxford: Clarendon Press). Law, R. (1979). Optimal life histories under age-specific predation. American Naturalist, 114, 399–417. Ljubuncic, P. & Reznick, A. Z. (2009). The evolutionary theories of aging revisited: a mini-review. Gerontology, 55, 205–16. Lundberg, A. S., Hahn, W. C., Gupta, P. & Weinberg, R. A. (2000). Genes involved in senescence and immortalization. Current Opinion in Cell Biology, 12, 705–9. Lutzoni, F., Kauff, F., Cox, C.J., et al. (2004). Assembling the fungal tree of life: progress, classification and evolution of subcellular traits. American Journal of Botany, 91, 1446–80. Martínez, D. E. (1998). Mortality patterns suggest lack of senescence in Hydra. Experimental Gerontology, 33, 217–25. Martins, E. P. & Hansen, T. F. (1997). Phylogenies and the comparative method: a general approach to incorporating phylogenetic information into the analysis of interspecific data. American Naturalist, 149, 646–67. Medawar, P. B. (1952). An Unsolved Problem of Biology (London: Lewis). Medina, M. (2005). Genomes, phylogeny, and evolutionary systems biology. Proceedings of the National Academy of Sciences of the United States of America, 102, 6630–5. Møller, A. P. (2007). Senescence in relation to latitude and migration in birds. Journal of Evolutionary Biology, 20, 750–7. Moorad, J. A. & Promislow, D. E. L. (2011). Evolutionary demography and quantitative genetics: age-specific survival as a threshold trait. Proceedings of the Royal Society B: Biological Sciences, 278(1702), 144–51. Motomura, H., Selosse, M.-A., Martos, F., et al. (2010). Mycoheterotrophy evolved from mixotrophic ancestors: evidence in Cymbidium (Orchidaceae). Annals of Botany, 106, 573–81. Motörhead. (1984). ‘Killed By Death’. No Remorse (London: Bronze Records). Nussey, D. H., Coulson, T., Delorme, D., et al. (2011). Patterns of body mass senescence and selective disappearance differ among three species of free-living ungulates. Ecology, 92, 1936–47. Nussey, D. H., Froy, H., Lemaitre, J.-F., et al. (2013). Senescence in natural populations of animals: widespread evidence and its implications for bio-gerontology. Ageing Research Reviews, 12, 214–25. Orive, M. E. (1995). Senescence in organisms with clonal reproduction and complex life histories. American Naturalist, 145, 90–108. Pianka, E. R. (1970). On r- and K-selection. American Naturalist, 104, 592–97.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:08:00, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.001

18

Introduction

Pigliucci, M. (2005). Evolution of phenotypic plasticity: where are we going now? Trends in Ecology & Evolution, 20, 481–86. Rabinow, P. (1997). Making PCR: A Story of Biotechnology (University of Chicago Press). Rasmussen, H. N., Dixon, K. W., Jersáková, J. & Těšitelová, T. (2015). Germination and seedling establishment in orchids: a complex of requirements. Annals of Botany, 116 (3), 391–402. Reznick, D., Bryant, M. & Holmes, D. (2006). The evolution of senescence and post-reproductive lifespan in guppies (Poecilia reticulata). PLoS Biology, 4, 136–43. Reznick, D., Nunney, L. & Tessier, A. (2000). Big houses, big cars, superfleas and the costs of reproduction. Trends in Ecology and Evolution, 15, 421–5. Reznick, D. N., Bryant, M. J., Roff, D., et al. (2004). Effect of extrinsic mortality on the evolution of senescence in guppies. Nature, 431, 1095–9. Ricklefs, R. E. (2000). Intrinsic aging-related mortality in birds. Journal of Avian Biology, 31, 103–11. Roach, D. A. (1993). Evolutionary senescence in plants. Genetica, 91, 53–64. Roach, D. A., Ridley, C. E. & Dudycha, J. L. (2009). Longitudinal analysis of Plantago: age by environment interactions reveal aging. Ecology, 90, 1427–33. Roff, D. A. (2002). Life History Evolution (Sunderland, MA: Sinauer Associates). Roper, M., Ellison, C., Taylor, J. W. & Glass, N. L. (2011). Nuclear and genome dynamics in multinucleate ascomycete fungi. Current Biology, 21, R786–93. Salguero-Gómez, R. & Casper, B.B. (2010). Keeping plant shrinkage in the demographic loop. Journal of Ecology, 98, 312–23. Salguero-Gómez, R., Jones, O. R., Archer, C. R., et al. (2015). The Compadre Plant Matrix Database: an open online repository for plant demography. Journal of Ecology, 103, 202–18. Salguero-Gómez, R., Jones, O. R., Jongejans, E., et al. (2016). Fast–slow continuum and reproductive strategies structure plant life-history variation worldwide. Proceedings of the National Academy of Sciences of the United States of America, 113, 230–5. Salguero-Gómez, R., Shefferson, R. P. & Hutchings, M. J. (2013). Plants do not count . . . or do they? New perspectives on the universality of senescence. Journal of Ecology, 101, 545–54. Schaible, R., Scheuerlein, A., Dańko, M. J., et al. (2015). Constant mortality and fertility over age in Hydra. Proceedings of the National Academy of Sciences of the United States of America, 112, 15701–6. Schweingruber, F. H. & Poschlod, P. (2005). Growth rings in herbs and shrubs: life span, age determination and stem anatomy. Forest, Snow and Landscape Research, 79, 195–415. Shefferson, R. P. (2009). The evolutionary ecology of vegetative dormancy in mature herbaceous perennial plants. Journal of Ecology, 97, 1000–9. Shefferson, R. P. & Roach, D. A. (2012). The triple helix of Plantago lanceolata: genetics and the environment interact to determine population dynamics. Ecology, 93, 793–802. Shefferson, R. P. & Roach, D. A. (2013). Longitudinal analysis in Plantago: strength of selection and reverse age analysis reveal age-indeterminate senescence. Journal of Ecology, 101, 577–84.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:08:00, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.001

References

19

Shefferson, R. P. & Tali, K. (2007). Dormancy is associated with decreased adult survival in the burnt orchid, Neotinea ustulata. Journal of Ecology, 95, 217–25. Silvertown, J. (2013). The Long and the Short of It: The Science of Life Span and Aging (University of Chicago Press). Silvertown, J., Franco, M. & Perez-Ishiwara, R. (2001). Evolution of senescence in iteroparous perennial plants. Evolutionary Ecology Research, 3, 393–412. Spitze, K. (1991). Chaoborus predation and life-history evolution in Daphnia pulex: temporal pattern of population diversity, fitness, and mean life history. Evolution, 45, 82–92. Stearns, S. C. (1992). The Evolution of Life Histories (Oxford University Press). Stearns, S. C. & Magwene, P. (2003). The naturalist in a world of genomics. American Naturalist, 161, 171–80. Tuomi, J., Crone, E. E., Gremer, J. R., et al. (2013). Prolonged dormancy interacts with senescence for two perennial herbs. Journal of Ecology, 101, 566–76. Vaupel, J. W., Baudisch, A., Dölling, M., et al. (2004). The case for negative senescence. Theoretical Population Biology, 65, 339–51. Vaupel, J. W. & Yashin, A. I. (1985). Heterogeneity’s ruses: some surprising effects of selection on population dynamics. American Statistician, 39, 176–85. Weismann, A. (1892). Das Keimplasma: eine Theorie der Vererbung (Jena, Germany: Fischer). Weismann, A. (1893). The Germ-Plasm: A Theory of Heredity (New York: Charles Scribner’s Sons). West-Eberhard, M. J. (2003). Developmental Plasticity and Evolution (Oxford University Press). Williams, G. C. (1957). Pleiotropy, natural selection, and the evolution of senescence. Evolution, 11, 398–411. Zhao, G., Guo, S., Somel, M. & Khaitovich, P. (2014). Evolution of human longevity uncoupled from caloric restriction mechanisms. PLoS ONE, 9, e84117. Zhu, Y., Tchkonia, T., Pirtskhalava, T., et al. (2015). The Achilles’ heel of senescent cells: from transcriptome to senolytic drugs. Aging Cell, 14(4), 644–58.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:08:00, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.001

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:08:00, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.001

2

The Disposable Soma Theory Origins and Evolution Thomas B. L. Kirkwood

Short Summary The disposable soma theory is a physiologically based evolutionary hypothesis to explain why and how senescence occurs in those species where it is present, why and how longevity has been acted upon by natural selection and why senescence may be absent or unclear in species that do not fulfil the requirements for its evolution. The theory was originally developed from the perspective of mechanisms of cellular ageing, where it had been shown that cells could increase their stability against disruption by molecular damage but at the expense of requiring a greater investment of energy in mechanisms for molecular proofreading and damage removal. The theory showed how a high investment in stability in the germ line was essential to preserve the lineage across generations but that in view of the mortality imposed by the environment, an energy-saving strategy of reduced error regulation in somatic cells would enhance fitness by allowing the organism to grow faster and reproduce at a higher rate. The theory provides support for the idea that senescence is caused by the gradual, progressive accumulation of faults in somatic cells. It also generalises to explain how the principles of optimal resource allocation can explain the great diversity in patterns of senescence not only across the phylogenetic spectrum (unicellular versus multicellular, asexual versus sexual reproduction, semelparity versus iteroparity, polyphenism versus monophenism and so on) but also in environments that may vary in quality across space and time and in which it may be adaptive to enable plasticity within the life history.

Introduction The aims and scope of this chapter are to describe the genesis and essence of the concept known as the ‘disposable soma theory of ageing’. It is written primarily as a descriptive account and does not seek to argue the case for the theory. As with any scientific theory, the disposable soma theory has value only to the extent that it advances understanding of its subject by delivering insight and by prompting experimental tests of its predictions. As the author was also the original proponent of the theory, it is a task chiefly for others to judge its usefulness. Nevertheless, it is evident from the literature that elements of the theory are sometimes misunderstood, and, since the theory originated not within the mainstream of evolutionary thinking but from studying mechanisms of cellular ageing Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:24:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.002

24

The Disposable Soma Theory

Resources (energy)

Figure 2.1

Growth Maintenance Reproduction Immune system Storage

Progeny

Metabolic resources must be allocated across a range of physiological functions in order for the organism to grow, survive and reproduce.

(and asking Why?), it is hoped that there may be value in describing how the theory came into being and what it does and does not do. The disposable soma theory is a physiologically based evolutionary hypothesis to explain why and how senescence occurs in those species where it is present, why and how longevity has been acted upon by natural selection and why senescence may be absent or unclear in species that do not fulfil the requirements for its evolution. The disposable soma theory was proposed by Kirkwood (1977), the name itself being first used in print by Kirkwood and Holliday (1979). The original description of the theory started with the observation that an evolutionary explanation for ageing was required since many organisms, notably higher plants, live and propagate indefinitely. The theory also acknowledged from the outset that it was fallacious to presume that older organisms would suffer progressive wear and tear, as had been done, for example, by Weismann (1891), who argued that selection would disfavour the retention of ‘old and worn-out’ individuals who would otherwise compete for resources with younger ones (but see Kirkwood and Cremer (1982) for a fuller examination of Weismann’s thinking as it developed through his successive writings on ageing). All organisms depend on the acquisition of resources – principally energy, but also including essential chemical elements – and on the deployment of these resources to survive and reproduce. For many organisms, the necessary resources are either scarce or hard won through activities such as foraging and hunting, which expose the individual to increased risk of death. Regardless, however, of whether resources are plentiful or scarce, every organism faces the challenge of how best to optimise its physiological allocation of resources in such a way that will maximise its fitness under natural selection. The concept of fitness is itself complex, an issue to which we shall return later. Ultimately, fitness is concerned with the successful production of viable offspring, and the problem of optimal allocation concerns the best way to convert resources into progeny (Figure 2.1). The list of physiological activities to which resources must be allocated is long, but for present purposes it can be reduced to a few primary headings. First is growth, which is a prerequisite for everything else. Second is ongoing maintenance and repair, to Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:24:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.002

Introduction

Non-ageing

Fitness

Ageing

25

Investment in maintenance Figure 2.2

The curve shows the predicted relationship between fitness and the investment in maintenance. The disposable soma theory predicts that the optimal investment in maintenance (that which maximises fitness) should be less than would be required to make the body potentially capable of remaining in sound condition indefinitely (i.e. non-ageing).

counteract the inevitable tendency for damage and errors to arise within molecules, cells, tissues and organs. Third is reproduction, which in many organisms draws a considerable fraction of the available resources. Other headings might include storage, to provide insurance against environmental fluctuation in resource availability, and defence, of which the immune system that protects against infections by microorganisms and parasites is one of the vital elements. The requirement for the organism to service the various physiological activities just listed means that there is an inevitable conflict of priorities. An organism might invest heavily in growth at the expense of maintenance only to grow a body that lacks the durability to live long enough to reproduce. Or an organism that invests heavily in maintenance might preserve a high degree of cellular integrity but, other things being equal, grow more slowly and produce fewer offspring. These trade-offs form the basis for the study of evolutionary physiology and life history theory (Stearns 1992; Townsend & Calow 1981). With regard to the evolution of ageing and longevity, the key question is how much the organism should allocate to maintenance and repair. The disposable soma theory asserts that in very general circumstances, the optimal level of investment in maintenance will be less than what would be required for the body to maintain its state indefinitely. Figure 2.2 illustrates the predicted relationship between fitness, here defined as the intrinsic rate of natural increase within the population (given by the root r of the EulerLotka equation ∫e−rx l(x)m(x)dx = 1, where l(x) and m(x) denote survivorship and fecundity at age x), and the investment in maintenance. In broad terms, it is predicted that too small an investment in maintenance results in low fitness through premature death. As the investment in maintenance increases, fitness increases too, but it reaches a maximum and thereafter declines as further increase in maintenance reduces fecundity. Mathematical models to substantiate this prediction were developed by Kirkwood and Rose (1991) and Drenos and Kirkwood (2005). Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:24:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.002

26

The Disposable Soma Theory

In purely verbal terms, the disposable soma concept may be understood according to the following sequence: (1) it is important for the organism to invest in sufficient maintenance that the body does not fall apart too soon; (2) however, most organisms in natural (wild) environments die young from extrinsic hazards, and there is little to be gained from investing in better maintenance than is required to keep the body in reasonably sound condition through the typical survival period experienced in the wild; and (3) therefore, under pressure of natural selection to make optimal use of resources, it was a higher evolutionary priority to invest in growth and reproduction than in maintaining a body well enough to last in good condition indefinitely, when the potential utility of such indefinite survival is extremely unlikely to be realised. The significance of the disposable soma theory is that it explains not only why ageing occurs but also how it is caused, the primary mechanisms of ageing being predicted to involve the accumulation of molecular and cellular defects. Indeed, the theory sprang from studies on mechanisms, rather than from a primarily evolutionary standpoint.

Origins At the core of the genesis of the disposable soma theory was the question of how a cell could maintain its molecular integrity in the face of the many sources of molecular error that would arise as an inevitable consequence of noise and damage within the fundamental biochemistry of living systems. During the mid-1970s, there was considerable interest in the idea that cells might become destabilised and thereby precipitated onto a pathway of progressive deterioration, in which not only would new errors continually arise but also there could be a self-amplifying ‘error catastrophe’ caused by the propensity for damaged components of the machinery for macromolecular biosynthesis to generate further faults (Orgel 1963, 1970, 1973). The same period had also seen the emergence of new ideas about how enhanced fidelity in macromolecular synthesis could be secured by mechanisms of kinetic proofreading (e.g. Hopfield 1974; for a comprehensive overview of this field, see also Kirkwood et al. (1986)). Theoretical analysis had shown that a cell might be stable or unstable in the face of error propagation, according not only to the specificity with which new building blocks were incorporated into nascent macromolecules but also to the extent that erroneous macromolecules were allowed to persist within the cell (Kirkwood & Holliday 1975). Specificity could be enhanced to an arbitrarily high level but only at the expense of either energy or time (the latter being of particular significance in schemes for kinetic amplification of accuracy by delaying the progression to the next step of macromolecular synthesis in order to allow a greater chance for mismatched units to dissociate, through a concept term ‘kinetic amplification’) (Ninio 1975). Erroneous macromolecules could be removed through active processes of targeted degradation (now termed ‘autophagy’), which also require expenditure of energy and/or time. The outcome of the theoretical analysis showed the property illustrated in Figure 2.3. The relationship of the molecular integrity of the cell at a time t + 1 would depend on the molecular integrity of the cell at the previous time point t, according to an S-shaped curve. Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:24:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.002

Molecular integrity, generation t+1

Origins

The margin of safety is increased (reduced) by increasing (reducing) the energy invested in molecular proofreading and error elimination

27

Reducing the margin of safety to a minimal level saves considerable energy but leaves the cell highly vulnerable to accumulation of defects

Molecular integrity, generation t Figure 2.3

The relationship between a cell’s molecular integrity at time t + 1 and at time t, according to theoretical analysis by Kirkwood and Holliday (1975). The dashed diagonal line indicates equality of molecular integrity at time t + 1 and at time t. The filled circles indicate points where the theoretical curves intersect the diagonal line. For each curve, the upper of these points is a point of potential stability; should molecular integrity fluctuate below this value, it will increase again during the next time unit, provided that it does not pass the lower of each pair of points, where the cell becomes doomed to error catastrophe.

In Figure 2.3, if the investments in specificity of macromolecular synthesis and/or clearance of damaged macromolecules were great enough, any cell that started with a high degree of molecular integrity would be likely to remain stable through time. From a state of complete integrity, errors would result in a fall to the upper stable point. Stochastic fluctuations would then occur around this stable point. Provided that these fluctuations remained small enough, the risk that the cell would pass the lower point of intersection of the curve with the 45-degree line of equality, beyond which decline to error catastrophe would be inescapable, would remain negligible. In this scenario, the cell would survive (at least in principle) indefinitely, or at least until some major disruptive event should destroy it. Such a scenario was suggested to be characteristic of cells of the germ line, for which progressive deterioration through time would doom the lineage to extinction. The theoretical analysis by Kirkwood and Holliday (1975) also showed how a cell could economise on its investments in specificity and/or degradation only at the expense of shrinking the margin of safety between the upper and lower intersection points of the curve with the 45-degree line of equality. If this strategy were taken too far, however, the cell would lose stability altogether or would operate in a quasi-stable condition, where stochastic fluctuation in macromolecular integrity could easily trigger an irreversible breakdown. Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:24:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.002

28

The Disposable Soma Theory

By recognising that the requirements for stability are different for germ-line and somatic cells in multicellular organisms where such a separation of cellular functions has evolved, the disposable soma theory suggested that senescence might be due to ‘an energy-saving strategy of reduced error regulation in somatic cells’ (Kirkwood 1977). The strategy would be evolutionarily stable in most ecological contexts because the alternative of maintaining high fidelity in somatic as well as germ cells would have caused an individual organism to grow or reproduce more slowly or to be subject to a greater risk of starvation or accidental death while seeking food.

Widening the Scope In the years following the original articulation of the disposable soma theory, the concept was developed to have wider scope both in terms of the molecular mechanisms thought to be involved and in terms of the kinds of life history patterns to which its core principle – optimising the allocation of resources between maintenance and other physiological functions such as reproduction, growth and so on – might be applied. The idea of cyclic propagation of errors in macromolecular synthesis, and especially in protein biosynthesis, was subjected to substantial experimental and theoretical test (Kirkwood 1980; Kirkwood et al. 1984; Rosenberger 1991). The results were provocative: although several experiments pointed to the possibility of error propagation, others indicated that Orgel’s idea of a propagation driven purely by feedback within the protein synthesising machinery was not supported. The advent of novel genomics technologies shifted the emphasis more towards the role in ageing of mutations in nuclear and mitochondrial DNA and away from proteins. However, recent renewal of interest in the role of accumulation of defective proteins and of protein clearance in ageing and agerelated diseases has indicated the importance of such mechanisms (Vilchez et al. 2014a, 2014b). In general, it is now recognised that ageing most likely involves multiple kinds of molecular damage and multiple mechanisms driving its accumulation. This has been advanced through the development of ‘network’ models that demonstrate the importance of interactions among various mechanisms (Kirkwood 2011; Kowald & Kirkwood 1996; Sozou & Kirkwood 2001). Nevertheless, the core recognition that holding the accumulation of damage in check is metabolically expensive is widely accepted. The second generalisation of the scope of the disposable soma theory has involved extension from the ‘classic’ metazoan model (e.g. human or mouse), where there is clear segregation between germ line and soma, to cases where there is much greater diversity in the physiological and anatomical form of the organism and in the arrangement of its life history (Kirkwood 1981). Some of the major dimensions of this diversity include unicellularity versus multicellularity, asexual versus sexual reproduction (which may not be an absolute distinction but may include opportunities for both kinds of reproduction within the individual life history), semelparity versus iteroparity of reproduction, polyphenism and monophenism, and so on. In all cases, there remains a fundamental shared issue of how the individual organism (a concept which itself may require careful consideration) (e.g. see Buss 1988) optimally Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:24:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.002

Relationship to Genetic Theories on Evolution of Ageing

29

allocates its resources between, on the one hand, maintenance and repair and, on the other hand, other essential functions. Thus, the core concept of the disposable soma theory is open to appropriate generalisation. For example, the observation by Martinez (1998) that Hydra appear to grow older without showing signs of senescence is entirely consistent with the principle that the evolution of disposability of the soma presupposes the existence of a clear germ–soma distinction. Of particular interest with regard to organismal diversity has been the discovery that unicellular organisms such as yeasts and bacteria can exhibit forms of senescence (Kirkwood 2005). At first sight, such discoveries were surprising, since the initial supposition was that unicellular organisms were necessarily endowed with germ-like immortality. In the case of the budding yeast Saccharomyces cerevisiae, it was possible to stretch a point, since the size asymmetry between mother and daughter cells made it possible to view the larger mother cell as a kind of soma, consistent with her observed replicative senescence, while the smaller daughter cell could be equated with the germ line (Lai et al. 2002). This was less easy, however, in the case of bacteria such as Escherichia coli, in which there was no difference in size between the twin products of cell division. The deeper enquiry that was prompted by these discoveries revealed, however, that there can be molecular asymmetry that differentiates outwardly identical daughters, since the age of the polar material of daughter cells is different (Stewart et al. 2005). Extension of the concept originally envisaged by contrasting immortal germ line with mortal soma is thus called for in order to extend the range of this line of thinking to the remarkable diversity in patterns of senescence across the phylogenetic spectrum (Finch 1990; Jones et al. 2014).

Relationship to Genetic Theories on Evolution of Ageing With its origin in mechanistic studies of cellular ageing, the disposable soma theory was proposed independently from two earlier concepts that had put forward genetic theories on the evolution of ageing. These are the ‘mutation accumulation theory’ of Medawar (1952) and the ‘antagonistic pleiotropy theory’ of Williams (1957) (note, however, that in neither case were these names used in the first articulations of the theories but were assigned to them at later dates). Both of the genetic theories shared the recognition that because of extrinsic mortality, there is a progressive weakening of the force of selection with advancing age, a principle that was analysed in greater detail by Hamilton (1966). For the mutation accumulation theory, the decline in survivorship means that by an age when very few individuals remain alive, the force of selection is too weak to counteract the accumulation of any germ-line mutations with late-acting deleterious effects. (It is important to emphasise that these mutations are germ-line mutations, since the theory is sometimes confused with the mechanistic somatic mutation theory, which is concerned only with the possible role of accumulating somatic mutations within the lifetime of an individual organism.) Over many generations, it was proposed that a wide range of alleles with late deleterious effects could accumulate within genomes. In the wild environment, such an accumulation would have little noticeable effect, since most Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:24:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.002

30

The Disposable Soma Theory

individuals would have died from other causes before reaching the age at which the deleterious effects would be felt. If organisms were moved, however, to protected environments, a greater fraction of individuals would live long enough to encounter the deleterious effects, whose cumulative impact would be to cause senescence. The antagonistic pleiotropy theory begins from the same logic as the mutation accumulation theory but instead of assuming alleles that simply have late deleterious effects (and are neutral in earlier life), it proposes the existence of alleles that have beneficial effects during growth and development but which produce harmful effects at later ages. As a hypothetical example, Williams (1957) considered an allele regulating calcium metabolism that promotes bone growth during development but which, if it remains continually active, causes calcification of arteries in later life. Thus, the antagonistic pleiotropy theory introduces the idea of a trade-off, but it should be noted that this is specifically a trade-off deriving from the differential effects in early and late life of an individual gene. It may readily be seen that the genetic theories share with the disposable soma theory the principle that it is the inevitability of mortality that means that the later portions of the life history have reduced evolutionary significance. There is therefore some broad overlap between their implications, but there are also major differences. The mutation accumulation theory sees senescence as driven chiefly by the neutrality that is imposed on late mutations by the ‘selection shadow’ that is cast by cumulative mortality. There is also a potentially serious difficulty for the mutation accumulation theory in that in order to regulate lateness of expression of the harmful effects of the postulated alleles, some kind of time-keeping mechanism is required but has not been satisfactorily identified; as pointed out by Kirkwood (1977), this renders the theory incomplete, or even circular. While subsequent theory has delivered considerable insight into the shaping of patterns of mutations with age-specific action and their effects on the life history (Charlesworth 2001; Wachter et al. 2013, 2014), this fundamental issue has not been fully resolved. Even when it is argued that deleterious alleles can have small deleterious effects early in life and that these simply accumulate with time, becoming eventually more deleterious, this scenario differs from the original proposition to explain the evolution of senescence in an organism in which old and young are physiologically the same. Kirkwood’s point that the mutation accumulation hypothesis was ‘either circular or incomplete’ retains its validity; indeed, it allowed for the possible construction of the necessary mechanism, but whether or not this has yet been done convincingly is still a moot point. The disposable soma theory is closer to the antagonistic pleiotropy theory in its implications, to the extent that it is sometimes described as a special case. This, however, is not correct. At a stretch, it could be argued that the genes responsible for regulating investments in maintenance are pleiotropic in that they boost survival but consume resources that might otherwise be used for growth and reproduction. But in order properly to fit the concept of antagonistic pleiotropy, it is necessary to invert these effects, since it is by depressing the action of maintenance genes that an early-life benefit of enhanced growth and reproduction is generated at the expense of late senescence. The two theories are perhaps more accurately seen as being complementary to each other rather than overlapping. Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:24:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.002

Evolution of Longevity

31

A limitation of the purely genetic theories of mutation accumulation and antagonistic pleiotropy is that they provide a rather inflexible framework for gene action in senescence. Leaving aside the interesting and important question of whether alleles of the relevant kinds can be detected in sufficient numbers to explain senescence, they cannot without substantial modification and addition explain the diversity of senescence considered earlier. For example, they do not easily accommodate the cases of polyphenism seen in important invertebrate models for the study of senescence (Flatt et al. 2013).

Evolution of Longevity It is important for an evolutionary theory of senescence to explain the divergence of species’ life spans. The concept of a life span is itself a matter for careful definition, since the distribution of the lengths of individual lives within the population will be influenced by the presence or absence of senescence, the nature of the senescence process, the quality of the environment, the sample size and so on. For present purposes, we will regard the species-specific life span as the age by which the great majority of individuals will have died; to make this more specific, we might take it to be the 99th centile of the distribution of the lengths of individual lives in the natural habitat of the species in question. By applying the essential logic of the disposable soma theory, it is straightforward to understand how, for example, a short-lived species such as a mouse might evolve the longevity of a longer-lived animal such as a bat. This is illustrated in Figure 2.4. For the mouse, the mortality pressure in the natural habitat is high and the life span correspondingly short. This makes investment in maintenance a relatively low priority so that the period through which the soma will, on average, remain in sound condition is also short. In consequence, individuals will undergo senescence quite soon, even if they are kept in a protected environment. Now suppose that the mouse evolves the means of flight, effectively evolving into a bat. Through this adaptation, the mortality pressure in the new natural habitat is reduced, and the survival curve shifts accordingly. If nothing else were to change, however, a significant fraction of individuals would expect to live beyond the period through which the soma remains sound. These individuals which might have lived longer will not do so because they succumb to senescence. Therefore, the full benefit of the adaptation may not be realised. Consequently, it makes sense for the bat that the investment in maintenance should evolve to be greater, with the result that the soma remains sound for longer, and senescence is postponed. The theory therefore predicts that longer-lived species acquire their greater longevity through investing more in somatic maintenance, for which there is growing support (e.g. Kapahi et al. 1999). At the same time, however, that the reduced mortality resulting from the adaptation of flight is creating the potential to evolve better somatic maintenance, any increase in maintenance is likely to impose a cost on the rates of reproduction and growth. Thus, there is expected to be a shift in several traits that make up the newly evolved life history. In order fully to understand the ways through which selection acts upon the determinants of longevity, the effects on the schedules of both survival and fecundity need to be Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:24:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.002

32

The Disposable Soma Theory

Survivorship

B

A

Sound soma: A Sound soma: B

A

B Age

Figure 2.4

Schematic representation of the possible steps in evolution of longevity (see text). The diagram shows a short-lived species (A) and a longer-lived species (B). For each species, the continuous curve indicates survivorship in the natural (wild) environment, and the dashed curve indicates survivorship in a protected environment where senescence is more readily apparent. If species A evolves an adaptation (e.g. flight) that reduces extrinsic mortality and thereby increases survivorship, becoming the new species B, this creates the potential for selection pressure to increase investment in maintenance. As a result, the soma of species B is kept in sound condition for longer, and senescence is postponed. This process could also operate in the reverse direction, as in the case of flightless birds which appear to have surrendered the benefits of flight (perhaps in favour of higher rates of reproduction) and reverted to having shorter life spans.

examined together. This has been done in a preliminary way for the disposable soma theory in a model described by Kirkwood and Rose (1991). Interestingly, it is not enough solely to lower the level of extrinsic mortality. If extrinsic mortality is considered a constant γ so that it enters into the survivorship term l(x) of the Euler-Lotka equation as a factor e−γx, this simply changes the intrinsic rate of natural increase r by a corresponding quantity, since mathematically γ and r are completely confounded with each other, and this produces no shift in the optimal investment in maintenance (or any other life history investment). This point was implicit in the analysis by Kirkwood and Rose (1991) and was made fully explicit by Caswell (2007). In order that a change in γ should result in a change in the optimal investment in maintenance, it was necessary to impose some kind of ecological constraint. Kirkwood and Rose (1991) achieved this by stipulating that the value of r should remain equal to zero at the optimum (i.e. highest point of the fitness curve in Figure 2.3), that is, that the optimal life history should be one that produces a population at equilibrium within its environment.

Plasticity and State Dependence The disposable soma theory and the two genetic theories on the evolution of ageing were originally developed with a view to organisms existing in given environments that were implicitly considered fixed. Such an approach was appropriate to address the primary Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:24:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.002

Plasticity and State Dependence

33

questions of why senescence might have evolved and what determines species-specific life spans. However, most organisms live in conditions in which the environment may vary either across time or across space (or both). A well-studied phenomenon in the biology of ageing concerns the effect of varying the food supply (dietary restriction). There will often be variability associated with seasonality, climatic fluctuation and many other causes. Such variability might simply have non-adaptive effects on ageing processes (e.g. by increasing or lowering damage-inducing stress), but there may also evolve a capacity for plastic responses in the life history that have an adaptive basis (e.g. see Bateson et al. 2004). It may also be the case that stochastic factors acting upon the organism during its life course will result in changes of state for which evolution may have fashioned statedependent responses, such as DNA damage-response or protein stress-response pathways, which have effects on the subsequent course of the physiology of senescence. These kinds of plastic or state-dependent responses are hard to fit within the framework of the specific kinds of alleles envisaged by the mutation accumulation and antagonistic pleiotropy theories, but they are relatively straightforward to encompass within the optimality principles that underlie the disposable soma theory. We can illustrate this with the case of dietary restriction as follows. At first sight, it might seem surprising that reducing the food supply should increase longevity, as was first observed in laboratory rodents and has now been recorded in several other species, when the initial expectation would be that less food means less energy for bodily functions, including maintenance. However, it was suggested by Harrison and Archer (1989) and Holliday (1989) that during periods of famine, the organism may do better to divert resources away from reproduction and towards extra maintenance, if such a strategy would preserve functionality until the period of famine has passed. While such an idea has intuitive appeal, its plausibility requires that it be shown that to divert resources from reproduction to maintenance during dietary restriction does actually produce a numerical increase in fitness. To date, this has been demonstrated only once, for the case of dietary restriction in mice. Shanley and Kirkwood (2000) used the framework of the disposable soma theory to test the hypothesis that it might be evolutionarily adaptive to redirect resources away from reproduction and towards increased somatic maintenance during short periods of famine. They developed a mathematical life history model of dynamic resource allocation in which senescence was modelled as a change in state that depended on the resources allocated to maintenance. Individuals were assumed to allocate the available resources to maximise the total number of descendants using the modelling technique of dynamic programming. The model showed that the evolutionary hypothesis was indeed plausible and identified two factors, both likely to exist, that favour such an outcome. The first of these factors was that survival of juveniles should be reduced during periods of famine, and the second was that the organism should have to pay an energetic ‘overhead’ before any litter of offspring can be produced; that is, an initial amount of resources would need to be committed to preparing the reproductive system before it becomes operational. If neither of these conditions held within the model, there was no evolutionary advantage to be gained from switching extra resources to maintenance. Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:24:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.002

34

The Disposable Soma Theory

By establishing pre-conditions for the adaptive plausibility of the dietary restriction response in rodents, this approach established a basis that might be used to evaluate whether the life-extending effects of calorie restriction could be adaptive also in other species.

Further Challenges In order to develop a full understanding of how senescence has been shaped through evolutionary processes, it is important that answers be provided not only to ultimate questions of why senescence occurs but also to the proximate factors that make up the mechanistic processes that drive ageing (and its diversity). The disposable soma theory has provided a starting point in developing this integrated perspective, and further developments along similar lines may prove fruitful in addressing further challenges. Testing the disposable soma theory itself should be a part of addressing such challenges. As we have seen, the theory has contributed insight into biology of ageing, and much data have been generated that are consistent with its broad predictive framework. Core predictions that are integral to the essence of the theory are these: germ cells are better protected against accumulation of damage than somatic cells, increased life span results from increased investment in somatic maintenance and increased life span involves evolutionary trade-off (note that it is important that the cost of such a trade-off on other components of the life history should be demonstrated through effects on fitness in an evolutionary context – costs that may be exposed by laboratory manipulation alone are not sufficient to test the theory). One of the intriguing questions in the biology of senescence concerns the distinction between immortality of germ line and mortality of soma. Although it can be taken as more or less axiomatic that the germ line is endowed with effective immortality, all extant organisms today being products of unbroken lineages of living cells that are several billion years old, individual germ cells are vulnerable to molecular damage during every second of their existence. The challenge is to understand how the immortality of the germ line is secured and to extend this understanding to embrace the many kinds of organism in which the distinction between germ line and soma is less clear-cut than it is in higher animals (Bell 1984; Kirkwood 1981). There are broadly three ways through which germ-line immortality can be secured: better maintenance, stringent cell selection in the germ-cell lineage (in order to delete cells that are compromised in their molecular integrity) and selection against progeny carrying faults that have arisen within the germ-cell lineage. It seems likely that all three of these mechanisms will operate together, although priorities may vary according to the anatomy and physiology of the species in question. In the case of mammals, Saretzki et al. (2004, 2008) reported that the induction of differentiation of embryonic stem cell lines (effectively germ line) into embryoid bodies was accompanied, within a very short time (about two days), by a generalised down-regulation of cell maintenance systems. This conforms strikingly to the prediction made within the original proposal of the disposable soma theory that ‘ageing may, therefore, be the result of . . . switching off the mechanisms responsible for Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:24:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.002

Further Challenges

35

high accuracy at or around the time of differentiation of somatic cells from the germ line’ (Kirkwood 1977). Intriguingly, recent data suggest that there may sometimes be communication between germ line and soma, operating apparently bi-directionally, through the discovery of a systemic response to DNA damage in germ cells of Caenorhabditis elegans that protects somatic tissues (Ermolaeva et al. 2013). If demonstrated to be a general phenomenon, such a response adds further complexity to this fascinating subject (Douglas & Dillin 2014) An issue that has been under-researched to date concerns the actual energy costs incurred by different physiological processes that shape the life history. While for an organism like an adult human it is clear that most of the everyday metabolic flux of energy is concerned with maintenance, since growth is over, daily additions to storage are modest and average daily expenditure on reproduction is small. However, the distribution of energy among different maintenance activities and the magnitudes of the expenditures on individual maintenance processes are largely unknown. In a classic study on the cost and evolution of accuracy in protein synthesis, Savageau and Freter (1979) calculated that the total cost merely of proofreading the aminoacylation of transfer RNAs is about 2 per cent of the energy required to synthesise a bacterial cell. Since this is but one of a very long list of molecular processes that underpin the maintenance of integrity of living cells, it can be expected that the overall cost of maintenance processes, most of which we simply take for granted, may be very high. It will be good to have better data not only on the costs of maintenance but also on other processes and to study how these may change according to variations within individuals and populations across time, state and space. Another issue needing more data concerns the nature and mechanisms of life history trade-offs. Along with much of life history theory, the disposable soma theory presumes that trade-offs exist. There is plenty of evidence that this is true, but we need better to understand how trade-offs operate, especially within the context of testing theoretical predictions by experiments. In the case of the disposable soma theory, the central argument is that selection has acted upon the investment in maintenance versus investments in growth or reproduction or other aspects of physiology. The theory does not require that any trade-off should be specifically between maintenance and reproduction, although this is a likely first candidate. It is important to appreciate when testing predictions from life history theory that the predictions concern trade-offs at the evolutionary life history level. They are silent on the question of whether or not there is direct mechanistic coupling between mechanisms regulating maintenance, reproduction or other functions. Therefore, an experiment that directly manipulates maintenance, reproduction or other mechanism will not necessarily have any relevance for testing whether an evolutionary trade-off exists, especially if such an experiment is conducted under laboratory conditions where resources are generally not limiting. Testing the existence of evolutionary trade-offs can, however, be undertaken using manipulation of selection forces in ecological contexts or through imposing artificial selection. A good example of the latter approach has been the use of artificial selection on age at reproduction in the fruitfly Drosophila melanogaster, in which ‘young’ lines of Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:24:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.002

36

The Disposable Soma Theory

flies were propagated in each generation from adults that had just emerged as adults, while ‘old’ lines were propagated from adults that had already been reproducing for some time. Compared with the ‘young’ lines, the ‘old’ lines were selected for increased life span because, in order to contribute to the next generation, they must have survived to the time of egg collection. Greater longevity of ‘old’ lines has consistently been found in these studies (e.g. Luckinbill et al. 1984; Roper et al. 1993; Rose 1984). In an alternative approach, direct selection for increased life span has been accomplished using an experimental design that exploited the effect of temperature on Drosophila longevity (Zwaan et al. 1995). In this experiment, the progeny from each pair of breeding flies were divided into two groups, one of which was maintained at 15°C and the other at 29°C. At 29°C, maximum life span was less than sixty days, whereas at 15° C, mean life span was greater than 120 days. This meant that life spans could be measured on the groups maintained at 29°C and the longest-lived families selected, at a time when the sibling groups kept at 15°C were still reproductively viable. If trade-offs are important in ageing, then we would expect the long-lived flies resulting from artificial selection experiments to pay for their late-life advantage at some point in their life history. Indeed, the long-lived flies were shown to have lower fecundity in early adult life (e.g. Luckinbill et al. 1984; Rose 1984; Zwaan et al. 1995). Underlying such experiments to demonstrate trade-offs through artificial selection is the requirement that the population should have standing genetic variation with respect to the determinants of the life history. From Figure 2.2 it is easy to see why the presence of such variation can be expected. The fitness peak at the optimal investment in maintenance has a very flat top. Therefore, within the region around this peak, the values of the intrinsic rate of natural increase r will change very slowly with small deviations from the actual optimum. An investment in maintenance somewhat less than the optimal will result in faster senescence and slower growth or reproduction, whereas an investment in maintenance somewhat more than the optimal will have the reverse effects. However, natural selection will be unlikely to be able to discriminate very effectively between these alternatives until the deviations from the optimal become larger. This will be the case particularly if multiple alleles determine the relevant investments. The possibility of exploiting standing genetic variation in life history determinants was used to test for life history trade-offs in a historical human population by Westendorp and Kirkwood (1998), who examined records of fertility and longevity in a large sample of British aristocrats. The data revealed that female aristocrats who died at the highest ages had significantly reduced fertility. Although confounding factors cannot be entirely ruled out when interpreting such data, it is noteworthy that, on average, it might be expected that phenotypic correlations (whereby healthier individuals should both have lived longer and been more successful at reproducing) would have tended to suppress the emergence of a trade-off between fertility and longevity. Thus, the observed relationship may well have underestimated any true biological trade-off. Looking to the future, it is clear that we have much still to learn about the factors that have shaped the complex biology of senescence. As Kirkwood (1977) noted, ‘[T]here Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:24:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.002

References

37

are three requirements for a satisfactory theory of ageing: it must be theoretically plausible; it must be supported by experimental evidence; and it must make evolutionary sense.’ The latter requirement has particular importance for the study of senescence since, without this condition being satisfied, it is all too easy to misinterpret the phenomenology.

References Bateson, P., Barker, D., Clutton-Brock, T., et al. (2004). Developmental plasticity and human health. Nature, 430, 419–21. Bell, G. (1984). Evolutionary and nonevolutionary theories of senescence. American Naturalist, 124, 600–3. Buss, L. W. (1988). The Evolution of Individuality (Princeton, NJ: Princeton University Press). Caswell, H. (2007). Extrinsic mortality and the evolution of senescence. Trends in Ecology and Evolution, 22, 173–4. Charlesworth, B. (2001). Patterns of age-specific means and genetic variances of mortality rates predicted by the mutation-accumulation theory of ageing. Journal of Theoretical Biology, 210, 47–65. Douglas, P. M. & Dillin, A. (2014). The disposable soma theory of aging in reverse. Cell Research, 24, 7–8. Drenos, F. & Kirkwood, T. B. L. (2005). Modelling the disposable soma theory of ageing. Mechanisms of Ageing and Development, 126, 99–103. Ermolaeva, M. A., Segref, A., Dakhovnik, A., et al. (2013). DNA damage in germ cells induces an innate immune response that triggers systemic stress resistance. Nature, 501, 416–20. Finch, C. E. (1990). Longevity, Senescence and the Genome (University of Chicago Press). Flatt, T., Amdam, G. V., Kirkwood, T. B. L. & Omholt, S. W. (2013). Life-history evolution and the polyphenic regulation of somatic maintenance and survival. Quarterly Review of Biology, 88, 185–218. Hamilton, W. D. (1966). The moulding of senescence by natural selection. Journal of Theoretical Biology, 12, 12–45. Harrison, D. E. & Archer, J. R. (1989). Natural selection for extended longevity from food restriction. Growth Development and Aging, 53, 3. Holliday, R. (1989). Food, reproduction and longevity: is the extended lifespan of calorie-restricted animals an evolutionary adaptation? BioEssays, 10, 125–7. Hopfield, J. J. (1974). Kinetic proofreading: a new mechanism for reducing errors in biosynthetic processes requiring high specificity. Proceedings of the National Academy of Sciences USA, 71, 4135–9. Jones, O. R., Scheuerlein, A., Salguero-Gómez, R., et al. (2014). Diversity of ageing across the tree of life. Nature, 505, 169–73. Kapahi, P., Boulton, M. E. & Kirkwood, T. B. L. (1999). Positive correlation between mammalian life span and cellular resistance to stress. Free Radicals in Biology and Medicine, 26, 495–500.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:24:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.002

38

The Disposable Soma Theory

Kirkwood, T. B. L. (1977). Evolution of aging. Nature, 270, 301–4. Kirkwood, T. B. L. (1980). Error propagation in intracellular information transfer. Journal of Theoretical Biology, 82, 363–82. Kirkwood, T. B. L. (1981). Repair and its evolution: survival vs reproduction. In Physiological Ecology: An Evolutionary Approach to Resource Use, ed. C. R. Townsend & P. Calow, pp. 165–89 (Oxford, UK: Blackwell Scientific). Kirkwood, T. B. L. (2005). Asymmetry and the origins of ageing. Mechanisms of Ageing and Development, 126, 533–4. Kirkwood, T. B. L. (2011). Systems biology of ageing and longevity. Philosophical Transactions of the Royal Society of London Series B, 366, 64–70. Kirkwood, T. B. L. & Cremer, T. (1982). Cytogerontology since 1881: a reappraisal of August Weismann and a review of modern progress. Human Genetics, 60, 101–21. Kirkwood, T. B. L. & Holliday, R. (1975). The stability of the translation apparatus. Journal of Molecular Biology, 97, 257–65. Kirkwood, T. B. L. & Holliday, R. (1979). The evolution of aging and longevity. Proceedings of the Royal Society of London Series B, 205, 531–46. Kirkwood, T. B. L. & Rose, M. R. (1991). Evolution of senescence: late survival sacrificed for reproduction. Philosophical Transactions of the Royal Society London, B332, 15–24. Kirkwood, T. B. L., Rosenberger, R. F. & Galas, D. J. (1986). Accuracy in Molecular Processes: Its Control and Relevance to Living Systems (London: Chapman & Hall). Kowald, A. & Kirkwood, T. B. L. (1996). A network theory of ageing: the interactions of defective mitochondria, aberrant proteins, free radicals and scavengers in the ageing process. Mutation Research, 316, 209–36. Lai, C. Y., Jaruga, E., Borghouts, C. & Jazwinski, S. M. (2002). A mutation in the ATP2 gene abrogates the age asymmetry between mother and daughter cells of the yeast Saccharomyces cerevisiae. Genetics, 162, 73–87. Luckinbill, L. S., Arking, R., Clare, M. J., et al. (1984). Selection of delayed senescence in Drosophila melanogaster. Evolution, 38, 996–1003. Martínez, D. E. (1998). Mortality patterns suggest lack of senescence in Hydra. Experimental Gerontology, 33, 217–25. Medawar, P. B. (1952). An Unsolved Problem of Biology (London: H.K. Lewis)(reprinted in Medawar, P. B. (1957). The Uniqueness of the Individual. (London: Methuen)). Ninio, J. (1975). Kinetic amplification of enzyme discrimination. Biochimie, 57, 587–95. Orgel, L. E. (1963). The maintenance of the accuracy of protein synthesis and its relevance to ageing. Proceedings of the National Academy of Sciences USA, 49, 517–21. Orgel, L. E. (1970). The maintenance of the accuracy of protein synthesis and its relevance to ageing: a correction. Proceedings of the National Academy of Sciences USA, 67, 1476. Orgel, L. E. (1973). Ageing of clones of mammalian cells. Nature, 243, 441–5. Roper, C., Pignatelli, P. & Partridge, L. (1993). Evolutionary effects of selection on age at reproduction in larval and adult Drosophila melanogaster. Evolution, 47, 445–55. Rose, M. R. (1984). Laboratory evolution of postponed senescence in Drosophila melanogaster. Evolution, 38, 1004–10.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:24:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.002

References

39

Rosenberger, R. F. (1991). Senescence and the accumulation of abnormal proteins. Mutation Research, 256, 255–62. Saretzki, G., Armstrong, L., Leake, A., et al. (2004). Stress defense in murine embryonic stem cells is superior to that of various differentiated murine cells. Stem Cells, 22, 962–71. Saretzki, G., Walter, T., Atkinson, S., et al. (2008). Downregulation of multiple stress defense mechanisms during differentiation of human embryonic stem cells. Stem Cells, 26, 455–64. Savageau, M. A. & Freter, R. R. (1979). On the evolution of accuracy and cost of proofreading tRNA aminoacylation. Proceedings of the National Academy of Sciences USA, 76, 4507–10. Shanley, D. P. & Kirkwood, T. B. (2000). Calorie restriction and aging: a life-history analysis. Evolution 54, 740–50. Sozou, P. D. & Kirkwood, T. B. L. (2001). A stochastic network model of cell replicative senescence based on telomere shortening, oxidative stress and somatic mutations in nuclear and mitochondrial DNA. Journal of Theoretical Biology, 213, 573–86. Stearns, S. C. (1992). The Evolution of Life Histories (Oxford University Press). Stewart, E. J., Madden, R., Paul, G. & Taddei, F. (2005). Aging and death in an organism that reproduces by morphologically symmetric division. PLoS Biology, 3, e45. Townsend, C. R. & Calow, P. (1981). Physiological Ecology: An Evolutionary Approach to Resource Use (Oxford, UK: Blackwell Scientific). Vilchez, D., Simic, M. S. & Dillin, A. (2014a). Proteostasis and aging of stem cells. Trends in Cell Biology, 24, 161–70. Vilchez, D., Saez, I. & Dillin, A. (2014b). The role of protein clearance mechanisms in organismal ageing and age-related diseases. Nature Communications, 5, 5659. Wachter, K. W, Evans, S. N. & Steinsaltz, D. (2013). The age-specific force of natural selection and biodemographic walls of death. Proceedings of the National Academy of Sciences USA, 110, 10141–8. Wachter, K. W, Steinsaltz, D. & Evans, S. N. (2014). Evolutionary shaping of demographic schedules. Proceedings of the National Academy of Sciences USA, 111, 10846–53. Weismann, A. (1891). Essays upon Heredity and Kindred Biological Problems, Vol. 1 (2nd edn.) (Oxford, UK: Clarendon Press). Westendorp, R. G .J. & Kirkwood, T. B. L. (1998). Human longevity at the cost of reproductive success. Nature, 396, 743–6. Williams, G. C. (1957). Pleiotropy, natural selection and the evolution of senescence. Evolution, 11, 398–411. Zwaan, B. J., Bijlsma, R., & Hoekstra, R. F. (1995). Direct selection of lifespan in Drosophila melanogaster. Evolution, 49, 649–59.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:24:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.002

3

A Hamiltonian Demography of Life History Michael R. Rose, Lee F. Greer, Kevin H. Phung, Grant A. Rutledge, Mark A. Phillips, Christian N. K. Anderson and Laurence D. Mueller

Short Summary We have been developing an approach to the study of life history based chiefly on evolutionary theories that depend on Hamilton’s forces of natural selection. After almost forty years of work in this area, we realised that we have been assembling the pieces of an overarching research programme for demography that is distinctively Hamiltonian. Here we attempt to sketch, in an admittedly somewhat allusive and summary form, that overall research programme. We have not completed our work on this fairly broad research programme, and indeed here we point out some components that remain incomplete or barely initiated. To summarise our overview, we delineate here the following elements of our Hamiltonian demography: (1) a priori evolutionary genetic theory, (2) numerically generating demographic patterns using such theory, (3) testing the underlying theory using experimental evolutionary biology, (4) applying the findings of this experimental research to wild populations and (5) inferring ageing phase transitions from demographic data. We do not claim that our research programme reflects a widespread consensus or the only practicable way forward for demography. We do claim, however, that it is one useful way forward for demography.

Introduction Demography, gerontology and their cognate fields have been influenced by the widespread notion that biological senescence of entire organisms involves some type of cumulative process of damage or accumulating physiological disharmony that is inherent to life. The influence of this assumption is to be found in the writings of most authors in these fields, starting with Aristotle (fourth century BCE) in his monograph, ‘On the Length and Brevity of Life’. It continues to prevail in most of the popular and academic publications on senescence, from Aubrey de Grey’s seven types of damage to innumerable technical works on the supposed molecular machinery of senescence (de Grey & Rae 2007). Dissenting from this view have been the works of many evolutionary biologists, starting with an obscure note from Alfred Russel Wallace (ca. 1870) and the early neoDarwinian works of August Weismann (1891). Such early evolutionary thinking on the topic began with group selectionist reasoning, in which senescence was conceived in Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:26:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.003

Introduction

41

terms of the benefits it might provide for the evolution of entire species or populations. But soon Weismann was developing early forerunners of more modern evolutionary genetic theories involving the weakening of natural selection (Kirkwood & Cremer 1982). From 1930 to 1957, R. A. Fisher, J. B. S. Haldane, P. B. Medawar and G. C. Williams articulated and expanded on verbal evolutionary theories of senescence based on the weakening of natural selection at later ages, with only rudimentary mathematical arguments (Fisher 1930; Haldane 1941; Medawar 1946, 1952; Williams 1957). From 1966 to 1980, W. D. Hamilton and Brian Charlesworth developed the first mathematically explicit versions of the idea that senescence evolves because the forces of natural selection decline during adulthood among populations with age-structured demography and no fissile reproduction (e.g. Charlesworth 1970, 1980; Hamilton, 1966). Starting in 1980 and continuing ever since, evolutionary biologists have published a variety of tests of the theoretical ideas of Hamilton, Charlesworth and others using quantitative genetics and experimental evolution, in effect working within what can now be seen as a Hamiltonian paradigm for ageing research, covering both senescence and post-senescence ageing (Luckinbill et al. 1984; Mueller et al. 2011; Rose 1984; Rose & Charlesworth 1980; Rose et al. 2002). Up to this point in time, the Hamiltonian approach to gerontological research has seen significant successes in its corroborative alignments of theory and experiment, at least by the standards of both evolutionary biology and gerontology. Although it remains controversial within gerontology, within evolutionary biology the Hamiltonian approach to senescence is seen as paradigmatic for the field as a whole. There is little controversy about it among evolutionists, at least with respect to the core theory and its direct tests. The basic result provided by Hamilton (1966) was to show the sensitivity of the intrinsic rate of increase r to changes in the log of age-specific survival P(x). Specifically in Hamiltonian terms, he showed that δr=δlnPð xÞ ¼ sð xÞ=T, where T is the generation time, and s ð xÞ ¼

X

ery lð yÞmð yÞ

y¼xþ1

and l(y) and m(y) are the familiar age-specific survival and fecundity parameters. Here we outline an overall programme for Hamiltonian research, with a particular focus on how it can be developed to provide a scientific paradigm for demography over all phases of ageing. We will not be dealing with the salience of Hamiltonian research for mechanistic work on the physiology or genomics of life history here; however, those are topics that we continue to work on (Rose & Burke 2011; Shahrestani et al. 2012a), as do many others (Forsberg et al. 2012; Hoffman et al. 2013; Polosak et al. 2010). More specifically, we wish to define an arc of connected research strategies that together we believe will eventually provide the foundations of a Hamiltonian demography: (1) a priori evolutionary genetic theory, (2) the alternative demographic patterns that such theory can generate, (3) testing the elements of such theory using experimental evolution and other empirical tools of evolutionary biology, (4) implications of experimental research for the predicted demographic patterns of wild populations and Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:26:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.003

42

A Hamiltonian Demography of Life History

(5) a look forward towards methods for inferring phase transitions from recently generated large bodies of demographic data. However, it is important to appreciate that this programme of research on the spectrum of life history patterns that arise over the course of ageing is one that is specifically our own. We are not attempting to enunciate a disciplinary consensus, not even for ‘evolutionary demography’, however that term might be construed. Nor do we wish to suggest that the kind of demography that we are interested in is the only interesting kind.

Hamiltonian Evolutionary Genetic Theory One way to view the theoretical line of attack pursued by Hamilton (1966) and Charlesworth (e.g. 1980) is that it amounts to a demonstration of the extent to which weakening effectiveness of natural selection at later ages leads to a loss of optimal tuning of age-specific life history characters at later ages. That is, while some theorists (e.g. Abrams & Ludwig 1995) continue to treat every aspect of life history from the standpoint of a Darwinian optimising demon of apparently infinite power and scope, our theoretical work has tended to focus on the extent to which natural selection is of progressively less benefit at later ages and may even produce harmful outcomes at later ages when there is antagonistic pleiotropy (Mueller & Rose 1996; Rose 1985), as both Medawar (1952) and Williams (1957) verbally conjectured. Hamiltonian evolutionary theory for life history has been developed for relatively simple cases to this point. Hamilton’s original 1966 paper did not offer explicit evolutionary genetic models, asymptotic state predictions or evolutionary trajectories. Charlesworth (1980) confined most of his theoretical analysis to evolutionary genetic systems with one segregating Mendelian locus. More recently, theory has been developed for the evolutionary stabilisation of mortality and fecundity rates at late ages (Charlesworth 2001; Mueller & Rose 1996; Mueller et al. 2011; Steinsaltz et al. 2005; Wachter et al. 2013), theoretical research motivated and underscored by the raw demographic phenomenon of late-life stabilisation uncovered in some notably large bodies of age-specific mortality-rate data (Carey et al. 1992; Curtsinger et al. 1992; Greenwood & Irwin 1939), age-specific fecundity data (Le Bourg & Moreau 2014; Rauser et al. 2003; Rauser, Mueller & Rose 2006) and age-specific male virility data (Shahrestani et al. 2012b). At this point, two basically distinct kinds of Hamiltonian evolutionary theory can be distinguished. The first, and simplest, involves models that assume age structure but generally clonal inheritance. In such models, Mendelian genetic transmission producing large-scale out-crossing and recombination is treated as either absent or rare. Instead, it is assumed that evolution proceeds by the successive fixation of beneficial mutants that modify life history in a well-defined manner. Examples of this type of theory are to be found in Mueller and Rose (1996) and Mueller et al. (2011). These studies have shown that natural selection will permit the establishment of genetic variants with antagonistic pleiotropic effects that increase early-life survival or fecundity as long as the negative fitness effects are relegated to later life. At sufficiently advanced ages when the force of Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:26:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.003

Hamiltonian Evolutionary Genetic Theory

43

selection becomes weaker than drift, the age specificity of the evolutionary forces acting on survival and fecundity are lost. This can then lead to the evolution of late-life plateaus for life history characteristics. Most importantly, from an experimental perspective, the age at which these plateaus start is predicted to be a function of the last ages of reproduction and survival and thus subject to evolution as these ages are varied (Rauser et al. 2006; Rose et al. 2002). The second kind of Hamiltonian evolutionary theory is that which is based on the full apparatus of Mendelian population genetics extended to age-structured populations. This is the work pioneered by Charlesworth (e.g. 1970) in the 1970s, to which he still contributes (e.g. Charlesworth 2001) and which has been recently extended (Wachter et al. 2013). Hamilton’s work was a major contribution to the theoretical study of life history evolution even though the theoretical centrepiece of his argument was largely heuristic (Hamilton 1966). Charlesworth extended this theory in important ways by showing that Hamilton’s forces of natural selection would properly predict the direction of selection in exact population genetic models (Charlesworth 1980). Recently, Baudisch (2005, 2008) has returned to the heuristic approach of Hamilton and pointed out that there exist alternative conceivable indicators of the effect of natural selection in an age-structured population – depending on what assumptions about the scale on which mortality effects at different ages should be used. More importantly, Baudisch showed that under certain circumstances, these alternative indicators gave different predictions than Hamilton’s. The problem is that there is no direct connection to Baudisch’s indicators and exact population genetic models as there is for Hamilton’s. For instance, see Charlesworth (1980: chap. 2), which develops such a connection for Hamilton’s sensitivity measure. Indeed, Mueller et al. (2011) show that one of Baudisch’s indicators gives incorrect predictions about the strength of selection. Recently, others have made contributions to this field as well (Steinsaltz et al. 2005; Wachter et al. 2013). For example, Wachter et al. (2013) studied conditions under which selection–mutation balance can produce late-life mortality plateaus. We would like to draw attention to new possibilities for theoretical research in this area. One of them is the transitional evolutionary dynamics that arise for life history when there are environmental changes. Some of these might be transitions in agespecific windows of reproduction, for which Rose et al. (2002) contribute some calculations. Other such transitions involve qualitative transitions in other environmental features that then have age-dependent heterogeneities in phenotypic impact. In particular, Phung et al. (in preparation) show that age-specific adaptation to a novel environment in which the optimal phenotype for survival has changed proceeds at different rates for genetic variants affecting early-life versus late-life phenotypes. The ages at which we expect to see a rapid shift from a maladaptive phenotype to a welladapted phenotype depend on a number of factors: the time since transitioning to the new environment, the strength of selection acting on new genetic variants, the magnitude of the phenotypic effects among new mutants and the effective population size. The application of the full apparatus of Mendelian population genetics to this spectrum of issues will be challenging but we believe very useful. Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:26:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.003

44

A Hamiltonian Demography of Life History

Theoretically Generated Hamiltonian Demographies There is an important bridging theory that has been somewhat neglected within the Hamiltonian research programme: generating predicted demographic patterns from explicit evolutionary genetic models. This type of task was first taken on by Mueller and Rose (1996), to the best of our knowledge, and since significantly elaborated on in Mueller et al. (2003, 2011). Initially, evolutionary genetic theory for age-structured populations, such as that of Charlesworth (1980), was focused more on how the novel Hamiltonian effects that arise with age structure would affect the evolution of allele frequencies. Interestingly, Hamilton (1966) was more concerned with making predictions about demographic patterns, although he did not supply the type of bridging theory that we are describing here. His inferences with respect to the demographic consequences of his declining forces of natural selection were essentially verbal and intuitive. And in any case, he did not develop the explicit dynamical equations for allele frequency change that Charlesworth (1980) developed, following somewhat the early pioneering work from Norton (1928) and Haldane (1927). Mueller has deliberately developed demographic predictions from Hamiltonian machinery, where his models have chiefly been of the clonal mutant substitution variety (Mueller & Rose 1996; Muller et al. 2011), as discussed previously. What clearly remains to be done is the development of full-scale Mendelian models in which the machinery of Norton and Charlesworth is extended to produce quantitative theoretical demographies. It is clear from experimental work that changes in the force of agespecific selection can result in relatively rapid changes in patterns of age-specific mortality and fecundity, including the age of onset of late-life plateaus (Rauser, Mueller & Rose 2006; Rauser et al. 2006; Rose 1984; Rose et al. 2002). This evolution is too rapid to be due to the effects of new mutations and is almost certainly a consequence of changes in the frequency of existing genetic variants. Thus, an important goal would be the development of theory based on standing Mendelian genetic variation and balanced polymorphisms in age-structured populations as a means of understanding the extensive experimental results that now exist.

Comparative and Laboratory Tests of Hamiltonian Demography Comparative Tests of Hamiltonian Theory It is rare for any type of evolutionary theory to make unequivocal predictions concerning comparative patterns of variation among species, but Hamiltonian evolutionary theory is just such a case. If the evolution of senescence indeed depends on whether or not Hamilton’s forces decline in an age-dependent manner, then species in which they do not decline should not exhibit the kind of persistent decline in survival probabilities and fecundities that are the demographic hallmarks of senescence, at least when cohorts are kept under favourable conditions, particularly in the laboratory.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:26:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.003

Comparative and Laboratory Tests of Hamiltonian Demography

45

Note that this is a falsifiable prediction that is specific to cohorts that are not undergoing ecological or genetic change. Despite the claims of Jones et al. (2014), there may be entirely exogenous ecological factors that generate patterns of declining survival with age in wild populations: the spread of infectious disease, secular environmental deterioration, inbreeding, depression as a result of declining effective population sizes, and so on. Under some laboratory conditions, cohorts can face declining conditions that artefactually emulate senescence too. Thus the stipulation that senescence almost always needs to be studied in cohorts protected from adversities is foundational, even for comparative research on the phenomenon (Comfort 1979; Rose 1991). There are anecdotal claims that some laboratory cohorts show no apparent senescence (Comfort 1979), but of greater value are mortality rate studies like those of Bell (1984) and Martinez (1998) using small invertebrates. Such research has found that demographic senescence does not occur among species that have well-established fissile reproduction that is roughly symmetrical: in these laboratory studies, the mortality rates of such cultures are stable and sometimes close to negligible. These are species in which it is likely that Hamilton’s forces of natural selection do not decline simply because there is no soma left after each act of reproduction. However, it should be noted that species with fissile or budding life cycles do not necessarily have the reproductive symmetry required to ensure that Hamilton’s forces do not decline. First, asymmetrical budding in the yeast Saccharomyces cerevisiae and visibly asymmetrical fission in some bacteria can both allow the evolution of senescence (Ackerman et al. 2007; Jazwinski 1990). Second, even in bacterial species that have morphologically symmetrical fission, such as Escherichia coli, asymmetrical partitioning of metabolic wastes can generate a ‘soma’ lineage that undergoes senescence (Steward et al. 2005). Thus, the central issue is less obvious than it might appear: the evolution of senescence hinges on whether or not there is the possibility of asymmetrical effects on the products of a vegetative act of reproduction. In some cases, determining whether or not such asymmetry has arisen will be tricky.

Quantitative Genetic Studies of Laboratory Senescence One of the first empirical avenues for Hamiltonian research on ageing was the study of the quantitative genetics of senescence, particularly in Drosophila melanogaster (Charlesworth & Hughes 1996; Promislow et al. 1996; Rose & Charlesworth 1981; Shaw et al. 1999; Tatar et al. 1996). This research was motivated by theoretical expectations that depended on two alternative population-genetic mechanisms: mutation accumulation and antagonistic pleiotropy. With respect to mutation accumulation, Charlesworth (1980) suggested that deleterious alleles with strictly age-specific effects on only one type of life history character, such as age-specific fecundity or age-specific mortality rate, should produce progressively increasing additive genetic variances for these characters as a function of age. Rose and Charlesworth (1980, 1981), however, found no such pattern for daily fecundity in D. melanogaster. In the course of research on age-specific mortality rates, Charlesworth and Hughes (1996) showed that varying patterns of dominance and Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:26:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.003

46

A Hamiltonian Demography of Life History

antagonistic pleiotropy could also produce increasing additive genetic variances for life history characters as a function of age. Further research has been done on this problem by Promislow et al. (1996), Tatar et al. (1996), Shaw et al. (1999) and Moorad and Promislow (2010). It has proven both problematic and demanding to define the conditions under which genetic variances will increase with age in theory and perhaps even harder to detect such patterns in quantitative genetic data. Thus, this line of research has been largely abandoned (for an exception, see Pujol et al. 2014). By contrast, research on patterns of genetic correlation has been more productive. Antagonistic pleiotropy is a genetic mechanism in which alleles with beneficial effects on some components of fitness have deleterious effects on other fitness components (Rose 1982, 1983). While it has been of particular interest for the evolution of senescence (Charlesworth 1980; Medawar 1952; Rose 1985; Williams 1957), it is a general evolutionary mechanism that is applicable to pleiotropic effects between types of life history characters at the same age as well as antagonistic effects on alternative sexes, although the latter case is often referred to as ‘sexual antagonism’. In the context of Hamiltonian quantitative genetic research, the hallmark of antagonistic pleiotropy as a mechanism for the evolution of senescence is negative genetic correlations between early-age life history characters and later-age life history characters, such as those found by Rose and Charlesworth (1981). However, there are several experimental design problems that interfere with the detection of such negative genetic correlations, among them inbreeding depression and genotype-by-environment interaction (Rose 1991). Either of these effects will tend to bias estimates of genetic correlations upward. Overall, quantitative genetic approaches to the analysis of the genetic foundations of the Hamiltonian evolution of senescence have been pursued relatively little over the last fifteen years. This is probably because of the significant theoretical and experimental challenges that such research poses.

Experimental Evolution of Senescence and Late Life Using Hamilton’s Forces A much less demanding and now more common experimental strategy is the evolutionary manipulation of laboratory populations by shifts in Hamilton’s forces of natural selection. This is most often achieved by shifting the timing of windows of reproduction in Drosophila populations maintained with discrete generations and then sustaining such shifts for significant periods of laboratory evolution (Chippindale et al. 1997; Luckinbill 1984; Rose 1980; Rose et al. 2002; Wattiaux, 1968a, 1968b). For example, it is a commonplace result that laboratory populations that have had their age of reproduction shifted to later ages soon evolve increased average life spans (Rose 1991). But such shifts in reproductive timing can also proceed by switching evolving populations from later ages of reproduction to earlier ages of reproduction, which results in progressively falling average life span over a number of generations of evolution (Chippindale et al. 1997; Rose et al. 2004; Service et al. 1988; Teotonio and Rose 2002). Other species have been subjected to comparable experimental shifts in reproductive timing (David & Bryant 2000; Nagai et al. 1995; Sokal 1970), with qualitatively similar effects on the evolution of senescence. Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:26:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.003

Comparative and Laboratory Tests of Hamiltonian Demography

47

One of the more interesting findings of Hamiltonian experimental evolution, from the standpoint of demography, is that both average longevity and age-specific mortality rates shift in ways that are qualitatively predictable from Hamiltonian theory (Rose et al. 2002). Likewise, patterns of age-specific fecundity shift in accordance with changes in the force of natural selection acting on fecundity (Rauser et al. 2003). Thus, patterns in the evolution of late-life mortality and fecundity plateaus can be predicted using Hamiltonian theory, and these predictions have been generally borne out in experimental evolutionary tests (Mueller et al. 2011; Rauser et al. 2003; Rose et al. 2002).

The Late-Life Controversy The causes of late-life deceleration in rates of mortality remain controversial within gerontology, demography and evolutionary biology (e.g. Mueller et al. 2011; Pletcher et al. 1998; Steinsaltz & Evans 2004; Wachter 1999). The most common and oldest theory for explaining it is lifelong heterogeneity, in which mortality rates slow at later ages entirely because of the elimination of those who have a lifelong greater risk of death due to a within-cohort attrition process (Greenwood & Irwin 1939). One of the earliest quantitative theories of lifelong heterogeneity was proposed by Vaupel et al. (1979). Under their model, the probability of an individual dying is described by a Gompertz equation. These theories further suppose the idea that due to heterogeneity in either the genetic composition of individuals and/or their development, the age-independent parameter of the Gompertz equation shows individual variation. This variation, if sufficiently pronounced, can cause the composite population to show a levelling of mortality rates at advanced ages even though every individual has an exponentially increasing chance of dying with age. We have pointed out several quantitative difficulties with the application of lifelong heterogeneity theories to the explanation of data from Drosophila experimental evolution (e.g. Mueller et al. 2003, 2011), especially the extreme levels of such heterogeneity required to explain the observed mortality-rate deceleration of demographically well-characterised large cohorts. Some recent experiments with Drosophila have claimed that the absence of a detectable mortality plateau in some Drosophila populations given different diets is consistent with heterogeneity theories and not with evolutionary theories (Zajitscheck et al. 2013). However, these experiments used only 300 individuals of each sex, so it is likely that these experiments were conducted below the minimal sample size needed to detect these phenomena reliably (cf. the 900–1,550 used in Curtsinger et al. (1992) or 1,000–2,800 per sex used in Rose et al. (2002)). A further difficulty with lifelong heterogeneity theories of mortality plateaus is their reliance on hypothetical underlying ‘robustness’ and ‘frailty’ variables, which are as inherently difficult to measure and study as ‘reproductive effort’ and other hypothetical variables in optimal life history theory. This makes the design of empirical tests of these models especially difficult. Fortunately, both fecundity and virility also show late-life plateaus (Rauser et al. 2003; Shahrestani et al. 2012a), in keeping with Hamiltonian theory (Mueller et al. 2011; Rauser, Mueller & Rose 2006). Lifelong heterogeneity theories for these two kinds of life history characters do not involve as many hidden Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:26:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.003

48

A Hamiltonian Demography of Life History

variables as mortality, so they can be used to test lifelong heterogeneity theory more readily. Regrettably for such theory, there is no evidence for lifelong heterogeneity effects that might produce late-life life history plateaus for these characters (Mueller et al. 2011;Rauser et al. 2005).

Further Topics for Hamiltonian Demographic Analysis There is evidence for transient heterogeneity within cohorts in our demographic data, a phenomenon quite distinct from the much-conjectured lifelong heterogeneity. Specifically, we and others have shown that a week or so before death, flies enter a period of physiological decline we have called the ‘death spiral’ (Mueller et al. 2007; Rauser et al. 2005; Shahrestani et al. 2012b). During this period, both female fecundity (Rauser et al. 2005; Rogina et al. 2007) and male virility (Shahrestani et al. 2012a) decline at a faster rate among dying individuals than among similarly aged individuals not in this final period of life. Consequently, every population will have heterogeneity due to the mixture of individuals that are and are not in the death spiral. This heterogeneity has been incorporated into models of age-specific fecundity (Mueller et al. 2007, 2011) with a noticeable improvement in model fit. One of the less noticed aspects of the Hamiltonian models offered in Mueller and Rose (1996) is the role of effective population size in the onset of late life. Specifically, inbreeding is expected to produce significant effects on the evolution of life history within the Hamiltonian paradigm. Not only can inbreeding impinge on the measurement of quantitative genetic parameters in life history research (Rose 1991), but effective population size also is a general scaling factor for the effectiveness of selection, much like Hamilton’s (1966) forces themselves. The obvious question is how do these two general types of scaling factors for life history evolution and demography interact with each other? This is a question that needs careful theoretical analysis and experimental investigation, particularly given the large population size fluctuations that often occur during laboratory domestication (Santos et al. 2012, 2013), and that have arisen in recent human evolutionary history (Cochran & Harpending 2009). Another demographic issue of importance for both experimental evolution and human evolution is the role of Hamiltonian scaling in evolutionary responses to environmental change. We have been exploring this issue both theoretically (Phung et al. in preparation) and experimentally (Rutledge et al. in preparation). To this point in our work, we find that early life history is evolutionarily more responsive to environmental change than later adult life, as we had conjectured previously (Mueller et al. 2011). When studying functional health in populations of D. melanogaster consuming both ancestral and more recent diets during laboratory domestication, at early adult ages we find that flies function as well or even better on an evolutionarily recent diet. Furthermore, at later adult ages, these same flies have better survival and reproduction on their ancestral diet compared to the recent diet (Rutledge et al. in preparation). Adaptation to the newer laboratory diet is apparently not sustained into later ages on an evolutionarily recent diet. At such later ages, the better diet for life history characters is apparently a diet that is more similar to their ancestral diet in nature. This may be theoretically attributed to the Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:26:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.003

From the Laboratory to Wild Populations

49

weakening of the forces of natural selection with adult age (Phung et al. in preparation), making later fruit fly dietary effects a relic of adaptation to the food available prior to laboratory domestication. However, much more work must be done, both theoretically and experimentally, to clarify and substantiate this pattern.

From the Laboratory to Wild Populations Recently, there have been attempts to characterise the demography of populations in the wild, with a view towards informing demographic and life history evolution theory (Jones et al. 2014). Our view is that data obtained from wild populations are unlikely to provide strong-inference tests of core theories of life history evolution because of issues of lack of replicability and control of factors such as gene-by-environment (G×E) interactions that have been discussed repeatedly (Mueller et al. 2005, 2011; Rose 1991). Then there is the issue of how well we can apply Hamiltonian demography to the interpretation of data from wild populations, which is a different scientific problem. While this too remains challenging, attempts at such application would be in keeping with a long scientific tradition in physics of going from laboratory experimentation to the modelling and exploration of possible mechanisms for astrophysical dynamics in far-off solar systems and even galaxies. For example, a long-standing question in Hamiltonian research is whether or not there is evidence of senescence in the demography of wild populations. Nesse (1988) made a pioneering attempt to find quantitative evidence for senescence in wild populations of animals. He concluded that it did in fact occur. Promislow (1991) applied a similar approach to a larger set of wild mammalian populations and again concluded that there was detectable senescence. The substantial problem not addressed by Jones et al. (2014), who report a lack of evidence for senescence in the wild, is that it is typically impossible to control environmental sources of extrinsic mortality in natural populations. To the extent that these extrinsic sources of mortality act in a strictly age-independent fashion, mathematical analysis shows that they will not be relevant to the evolution of senescence (Caswell 2007). Thus, a cohort followed over time may experience extrinsic sources of mortality that also vary over time. If these extrinsic mortality sources cannot be quantified, then any pattern of mortality is possible, and the inference of senescence will be difficult. In a similar fashion, mortality data collected over a short period of time from individuals of many ages in natural populations cannot control for the past heterogeneous histories of these individuals, making mortality patterns collected in this fashion ambiguous. For instance, in Drosophila, it has been known for nearly ninety years that larval rearing density can affect later age-specific mortality rates (Pearl et al. 1927). However, efforts to analyse data from populations in nature continue. Ricklefs (2008) compared age-related mortality patterns in birds in captivity and wild populations. He found that the rate of senescence is similar to that in wild populations despite the absence of exogenous forces of mortality (i.e. predation, disease) in captivity. Although evolutionary theory suggests that increasing mortality and decreasing fertility with age Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:26:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.003

50

A Hamiltonian Demography of Life History

are expected, Jones et al. (2014) compared several demographic age trajectories across a number of species including vertebrates, invertebrates, vascular plants, and a green algae, which showed considerable variation in mortality trajectories. Caleb E. Finch, other gerontologists, and some demographers have argued that there is evidence of negligible senescence in natural populations of a number of species, from rockfish to tortoises (Finch 1998, 2009). However, patterns of mortality found in wild populations using ‘marked’ individuals cannot easily account for extrinsic forces of mortality (Jones et al. 2014; Ricklefs 2008). For example, an observation of constant mortality may be due to increased intrinsic mortality during a period when the environment also became more favourable during the time span of the study, the two effects thereby cancelling out. Without conducting controlled experiments, interpreting the patterns of senescence in wild populations, especially when only one mortality-rate survey with a unique set of temporal data is used, is at a minimum challenging. What are we to make of these findings? Does evidence of negligible senescence in wild populations suggest that their ageing is of negligible importance for their demography? Or, alternatively, do such data suggest that late-life mortality-rate plateaus arise early enough in the adult lives of wild populations to be detectable in their demography? Perhaps further progress hinges on the development of new methods of analysing demographic data, to which we now turn.

Inferring Life History Phase Transitions from Demographic Data One of the core problems impinging on Hamiltonian and other demographic research is identifying the ages at which life histories transition between the successive phases of ageing: development, senescence, late life, and dying. Hamiltonian theory predicts a priori three distinct phases of life history that arise from the decline of the forces of natural selection with age in a demographic cohort after the onset of reproduction: (1) early high function, (2) senescence and (3) late life (Mueller et al. 2011). In addition, as already mentioned, we and others have found extensive evidence of a distinctive fourth phase of dying in our data (Mueller et al. 2007, 2011; Rauser et al. 2005; Rogina et al. 2007). If these or any subset of these phases correspond to real phenomena, under sufficiently good conditions they should be detectable directly from abundant highquality demographic data in cohorts from species that are subject to declines in Hamilton’s forces. Thus, an important problem in Hamiltonian demography is how best to detect these four demographic phase transitions in mortality, survival, fecundity and virility as they vary with age. The first significant phase transition is the onset-of-senescence boundary. The second phase transition is the end-of-senescence plateau boundary. The third and final phase transition is the start of the death spiral. We are now seeking empirical estimators of the timings of these phase transitions for two kinds of data: (1) new, yet-tobe published replicated cohort studies of mortality and fecundity more than ten times larger than any of our experiments heretofore and (2) simulated data sets generated from a priori Hamiltonian models, including the crude two-stage Gompertz models, with and Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:26:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.003

References

51

without heterogeneity correction, which have been our standard tools of analysis (Mueller et al. 2011; Rose et al. 2002). When we can reliably infer the phase transitions of Hamiltonian demography from cohort data, we may, in turn, be better able to apply Hamiltonian reasoning to data collected from wild populations (cf. Jones et al. 2014, Promislow 1991; Ricklefs 2008), the trickiest frontier of all in demographic research. But that remains to be seen.

Conclusion We have argued here that there are four phases of ageing: development, senescence, late life and dying. Age-structured population genetics theory shows how Hamilton’s forces of natural selection lead to three of these distinct phases of life history. We are endeavouring to develop an overarching scientific programme for this alternative Hamiltonian demography, focusing on the following elements: (1) a priori evolutionary genetic theory, (2) the alternative demographic patterns that such theory can generate, (3) testing the elements of such theory using experimental evolution and other empirical tools of evolutionary biology, (4) implications of Hamiltonian experimental research for the predicted demographic patterns of wild populations and (5) seeking empirical methods of inferring phase transitions directly from demographic data. Our view is that this is a promising way forward for demography conceived as the general scientific study of life history characters in biological populations. However, it remains to be seen if this approach is of much value for the most specific applications of demography to the case of human populations, for which many of our scientific research tactics will not be feasible.

References Abrams, P. A. & Ludwig, D. (1995). Optimality theory, Gompertz’ law, and the disposable soma theory of senescence. Evolution, 49, 1055–66. Ackerman, M., Schauerte A., Stearns S. C. & Jenal U. (2007). Experimental evolution of ageing in a bacterium. BMC Evolutionary Biology, 7, 126. Baudisch, A. (2005). Hamilton’s indicators of the force of selection. Proceedings of the National Academy of Science of the United States of America, 102(23), 8263–8. Baudisch, A. (2008). Inevitable Ageing? Contributions to Evolutionary-Demographic Theory (Berlin: Springer). Bell, G. (1984). Evolutionary and non-evolutionary theories of senescence. American Naturalist, 124, 600–3. Carey, J. R., Liedo, P., Orozdo, D. & Vaupel, J. W. (1992). Slowing of mortality rates at older ages in large medfly cohorts. Science, 258, 447–61. Charlesworth, B. (1970). Selection in populations with overlapping generations: I. The use of Malthusian parameters in population genetics. Theoretical Population Biology, 1, 352–70.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:26:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.003

52

A Hamiltonian Demography of Life History

Charlesworth, B. (1980). Evolution in Age-Structured Populations (New York: Cambridge University Press). Charlesworth, B. (2001). Patterns of age-specific means and genetic variances of mortality rates predicted by the mutation-accumulation theory of ageing. Journal of Theoretical Biology, 210, 47–65. Charlesworth, B. & Hughes, K. A. (1996). Age-specific inbreeding depression and components of genetic variance in relation to the evolution of senescence. Proceedings of the National Academy of Science of the United States of America, 93(12), 6140–5. Chippindale, A. K., Alipaz, J. A., Chen, H.-W. & Rose, M. R. (1997). Experimental evolution of accelerated development in Drosophila: 1. Larval development speed and survival. Evolution, 51, 1536–51. Cochran, G. & Harpending, H. (2009). The 10,000 Year Explosion: How Civilization Accelerated Human Evolution (New York: Basic Books). Comfort, A. (1979). The Biology of Senescence (Edinburgh: Churchill Livingstone). Caswell, H. (2007). Extrinsic mortality and the evolution of senescence. Trends in Ecology and Evolution, 22(4), 173–4. Curtsinger, J. W., Fukui, H. H., Townsend, D. R. & Vaupel, J. W. (1992). Demography of genotypes: failure of limited life span. Science, 258, 461–3. David, R. H. & Bryant, E. H. (2000). The evolution of senescence under curtailed life span in laboratory populations of Musca domestica (the housefly). Heredity, 85, 115–21. De Grey, A. & Rae, M. (2007). Ending Ageing: The Rejuvenation Breakthroughs that Could Reverse Human Ageing in Our Lifetime (New York: St Martin’s Press). Finch, C. E. (1998). Variations in senescence and longevity include the possibility of negligible senescence. Journals of Gerontology: Medical Science, 53A(4), B235–9. Finch, C. E. (2009). Update on slow ageing and negligible senescence: a mini-review. Gerontology, 55(3), 307–13. Fisher, R. A. (1930). The Genetical Theory of Natural Selection (Oxford University Press). Forsberg, L. A., Rasi, C., Razzaghian, H. R., et al. (2012). Age-related somatic structural changes in the nuclear genome of human blood cells. American Journal of Human Genetics, 90, 217–28. Greenwood, M. & Irwin, J. O. (1939). The biostatistics of senility. Human Biology, 11, 1–23. Haldane, J. B. S. (1927). A mathematical theory of natural and artificial selection, part IV. Proceedings of the Cambridge Philosophical Society, 23, 607–15. Haldane, J. B. S. (1941). New Paths in Genetics (London: Allen & Unwin). Hamilton, W. D. (1966). The moulding of senescence by natural selection. Journal Theoretical Biology, 12, 12–45. Hoffman, J. M., Creevy, K. E. & Promislow, D. E. L. (2013). Reproductive capability is associated with lifespan and cause of death in companion dogs. PLOS One, doi: 10.1371/journal.pone.0061082. Jazwinski, S. M. (1990). Ageing and senescence of the budding yeast Saccharomyces cerevisiae. Molecular Microbiology, 4(3), 337–43. Jones, O. R., Scheuerlein, A., Salguero-Gómez, R., et al. (2014). Diversity of ageing across the tree of life. Nature, 505, 169–73.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:26:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.003

References

53

Kirkwood, T. B. & Cremer, T. (1982). Cytogerontology since 1881: a reappraisal of August Weissmann and a review of modern progress. Human Genetics, 60, 101–21. Le Bourg, E. & Moreau, M. (2014). Individual late-life fecundity plateaus do exist in Drosophila melanogaster and are very common at old age. Experimental Gerontology, doi: 10.1016/j.exger.2014.04.001. Luckinbill, L. S., Arking, R., Clare, M., et al. (1984). Selection for delayed senescence in Drosophila melanogaster. Evolution, 38, 996–1003. Martinez, D. E. (1998). Mortality patterns suggest lack of senescence in Hydra. Experimental Gerontology, 33, 217–25. Moorad, J. A. & Promislow, D. E. L. (2010). Evolutionary demography and quantitative genetics: age-specific survival as a threshold trait. Proceedings of the Royal Society of London Series B: Biological Sciences, 278, 144–51. Medawar, P. B. (1946). Old age and natural death. Modern Quarterly, 1, 30–56. Medawar, P. B. (1952). An Unsolved Problem of Biology (London: Lewis). Mueller, L. D., Drapeau, M. D., Adams, C. S., et al. (2003). Statistical tests of demographic heterogeneity theories. Experimental Gerontology, 38, 373–86. Mueller, L. D., Rauser, C. L. & Rose, M. R. (2005). Population dynamics, life history and demography: lessons from Drosophila. In Advances in Ecological Research: Population Dynamics and Laboratory Ecology, Ed. L. Yiqi (New York: Academic Press). Mueller, L. D., Rauser, C. L. & Rose, M. R. (2007). An evolutionary heterogeneity model of late-life fecundity in Drosophila. Biogerontology, 8, 147–61. Mueller, L. D., Rauser, C. L. & Rose, M. R. (2011). Does Aging Stop? (New York: Oxford University Press). Mueller, L. D. & Rose, M. R. (1996). Evolutionary theory predicts late-life mortality plateaus. Proceedings of the National Academy of Sciences of the United States of America, 93, 15249–53. Nagai, J., Lin, C. & Sabour, M. P. (1995). Lines of mice selected for reproductive longevity. Growth, Development, and Ageing, 59(3), 79–91. Nesse, R. M. (1988). Life table tests of evolutionary theories of senescence. Experimental Gerontology, 23, 445–53. Norton, H. T. J. (1928). Natural selection and Mendelian variation. Proceedings of the London Mathematical Society, 28, 1–45. Partridge, L. & Fowler, K. (1992). Direct and correlated responses to selection on age at reproduction in Drosophila melanogaster. Evolution, 46, 76–91. Pearl, R., Miner, J. R. & Parker, S. L. (1927). Experimental studies on the duration of life: XI. Density of population and life duration in Drosophila. American Naturalist, 61, 289–317. Polosak, J., Roszkowska-Gancarz, M., Kurylowicz, A., et al. (2010). Decreased expression and the Lys751Gln polymorphism of the XPD gene are associated with extreme longevity. Biogerontology, 11, 287–97. Phung, K. H, Rose, M. R. & Mueller, L. D. (In preparation). Transient age-specific adaptation to a novel environment. Pletcher, S. D. & Curtsinger, J. W. (1998). Mortality plateaus and the evolution of senescence: why are old-age mortality rates so low? Evolution, 52, 454–64.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:26:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.003

54

A Hamiltonian Demography of Life History

Promislow, D. E. L. (1991). Senescence in natural populations of mammals: a comparative study. Evolution, 45, 1869–87. Promislow, D. E. L, Tatar, M., Khazaeli, A. & Curtsinger, J. W. (1996). Age-specific patterns of genetic variance in Drosophila melanogaster: I. Mortality. Genetics, 143, 839–48. Pujol, B., Marrot, P. & Pannell, J. R. (2014). A quantitative genetic signature of senescence in a short-lived perennial plant. Current Biology, 24, 744–7. Rauser, C. L., Abdel-Aal, Y., Sheih, J. A., et al. (2005). Lifelong heterogeneity in fecundity is insufficient to explain late-life fecundity plateaus in Drosophila melanogaster. Experimental Gerontology, 40(8–9), 660–70. Rauser, C. L., Mueller L. D. & Rose, M. R. (2003). Ageing, fertility and immortality. Experimental Gerontology, 38, 27–33. Rauser, C. L., Mueller, L. D. & Rose, M. R. (2006). The evolution of late life. Ageing Research Reviews, 5, 14–32. Rauser, C. L., Tierney, J. J., Gunion, S. M., et al. (2006). Evolution of late-life fecundity in Drosophila melanogaster. Journal of Evolutionary Biology, 19, 289–301. Ricklefs, R. E. (2008). The evolution of senescence from a comparative perspective. Functional Ecology, 22, 379–92. Rogina, B., Wolverton, T., Bross, T. G., et al. (2007). Distinct biological epochs in the reproductive life of female Drosophila melanogaster. Mechanisms of Aging and Development, 128, 477–85. Rose, M. R. (1982) Antagonistic pleiotropy, dominance, and genetic variation. Heredity, 48, 63–78. Rose, M. R. (1983). Further models of selection with antagonistic pleiotropy. In Population Biology, ed. H. I. Freedman and C. Strobeck (pp. 47–53)(Berlin: Springer). Rose, M. R. (1984). Laboratory evolution of postponed senescence in Drosophila melanogaster. Evolution, 38, 1004–10. Rose, M. R. (1985). Life-history evolution with antagonistic pleiotropy and overlapping generations. Theoretical Population Biology, 28, 342–58. Rose, M. R. (1991). Evolutionary Biology of Ageing (New York: Oxford University Press). Rose, M. R. & Burke, M. K. (2011). Genomic Croesus: experimental evolutionary genetics of ageing. Experimental Gerontology, 46, 397–403. Rose, M. R. & Charlesworth, B. (1980). A test of evolutionary theories of senescence. Nature, 287, 141–2. Rose, M. R. & Charlesworth B. (1981). Genetics of life-history in Drosophila melanogaster: I. Sib analysis of adult females. Genetics, 97, 173–85. Rose, M. R., Drapeau, M. D., Yazdi, P. G., et al. (2002). Evolution of late-life mortality in Drosophila melanogaster. Evolution, 56, 1982–91. Rose, M. R., Passananti, H. B. & Matos, M. (eds.) (2004). Methuselah Flies: A Case Study in the Evolution of Ageing (Singapore: World Scientific Publishing). Santos, J., Pascual, M., Simões, P., et al. (2012). From nature to the lab: the impact of founder effects on adaptation. Journal of Evolutionary Biology, 25, 2607–22. Santos, J., Pascual, M., Simões, P., et al. (2013). Fast evolutionary genetic differentiation during experimental colonizations. Journal of Genetics, 92, 183–94.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:26:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.003

References

55

Service, P. M., Hutchinson, E. W. & Rose, M. R. (1988). Multiple genetic mechanisms for the evolution of senescence in Drosophila melanogaster. Evolution, 42, 708–16. Shahrestani, P., Tran, X. & Mueller, L. D. (2012a). Physiological decline prior to death in Drosophila melanogaster. Biogerontology, 13, 537–45. Shahrestani, P., Tran, X. & Mueller, L. D. (2012b). Patterns of male fitness conform to predictions of evolutionary models of late life. Journal of Evolutionary Biology, 25 (6), 1060–5. Shaw, F. H., Promislow, D. E. L, Tata, R. M., et al. (1999). Toward reconciling inferences concerning genetic variation in senescence in Drosophila melanogaster. Genetics, 152, 553–66. Sokal, R. R. (1970). Senescence and genetic load: evidence from Tribolium. Science, 167, 1733–4. Steinsaltz, D. & Evans, S. N. (2004). Markov mortality models: implications of quasistationarity and varying initial conditions. Theoretical Population Biology, 65: 319–37. Steinsaltz, D., Evans, S. N. & Wachter, K. W. (2005). A generalized model of mutation-selection balance with applications to ageing. Advanced Applied Mathematics, 35, 16–33. Tatar, M., Promislow, D. E. L, Khazaeli, A. & Curtsinger, J. W. (1996). Age specific patterns of genetic variance in Drosophila melanogaster: II. Fecundity and its genetic correlation with agespecific mortality. Genetics, 143, 849–58. Vaupel, J. W., Manton, K. G. & Stallard, E. (1979). The impact of heterogeneity in individual frailty on the dynamics of mortality. Demography, 16, 439–54. Wachter, K. W. (1999). Evolutionary demographic models for mortality plateaus. Proceedings National Academy of Sciences of the United States of America, 96, 10544–7. Wachter, K. W., Evans, S. N. & Steinsaltz D. (2013). The age-specific force of natural selection and biodemographic walls of death. Proceedings of the National Academy of Sciences of the United States of America. 110(25), 10141–6. Wattiaux, J. M. (1968a). Cumulative parental age effects in Drosophila subobscura. Evolution, 22, 406–21. Wattiaux, J. M. (1968b). Parental age effects in Drosophila pseudobscura. Experimental Gerontology, 3, 55–61. Weismann, A. (1891). On Heredity (Oxford: Clarendon Press). Williams, G. C. (1957). Pleiotropy, natural selection, and the evolution of senescence. Evolution, 11, 398–411. Zajitschek, F., Jin, T., Colchero, F. & Maklakov, A. A. (2013) Ageing differently: dietand sex-dependent late-life mortality patterns in Drosophila melanogaster. Journal of Gerontology, 69(6), 66–74.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:26:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.003

4

Senescence, Selection Gradients and Mortality Hal Caswell and Esther Shyu

Short Summary Evolutionary explanations of senescence are based on the established fact that the selection gradients (the rate of change of fitness with respect to a trait) on age-specific survival and fertility decline with age. Improvements in survival or fertility at older ages have less of an effect than the same improvements at younger ages. As a result, traits with small positive effects at early ages may accumulate even if they are accompanied by larger negative effects at late ages. It is often claimed that these results imply that additional mortality should make it easier for senescence to evolve, a claim that has led to a great deal of empirical study. However, the actual effect of additional mortality depends on the age dependence of that mortality. We show in detail that age-independent mortality has no effect on the selection gradients in time-invariant, periodic, stochastic and density-dependent life cycles. When the additional mortality is age dependent, its effects on the evolution of senescence depend on where in the life cycle it acts. We develop indices Mp and Mf to measure the shape of the selection gradient on survival and fertility, respectively. These indices give the mean age calculated over the selection gradient; larger values indicate a life history that is more resistant to the evolution of senescence. For several model life tables and for data on human life tables from Sweden, we show that additional early mortality increases Mp and Mf, while mortality later in life has the opposite effect.

Introduction Our theoretical understanding of the evolution of senescence is perhaps more intimately tied to selection gradients than that of any other life history trait. The selection gradient on a trait is the slope of a line relating fitness to the trait value. Imagine a diagram with fitness plotted against the trait; given the appropriate mathematical and demographic tools, the slope of that line can be written as the derivative of fitness with respect to the value of the trait Selection gradient ¼

dfitness ; dθ

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:28:05, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.004

Introduction

57

where θ is the value of the trait. If the trait is multivariate, then the selection gradient is a vector whose entries give the slopes of the relationship between fitness and each of the traits considered alone (Lande 1982). Therefore, at any step in constructing hypotheses about evolution through natural selection – for example, about why human canines do not protrude, why deer antlers are annually shed and renewed, why parrots mimic, why dolphins play – one can visualise such a diagram and consider whether the slope really would be appreciably non-zero under the assumptions of the theory. If there is no slope, then there is no frequency change . . . and the hypothesis is probably wrong. (Price 1970)

The selection gradient appears, in various guises: in classical population genetics, where θ is a gene frequency and fitness is the mean genotypic fitness (Wright 1937); in quantitative genetics, where θ is the mean of a phenotypic trait (Robertson 1968; Lande 1982); and in the theory of adaptive dynamics, where the gradient is the derivative of invasion fitness (a long-term growth rate) with respect to the value of a quantitative trait (Diekmann 2004; Dieckmann & Law 1996). In all these cases, the consequences of the selection gradient are the same: if fitness is positively correlated with the value of the trait, then the trait is expected to increase under natural selection given the existence of appropriate genetic variation. If the correlation is negative, the trait should decrease. The history of ideas about the evolution of senescence is a long variation on the theme of age-specific selection gradients. From the classical works of Fisher (1930), Medawar (1952), Williams (1957) and Hamilton (1966) to a more recent literature too vast to begin to review here (e.g. Baudisch 2008; Charlesworth 1994, 2000; Finch 1990; Kirkwood 1990; Kirkwood & Holliday 1986; Lee 2003; Rose 1990; Silvertown 2013; Wachter, Evans & Steinsaltz 2013; Wachter & Finch 1997; Wachter, Steinsaltz & Evans 2014), attention has focused on why effects at older ages are less important to fitness than effects at younger ages. ‘Our understanding of the evolution of senescence is, at one level, very complete; we know that senescence is an evolutionary response to the diminishing effectiveness of selection with age’ (Charlesworth 2000). Fisher (1930) suggested that the decline might be related to the shape of the reproductive value function. Medawar (1952) focused on the shape of the stable age distribution. Hamilton (1966) presented a complete demographic calculation of the selection gradient dr/dμ(x), where r is the rate of increase from the Euler-Lotka age-classified equation, and μ(x) is the mortality rate at age x.1 Hamilton obtained the selection gradients on age-specific mortality and age-specific fertility from the Euler-Lotka equation, which implicitly defines fitness r as the solution to 1¼

ð∞

ℓðxÞmðxÞerx dx;

ð4:1Þ

0

1

It turns out that Hamilton’s result, which can be generalised to stage-classified demography, involves the product of the stable age distribution and the reproductive value distribution, combining the perspectives of both Fisher and Medawar(Caswell 1978, 2010). Thus, both Fisher and Medawar were partly, but only partly, right in their explanations.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:28:05, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.004

58

Senescence, Selection Gradients and Mortality

where ℓ(x) is the survivorship to age x, and m(x) is the fertility at age x. As is standard practice, this equates the fitness of a trait with the rate at which a life cycle defined by that trait can propagate itself into the future (e.g. Charlesworth 1994, 2000; Fisher 1930; Hamilton 1966; Metz 2008; Metz, Nisbet & Geritz 1992). The selection gradient on a trait is the derivative of r with respect to the trait. By implicit differentiation of (4.1), Hamilton obtained the selection gradients on agespecific mortality and age-specific fertility ∂r and ∂μð xÞ

∂r : ∂mð xÞ

Hamilton showed that the selection gradient on mortality is always non-increasing and, after the age of first reproduction, is monotonically decreasing with age. The selection gradient on fertility declines with age, provided that r is not too negative.2 Hence, selection against mortality (or in favour of survival) and selection for increased fertility act less intently on them at older ages than at younger ages. Thus, the balance between mutation and selection will leave more deleterious genes with effects at older ages and pleiotropic genes that have small positive effects at early ages and larger negative effects at older ages and may still have a net positive effect on fitness. This decline in the selection gradient with age is not subtle; it can encompass many orders of magnitude (Caswell 1978), especially in rapidly growing populations (e.g. ten orders of magnitude in laboratory populations of the nematode Caenorhabditis elegans) (Chen et al. 2006). In more slowly growing populations, the declines may be less severe; see Figure 4.1 for an example for the human population of Sweden.

A Prediction: Mortality and Senescence A repeated prediction of senescence theory is that increasing mortality should make selection for senescence stronger (e.g. Charlesworth 1994; Finch 1990; Rose 1990; Williams 1957; Williams et al. 2006). The intuition behind the prediction is that if mortality is higher, the probability of living to old age is lower. If the probability of living to old age is lower, selection on traits expressed at old ages should be weaker. If selection on traits expressed at old ages is weaker, senescence should evolve more easily. This prediction has inspired a huge literature of studies (see review by Williams et al. 2006). Some of these have involved inter-taxon comparisons (e.g. bats versus other mammals; bats might experience lower mortality because they can fly). Others have focused on comparisons of different environmental conditions (e.g. the presence or

2

The sensitivity of λ or r to fertility is proportional to the stable age distribution. As is well known, if r ≥ 0, the stable age distribution always declines with age; there are more individuals in any age class than in the subsequent older age class. If r < 0, the stable age distribution may increase with age. That is, over part or all of the age range there may be more older individuals than younger individuals. The conditions for this to happen depend on the survival schedule; see Caswell (1978) for details.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:28:05, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.004

Selection Gradients from Matrix Population Models

10−1

59

10−1

10−3

Gradient on fertility

Gradient on survival

10−2

1891 2000 10−4

10−5

10−2

10−6

10−7

0

20

40

60

10−3

0

Age Figure 4.1

20

40 Age

The selection gradients on age-specific survival probabilty Pi and age-specific fertility Fi for Sweden in 1891 (at which point, log λ = 0.013) and in 2000 (log λ = –0.009).

absence of predators) and other kinds of mortality (e.g. Reznick, Bryant & Holmes 2006; Reznick et al. 2004). But the prediction itself is not valid. Abrams (1993) and Caswell (2007a) showed that the effects of mortality depend on whether it is focused on part of the life cycle or affects individuals of all ages or stages equally. Mortality that is independent of age or stage has no effect. This conclusion seems surprising because even stage-independent mortality reduces the chances of living to old age, but as we will see in this chapter, it applies very generally (see also Wensink, Caswell, and Bandisch 2016).

Selection Gradients from Matrix Population Models Because the evolution of senescence depends on the shape (increasing or decreasing and how steeply) of the selection gradients, we can examine the effect of additional mortality by exploring how selection gradients change in response to a change in mortality. To do so, we need a demographic model appropriate to the life cycle and ecology of the organism and from which we can calculate selection gradients. Here we will use matrix population models, which are a powerful and general demographic model available. They make no assumptions about the nature of the life-cycle stages, in particular, about whether those stages are continuous or discrete, ordered or not, and have the most

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:28:05, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.004

60

Senescence, Selection Gradients and Mortality

extensively developed analytical tools of any demographic model framework. Developing similar theory for other classes of demographic models remains an open research problem. Because he relied on the Euler-Lotka equation (4.1), Hamilton’s results on selection gradients apply only to age-classified demography in time-invariant environments. Thus, they tell us nothing about the effects of extrinsic mortality on life cycles in which the vital rates depend on i-states such as individual size, developmental stage, physiological condition or spatial location. Here we will extend the results on selection gradients to stage-classified models and to time-varying and density-dependent models. Consider a linear, time-invariant matrix population model nðt þ 1Þ ¼ AnðtÞ;

ð4:2Þ

where n is a vector of stage abundances, and A is a population projection matrix (e.g. Caswell 2001). The age-classified model is a special case in which A contains survival probabilities on the sub-diagonal, fertilities in the first row and zeros elsewhere (the socalled Leslie matrix) (Leslie (1945). The discrete-time invasion fitness λ is given by the dominant eigenvalue of A; the stable stage distribution and reproductive value vectors are given by the right and left eigenvectors w and v, respectively. The continuous-time fitness is r = log λ; either measure can be used for senescence calculations. Now consider a trait θ that influences the vital rates that appear in A and thus has an effect on λ. The selection gradient on θ is dλ dθ

or

dr 1 dλ : ¼ dθ λ dθ

ð4:3Þ

To obtain this selection gradient, we must find the derivative of the dominant eigenvalue λ with respect to a variable that affects the entries of the matrix. Fortunately, we know how to do that; in the last decade, the initial formulations for time-invariant models (Caswell 1978) and stochastic models (Tuljapurkar 1990) have been generalised using tools from matrix calculus (Caswell 2006, 2007b, 2008, 2009; Caswell & Shyu 2012; Shyu & Caswell 2014). We will use these results throughout this chapter. The derivations have been collected together in Appendix 4A. First, we generalise the trait θ to a vector of traits θ; this might be an entire schedule of mortality rates or fertilities or a set of parameters that determine those rates. The selection gradient on θ is  dvec A dr 1 ¼ wT ⊗ vT : T λ dθ dθT

ð4:4Þ

Figure 4.1 shows the selection gradients on survival (the sub-diagonal elements Pi) and fertility (the first-row elements Fi) in an age-classified model for Sweden in 1891 and 2000. In the first case, population growth was positive, and selection gradients on both

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:28:05, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.004

Mortality, Stage Dependence and Senescence

61

survival and fertility decline with age. In the second case, population growth was slightly negative, and the selection gradient on fertility actually increases with age. Our next step is to modify the population projection model (4.2) to include an extra source of mortality and see how it affects the selection gradients.

Mortality, Stage Dependence and Senescence We will examine the effects of additional mortality in four different cases: constant environments, periodic environments, stochastic environments and density-dependent models. In each case the model corresponding to (4.2) will be presented and then modified to include the additional mortality. The selection gradients will be derived for the special case where the additional mortality is stage-independent.

Notation for the Projection Matrix We will use the following symbols. We define μ(x) as a vector of age-specific mortality rates (also called ‘mortality hazards’). The vector of survival probabilities, from t to t + 1, obtained by these mortality hazards is p ¼ expðμÞ;

ð4:5Þ

where the exponential function is applied to each element of μ. Age-specific fertilities are contained in a vector f. The projection matrix A in (4.2) contains the survival vector p on the subdiagonal and the fertility vector fT in the first row. When additional mortality is imposed, the probability that an individual in stage i survives that mortality is the scalar ϕi; the probabilities for all stages are contained in the vector ϕ, which appears on the diagonal of a survival matrix Φ. See Appendix 4A for more detailed description of notation.

Stage-Independent Mortality in Constant Environments The matrix model (4.2) projects the population vector n from t to t + 1 and implies the selection gradient (4.4), where θ could be any vector of parameters. Suppose that an additional mortality hazard is imposed between t and t + 1. How does this affect the selection gradient on θ? Let ϕi be the probability that an individual of stage i survives the new hazard, and let ϕ be a vector containing these probabilities. We create a survival matrix Φ by putting the ϕ on the diagonal. The vector of mortality hazards ϕ can incorporate any age- or stage-dependent pattern of extrinsic mortality. The population is now described by nðt þ 1Þ ¼ ΦAnðtÞ e ðtÞ; ¼ An

ð4:6Þ

e ¼ ΦΑ contains the additional mortality. The new where the new projection matrix A e and e matrix will have invasion fitness er ¼ logeλ; eigenvectors w v , and selection gradient Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:28:05, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.004

62

Senescence, Selection Gradients and Mortality

 e der 1 T T dvec A e e ¼ ⊗ v : w T T eλ dθ dθ

ð4:7Þ

This new selection gradient, calculated from (4.7) using the new eigenvalue and e could differ from the old selection gradient eigenvectors, and the new derivative of A in many ways. But in one special case we can obtain the new gradient directly from the original one. Suppose that the new mortality hazard is stage-independent, with the same survival probability ϕ applying to every stage. Then the survival matrix becomes Φ ¼ ϕI;

ð4:8Þ

e ¼ ϕA: The derivative of A e where I is the identity matrix. The new projection matrix is A is e dvec A dvec A ¼ϕ : T dθ dθT

ð4:9Þ

It is known that multiplying a matrix by a scalar multiplies the eigenvalues by the same scalar and does not change the eigenvectors. Thus, eλ ¼ ϕλ;

ð4:10Þ

e ¼ w; w

ð4:11Þ

e v ¼ v:

ð4:12Þ

Substituting (4.10), (4.11), (4.12) and (4.9) into the selection gradient (4.7), it follows immediately that der dr ¼ T: T dθ dθ

ð4:13Þ

That is, the selection gradient on θ is unchanged by the addition of stage-independent mortality. This result applies very generally to age- or stage-dependent demography and to selection on any traits, not only mortality schedules. However, it is very specific to the linear, time-invariant model shown in Equation (4B.2). In the sections that follow we will show how to formulate the same type of additional mortality for other kinds of timevarying or density-dependent populations. The analyses of these models are presented in Appendix 4A.

Stage-Independent Mortality in Periodic Environments Consider a time-varying environment that oscillates with a period of k. For simplicity, we will discuss the model in the context of periodic seasonal variation (e.g. Smith,

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:28:05, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.004

Mortality, Stage Dependence and Senescence

63

Caswell & Mettler-Cherry 2005), although the models could also approximate interannual variation. In this model, a projection matrix Bi is associated with season i. The number of stages, and their identity, may change over the cycle (e.g. reproductive stages may be present only at some seasons of the year), so Bi may in general be rectangular (Caswell 2001: chap. 13). The model (4.2) now projects the population from year t to year t + 1, with the projection matrix A given by the product of the Bi A ¼ Bk . . . B1 ;

ð4:14Þ

and the invasion fitness is given by r = log λ(A). Now we add additional mortality, which we allow to differ depending on the season so that e i ¼ Φi Bi B

ð4:15Þ

e ¼ ðΦk Bk Þ . . . ðΦ1 B1 Þ: A

ð4:16Þ

and

If mortality is independent of age or stage, regardless of what season it occurs in, then e i ¼ ϕi Bi B

ð4:17Þ

e ¼ ðϕk . . . ϕ1 ÞA: A

ð4:18Þ

and

e are the same as those of A, and the eigenvalue of A e is Once again, the eigenvectors of A that of A multiplied by ðϕk . . . ϕ1 Þ. So, as in the time-invariant case in the previous section, the selection gradient is unchanged by the additional mortality. No matter what the seasonal pattern of the additional mortality, no matter how the stages in the population change over seasons, and no matter what traits are under selection, the addition of stage-independent mortality has no effect on the selection gradients.

Stage-Independent Mortality in Stochastic Environments In a stochastic environment, demography is given by nðt þ 1Þ ¼ At nðtÞ;

ð4:19Þ

where At is generated by a stochastic environmental process. If the environmental process and the matrices it produces satisfy a few reasonable requirements (Tuljapurkar 1990), then the population will eventually grow at a rate that is the longterm average growth rate from any non-zero initial population vector n0 log λs ¼ lim

N→∞

1 log kAN1 . . . A0 n0 k: N

ð4:20Þ

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:28:05, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.004

64

Senescence, Selection Gradients and Mortality

The stochastic growth rate log λs is the appropriate measure of fitness in a stochastic environment (Tuljapurkar 1990). The selection gradient on a trait vector θ that affects the matrices Ai in the stochastic environment is dlog λs : dθT Additional mortality is added by writing e t ¼ Φ t At A

ð4:21Þ

where Φt is a survival matrix that may depend on time (e.g. if the matrices represent different moisture conditions, the additional mortality might be stronger in dry years than in wet years). If the mortality is stage-independent, then Φt ¼ ϕt I: Again, our task is to compare the selection gradients on θ with and without the additional mortality, in the case where the mortality is stage independent (Appendix 4A). The conclusion matches those for the time-invariant and the periodic models: no matter how the mortality is applied, if it is stage independent, it has no effect on the selection gradients.

Stage-Independent Mortality in Density-Dependent Models Density-dependent models add an extra complication because the demographic rates depend not only on the parameters but also on density, which, in turn, is affected by the parameters. We can write a non-linear, density-dependent model as nðt þ 1Þ ¼ A½n; θnðtÞ;

ð4:22Þ

where the matrix depends on the current state of the population and the trait vector θ. Such models can take many forms, depending on how A depends on the population e: vector n. We will assume that the model leads to a stable equilibrium population n The selection gradient on θ in density-dependent models is best described in the context of adaptive dynamics (e.g. Diekmann 2004), in which evolution is viewed as a sequence of invasions of new phenotypes until the population arrives at a phenotype that is un-invadable by any other. The invasion fitness of a trait is calculated as the growth rate of an invader with that trait, at low density, living in the environment defined by the equilibrium of the resident type. The invasion fitness is r ¼ log λðA½ˆn ; θÞ;

ð4:23Þ

and the selection gradient is given by the derivative of r with respect to θ. Imposing additional mortality leads to a new projection matrix e θ ¼ ΦA½n; θ: A½n;

ð4:24Þ

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:28:05, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.004

Mortality, Stage Dependence and Senescence

65

e θ: In this The invasion exponent er is now the log of the dominant eigenvalue of A½n; non-linear model, however, the additional mortality affects the selection gradient in two ways: first, as the mortality factor incorporated in Φ, and second, through the equilibrium population nˆ ; which, in turn, affects the entries in A. There is almost no limit on how these density effects might operate. However, if the density effects are also stage independent, then we can determine the results. In this case A½n; θ ¼ f ðnÞA½0; θ;

ð4:25Þ

where A½0;θ is the projection matrix at low density, and the scalar function f (n) reduces it as density increases, affecting every stage equally. Thus, we assume that f (0) = 1 and that f (n) decreases with density. Although (4.25) is unlikely given the details of how density effects can operate, it has a long theoretical history (Leslie 1948; Liu & Cohen 1987) and has occasionally been applied to empirical studies (Allen 1989; Rabinovitch 1969). In this very special case, we can show that the selection gradient is unchanged by the imposition of stage-independent mortality (Appendix 4A). The conclusion is general to any kind of age- or stage-classified demography and any form of density dependence provided that it is stage independent. If the density dependence is stage dependent, then it will produce stage-dependent effects even if the additional mortality is stage independent, leading to changes in the selection gradients.

The Logic of Counter-Examples The claim that additional mortality changes selection gradients in a way that makes senescence more likely has been made repeatedly. From a logical point of view, such a claim is refuted by the demonstration of even a single case where it is not true. We have done this here, with multiple cases: age- or stage-classified, time-invariant, periodic, stochastic or non-linear dynamics, with additional mortality that may vary with seasons or environmental states and with no limits on the kind of traits under consideration. From a logical point of view, this is the end of the story. The likelihood of stageindependent mortality is irrelevant. Failure to appreciate the logic of counter-examples is not uncommon. But, from a scientific point of view, the counter-example is just the first step. The claim about the effect of mortality on senescence was made not as an exercise in logic but because of its biological content. We know that it would be unlikely to find a mortality source in nature that is genuinely stage independent. Predators or herbivores target the large, or the small, or the young, or the old. Disease is more of a danger to the non-immune than to the immune. Extremes of temperature, moisture or other physical conditions will have effects that depend on the activities, morphology and physiological condition of the individuals experiencing them, and those vary by age or stage. This suggests that the counter-example should lead to refinements of the theory – to asking how stage-dependent mortality will affect

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:28:05, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.004

66

Senescence, Selection Gradients and Mortality

selection gradients. We turn to this topic next, using some very new results in demographic sensitivity analysis.

Age-Specific Mortality: How Does It Change Selection Gradients? The evolution of senescence depends on the decline with age in the selection gradients on mortality or survival, and we now know that age- or stage-independent mortality does not change the shape of the selection gradients. But age-dependent mortality can change the shape and thus, potentially, the resistance of the life cycle to the evolution of senescence. Caswell (2007a) made a qualitative conjecture that Increased early mortality ) less senescence; and Increased late mortality ) more senescence: Our knowledge of selection gradients makes it possible to evaluate and improve on this conjecture. To do so, we need a way to measure the shape of a selection gradient, and we need to see how the shape changes in response to mortality. Because the selection gradient is a derivative (the derivative of fitness with respect to the trait in question), to measure the changes in the selection gradient requires the second derivatives of fitness. Fortunately for us, those second derivatives are now available (Shyu & Caswell 2014).

Indices of Resistance to Senescence: Mp and Mf The shape of the selection gradient determines how easy, or how difficult, it is for senescence to evolve in a given life history. Because it is the decline in the senescence gradient with age that appears in the theories of senescence, some measure of that decline will determine how resistant, or how susceptible, the life history is to the evolution of senescence. Recall that the survival probability for age x is p(x); the selection gradient on p(x) is sð xÞ ¼

∂λ : ∂pð xÞ

ð4:26Þ

The result of selection depends on how the trait affects survival probabilities and then how the selection gradient maps those effects to fitness. So let us probe the selection gradient with a trait θ that is a poster child for the evolution of senescence: it affects survival at every age, with dpð xÞ ¼ a  bx: dθ

ð4:27Þ

That is, the trait θ increases survival at young ages (from x = 0 to x ¼ xˆ ¼ a=b), but at the cost of a reduction in survival at ages older than xˆ . When xˆ is large, the effect of θ is

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:28:05, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.004

Age-Specific Mortality: How Does It Change Selection Gradients?

67

positive over a large portion of the life cycle; in the limit as b→0, the effect of θ on survival is positive at every age. As b becomes large, θ has negative effects on survival at almost every age. The selection gradient on this senescence-related trait θ is dλ X ∂λ dpð xÞ ¼ dθ ∂pð xÞ dθ x X ¼ sð xÞða  bxÞ: x

ð4:28Þ ð4:29Þ

Senescence is favoured if and only if dλ=dθ is positive, which requires X xsð xÞ a > X ¼ M: b sð xÞ

ð4:30Þ

The quantity on the right-hand side of (4.30) is the weighted mean age, calculated over the selection gradient on survival. We will denote such means by M (think of it as inspired by Medawar). The larger this mean age, the more slowly s(x) declines, the more difficult it is for (4.30) to be satisfied and the more resistant the life history is to senescence.3 The opposite is true when the mean age over the selection gradient is small, in which case the gradient falls off steeply with age, and traits affecting later life have little effect; the life history is senescence prone. Thus, the mean age is a shape parameter capturing some of the senescence-proneness of a life cycle. We can define the mean age in relation to any of the selection gradients; for example, Mp ¼ mean age over the selection gradient on survival; Mμ ¼ mean age over the selection gradient on mortality; Mf ¼ mean age over the selection gradient on fertility:

3

A similar analysis shows that if θ affects mortality rate μ rather than survival, with dμðxÞ ¼ a þ bx; dθ

ð4:31Þ

then X ∂λ x dλ a ∂μðxÞ > 0⇔ > X ∂λ dθ b ∂μðxÞ X

∂λ ∂PðxÞ : ¼ X ∂λ Pð x Þ ∂PðxÞ

ð4:32Þ

xPð xÞ

ð4:33Þ

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:28:05, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.004

68

Senescence, Selection Gradients and Mortality

Effects of Age-Dependent Mortality on M Consider a population with age classes 1, . . ., ω. If we write the selection gradient on the trait vector θ as sT ¼

dλ ; dθT

ð4:34Þ

and define a vector containing the midpoints of the age intervals x ¼ ð0:5

1:5

. . . ω  0:5 ÞT ;

ð4:35Þ

then the index of resistance to senescence is M¼

xT s : 1T s

ð4:36Þ

To find the effect of additional mortality, we need to calculate the derivative of M with respect to that mortality. That derivative is dM dM ds ¼ : dμT dsT dμT

ð4:37Þ

Evaluating the terms in (4.37) gives " # dM xT sTx ⊗ 1T ds ¼ T   2 dμT dμT 1 s 1T s

ð4:38Þ

(for derivation, see Appendix 4B). The derivative ds=dμT is the key to this result. Because the selection gradient s is a vector of first derivatives of λ, differentiating it requires methods for second derivatives, given in Shyu and Caswell (2014). The end result depends not only on the sensitivity of A to the parameter vector θ but also on both eigenvectors (w and v) and their sensitivities to the entries of A      ds dvec A T dw dv dvec A þ ðw ⊗ Iω Þ ¼ ð I ω ⊗ vÞ DðpÞ: ð4:39Þ dμT dvecT A dvecT A dpT dθT See Appendix 4B for definitions of terms and derivations of results.

Applications According to the additional mortality hypothesis (Williams 1957; Williams et al. 2006), increased mortality should reduce M, facilitating the evolution of senescence. That is, the derivative of M with respect to μ should be negative. We have already seen that

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:28:05, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.004

Applications

69

age-independent mortality has no such effect. Now we will examine age-dependent mortality. The effects, it turns out, depend strongly on where in the life cycle the mortality is added. We will consider several cases. First, we will examine some artificial life histories produced by combining simple models for mortality and fertility. Then we will use a long time series of life tables for the population of Sweden to see what the effects look like in a real (human) population.

Model Life Cycles To create a model life cycle, we specify a mortality schedule and a fertility schedule. The simplest mortality schedule is age invariant, with a constant mortality rate; it provides a non-senescent baseline against which to compare other mortality schedules. As an alternative, we will explore the commonly used Gompertz model, in which mortality increases exponentially with age μð xÞ ¼ μ

age-invariant;

ð4:40Þ

Gompertz :

ð4:41Þ

μð xÞ ¼ aebx

The simplest fertility schedule would also be age-independent, exhibiting no senescence of any kind so that f ð xÞ ¼ c: ð4:42Þ A model fertility schedule with age dependence is provided by the Brass polynomial (Brass 1960), which concentrates fertility into an interval between a first reproductive age s and a final age s + w, outside of which fertility is 0 f ð xÞ ¼ cð x  sÞðs þ w  xÞ2 s ≤ x ≤ s þ w

ð4:43Þ

and 0 otherwise. The four combinations of these functions provide examples of life cycles with and without senescence in mortality and fertility. In our examples, we use parameter values4 that approximate human life cycles so that ages range up to x = 100. In each case, total fertility was rescaled so that λ = 1. Figure 4.2 shows the sensitivity of Mp to changes in age-specific mortality for each of the four model life cycles. It is worthwhile to examine this figure closely. When fertility is age invariant, regardless of the mortality schedule, the sensitivity of dMp =dμð xÞ is positive up to x ≈ 20 and then becomes negative, eventually approaching zero at old ages. This means that additional age-specific mortality that affects young ages will increase Mp, reducing the tendency to evolve senescence. Conversely, additional mortality at later ages reduces Mp, increasing the tendency to senescence. When fertility is age dependent, once again, additional mortality at young ages increases Mp, and mortality at older ages reduces Mp (Figure 4.2). However, the 4

The results are not strongly dependent on parameter values. In the figures presented here, we set μ = 0.1 and c = 0.007 for the age-independent models. For the Gompertz mortality function, a = 0.01 and b = 0.03. The Brass polynomial parameters were c = 1, s = 15 and w = 35. Fertility was then rescaled to achieve λ = 1.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:28:05, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.004

70

Senescence, Selection Gradients and Mortality

1

8

0.5 dMp/dµ(x)

dMp/dµ(x)

6 4 2

−0.5

0 −2

0

0

−1

20

40 60 80 100 Age x (a) Mortality constant, fertility constant

8

0

20

40 60 80 100 Age x (b) Mortality constant, fertility Brass

1

6 dMp/dµ(x)

dMp/dµ(x)

0.5 4 2 0

0

−0.5 −2 −4

Figure 4.2

0

20

40 60 80 100 Age x (c) Mortality Gompertz, fertility constant

−1

0

20

40 60 80 100 Age x (d) Mortality Gompertz, fertility Brass

Sensitivity of the senescence index Mp, defined in (4.30), to changes in age-specific mortality μ(x) for several model life tables. Mortality rates are constant (first row) or follow the Gompertz model (second row). Fertilities are either constant (first column) or follow the Brass polynomial model fertility schedule (second column).

sensitivity of Mp to age-specific mortality first increases at early ages before declining and becoming negative. The sensitivity of Mp to increased mortality is zero after fertility falls to zero because changes in mortality have no effects after that age. Figure 4.3 shows the sensitivities of Mf to changes in age-specific mortality. The cases with age-independent fertility show similar patterns to the pattern of sensitivity of Mp: mortality at early ages leads to reduced senescence; mortality at later ages leads to increased senescence.

An Historical Series for Sweden Although the principles are based on demography, not taxonomic identity, and although a diversity of ageing patterns exist (Jones et al. 2014), human populations remain a valuable source of detailed age-specific mortality and fertility data (Human Fertility Database 2013; Human Mortality Database 2013). Consider the historical series for Sweden from 1891 to 2007. During this period, Swedish demography

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:28:05, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.004

8

4

6

3 dMf /dμ(x)

dMf /dμ(x)

Applications

4 2 0 −2

2 1 0

0

20

40 60 Age x

80

−1

100

0

(a) Mortality constant, fertility constant

80

100

10 dMf /dμ(x)

dMf /dμ(x)

40 60 Age x

15

10 5 0

Figure 4.3

20

(b) Mortality constant, fertility Brass

15

−5

71

5 0 −5

0

20

40 60 80 100 Age x (c) Mortality Gompertz, fertility constant

−10

0

20

40 60 80 100 Age x (d) Mortality Gompertz, fertility Brass

Sensitivity of the senescence index Mf, defined in (4.30), to changes in age-specific mortality μ(x) for several model life tables. Mortality rates are constant (first row) or follow the Gompertz model (second row). Fertilities are either constant (first column) or follow the Brass polynomial model fertility schedule (second column).

underwent significant changes (Figure 4.4). Life expectancy increased by almost 60 per cent, from fifty-three to eighty-three years; total fertility rate fell by half; and the net reproductive rate declined from R0 ≈ 1.5 to a value below 1. The population growth rate fluctuated from increases on the order of 1 per cent per year to declines of the same magnitude. These changes, over the span of just a few generations, would be regarded as significant demographic variation in a natural population of large mammals. How much did they change the shape of the selection gradients and the way in which those gradients would respond to mortality changes? Figure 4.5 shows that Mp declined smoothly from 1891 to about 1975, at which point it began to increase. The range of values, from about 15.5 to 13 years, gives the age at which our hypothetical test trait, with its linear decline in marginal effect on survival probability given by (4.27), would switch from positive to negative. The value of Mf increased from approximately thirty to approximately forty years, with fluctuations that are negatively correlated with those of the population growth rate λ (Figure 4.4). Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:28:05, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.004

72

Senescence, Selection Gradients and Mortality

0.015

90 Life expectancy

Growth rate

0.01 0.005 0 −0.005 1900

1950

70 60 50

2000

5

1.6

4

1.4

1900

1950

2000

1900

1950 Year

2000

1.2 R0

TFR

−0.01

80

3

1 2

0.8

1 1900 Figure 4.4

2000

1950 Year

Demographic trends in Sweden, 1891–2007. The population growth rate λ, the life expectancy, the total fertility rate and the net reproductive rate R0.

16

Mp

15

14

13

1900

1920

1940

1900

1920

1940

1960

1980

2000

1960

1980

2000

50

Mf

45 40 35 30 25

Years Figure 4.5

Changes in senescence indices on survival (Mp) and on fertility (Mf), as defined in (4.30), in Sweden, 1891–2007.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:28:05, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.004

Conclusion

dMp /dμ(x)

Year 2000

Years 1891−2000

1

1

0.5

0.5

0

0

−0.5

−0.5

−1

−1

dMf /dμ(x)

0

50

100

30

30

20

20

10

10

0

0

−10

−10

0

50

100

0

50 Age

100

−20

−20 0 Figure 4.6

73

50 Age

100

Sensitivity of the senescence indices Mp and Mf to changes in age-specific mortality μ(x) in Sweden. The left column shows the sensitivities for the year 2000; the right column shows the sensitivities for all years 1891–2007.

This is not surprising because the slope of the selection gradient on fertility mirrors that of the stable age distribution, which is affected by both the mortality schedule and the value of λ. This being the first time these indices have been computed, it remains to be seen what patterns comparative studies will reveal. The results of calculating dM=dμT from (4.37) are shown in Figure 4.6. In the year 2000, the sensitivities of Mp and Mf show patterns very similar to those shown in Figures 4.2 and 4.3. Looking at all 117 years combined shows that the figures have changed quantitatively but not at all qualitatively.

Conclusion We have now examined the so-called central prediction (e.g. Williams 2006) of senescence theory, that mortality should influence the selection gradients, from many angles. The picture is both more complicated and more interesting than it appeared at first. It is not the case that additional mortality automatically favours the evolution of senescence. If the additional mortality is age or stage independent, it has no effect on the selection gradient. We have extended earlier results (Abrams 1993; Caswell 2007a) to show that this remains true for

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:28:05, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.004

74

Senescence, Selection Gradients and Mortality

• Both age- and stage-classified models; • Periodic or stochastic time-varying models, regardless of how the stage-independent mortality changes over time; and • Non-linear density-dependent models, providing that both the additional mortality and the effects of density are age or stage independent. The conclusion applies to selection on any trait, not only traits (i.e. mortality, survival or fertility) usually associated with discussions of senescence. Given that stage-independent mortality has no effect, it is natural to wonder what effect stage-dependent mortality might have. We find that it has a consistent effect on the shape of the selection gradient, where we measure shape by an index of the age pattern of selection gradients on the trait, and the conditions under which a pleiotropic, senescence-creating trait that improves early performance at the expense of later performance can invade. In both model life cycles and in a historical series from Sweden, it turns out that • Additional mortality acting early in life changes the selection gradient to make the life cycle more resistant to senescence, and • Additional mortality imposed late in life has the opposite effect; it changes the selection gradient to increase the tendency to evolve senescence. These analyses do not require that the additional mortality be in any sense ‘extrinsic’. They apply equally to an increase in the hazard of being struck by lightning or an increase in the hazard of death due to cardiovascular disease; the former would usually be regarded as extrinsic and the latter as intrinsic. The ecological setting in which a population lives, both the biotic and the abiotic environment, determine the selection pressures to which it is subject. The implicit assumption of the structured models that underlie demography is that processes act in a stage-dependent way: the old differ from the young, the large from the small, the mature from the immature, the reproductive from the non-reproductive. If there is a message from this chapter, it is that the rich ecological and demographic consequences of stage dependence must be taken into account in analysing the selection on senescence.

Acknowledgements Support for this research has been provided by the European Research Commission (ERC Advanced Grant 322989), the US National Science Foundation (NSF Grant DEB1145017), the Alexander Humboldt Foundation and the Academic Programs Office of the Woods Hole Oceanographic Institution. Comments from Dmitrii Logofet and an anonymous reviewer helped to improve an earlier version. HC is grateful for the hospitality of the Max Planck Institute for Demographic Research, where these ideas were originally developed.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:28:05, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.004

Appendix 4A: Derivations

75

Appendix 4A: Derivations In this appendix we present the derivations of the selection gradient results. These analyses use matrix calculus methods, described in detail in Caswell (2007b, 2008, 2009, 2012) and Shyu and Caswell (2014). In this approach, the derivative of a vector y with respect to a vector x is a matrix whose (i, j) entry is the derivative of yi with respect to xj   dyi dy ¼ : ð4A:1Þ dxj dxT When matrices appear (e.g. in differentiating a projection matrix A), they are transformed to vectors by the vec operator, where vec X stacks each column of X on top of the following column. The Kronecker product is denoted by ⊗ . The identity matrix is I, and when necessary, subscripts are used to indicate its dimension, so that Iω is an ω × ω identity. The diagonal matrix with x on the diagonal is DðxÞ. All vectors are column vectors; xT is the transpose of x. The 1-norm of a vector x is denoted ‖x‖. Periodic Environments The population in a periodic environment is described by the matrix product A ¼ Bk . . . B1 ;

ð4A:2Þ

and fitness is given by r, the log of the dominant eigenvalue λ of A. We suppose that the matrices Bi are functions of a parameter vector θ. The selection gradient on θ is dr 1 dλ dvec A ¼ T λ dvecT A dθT dθ k 1 dλ X ∂vec A dvec Bi T λ dvec A i¼1 ∂vecT Bi dθT

ð4A:4Þ

k h i dvec B X T 1 T i w ⊗ vT Ci1 ⊗ Ckiþ1 ; 1 T λ dθ i¼1

ð4A:5Þ

¼ ¼

ð4A:3Þ

where Cyx is the product of the B matrices from Bx to By; that is Cyx ¼ By . . . Bxþ1 Bx (Caswell 2012). Now impose an additional mortality source. The mortality may vary from season to season so that e i ¼ Φi Bi ; B

ð4A:6Þ

where Φi ¼ Dðϕi Þ incorporates the vector of survival probabilities in season i. In the e ¼e e i ¼ ϕi Bi and A special case of stage-independent mortality, B ϕA; where

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:28:05, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.004

76

Senescence, Selection Gradients and Mortality

e ϕλ but ϕ ¼ ðϕ1 . . . ϕk Þ. The additional mortality changes the growth rate from λ to e leaves the eigenvectors unchanged. The derivative of Bi is ei dB dBi ¼ ϕi T : dθT dθ

ð4A:7Þ

The selection gradient on θ under the new mortality regime is k  X  T der 1  T dvec Bi ¼ ½ Ci1 ⊗ Ckiþ1  w ⊗ vT e ϕ 1 T e dθ dθT ϕλðAÞ i¼1

¼

dr : dθT

ð4A:8Þ ð4A:9Þ

In other words, no matter how the second mortality factor may vary with seasons, as long as it is stage independent (no matter how the identity and number of stages may vary over the seasons), it has no effect on the selection gradient. Stochastic Environments Suppose that the matrices At that appear in the stochastic model (4.19) depend on a vector of parameters θ. The invasion fitness is given by the stochastic growth rate log λs (Tuljapurkar 1990), and the selection gradient on θ is given by N 1 dlog λs 1X ½wT ðtÞ ⊗ vT ðt þ 1Þ dvec At ¼ lim ; N→∞ N RðtÞvT ðt þ 1ÞwT ðt þ 1Þ dθT dθT t¼0

ð4A:10Þ

which is the matrix calculus version of Tuljapurkar’s (1990) sensitivity formula, as presented in Caswell (2010). In (4A.10), R(t) is the one-step growth rate at time t, and the vectors w(t) and v(t) are the analogues of the right and left eigenvectors of the time-invariant matrix A; these are given by At wð t Þ ; ∥At wðtÞ∥

ð4A:11Þ

vT ðtÞAt1 ; ∥vT ðtÞAt1 ∥

ð4A:12Þ

wð t þ 1Þ ¼ vT ðt  1Þ ¼ RðtÞ ¼

∥At wðtÞ∥ : ∥wðtÞ∥

ð4A:13Þ

See section 14.4 of Caswell (2001). We now impose an additional mortality, replacing At by

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:28:05, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.004

Appendix 4A: Derivations

e t ¼ Φ t At ; A

77

ð4A:14Þ

where Φt ¼ Dðϕt Þ incorporates an arbitrarily time-varying vector of survival probabilities. If the additional mortality is stage independent, then Φt ¼ ϕt I. The selection gradient on θ is now given by (4A.10), but with At, w(t), v(t) and R(t) e t, w e e ðtÞ, e replaced by A v ðtÞ and RðtÞ, and e t dvec At dvec At dvec A ¼ : T T d vec At dθT dθ

ð4A:15Þ

When the additional mortality is stage independent, the vectors and growth rates satisfy e ðtÞ ¼ ϕt RðtÞ; R

ð4A:16Þ

e ðtÞ ¼ wðtÞ; w

ð4A:17Þ

e v ðtÞ ¼ vðtÞ;

ð4A:18Þ

dvec At ¼ ϕt Is2 : dvecT At

ð4A:19Þ

and

Substituting these results into (4A.10) reveals that dlog eλ s dlog λs ¼ : dθT dθT

ð4A:20Þ

That is, the selection gradient on any trait is unaffected by an arbitrarily time-varying additional mortality, provided that mortality is stage independent. Density-Dependent Models We suppose that the density-dependent growth model (4.22) leads to an equilibrium n. From (4.4) the selection gradient on the trait vector θ is  dvec A½ˆn  dr 1  T ¼ w ⊗ vT dθ λ dθ

ð4A:21Þ

When we impose additional mortality with the diagonal matrix Φ, the density-dependent projection matrix becomes e θ ¼ ΦA½n; θ: A½n;

ð4A:22Þ

e n ;θ. The extra The invasion exponent is the log of the dominant eigenvalue of A½ˆ mortality affects the selection gradient in two ways: first, as the mortality factor incorporated in Φ and, second, through the equilibrium population nˆ : Even if the mortality is stage independent, the density feedbacks may be stage dependent.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:28:05, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.004

78

Senescence, Selection Gradients and Mortality

If both the additional mortality and the density effects are stage independent, A½n; θ ¼ f ðnÞA½0; θ;

ð4A:23Þ

where the density effects operate through the scalar function f (n), affecting every stage equally. At an equilibrium nˆ ; the dominant eigenvalue of A½ˆn ;θ must be 1. Thus, nˆ must satisfy f ðnˆ Þ ¼

1 : λðA½0; θÞ

ð4A:24Þ

If the extra mortality is stage-independent, with Φ ¼ ϕI, then the new equilibrium n^e satisfies eÞ ¼ f ðn

1 ϕλðA½0; θÞ

ð4A:25Þ

and the projection matrix for the invader, from which the invasion exponent is calculated, is 1 ^; θ ¼ e e A½n ϕA½0; θ ð4A:26Þ ϕλðA½0; θÞ ¼ A½ˆn ; θ

ð4A:27Þ

Since the matrix is unchanged by the imposition of the extra mortality, the selection gradients calculated from the eigenvalues of that matrix are also unchanged.

Appendix 4B: Mortality and the Shape of the Selection Gradient The shape index M describes the selection gradient, which is the first derivative of λ. Differentiating M thus requires calculation of second derivatives of λ. To do so, we use the results of Shyu and Caswell (2014). By the chain rule of matrix calculus, we can write dM dM ds ¼ dμT dsT dμT

ð4B:1Þ

where s ¼ dλ=dpT and s ¼ dλ=df T are the selection gradients on survival and on fertility, respectively. Consider the two components in turn. Differentiating M in (4.36) yields 

 1T s xT ðdsÞ  1T ðdsÞxT s dM ¼ :  T 2 1 s

ð4B:2Þ

Applying the vec operator to both sides gives

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:28:05, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.004

Appendix 4B: Mortality and the Shape of the Selection Gradient

79

T  s x ⊗ 1T xT dM ¼ T ds   2 ds; 1 s 1T s

ð4B:3Þ

T  s x ⊗ 1T dM xT ¼   2 : dsT 1T s 1T s

ð4B:4Þ

so

To differentiate the selection gradient, let θ denote the trait (p or f) under consideration. Then  sT ¼

dλ dθT

T

  dvec A ¼ wT ⊗ vT : dθT

Following equations (12ff) in Shyu and Caswell (2014), we differentiate sT  

 T   T  dvec A  T  dvec A T T T T ds ¼ dw ⊗ v þ w ⊗ dv þ w ⊗v d : dθT dθT

ð4B:5Þ

ð4B:6Þ

Because it is the derivative of a matrix with respect to (some of) its entries, dvecA=dθT is a matrix of zeros and ones. Thus, its differential is zero, and the second term in (4B.6) disappears. Transposing both sides of (4B.6), rearranging, and invoking the chain rule to go from differentials to derivatives with respect to μ gives ds ¼ dμT



    T  dv dvec A T  T dw I ⊗ v þ w ⊗ I : ω ω dμT dμT dθT

ð4B:7Þ

The derivatives of the eigenvectors w and v are dw dw dvec A dp ¼ T dμ d vecT A dpT dμT

ð4B:8Þ

with a corresponding expression replacing w with v. The eigenvector derivatives dw=dvecT A and dv=d vecT A are given in equations (21) and (22) of Shyu and Caswell (2014). Because p ¼ expðμÞ; dp ¼ DðpÞ: dμT

ð4B:9Þ

It remains only to obtain dvec A=dθT . If θ = p, define Zp(i) as a matrix of size A with a 1 in the (i + 1, i) entry and zeros elsewhere. Then column i of dvec A=dpT is

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:28:05, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.004

80

Senescence, Selection Gradients and Mortality

dvec A ð: ;iÞ ¼ vec ZpðiÞ : dpT

ð4B:10Þ

If θ = f, define Zf(i) as a matrix of size A with a 1 in the (1, i) entry and zeros elsewhere. Then dvec A ð: ;iÞ ¼ vec Zf ðiÞ : df T

ð4B:11Þ

Combining all the pieces gives the final result      ds dvec A T dw dv dvec A þ ðw ⊗ I ω Þ ¼ ðIω ⊗ vÞ DðpÞ: dμT dvecT A d vecT A dpT dθT ð4B:12Þ

References Abrams, P. A. (1993). Does increased mortality favor the evolution of more rapid senescence? Evolution, 877–87. Allen, L. J. S. (1989). A density-dependent Leslie matrix model. Mathematical Biosciences, 95, 179–87. Baudisch, A. (2008). Inevitable Aging? Contributions to Evolutionary-Demographic Theory (Berlin: Springer Science and Business Media). Brass, W. (1960). The graduation of fertility distributions by polynomial functions. Population Studies, 14(2), 148–62. Caswell, H. (1978). A general formula for the sensitivity of population growth rate to changes in life history parameters. Theoretical Population Biology, 14(2), 215–30. Caswell, H. (2001). Matrix Population Models (2nd edn.) (Sunderland, MA: Sinauer Associates). Caswell, H. (2006). Applications of Markov chains in demography. In MAM2006: Markov Anniversary Meeting (pp. 319–34) (Raleigh, NC: Boson Books). Caswell, H. (2007a). Extrinsic mortality and the evolution of senescence. Trends in Ecology and Evolution, 22(4), 173–4. Caswell, H. (2007b). Sensitivity analysis of transient population dynamics. Ecology Letters, 10(1), 1–15. Caswell, H. (2008). Perturbation analysis of non-linear matrix population models. Demographic Research, 18(3), 59–116. Caswell, H. (2009). Stage, age and individual stochasticity in demography. Oikos 118 (12), 1763–82. Caswell, H. (2010). Reproductive value, the stable stage distribution, and the sensitivity of the population growth rate to changes in vital rates. Demographic Research, 23, 531–48. Caswell, H. (2012). Matrix models and sensitivity analysis of populations classified by age and stage: a vec-permutation matrix approach. Theoretical Ecology, 5(3), 403–17.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:28:05, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.004

References

81

Caswell, H. & Shyu, E. (2012). Sensitivity analysis of periodic matrix population models. Theoretical Population Biology, 82(4), 329–39. Charlesworth, B. (1994). Evolution in Age-Structured Populations (2nd edn.) (Cambridge University Press). Charlesworth, B. (2000). Fisher, Medawar, Hamilton and the evolution of aging. Genetics, 156(3), 927–31. Chen, J., Lewis, E. E., Carey, J. R., et al. (2006). The ecology and biodemography of Caenorhabditis elegans. Experimental Gerontology, 41, 1059–65. Dieckmann, U. & Law, R. (1996). The dynamical theory of coevolution: a derivation from stochastic ecological processes. Journal of Mathematical Biology, 34(5–6), 579–612. Diekmann, O. (2004). A beginner’s guide to adaptive dynamics. Polish Academy of Sciences, Banach Center Publications, 63, 47–86. Finch, C. E. (1990). Longevity, Senescence, and the Genome (University of Chicago Press). Fisher, R. A. (1930). The Genetical Theory of Natural Selection (revised 1956) (Oxford University Press). Hamilton, W. D. (1966). The moulding of senescence by natural selection. Journal of Theoretical Biology, 12(1), 12–45. Human Fertility Database. Max Planck Institute for Demographic Research (Germany) and Vienna Institute of Demography (Austria). Available at www.humanfertility.org Human Mortality Database. University of California, Berkeley (USA), and Max Planck Institute for Demographic Research (Germany). Available at www.mortality.org or www.humanmortality.de Jones, O. R., Scheuerlein, A., Salguero-Gómez, R., et al. (2014). Diversity of ageing across the tree of life. Nature, 505(7482), 169–73. Kirkwood, T. B. L. (1990). The disposable soma theory of aging. In Genetic Effects on Aging, Vol. II (pp. 9–19), ed. D. E. Harrison (Caldwell, NJ: Telford Press). Kirkwood, T. B. L. & Holliday, R. (1986). Ageing as a consequence of natural selection. In The Biology of Human Ageing (pp. 1–16), ed. A. H. Bittles & A. J. Collins (Cambridge University Press). Lande, R. (1982). A quantitative genetic theory of life history evolution. Ecology, 63(3), 607–15. Lee, R. D. (2003). Rethinking the evolutionary theory of aging: transfers, not births, shape senescence in social species. Proceedings of the National Academy of Sciences of the United States of America, 100(16), 9637–42. Leslie, P. H. (1945). On the use of matrices in certain population mathematics. Biometrika, 33, 183–212. Leslie, P. H. (1948). Some further notes on the use of matrices in population mathematics. Biometrika, 35, 213–45. Liu, L. & Cohen, J. E. (1987). Equilibrium and local stability in a logistic matrix model for age-structured populations. Journal of Mathematical Biology, 25, 73–88. Medawar, P. (1952). An Unsolved Problem in Biology (London: Lewis). Metz, J. A. J. (2008). Fitness. In Encyclopedia of Ecology, Vol. 2 (pp. 1599–1612) (Amsterdam: Elsevier). Metz, J. A. J., Nisbet, R. M. & Geritz, S. A. H. (1992). How should we define ‘fitness’ for general ecological scenarios? Trends in Ecology and Evolution, 7(6), 198–202.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:28:05, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.004

82

Senescence, Selection Gradients and Mortality

Price, G. R. (1970). Selection and covariance. Nature, 227, 520–1. Rabinovitch, J. E. (1969). The applicability of some population growth models to a single-species laboratory population. Annals of the Entomological Society of America, 62(2), 437–42. Reznick, D., Bryant, M. & Holmes, D. (2006). The evolution of senescence and post-reproductive lifespan in guppies (Poecilia reticulata). PLoS Biology, 4(1), e7. Reznick, D. N., Bryant, M. J., Roff, D., et al. (2004). Effect of extrinsic mortality on the evolution of senescence in guppies. Nature, 431(7012), 1095–9. Robertson, A. (1968). The spectrum of genetic variation. In Population Biology and Evolution, ed. R. C. Lewontin (pp. 5–16 ) (Syracuse University Press). Rose, M. R. (1990). Evolutionary Biology of Aging (Oxford University Press). Shyu, E. & Caswell, H. (2014). Calculating second derivatives of population growth rates for ecology and evolution. Methods in Ecology and Evolution, 5, 473–82. Silvertown, J. (2013). The Long and the Short of It: The Science of Life Span and Ageing (University of Chicago Press). Smith, M., Caswell, H. & Mettler-Cherry, P. (2005). Stochastic flood and precipitation regimes and the population dynamics of a threatened floodplain plant. Ecological Applications, 15(3), 1036–52. Tuljapurkar, S. (1990). Population dynamics in variable environments. in Lecture Notes in Biomathematics, no. 85 (New York: Springer-Verlag). Wachter, K. W., Evans, S. N. & Steinsaltz, D. (2013). The age-specific force of natural selection and biodemographic walls of death. Proceedings of the National Academy of Sciences of the United States of America, 110(25), 10141–6. Wachter, K. W. & Finch, C. E. (1997). Between Zeus and the Salmon: The Biodemography of Longevity (Washington, DC: National Academies Press). Wachter, K. W., Steinsaltz, D. & Evans, S. N. (2014). Evolutionary shaping of demographic schedules. Proceedings of the National Academy of Sciences of the United States of America, 111(Suppl. 3), 10846–53. Wensink, M. J., Caswell, H., & Baudisch, A. (2016). The rarity of survival to old age does not drive the evolution of senescence. Evolutionary Biology. doi 10.1007/s 11692-016-9385-4. Williams, G. C. (1957). Pleiotropy, natural selection, and the evolution of senescence. Evolution, 11(4), 398–411. Williams, P. D., Day, T., Fletcher, Q. & Rowe, L. (2006). The shaping of senescence in the wild. Trends in Ecology and Evolution, 21(8), 458–63. Wright, S. (1937). The distribution of gene frequencies in populations. Proceedings of the National Academy of Sciences of the United States of America, 23(6), 307–20.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:28:05, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.004

5

Taxonomic Diversity, Complexity and the Evolution of Senescence Alan A. Cohen

Short Summary One of the largest outstanding questions in the biology of senescence is why senescence differs as it does across species. New information on the diversity of demographic senescence patterns increases the need for a clear explanation. Such an explanation must also incorporate what is known about biological mechanisms, particularly the possibility that senescence may be due in part to the complex dynamics of regulatory networks. Classical theory on senescence cannot explain why some species with distinct somas do not senesce, why many species without distinct somas do senesce and how physiological complexity may affect the evolution of senescence. This chapter outlines what is needed for a satisfactory explanation of the evolution of senescence and its underlying physiological mechanisms.

Introduction Senescence has long been one of the major evolutionary puzzles. Given that each cell contains a full complement of DNA and that organisms develop from a single cell, it is not immediately obvious why regeneration is insufficient to keep all organisms eternally young, as indeed happens in some species (Martínez 1998). Evolutionary theory has long provided plausible answers to this question: the three ‘classical’ senescence theories: mutation accumulation (MA hereafter), antagonistic pleiotropy (AP) and the disposable soma (DS), reviewed in more detail later (Kirkwood 1977; Medawar 1952; Williams 1957; see also Chapter 2 by Kirkwood for further details on DS). All three theories are based on a shared principle: that the force of natural selection gets weaker with age due to the increasing probability that an individual will have died due to random (i.e. not age-related) causes. Since their formulation, these theories have together provided a sufficient framework to understand the evolution of senescence. Several recent developments, however, are starting to call into question the adequacy of these theories. On the demographic side, recent models show that senescence is less inevitable than previously proposed by Hamilton (1966) in the context of the classical theories (Baudisch 2008; Horvitz and Tuljapurkar 2008; Vaupel et al. 2004). On the taxonomic side, evidence for non-senescing species could previously have been considered anecdotal or present only in certain taxa with particular conditions (Finch 1990; Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:30:17, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.005

84

Taxonomic Diversity, Complexity and the Evolution of Senescence

Kirkwood 1985). Recent work now shows clearly that there is a wide diversity of patterns of senescence and non-senescence broadly but not randomly distributed across the Tree of Life (Baudisch et al. 2013; Jones et al. 2014). These patterns are not explained by the classical theories and may sometimes contradict them. Lastly, on the mechanistic front, there is increasing evidence that senescence is related to emergent properties of complex adaptive systems (Cohen et al. 2013; Lipsitz 2004). If complex system dynamics are important in determining senescence, there are certain challenges the classical theories must overcome. The goal of this chapter is to review the new evidence and to evaluate to what extent it may be in conflict with the classical theories. The theories are described in substantial detail elsewhere, so I will focus mainly on the aspects that are most pertinent for the new evidence. Likewise, I will concentrate on the taxonomic and mechanistic evidence more than the demographic evidence, which is treated by other authors in this book. DS is the most fully developed of the three theories mechanistically, so I will spend substantial time outlining the mechanistic implications of this theory. I will show that there is nothing in the new evidence that disproves the classical theories; nonetheless, these theories may not be sufficient to explain the full range of what is known about senescence. In particular, while it is possible to incorporate all the current evidence into DS, to do so requires a number of suppositions and models that are somewhat unlikely. A note on terminology: throughout this book age-related changes in fitness components (e.g. demographic ageing) are termed ‘senescence’ rather than ‘ageing’. I conform to this, but given the substantial physiological component of this chapter, I use ‘ageing’ to refer to age-related declines in physiological functioning and to distinguish this from the fitness consequences of these declines.

The Three Classical Theories Mutation Accumulation (MA) MA, originally proposed by Medawar (1952), posits that the declining force of natural selection with age (‘selection’s shadow’) results in weakened selection against mutations that primarily have negative effects later in life. At some age, selection will become effectively impotent to weed out late-acting deleterious alleles; the accumulation of multiple such alleles could be responsible for ageing. MA is the only one of the three classical theories to suggest that ageing is completely non-adaptive. The other two theories, AP and DS, suggest that while ageing does decrease fitness, it can still be selected for in the context of trade-offs. MA has also proven the hardest of the three theories to test empirically; some tests that use a combination of population genetics and demography in model organisms appear to provide support for MA (Hughes et al. 2002), but the evidence is not generally considered conclusive. In fact, it is difficult to find predictions of MA that differ substantially from those for AP absent evidence of how selection acts on specific genes and their mutations (Hughes & Reynolds 2005; Moorad & Promislow 2009). While it might be

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:30:17, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.005

The Three Classical Theories

85

feasible to identify such genes, it is not likely that we could identify all of the many that would be needed to cause senescence. Thus, MA remains a plausible explanation for senescence and likely plays at least some role, but the lack of testable predictions has made it difficult to evaluate the degree to which it actually is responsible for senescence. However, several predictions can be made that relate to the new evidence reviewed later. First and most importantly, MA does not suggest any mechanism by which any organism might avoid senescence and thus predicts that senescence is universal. Also, MA suggests that many mutations make minor contributions to senescence and that these mutations may be more or less independent. This implies that ageing mechanisms should be diverse, cropping up in various aspects of physiological functioning. If we also assume that selection does have some power to weed out deleterious mutations, just not enough power to weed them out as fast as they appear, there should be a relatively high turnover rate of mutations (and thus of biological mechanisms) over evolutionary time. Thus, MA could be construed as predicting that (1) all organisms subject to selection’s shadow would evolve senescence, (2) there are few, if any, shared ageing mechanisms across all species and (3) ageing mechanisms should vary at a fine taxonomic scale such that even species within a family or genus could have quite different mechanisms. (This could be mitigated to some extent by consistent variation in the impact of ageing mechanisms on fitness components.) Predictions (2) and (3) are weaker because certain aspects of physiological functioning or maintenance may be inherently more unstable and thus more subject to perturbation by mutations. Nonetheless, a detailed look at the diversity of ageing mechanisms across the Tree of Life should provide some insight into whether MA is a plausible explanation.

Antagonistic Pleiotropy (AP) AP, proposed in 1957 by Williams, is superficially similar to MA in that it is based on how the declining force of selection with age affects selection on individual genes. However, AP proposes that the same alleles that have deleterious effects late in life also have positive effects early in life and thus that such alleles are under positive selection pressure because, all else being equal, selection is stronger at younger ages than at more advanced ages. Again, the accumulation of many such genes could result in senescence. Unlike MA, AP suggests that ageing – though itself decreasing fitness – is actively selected for because early reproduction can be increased by paying the price of accepting senescence. In other words, AP implies trade-offs between survival and reproduction. One point of contention is the extent to which AP encompasses all trade-offs. For example, Lambeth (2007) has suggested that Nox genes that produce free radicals as part of the immune system and for signalling purposes are an example of AP. These genes protect an organism against pathogens and improve signalling but over the course of time may cause damage to macro-molecules and lead to ageing. It is thus a candidate example of a trade-off: reducing the activity of these genes might slow senescence, but at a cost to early vigour and reproduction. It is also possible that reducing the activity of Nox genes would simply reduce survival at young ages so much that there would not be a trade-off: accepting the damage they imply is simply making the best of an imperfect Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:30:17, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.005

86

Taxonomic Diversity, Complexity and the Evolution of Senescence

world. However, the activity of Nox genes is not necessarily age specific, nor are their general physiological effects. It is thus a question of semantics whether such an example is included in a larger definition of AP encompassing all trade-offs or whether AP is restricted to genes that have true age-specific effects at a molecular level. There is a fair amount of support for a more broadly defined AP based on demographics and selection experiments in model organisms, as well as based on a number of candidate genes moderating trade-offs (Bochdanovits & de Jong 2004; Hughes & Reynolds 2005; Promislow 2004). Even testosterone levels in songbirds appear to moderate such a trade-off: high levels induce greater activity levels in males, which lead to more reproduction and greater exposure to predation risk (Reed et al. 2006). Note that this example may explain demographic senescence independent of physiological ageing. Thus, it is clear that there are trade-offs between survival and reproduction and that selection can act on these to optimise overall fitness. It is much less clear to what extent individual genes may have agespecific effects that hew precisely to Williams’ original vision or to what extent such genes are responsible for senescence as we know it (Moorad & Promislow 2008; Wensink 2013). As with MA, AP has not traditionally been construed to make predictions about ageing mechanisms, nor does it provide much basis to explain a diversity of senescence patterns. Nonetheless, it is possible to make some predictions. Unlike with MA, where mutations might theoretically weaken the durability of almost any aspect of physiology, it is unlikely that many genes moderate trade-offs via AP. Because gene functions are relatively conserved across taxa, AP predicts greater conservation of ageing mechanisms and more phylogenetic signal in them than MA. While AP might be reconciled with some non-senescing species, this would imply that trade-offs and AP genes are absent or sufficiently weak in these species (e.g. Wensink et al. 2014). AP does not provide a framework to predict or understand taxonomic variation in these trade-offs.

Disposable Soma Disposable soma (DS), originally proposed by Kirkwood (1977) and developed in substantial detail since then (Flatt et al. 2013; Kirkwood 1981, 1985, 2002, 2005; Kirkwood & Holliday 1979; Kirkwood & Rose 1991), is founded on the premise that ageing should not affect the germ line, the cells that produce the next generation. If it did, all lineages of germ-line cells (and thus all life) would eventually age and die. The soma – all cells other than the germ line – thus serves merely as a vehicle for propagating the germ line into the next generation. In this context, the soma will age or not as necessary to optimise fitness, balancing current reproduction with future reproduction and expected survival. Cellular maintenance is complicated, involving processes such as protection of macro-molecules from damage, repair or elimination of damaged molecules, removal of toxins, apoptosis of compromised cells and protection and elongation of telomeres while balancing cancer risk. Consequently, substantial energetic investment is necessary to maintain function and avoid macro-molecular damage that may be associated with Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:30:17, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.005

The Three Classical Theories

87

senescence. Thus, the limited energetic resources available to any species will force trade-offs between maintenance and reproduction. Such trade-offs may be physiological or at higher organisational levels: increased foraging to improve energy intake may increase predation risk, for example. Life span will evolve as a function of maintenance costs, costs of reproduction and expected survival. Both DS and AP thus predict trade-offs, and it might be argued that there is relatively little difference between DS and a broad interpretation of AP. Nonetheless, DS appears to better reflect our intuitive understanding of damage accumulation and maintenance costs, generally considered to be an important part of ageing biology. Additionally, DS provides explicit predictions about the taxonomic distribution of senescence (Kirkwood 1992, 2005). Senescence is expected when there is a distinct soma and germ line, and senescence is not expected in the germ line. Indeed, this prediction appears to agree with a number of observations about the taxonomic distribution of senescence: many non-senescing species are plants or animals with modular body plans (e.g. corals, anemones, sponges), which generally have no distinct soma (Finch 1990; Peñuelas 2005; see also Chapter 11). One of the best examples of a non-senescing species is the Hydra, a species that clearly lacks the germ line–soma distinction (Martínez 1998). At least at a broad scale, DS appears to be supported by observed taxonomic variation in senescence. It is less clear exactly how strong these predictions are: does DS predict that all species with a distinct soma age and that all that do not have a distinct soma do not? Kirkwood explicitly predicts that senescence will always evolve in iteroparous species with distinct somas and develops a simple mathematical model to show this (Kirkwood & Rose 1991). However, this model assumes a continuous trade-off between investment in maintenance and reproduction. It is certainly possible to imagine more complex dynamics, such as threshold effects, that might allow for the evolution of non-senescence. This is discussed in more detail later in this chapter. The implications of DS for ageing mechanisms depend again on how specifically we interpret the theory. The original formulation of the theory (Kirkwood 1977) does not specify which mechanisms might operate, and a broad construal might imply that almost any trade-off is consistent with DS. However, since the original proposition of the theory, most development of its mechanistic implications has focused on the energetic costs involved in cellular maintenance (Kirkwood 1981, 2005). Accordingly, oxidative stress, mitochondrial ageing and other cellular processes have been examined in substantial detail. These processes are shared across all eukaryotes and tend to reinforce the preceding prediction that senescence is inevitable in organisms with distinct somas. Moreover, these mechanisms predict that metabolic rate should be the main driver of species differences in life span because higher metabolism is likely to be the major determinant of the cost of maintenance for both mitochondrial function and oxidative damage. If these are the underlying mechanisms, the conserved eukaryotic cellular machinery may leave relatively little room for major differences in senescence patterns across taxa, excepting some higher-order trade-offs on things like predation risk during foraging. Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:30:17, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.005

88

Taxonomic Diversity, Complexity and the Evolution of Senescence

The Integrated Classical Framework Current practice has integrated the three theories into a more general framework to understand the evolution of senescence. Hamilton (1966) provided a mathematical framework demonstrating the inevitability of senescence, and current discussion refers more to trade-offs than to specific pleiotropic genes or maintenance mechanisms. This integrated framework can include some mechanisms that might not fit, strictly speaking, into any of the three theories. For example, the Nox example earlier does not necessarily imply a direct maintenance cost; there may still be indirect costs based on repair of damage to macro-molecules. Thus, Nox genes are not pleiotropic in the sense of having different effects at different ages, nor do they directly represent a trade-off in energyallocation strategies. Nonetheless, no one would doubt that the ageing mechanism represented by Nox genes fits squarely within the more general integrated framework. The goal of this chapter is more to assess the sufficiency of this integrated framework than to separately compare the three classical theories. Accordingly, it is worth asking if there are general predictions of the integrated framework beyond the specific ones listed earlier. Perhaps most important, while universal senescence is not necessarily an explicit prediction of this framework (despite Hamilton’s models), there is no basis in the framework to explain why any species with a distinct soma would not age. Second, senescence is explained as either the result of trade-offs or mutation accumulation. Thus, the classical framework would be contradicted by evidence that senescence is an inevitable consequence of physiological organisation, for example. All three classical theories suggest substantial room for a multiplicity of mechanisms; indeed, they predict that there should not be any single universal mechanism, which would contradict the idea of many genes with small effects on multiple aspects of physiological/cellular maintenance. It is worth noting that a substantial body of recent work has examined the limits of this integrated framework from various perspectives. Much of this has been based on limits of and challenges to Hamilton’s initial models, particularly from a mathematical perspective. For example, in many organisms, stage is more important than age. Changes in mortality across stages thus can give the appearance of ageing patterns while being mechanistically unrelated to age per se (Horvitz & Tuljapurkar 2008). In another line of research, allocation theory suggests that it is not just the declining force of selection with age but also the ‘option set’ of allocation choices available to each species at each moment: will it invest in growth, reproduction or maintenance, and what consequences do these choices have for the options less favoured (Baudisch & Vaupel 2012; Vaupel et al. 2004; Wensink et al. 2014)? This framework presents a promising direction for research on the diversity of ageing patterns across the Tree of Life, but two important questions remain. Firstly, why and how do option sets differ across species, that is, what are the physiological mechanisms that would explain allocation theory (Wensink et al. 2014)? Secondly, to what extent does the ability of allocation theory to explain ageing patterns require agespecific choices in allocation, and to what extent does this agree with known correlations in physiology across ages (Wensink 2013)? There does not appear to be any

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:30:17, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.005

The Emerging Evidence

89

biological clock that could be used by organisms to time changes in allocation strategies, and thus such changes would likely need to be stage or state dependent rather than age-dependent. For example, it would be erroneous to view human menopause as a timed shift in allocation from one’s own reproduction to aiding one’s descendants; menopause is rather the culmination of a gradual process of oocyte loss, and changes in the timing of menopause would almost certainly necessitate changes in the entire trajectory of fertility across the life course (Cohen 2004). This does not imply that allocation strategies are irrelevant for determining age at menopause, but it does imply that the allocation strategies at one age are constrained by those at other ages.

The Emerging Evidence Genetic Control of Ageing Over the last twenty years, genetic studies on model organisms have identified many genes that can increase life span when down-regulated (Guarente & Kenyon 2000). Yet, more surprising, many of these genes are highly evolutionarily conserved, having similar effects in yeast, nematode worms, fruit flies and mammals (e.g. Holzenberger et al. 2003). Many of these genes are involved in pathways that regulate energy usage and metabolism and are connected to changes produced by caloric restriction, the only environmental factor known to extend life span across a wide range of organisms (Weindruch & Sohal 1997). On the surface, this finding appears to contradict the classical framework in two ways. First, it suggests active selection for specific ageing mechanisms (otherwise, why not just down-regulate expression of these genes, for example?). Second, it implies a single, unified mechanism rather than the diversity of mechanisms predicted by the evolutionary theories. However, both of these apparent contradictions can be relatively easily reconciled. What is conserved is not a mechanism of ageing per se but rather control mechanisms that adjust the investment in maintenance versus reproduction within the life span of an organism depending on its conditions. When resources are scarce, it is better to invest in maintenance and wait for another opportunity to reproduce: this is the effect produced by caloric restriction. Thus, evolutionarily conserved ‘ageing’ genes are genes responsible for upstream control over many mechanisms (which may differ across taxa) involved in the trade-off between maintenance/longevity and immediate reproduction (Kirkwood 2005; Partridge & Gems 2002). Viewed in this light, the presence of these genes actually confirms rather than contradicts the classical framework in general and DS in particular. An interesting test of this would be in species that take the opposite strategy: when faced with harsh conditions, they put all their effort into a last burst of reproduction (Froy et al. 2013; Velando, Drummond & Torres 2006; see also Chapter 14 on semelparous species). One might predict that, in such species, the same genes that protect against ageing in other species would be responsible for accelerating it.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:30:17, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.005

90

Taxonomic Diversity, Complexity and the Evolution of Senescence

Taxonomic Diversity in Senescence Differences in life span and senescence patterns across species have long been recognised and integrated into the classical framework (Comfort 1979; Finch 1990; Kirkwood 1985). For example, both birds and mammals show strong correlations between intrinsic and extrinsic mortality rates, as predicted by trade-off theory (Promislow & Harvey 1990; Ricklefs 1998; Turbill & Ruf 2010). Finch (1990) produced a detailed look at the diversity of senescence patterns, identifying three main patterns: negligible senescence, gradual senescence and rapid senescence. Gradual senescence is what we colloquially think of as ageing. Rapid senescence occurs in relatively few species that reproduce once and die (semelparous). Negligible senescence appears to be most widespread among plants and other organisms without a distinct soma. Finch (1990) presents a few examples of non-senescing species with a distinct soma, but the weight of these examples is somewhat diminished for two reasons. Firstly, the bar for proving nonsenescence or negligible senescence is relatively high. Just living a long time is not sufficient, since the species may just have a slow pace of life. How far must one follow out a species to show that it truly does not senesce? 100 years? 1,000 years? 1 million years? I propose that a reasonable working definition is that a species shows no increase in mortality rates with age until well after the force of selection approaches zero. Some of Finch’s examples likely met this criterion, but it is not specifically addressed; negligible senescence appears to be identified based on the presence of a long life span and the absence of evidence of senescence. Secondly, it is hard to measure demographic processes such as survival and reproduction in long-lived species. Nonstationary populations, populations under conditions atypical for their species, and measurement challenges all present substantial barriers to a high level of confidence in any given data set (Salguero-Gómez, Shefferson et al. 2013). It is thus not surprising that among hundreds of species with some data, a few might appear not to age. For these reasons, until recently, knowledge of taxonomic diversity in senescence gave very little reason to doubt the classical framework. Recent work by Jones, Scheuerlein, Salguero-Gómez, Vaupel, and colleagues at the Max Planck Institute for Demographic Research and collaborators now permits us to say with certainty that the examples of negligible senescence shown by Finch are not erroneous exceptions but rather under-represent the diversity of species that do not age. Jones and colleagues painstakingly collected and calculated life tables for thousands of species from across Animalia and Plantae, conducting extensive quality control and standardisation to create the DatLIFE, COMPADRE and COMADRE databases (Baudisch et al. 2013; Jones et al. 2014; Salguero-Gómez et al. 2015, 2016; see chapter 20). Unlike much of the previous data, there is now good representation of many traditionally neglected taxa, including a wide diversity of plants and invertebrates. These data show that even among organisms with distinct somas, there is a wide variety of demographic patterns, including non-senescence. All mammals and probably all birds appear to senesce, but not necessarily with the stereotyped demography predicted by Hamilton (1966). Among other vertebrates and distinct-soma invertebrates, some species clearly age, but others clearly do not. Conversely, lack of senescence is far from universal among animals lacking

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:30:17, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.005

The Emerging Evidence

91

a distinct soma. Many of these species, most notably plants, do show clear senescence and a variety of senescence patterns. In some cases, fairly close relatives of a non-senescing species show clear senescence. For simplicity, I do not consider here species with an intermediate level of soma/germ-line distinction, such as unicellular organisms with asymmetric cell division. See further Kirkwood (2005). One caveat to these comparative data is that they present a convincing portrait of demographic senescence (or lack thereof) in the wild but cannot provide a definitive proof of the absence of physiological ageing. For example, demographic negative senescence may be due to changing predation rates with size; this does not preclude a weaker underlying effect of physiological ageing. Conversely, tooth wear due to food type may explain demographic senescence in laboratory rodents without implying that the demography has a broad physiological basis (Simons et al. 2013). For this reason, it will be critical to assess whether there is significant survival without increasing mortality well past when the force of selection approaches zero, as noted earlier. Taken together, these findings can be summarised as follows: (1) there is a surprising diversity of demographic patterns across the Tree of Life; (2) many species from a wide array of taxa do not show any evolutionarily meaningful senescence, including many species with distinct somas; (3) there is a strong taxonomic signal in senescence patterns within some clades, whereas other clades show substantial diversity; and (4) there appears to be support for less senescence in species lacking a distinct soma, though formal tests have not yet been conducted. These data, which for the moment must be considered the definitive portrait of diversity in senescence, present a substantial challenge for the classical framework. Both Hamilton’s models and the DS explicitly predict that senescence should always evolve, at least in species with a distinct soma (Hamilton 1966; Kirkwood & Rose 1991). Even MA and AP, which do not make such an explicit prediction, include no framework to explain why some species might age and others might not, and MA seems to imply that all species would be subject to senescence. This does not necessarily imply that the classical framework is wrong, but it suggests that either (1) the classical framework is insufficient, (2) there is an error of some kind in the demographic analyses or (3) there is some way to reconcile these apparent contradictions that is yet to be found (but see later in this chapter on resource allocation). One potential way to reconcile them would be to suppose a non-continuous model of how trade-offs work. For example, rather than supposing that a unit of maintenance might be achieved for every unit of energy invested, it is possible that once a certain threshold of energy is invested, there is a jump to a much higher state of maintenance such that the organism could live much longer than predicted by a continuous model. Taxonomic diversity in senescence might then be explained by which taxa find their optima close to or far from such a threshold (Figure 5.1). Such a model is not particularly far-fetched: there is no particular general reason to assume a continuous function relating maintenance benefit to energy invested. Nonetheless, it does contradict the particular model of physiological maintenance generally associated with DS. Cellular maintenance is generally assumed to be largely a question of damage prevention and repair; in such a situation, it is normal to suppose Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:30:17, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.005

92

Taxonomic Diversity, Complexity and the Evolution of Senescence

Senescence rate

Linear Exponential Threshold

No senescence

Investment in maintenance Figure 5.1

Various models for how investment in physiological maintenance might affect senescence rate. In a linear model, each unit of investment in maintenance produces an identical benefit for senescence rate, regardless of the level of investment. This is not particularly realistic biologically because it should become increasingly harder to achieve good maintenance at higher levels. The linear model also implies that sufficient maintenance could be extrapolated to cause negative senescence at the extreme. An exponential model of diminishing returns is much more intuitive and realistic and is in agreement with DS. Such a model predicts that an infinite investment would merely succeed in eliminating senescence, not in making it negative. Lastly, a threshold model predicts that there is some return on investment up to a certain point (shown here as linear, but it could be any form) but that beyond a certain threshold the physiological state changes abruptly, and senescence becomes negligible or much lower. Obviously, many other examples of noncontinuous dynamics might be imagined; while the physiological basis of such models is not necessarily clear, the possibility cannot be excluded, particularly if there are complex system dynamics. Such a threshold model may be necessary to explain non-senescence in organisms with distinct somas.

a model of exponentially increasing costs, similar to diminishing marginal returns in economics. This is so because the ‘low-hanging fruit’ of maintenance is used first, and the more optimised the maintenance, the more effort/energy is required per unit reduction in senescence rate. An intuitive example is sweeping a very dirty floor: a lot of dirt can be swept up with little effort at the beginning, but the cleaner the floor becomes, the more energy is required to get an increasingly small amount of dirt. Gems’ ‘green’ theory of ageing lays this out explicitly (Gems & McElwee 2005). Such a model of exponentially increasing costs strongly predicts that senescence would always evolve: at higher ages, the exponentially increasing costs bump up against the fast-declining force of selection. Thus, the taxonomic data presented earlier suggest that at the least such exponentially increasing costs are not universal in determining senescence. One promising way to reconcile taxonomic diversity with classical senescence theories is based on continuing resource-allocation choices by an organism, as noted earlier Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:30:17, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.005

The Emerging Evidence

93

(Baudisch & Vaupel 2012; Wensink et al. 2014). Rather than assume that organisms face a single, lifelong choice of how to allocate energy, nutrients and other resources, substantial variety in senescence patterns can be explained by understanding how organisms make continual choices to allocate such resources to different tasks such as growth, reproduction and maintenance. Indeed, this has long been the prevailing paradigm for understanding how plants use nutrients, and it nicely explains the data on caloric restriction in animals and yeast. There can be little doubt that such principles of resource allocation play a big role in explaining taxonomic diversity in senescence. However, this begs another question: why do resource-allocation choices differ across taxa, thereby creating a taxonomic framework for how senescence may evolve? For example, no mammal appears to be able to make a resource-allocation choice that will prolong its life span indefinitely while waiting for better conditions, but many other species can. The apparently indefinite life span of some turtles (see Warner et al. 2016) also does not appear to be conditional on resource-allocation choices within an individual. Such taxonomic differences may well be due to species-specific allocation choices, but more information is needed on what constrains these choices.

Complex System Dynamics in Ageing Mechanistically speaking, ageing is one of the least understood aspects of biology. A 1990 review identified more than 300 mechanistic theories of ageing (Medvedev 1990). In some sense, this number is an exaggeration: many of these theories are not mutually exclusive, and many have very little support as key ageing mechanisms. Nonetheless, this diversity of plausible mechanisms has prompted many researchers to suggest that there are many underlying causes of senescence (Kirkwood 2005). This would explain taxonomic diversity in mechanisms: for example, inflammation may be an important mechanism in mammals but probably not in birds and certainly not in invertebrates or plants, which do not have inflammatory systems. While most ageing researchers would probably agree that there is no single mechanism, there is not much consensus on how to understand a multifactorial theory of ageing. Over the last decade, there is an increasing focus on complex systems theories of ageing, particularly among geriatricians and other experts in human ageing (Ferrucci 2005; Fried et al. 2005; Lipsitz 2004; Seplaki et al. 2006; Taffett 2003). Similar theories have been given names such as ‘allostatic load’ (McEwen & Wingfield 2003), ‘homeostenosis’ (Taffett 2003) and (multisystem) ‘physiological dysregulation’ (Fried et al. 2009; Seplaki et al. 2006). However, such theories run a substantial risk of misinterpretation due to imprecise terminology and a lack of understanding of complex systems theory among most researchers. The root of such theories is that homeostasis is maintained and adjusted by sophisticated biological networks that have been shaped by evolution (Cohen et al. 2012). Such networks are composed of large numbers of molecules that regulate each other (‘physiological regulatory networks’ (PRNs)). In order to maintain homeostasis, they rely on specific network structures that cause feedback effects (usually negative feedback). In other words, such networks are, in general, specifically structured by selection so as not to have the simple linear properties Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:30:17, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.005

94

Taxonomic Diversity, Complexity and the Evolution of Senescence

often supposed by reductionist approaches to assessing which molecule regulates which other. A may up-regulate B, but this does not mean that increasing A in the short term will cause B to increase in the long term. The dynamics of such regulatory networks are generally non-linear, complex and largely unknown, particularly at longer timescales. Biological networks are considered a prime example of a ‘complex adaptive system’ (Holland 1992). A complex systems understanding of ageing suggests that such networks and the multiple ageing mechanisms in play are not merely complicated (i.e. having many parts) but also complex, having dynamics that cannot be predicted based on understanding the parts individually (Kier & Witten 2005). The field of systems biology focuses on these kinds of complex systems and is concerned with emergent biological properties at the intra-organismal level (Kitano 2002). Its application in the context of complex adaptive systems has been proposed for ageing (West & Bergman 2009). Mathematically, the major approach to understanding networks has been through graph theory (West 2001), but this bottom-up approach demands a full knowledge of network structure. Top-down approaches can also be used to understand some aspects of system structure and function even in the absence of the complete knowledge of network structure. How exactly might ageing relate to the dynamics of complex systems? The most common explanation is as a result of dysregulation (Cohen et al. 2013): the ability of biological networks to maintain homeostasis is imperfect, and over time small deviations from perfect homeostasis will, through feedback effects, cause biological networks to function less and less efficiently. The best known example of this is the vertebrate corticosteroid stress response (McEwen & Wingfield 2003; Sapolsky et al. 2002). Ideally, stress causes an augmentation in the levels of corticosteroid hormone, which triggers a cascade of effects appropriate to a species’ needs when stressed. Over time, negative-feedback mechanisms bring levels of the hormone down to baseline. However, when organisms are chronically stressed, levels of the hormone can fail to return to baseline even after the stress is no longer present, and this can have numerous long-term adverse health effects. Other than this example, empirical support for a complex systems theory of ageing has been sparse. This is probably because of a lack of appropriate methods to test the theory: our knowledge of the structure of even the simplest biological networks is generally incomplete. New human hormones continue to be discovered (Ganz 2003). Even when the structure is known, we have very little understanding of long-term dynamics. Nonetheless, a few recent studies are starting to provide clear support for the importance of complex system dynamics in ageing. Lipsitz (2004) has shown that complexity is a sign of health in patterns of heart rates at multiple timescales. Cohen et al. (2013) showed that the statistical distance of a biomarker profile is a robust measure of dysregulation that increases with age and predicts multiple health outcomes. Importantly, a similar signal of dysregulation can be detected using a wide variety of biomarker combinations, suggesting that the signal is diffuse throughout the regulatory networks and not an artefact of one or two molecules (Cohen et al. 2014). The same measure has been shown to work in humans and a migratory shorebird (Milot et al. Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:30:17, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.005

The Emerging Evidence

95

2014). Additionally, analysis of human biomarkers has identified an ageing-related, highly stable emergent process implicating anaemia, inflammation, calcium, and protein transport (Cohen et al. 2015). Such examples clearly demonstrate the presence of complex dynamics during ageing but cannot definitely prove whether the dysregulation itself is the primary mechanism of ageing or is a result of other, more basic ageing mechanisms. It is increasingly common to see articles on ageing referring to the systems biology of ageing (Kirkwood 2008, 2011; Kirkwood et al. 2003), network theories of ageing (Kowald & Kirkwood 1996; Soltow et al. 2010) and bioinformatics (de Magalhães & Toussaint 2004). However, such articles rarely take a complex systems approach in the sense I referred to earlier. For example, the first article (to my knowledge) on a ‘network theory of ageing’ describes how interactions among multiple ageing mechanisms could affect our understanding of ageing (Kowald & Kirkwood 1996). However, the model used is largely deterministic and does not incorporate complex dynamics or system-level properties. Likewise, many recent articles on the systems biology of ageing refer largely to the collection and analysis of large numbers of molecules through ‘-omics’ and bioinformatics approaches but do not consider emergent properties or dynamics (de Magalhães & Toussaint 2004; de Magalhães et al. 2009; Kirkwood 2011). Substantial effort has been put into systems biology approaches that incorporate interactions among mechanisms and across organisational levels in order to make predictions about the consequences of interactions within regulatory networks (Kirkwood 2008; Kirkwood et al. 2003). Some recent articles on network aspects of ageing do use network theory but do not explore how dysregulation or other emergent properties might cause ageing beyond the contributions of the individual molecules (Csermely & Sőti 2006; Hoffman et al. 2014; Xue et al. 2007). My goal is not to disparage these studies, all of which make important contributions; it is merely to point out that use of the terms ‘network’, ‘systems biology’ and ‘complexity’ can mean different things in different contexts, and much of the work that falls under these banners does not address the structural stability of the organism-level regulatory networks that I referred to earlier. This distinction is important because Kirkwood, who authored the disposable soma theory, has an extensive research programme in systems biology. Kirkwood is a proponent of multifactorial theories of ageing but in a sense that suggests that ageing is complicated rather than complex: that we can add up the effects of multiple molecular mechanisms, taking into account that they may feedback onto each other and, in theory, fully understand ageing (Kirkwood et al. 2003). Indeed, if there were formal complexity in ageing, this could present a substantial challenge to the sufficiency of the DS theory and the rest of the classical framework. A complex systems (CS) approach to ageing implies that senescence could result solely from the imperfect robustness of biological networks to the insults that life will subject them to. In theory, though not in practice, there might be no need to refer to maintenance or cellular processes: ageing might result largely from the organismal networks composed primarily of hormones, the immune system, lipids and so forth, as well as their interactions with the central nervous system (Cohen et al. 2012; Cohen 2016). Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:30:17, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.005

96

Taxonomic Diversity, Complexity and the Evolution of Senescence

This is not to suggest that DS or the classical framework more generally is necessarily in conflict with a CS approach. Both may operate simultaneously and may interact with each other. For example, poor cellular maintenance could accelerate dysregulation, and dysregulation could cause problems in cellular maintenance. Either alone might be sufficient to cause senescence. But CS cannot necessarily be subsumed within DS and the classical framework: under some circumstances, it may represent an independent ageing mechanism and may imply different evolutionary consequences. For example, the types of evolutionary constraints implied by CS and DS may differ substantially and might produce different patterns of life span distributions across species. At this point, although evidence for CS dynamics in ageing is relatively strong, the evidence does not distinguish whether or not these dynamics are a downstream effect of DS or other mechanisms. Unfortunately, it is relatively hard to identify tests that could clarify this, and we may not have a definitive answer in the near future. What may be possible is to continue to gather evidence separately for both CS and DS, with a view towards identifying aspects of ageing not explained by them, singly or jointly. It is possible that even the combination of CS and DS is not a full explanation of ageing.

Conclusions The review of the evidence presented here is hardly comprehensive, but it is sufficient to draw several clear conclusions. Firstly, the classical framework generally and the DS theory in particular have a substantial amount of evidence in their favour, including a disproportionate number of non-senescing species among taxa without distinct somas, correlations between intrinsic and extrinsic mortality within taxa such as birds and mammals, agreement with data on caloric restriction and genetic modulation of ageing rate and a number of clear mechanisms that appear to support an investment in maintenance and cellular function as part of a reproduction–survival trade-off. There is little doubt that any comprehensive understanding of senescence will include the classical framework to explain trade-offs and certain ageing mechanisms. Secondly, recent evidence presents at least three major challenges for the comprehensiveness of the DS theory in particular. (1) Many species with distinct somas show clear evidence of non-senescence, and the cellular and molecular mechanisms generally associated with DS are hard to reconcile with the sorts of non-linear trade-off functions that would be necessary to explain this in the context of maintenance costs. (2) Many species without distinct somas show clear evidence of senescence; the mechanism underlying this senescence is unlikely to be related to the sorts of maintenance costs associated with DS. (3) There is substantial theoretical reason to believe that complex systems dynamics are important in ageing, and recently there is at least some empirical evidence to support this as well. Such dynamics, if real, may not contradict DS but do present a complementary approach to understanding senescence, one that cannot simply be viewed as part of DS. Thirdly, the broader classical framework presents no real way to understand the broad taxonomic diversity in senescence demography recently observed. Unless there is Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:30:17, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.005

References

97

a major systematic problem with these demographic analyses – and it is hard to imagine what that might be – some framework beyond the classical one will be needed to explain why some taxa always age, some taxa sometimes age and some taxa never age. Fourthly (covered in more detail elsewhere), Hamilton’s formulation on the inevitability of senescence has proven incomplete, and an increasing focus on diverse taxa, particularly those with mortality rates more determined by stage than age (Baudisch 2008; Horvitz & Tuljapurkar 2008; Vaupel et al. 2004), suggests that a new demographic paradigm may be emerging to explain the full diversity of senescence patterns outside the classical framework (see Chapters 1 and 20). Overall, it thus appears that a new synthesis of senescence evolution is emerging, one that includes and does not contradict the classical framework but goes farther and explains many additional aspects of senescence. This is similar to the incorporation of Darwin into the neo-Darwinian synthesis and, more recently, the incorporation of the neo-Darwinian synthesis into the extended synthesis (Pigliucci & Muller 2010). The demographic side of the emerging senescence synthesis is best developed, but there is a substantial need to expand the mechanistic side and to explain why we observe certain patterns of variation across the Tree of Life. It is increasingly clear that a comprehensive understanding of the evolution of senescence must integrate a mechanistic understanding as well and thus that the final synthesis will include evolutionary biology, demography and physiology/cellular biology. Again, this is parallel to the need for the neo-Darwinians to include genetic mechanisms and for the extended synthesists to include developmental constraints, aspects of biology that had been unnecessary for understanding evolution before their respective syntheses. It will be interesting to see the understanding that emerges over the next several years.

Acknowledgements I thank A. Baudisch, M. Simons, T. B. L. Kirkwood, O. R. Jones and R. Salguero-Gómez for comments on the manuscript. AAC is a member of the FRQ-S-supported Centre de recherche sur le vieillissement and Centre de recherche cinique du CHUS and is a funded Research Scholar of the FRQ-S. This research was supported by NSERC Discovery Grant 402079–2011.

References Baudisch, A. (2008). Inevitable Aging? Contributions to Evolutionary-Demographic Theory (Berlin: Springer). Baudisch, A., Salguero-Gómez, R., Jones, O. R., et al. (2013). The pace and shape of senescence in angiosperms. Journal of Ecology, 101(3), 596–606. Baudisch, A. & Vaupel, J. W. (2012). Getting to the root of aging. Science, 338(6107), 618–19.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:30:17, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.005

98

Taxonomic Diversity, Complexity and the Evolution of Senescence

Bochdanovits, Z. & de Jong, G. (2004). Antagonistic pleiotropy for life-history traits at the gene expression level. Proceedings of the Royal Society Series B: Biological Sciences, 271(Supp. 3), S75–8. Cohen, A. A. (2004). Female post-reproductive life span: a general mammalian trait. Biological Reviews, 79(4), 733–50. Cohen, A. A., Martin, L. B., Wingfield, J. C., et al. (2012). Physiological regulatory networks: ecological roles and evolutionary constraints. Trends in Ecology and Evolution, 27(8), 428–35. Cohen, A. A., Milot, E., Yong, J., et al. (2013). A novel statistical approach shows evidence for multi-system physiological dysregulation during aging. Mechanisms of Ageing and Development, 134(3–4), 110–17. Cohen, A. A., Milot, E., Li, Q., et al. (2014). Cross-population validation of statistical distance as a measure of physiological dysregulation during aging. Experimental Gerontology, 57, 203–210. Cohen, A. A., Poirier, R., Dusseault-Bélanger, F., et al. (2015). Detection of a novel, integrative aging process suggests complex physiological integration. PloS one, 10(3), p.e0116489. Cohen, A. A. (2016). Complex systems dynamics in aging: new evidence, continuing questions. Biogerontology, 17(1), 205–220. Comfort, A. (1979). The Biology of Senescence (Edinburgh: Churchill Livingstone). Csermely, P. & Sőti, C. (2006). Cellular networks and the aging process. Archives of Physiology and Biochemistry, 112(2), 60–4. de Magalhães, J. P., Curado, J. & Church G. M. (2009). Meta-analysis of age-related gene expression profiles identifies common signatures of aging. Bioinformatics, 25 (7), 875–81. de Magalhães, J. P. & Toussaint O. (2004). How bioinformatics can help reverse engineer human aging. Ageing Research Reviews, 3(2), 125–41. Ferrucci, L. (2005). An exciting thought. Journals of Gerontology Series A: Biological Sciences and Medical Sciences, 60(1), 56. Finch, C. E. (1990). Longevity, Senescence, and the Genome (University of Chicago Press). Flatt, T., Amdam, G. V., Kirkwood, T. B. L. & Omholt S. W. (2013). Life-history evolution and the polyphenic regulation of somatic maintenance and survival. Quarterly Review of Biology, 88(3), 185–218. Fried, L. P., Hadley, E. C., Walston, J. D., et al. (2005). From bedside to bench: research agenda for frailty. Science of Aging Knowledge Environment, 2005(31): 24. Fried, L. P., Xue, Q.-L., Cappola, A. R., et al. (2009). Nonlinear multisystem physiological dysregulation associated with frailty in older women: implications for etiology and treatment. Journals of Gerontology Series A: Biological Sciences and Medical Sciences, 64(10), 1049–57. Froy, H., Phillips, R. A., Wood, A. G., et al. (2013). Age-related variation in reproductive traits in the wandering albatross: evidence for terminal improvement following senescence. Ecology Letters, 16(5), 642–9. Ganz, T. (2003). Hepcidin, a key regulator of iron metabolism and mediator of anemia of inflammation. Blood, 102(3), 783–8. Gems, D. & McElwee, J. J. (2005). Broad spectrum detoxification: the major longevity assurance process regulated by insulin/IGF-1 signaling? Mechanisms of Ageing and Development, 126(3), 381–7.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:30:17, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.005

References

99

Guarente, L. & Kenyon, C. (2000). Genetic pathways that regulate ageing in model organisms. Nature, 408(6809), 255–62. Hamilton, W. D. (1966). The moulding of senescence by natural selection. Journal of Theoretical Biology, 12, 12–45. Hoffman, J. M., Soltow, Q. A. Li, S., et al. (2014). Effects of age, sex, and genotype on high-sensitivity metabolomic profiles in the fruit fly, Drosophila melanogaster. Aging Cell. Holland, J. H. (1992). Complex adaptive systems. Daedalus, 17–30. Holzenberger, M., Dupont, J., Ducos, B., et al. (2003). IGF-1 receptor regulates life span and resistance to oxidative stress in mice. Nature, 421, 182–7. Horvitz, C. C. & Tuljapurkar, S. (2008). Stage dynamics, period survival, and mortality plateaus. American Naturalist, 172(2), 203–15. Hughes, K. A., Alipaz, J. A., Drnevich, J. M. & Reynolds, R. M. (2002). A test of evolutionary theories of aging. Proceedings of the National Academy of Sciences of the United States of America, 99(22), 14286–91. Hughes, K. A. & Reynolds, R. M. (2005). Evolutionary and mechanistic theories of aging. Annual Review of Entomology, 50(1), 421–45. Jones, O. R., Scheuerlein, A., Salguero-Gómez, R., et al. (2014). Diversity of ageing across the tree of life. Nature, 505(7482): 169–73. Kier, L. & Witten, T. (2005). Cellular automata models of complex biochemical systems. In Complexity in Chemistry, Biology, and Ecology, ed. D. Bonchev & D. Rouvray (pp. 237–301) (New York: Springer). Kirkwood, T. B. L. (1977). Evolution of ageing. Nature, 270, 301–4. Kirkwood, T. B. L. (1981). Repair and its evolution: survival versus reproduction. In Physiological Ecology: An Evolutionary Approach to Resource Use, ed. C. R. Townsend & P. Calow (pp. 165–89) (Oxford, Blackwell Scientific). Kirkwood, T. B. L. (1985). Comparative and evolutionary aspects of longevity. In Handbook of the Biology of Aging, ed. C. E. Finch & E. L. Schneider (pp. 27–44) (New York, Van Nostrand Rheinhold). Kirkwood, T. B. L. (1992). Comparative life spans of species: why do species have the life spans they do? American Journal of Clinical Nutrition, 55, 1191S–5S. Kirkwood, T. B. L. (2002). Evolution of ageing. Mechanisms of Ageing and Development, 123(7), 737–45. Kirkwood, T. B. L. (2005). Understanding the Odd Science of Aging. Cell, 120(4), 437–47. Kirkwood, T. B. L. (2008). A systematic look at an old problem. Nature, 451, 644–7. Kirkwood, T. B. L. (2011). Systems biology of ageing and longevity. Philosophical Transactions of the Royal Society B: Biological Sciences, 366(1561), 64–70. Kirkwood, T. B. L., Boys, R. J., Gillespie, C. S., et al. (2003). Towards an e-biology of ageing: integrating theory and data. Nature Reviews Molecular Cell Biology, 4(3), 243–9. Kirkwood, T. B. L. & Holliday, R. (1979). The evolution of ageing and longevity. Proceedings of the Royal Society of London, B205, 531–46. Kirkwood, T. B. L. & Rose, M. R. (1991). Evolution of senescence: late survival sacrificed for reproduction. Philosophical Transactions of the Royal Society of London Series B, 332, 15–24. Kitano, H. (2002). Systems biology: a brief overview. Science, 295(5560), 1662–4.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:30:17, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.005

100

Taxonomic Diversity, Complexity and the Evolution of Senescence

Kowald, A. & Kirkwood, T. B. L. (1996). A network theory of ageing: the interactions of defective mitochondria, aberrant proteins, free radicals and scavengers in the ageing process. Mutation Research/DNAging, 316(5–6), 209–36. Lambeth, J. D. (2007). Nox enzymes, ROS, and chronic disease: an example of antagonistic pleiotropy. Free Radical Biology and Medicine, 43(3), 332–47. Lipsitz, L. A. (2004). Physiological complexity, aging, and the path to frailty. Science of Aging Knowledge Environment, 2004(16), 16. Martı́nez, D. E. (1998). Mortality patterns suggest lack of senescence in Hydra. Experimental Gerontology, 33(3), 217–25. McEwen, B. S. & Wingfield, J. C. (2003). The concept of allostasis in biology and biomedicine. Hormones and Behavior, 43(1), 2–15. Medawar, P. B. (1952). An Unsolved Problem of Biology (London, Lewis). Medvedev, Z. A. (1990). An attempt at a rational classification of theories of ageing. Biological Reviews, 65(3), 375–98. Milot, E., Cohen, A. A., Vézina, F., et al. (2014). A novel integrative method for measuring body condition in ecological studies based on physiological dysregulation. Methods in Ecology and Evolution, 5(2), 146–55. Moorad, J. A. & Promislow, D. E. (2009). What can genetic variation tell us about the evolution of senescence? Proceedings of the Royal Society of London Series B: Biological Sciences, 276(1665), 2271–8. Moorad, J. A. & Promislow, D. E. L. (2008). A theory of age-dependent mutation and senescence. Genetics, 179(4), 2061–73. Partridge, L. & Gems, D. (2002). Mechanisms of aging: public or private? Nature Reviews Genetics, 3(3), 165–75. Peñuelas, J. (2005). Plant physiology: a big issue for trees. Nature, 437(7061), 965–6. Pigliucci, M. & Muller, G. (2010). Evolution – the extended synthesis. Promislow, D. E. L. (2004). Protein networks, pleiotropy and the evolution of senescence. Proceedings of the Royal Society of London Series B: Biological Sciences, 271(1545), 1225–34. Promislow, D. E. L. & Harvey, P. H. (1990). Living fast and dying young: a comparative analysis of life-history variation among mammals. Journal of Zoology, 220(3), 417–37. Reed, W. L., Clark, M. E., Parker, P. G., et al. (2006). Physiological effects on demography: a long-term experimental study of testosterones effects on fitness. American Naturalist, 167(5), 667–83. Ricklefs, R. E. (1998). Evolutionary theories of aging: confirmation of a fundamental prediction, with implications for the genetic basis and evolution of life span. American Naturalist, 152(1), 24–44. Salguero-Gómez, R., Jones, O. R., Archer, C. R., et al. (2015). The COMPADRE Plant Matrix Database: an open online repository for plant demography. Journal of Ecology 103(1), 202–218. Salguero-Gómez, R., Jones, O. R., Archer, C. A., et al. 2016. COMADRE: a global database of animal demography. Journal of Animal Ecology 85, 371–384. Salguero-Gómez, R., Shefferson, R. P. & Hutchings, M. J. (2013). Plants do not count . . . or do they? New perspectives on the universality of senescence. Journal of Ecology, 101(3), 545–54.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:30:17, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.005

References

101

Sapolsky, R. M., Krey, L. C. & McEwen, B. S. (2002). The neuroendocrinology of stress and aging: the glucocorticoid cascade hypothesis. Science of Aging Knowledge Environment, 2002(38), 21. Seplaki, C. L., Goldman, N., Weinstein, M. & Lin, Y.-H. (2006). Measurement of cumulative physiological dysregulation in an older population. Demography, 43(1), 165–83. Simons, M. J., Koch, W. & Verhulst, S. (2013). Dietary restriction of rodents decreases aging rate without affecting initial mortality rate – a meta-analysis. Aging Cell, 12(3), 410–14. Soltow, Q. A., Jones, D. P. & Promislow, D. E. L. (2010). A network perspective on metabolism and aging. Integrative and Comparative Biology, 50(5), 844–54. Taffett, G. E. (2003). Physiology of aging. In Geriatric Medicine, ed. C. K. Cassel, R. M. Leipzig, H. J. Cohen et al. (pp. 27–35) (New York: Springer). Turbill, C. & Ruf, T. (2010). Senescence is more important in the natural lives of long-than short-lived mammals. PLoS ONE, 5(8), e12019. Vaupel, J. W., Baudisch, A., Dölling, M., et al. (2004). The case for negative senescence. Theoretical Population Biology, 65(4), 339–51. Velando, A., Drummond, H. & Torres, R. (2006). Senescent birds redouble reproductive effort when ill: confirmation of the terminal investment hypothesis. Proceedings of the Royal Society of London Series B: Biological Sciences, 273(1593), 1443–8. Warner, D. A., Miller, D. A., Bronikowski, A. M., & Janzen, F. J. (2016). Decades of field data reveal that turtles senesce in the wild. Proceedings of the National Academy of Sciences, 201600035. Weindruch, R. & Sohal, R. S. (1997). Caloric intake and aging. New England Journal of Medicine, 337(14), 986–94. Wensink, M. (2013). Age-specificity and the evolution of senescence: a discussion. Biogerontology, 14(1), 99–105. Wensink, M. J., Wrycza, T. F. & Baudisch, A. (2014). No senescence despite declining selection pressure: Hamilton’s result in broader perspective. Journal of Theoretical Biology, 347, 176–81. West, D. B. (2001). Introduction to Graph Theory (Upper Saddle River, NJ: PrenticeHall). West, G. B. & Bergman, A. (2009). Toward a systems biology framework for understanding aging and health span. Journals of Gerontology Series A: Biological Sciences and Medical Sciences, 64A(2), 205–8. Williams, G. C. (1957). Pleiotropy, natural selection and the evolution of senescence. Evolution, 11, 398–411. Xue, H., Xian, B., Dong, D., et al. (2007). A modular network model of aging. Molecular Systems Biology, 3(1).

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:30:17, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.005

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:30:17, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.005

6

Evolutionary Demography of the Human Mortality Profile Oskar Burger

Short Summary Classic evolutionary theories of ageing posit relationships between either mortality or selection pressure and the rate of ageing. In many cases, empirical tests of these relationships among populations of a single species or a group of related species have failed to support them. Such failures are part of a rising tide of tension between theory and data across evolutionary studies of ageing and life span. Here human populations that differ greatly in mortality and economy are analysed as an illustrative case for some current sources of this tension. Most of the historical progress in lowering mortality is a direct (lowering infectious disease) or indirect (lowering extrinsic mortality with infrastructure) result of economic development put in motion by the industrial revolution, which is here considered to be ecological change that alleviated density dependence and changed the condition dependence of human mortality rates. Despite orders of magnitude of change in environmental quality and mortality, the measures of the force of selection that were designed by classical theory change very little, essentially not making a prediction about the direction of change for the rate of ageing that should occur in response to the large reduction in mortality. The reasons behind this predictive ambiguity are interpreted as further evidence that the evolutionary theory of ageing needs some adjustment. This chapter argues that density- and condition-dependent factors are especially important for understanding changes in life span and ageing of humans of the last two to three centuries.

Introduction Life spans have nearly doubled in the last 200 hundred years in many human populations. In fact, the distributions of country-level life expectancies in the year 2000 are almost non-overlapping with those of 1800 (Figure 6.1). The best-surviving country in 1800 had poorer survival than the worst-surviving country today, with just one exception, the impoverished nation of Sierra Leone. This is a big shift in just 200 years (Finch 2012). If the life-expectancy distributions in Figure 6.1 were for a hypothetical physical trait of a different organism, say, beak length on a Galapagos finch, we’d likely conclude that we were studying different species or that a speciation event had occurred. However, this rapid improvement is driven almost entirely by environmental change and Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:31:53, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.006

Evolutionary Demography of the Human Mortality Profile

100

106

Number of countries 20 40 60 80

1800

0

2000

20 Figure 6.1

30

50 60 70 40 Life expectancy, years

80

90

Life expectancy at birth for countries in 1800 (light bars, left) and 2000 (dark bars, right). The distributions are almost totally non-overlapping.

technological development at the near exclusion of genetic causes (Finch 2012). The remarkable feature of both life span and its reciprocal, adult mortality, is the variation we observe across populations today and within populations during the last 100 years (Burger et al. 2012). The speed and magnitude of mortality reduction over the time interval captured by Figure 6.1 pose interesting problems for evolutionary theories of ageing. For instance, one of the most widely applied predictions from the classical theories of ageing is Williams’ hypothesis that reduced extrinsic mortality should lead to slower ageing because lower mortality selects for slower growth and higher-quality tissues and cells (Williams 1957). This prediction has met with mixed results in several organisms, including water fleas (Daphnia ambigua) (Walsh et al. 2014), guppies (Poecilia reticulata) (Reznick et al. 2004) and humans (Gurven & Fenelon 2009; Hawkes 2010). Contrary to the ‘prediction’, it often seems that removing major sources of mortality does not change the rate of ageing (Finch et al. 1990; Gurven & Fenelon 2009; Maklakov et al. 2015). Yet, there is little consensus on what failures of this classic prediction mean for how ageing and mortality patterns are measured and interpreted, nor has there been a concerted effort to modify the existing theory to accommodate such problematic empirical findings (Williams et al. 2006), although some important steps have been made (Chen & Maklakov 2012; Tuljapurkar et al. 2007). For instance, does the prediction fail for humans simply because of a distinctive ecology or due to the short time scale involved, or the lack of a wider taxonomic comparison (e.g. Ricklefs 2010)? The lack of clear support for Williams’ hypothesis suggests a need to investigate the links between environmental quality, extrinsic mortality and selection against ageing more closely. Humans are an ideal focal organism for this because they have experienced dramatic mortality decline, and there is a wealth of information available for populations that differ greatly in background mortality levels and ecological or economic setting.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:31:53, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.006

Introduction

107

Numerous social and economic changes accompany the recent and rapid increase in human longevity, making the attribution of causality rather challenging. From an evolutionary perspective, researchers would like to isolate a relationship between ecology and energy allocation that affects maintenance- and growth-related trade-offs underlying the rate of somatic decay, but in addition to the many complicated covariates of the mortality-related ‘revolutions’ (epidemiological, health, demographic, industrial and so on) that have occurred in the last couple of centuries, researchers must also contend with the non-equilibrium state of human demographics as populations change in composition, mortality and fertility as part of a widely studied and debated process called the ‘demographic transition’ (Borgerhoff Mulder 1998; Kirk 1996). Because of such difficulties, it seems fair to say that many would consider industrial human populations as a totally inappropriate test case for biological or evolutionary theories of ageing. Yet, concomitantly, it is fair to say that the majority of research done on ageing is at least distantly, if not directly, related to the goal of understanding ageing patterns in these highly unusual populations (‘these populations’ here is shorthand for ‘ourselves’ and must surely include anyone who might pick up this book, for example). The classic theories of ageing have faced growing empirical challenges whether applied to humans, guppies or flatworms, but interestingly the resolution to these challenges may be broadly similar across these and other species of interest to evolutionary theorists who study ageing, mortality or life span. In addition to some problems supporting Williams’ hypothesis, other apparent shortcomings of the classic theories have been recently pointed out in the literature. Jones et al. (2014) demonstrate that mortality rates for some species may be constant or even improve with age, and yet the evolutionary theory of ageing claims that senescence should be ‘inevitable’ (Hamilton 1966). Because of pressures from natural selection, long periods of survival past the end of reproduction are predicted to be exceedingly rare and yet are found in numerous species (Levitis et al. 2013). Antagonistic pleiotropy, or alleles with a positive effect early in life but a negative effect late in life, are classically predicted to drive ageing, yet genes with these effects seem to be rare, and there is increasing evidence for the reverse, or alleles that improve fitness early in life and late in life (Maklakov et al. 2015). In retrospect, it seems intuitive that many alleles should have effects of positive pleiotropy because organisms that invest more in growth often live longer, and extremely fit individuals are often fit at early and late ages. Such alleles are inconsistent with the classic theory, and evidence for them was initially dismissed as error (Rose 1991). After conducting an empirical test of Williams’ hypothesis that failed to support it, Walsh et al. (2014) called ‘for a greater consideration of models that more explicitly consider the ecology of focal organisms’. Further, they argued that evolutionary studies of senescence still ‘lack a general understanding of the relationship between ecologically driven mortality and senescence’. The study organism for Walsh et al. (2014) was Daphnia, a species of small aquatic crustacean, but it is argued here that this call to increase emphasis on the link between ecology and senescence is at least as valid for the humans as for any other organism. This chapter views human mortality changes in light of these and other empirical challenges to the canonical evolutionary theories of ageing. On the one hand, insight Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:31:53, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.006

108

Evolutionary Demography of the Human Mortality Profile

gained from the study of non-human species can inform understanding of human life span and ageing. On the other, the study of human populations can and should play a greater role as a test case for predictions about life span and ageing in general. Williams’ hypothesis is about change in ageing that occurs in response to change in mortality. For this reason, human mortality improvement is examined in a large-scale, macro-ecological framework using a comparative evolutionary approach (Carey & Judge 2001). Firstly, economic correlates of life span improvement are described and interpreted as essential contributors to the changes in density and condition-dependent feedbacks responsible for the increase in life span of the last 200 years. Researchers working on other organisms have noted the importance of density and condition, both by age and across populations, and the role of economic inputs in mortality reduction for humans should be understood in similar terms. Examining links between economic development and mortality reduction is not typical in evolutionary-motivated studies of ageing, but it is increasingly clear that economic developments are a large part of the environmental changes that allowed human life history traits, ageing and mortality in particular, to change so drastically in the last 100 to 200 years (Carey & Judge 2001; Fogel 1997). Secondly, descriptive measures are used to frame the reduction of mortality at all ages, to place human mortality improvement in species-wide historical context and to demonstrate the progress made relative to the estimated ‘typical’ mortality pattern of human ancestors. Because such large changes in economically conditioned mortality should, under expectations of canonical theory, be expected to change selection pressures in favour of slower ageing, indicators of the force of selection are calculated across a range of societies. This allows the observations on mortality to be linked to theories of ageing. Also, it heeds Caswell’s (2007) observation that the common perception of Williams’ hypothesis as a prediction regarding extrinsic mortality and the rate of ageing is incongruent with Hamilton’s main results, which are specific indicators for how selection pressure should change with age given schedules of mortality and fertility. The following sections examine three separate areas where evolutionary theory of ageing seems to conflict with observations on human demography. Overall, a case is made that the standard theory needs updating. The development of derived theories of ageing, as proposed by Reznick et al. (2004), Williams et al. (2006) and Walsh et al. (2014), among others (e.g. Chen & Maklakov 2012), is proposed as the strongest candidate solution for rectifying the growing tensions between theory with data in ageing research on humans and across the Tree of Life.

Externalities Lower Mortality Studying ecological change of the past 100 years raises the problem of co-linearity because many changes occurred that are inter-related in complex ways, and the relevant variables will depend on scale and region and many other factors. For the sake of simplicity, consider the relationship between gross domestic product per person (GDPpc) and life expectancy. It is primarily the case that the higher a country’s Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:31:53, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.006

Externalities Lower Mortality

109

Female life expectancy at birth, years

Global leaders, 1840 – 2000

80

70

60

50

2500 Figure 6.2

5000 10000 GDP per capital, log scale

25000

The world’s highest female life expectancy at birth as a function of the world’s highest GDPpc for each year 1840 to 2010 (outlying countries removed, all oil-producing nations with very high GDPpc). The ‘forefront’ of development and mortality progress are very closely related. The bestfit line explains 95 per cent of the variance in life span (slope = 15.2 (0.28), intercept = −69.1 (2.46)). (Life-expectancy data are from Oeppen & Vaupel 2002; GDPpc data are from Angus Maddison.)

GDPpc, the lower its typical rates of mortality and the more developed its infrastructure (with notable exceptions, of course, such as some of the economies particularly reliant on oil production). For the sake of argument, let us further assume that as a country’s GDPpc rises, so does the diversity of the social and economic niches available to members of the society. Oeppen and Vaupel (2002) famously showed that life expectancy at birth has been increasing linearly as a function of time since about 1840 when measured for the world’s life-expectancy leaders, defined as the country with the highest life expectancy at birth for any given year. How much of this well-known result could be attributable to improved environmental quality, as indexed by GDPpc (Preston 1975, 2007)? The relationship between the upper limit of the phenotypic response, life span, and a candidate environmental driver is noticeably linear and statistically strong; a strikingly high degree of the variance in the world’s best life expectancies is explained by the logarithm of world’s best GDPpc (Figure 6.2). The time required to make each jump in life expectancy is variable, but the concurrent economic increase is consistently escalating (Table 6.1). For example, it took thirtyseven years (from 1890 to 1927) for life expectancy to increase from fifty-five to sixty but only twelve years (from 1961 to 1973) for it to increase from seventy-five to eighty. In contrast, each five-year increment of increased life expectancy was accompanied by a progressively larger increase in GDPpc. GDPpc increased by about $2,600 as the leading life expectancy rose from sixty-five to seventy years and about $5,300 as it rose

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:31:53, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.006

110

Evolutionary Demography of the Human Mortality Profile

Table 6.1 Cost of Life-Expectancy Increase for Each Five Years of Improvement from Fifty-Five to Eighty-Five Years, with Costs and Annual Growth Rate in GDPpc Life expectancy

Calendar year

GDPpc a

From

To

From

To

Years elapsed

Annual

Increaseb

55 60 65 70 75 80

60 65 70 75 80 85

1872 1890 1927 1947 1961 1973

1890 1927 1947 1961 1973 1996

18 37 20 14 12 23

0.020 0.009 0.017 0.022 0.030 0.014

1,415 1,855 2,595 3,260 5,317 6,829

Source: Based on fitted values from Figure 6.2. Units for GDPpc are 1990 international Geary-Khamis dollars. a Mean annual growth rate in GDPpc for the time interval log(GDPpct+1/GDPpct)/(years elapsed). b The absolute increase in GDPpc for each five-year increase in life expectancy.

from seventy-five to eighty (Table 6.1). This implies that each additional year of life expectancy requires more economic growth than the previous year. This would happen if part of economic growth, either intentionally or inadvertently, removes sources of mortality from, roughly, least to most costly as economies evolve. These calculations suggest that the continual lowering of mortality requires constantly improving the environment to remove shocks, stresses and causes of death. That is, the causes of mortality reduction, as complicated as they may be to individually pinpoint through time and by age, are wholly extrinsic in nature. Moreover, there is a diminishing return to the amount of economic growth required to extend life span such that each additional year of life costs more than the previous. That said, the ability to stretch the mortality profile to ever lower age-specific risks of death by improving the environment is an incredible feat, exhibiting a capacity for variation in mortality due to environmental improvement. Put another way, these data provide strong evidence for an evolved demographic plasticity that current evolution theory neither predicts nor accounts for. It also shows that human populations are changing greatly in composition and condition and in ways that are almost certainly not genetic.

The Magnitude of Human Mortality Reduction: Comparative View Comparing several populations that differ greatly in mortality (Table 6.2), the dramatic recent reductions are clearly evident in age-specific patterns of survivorship by age lx and the annual probability of death qx (Figure 6.3). Figure 6.3A shows the proportion surviving for several periods in the long history of Sweden (long in terms of data availability), hunter-gatherers, acculturated hunter-gatherers and wild chimpanzees (Pan troglodytes). The survival curves in Figure 6.3A clearly illustrate the process of rectangularisation, where mortality at early ages is reduced, making the curve more boxlike as most of the mortality becomes compressed to later ages (compare Sweden 2010 as a highly rectangularised population to early Sweden or hunter-gatherers, where the Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:31:53, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.006

The Magnitude of Human Mortality Reduction: Comparative View

111

Table 6.2 Life Span Statistics that Track Mortality Change and Improvement

Population Wild chimps Captive chimps Hunter-gatherers Sweden 1800 Sweden 1751 Acculturated huntergatherers Sweden 1850 Sweden 1900 Japan 1950 Sweden 1950 Sweden 2010 Japan 2010 a b

Median life spana

Age when ‘old’b

Age when ‘very old’b

Life expectancy Life expectancy when ‘very when ‘old’ old’

8 10 25 29 42 45

Always Always Always 30 29 39

36 53 74 71 78 78

na na na 30 33 29

7 6 6 6 6 6

15 21 32 32 38 40

53 63 68 76 85 86

35 49 50 57 65 67

75 78 77 79 86 89

30 24 22 20 20 20

6 6 6 6 6 6

45 52 59 71 82 83

Life expectancy at birth

Median life span defined as age when half the population is deceased, age when lx = 0.5. Age when ‘old’ and ‘very old’ are ages when qx is approximately 1 and 10 per cent, respectively.

survival curve cuts more diagonally through the plot). Researchers are concerned with evaluating how much of the extension in life expectancy is due to rectangularisation as opposed to actual extensions of the intrinsic life span, indicated by a migration to the right at the point where the survival curve intersects the x-axis of the figure. The former would be due to reducing mortality, whereas the latter indicates change to the make-up of the population and an increase in the maximum possible life span. Rossi et al. (2012) have shown that the secular increase in life expectancy is in fact not just a function of reduced early life mortality, rectangularisation, but also due to increases in the possible life span. The horizontal dotted line is the median life span, or the age at which 50 per cent of the population is still surviving (Table 6.2). This indicates the amazing jump in survival seen for the healthy and wealthy countries like Sweden. The median life span for wild chimpanzees is shy of ten years, is just over twenty-two years for hunter-gatherers, and over eighty years for 2010 Sweden. Most industrialised countries would be broadly similar to Sweden in this plot, especially from the macro-demographic perspective used here. The already amazing feet of doubling median life span with the life history evolution from chimpanzees to humans is eclipsed by the almost four-fold increase from hunter-gatherers to modern Sweden, most of which occurred in just the last century and a half (Burger et al. 2012). Early in the sequence available for Sweden, the survival experience was fairly similar to that of hunter-gatherers (Figure 6.3A). For instance, Sweden 1751 and Sweden 1800 are between the hunter-gatherer and acculturated hunter-gatherer curves. The relatively small reductions in mortality gained by acculturated hunter-gatherers comes from

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:31:53, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.006

112

Evolutionary Demography of the Human Mortality Profile

Proportion surviving, Ix

(a) Sweden 1751–2010 Hunter–Gatherers Acc. Hunter–Gatherer Wild Chimps

1.00 0.75 0.50 0.25 0.00 0

10

30

50

70

90

Annual probability of death, qx

(b)

10−1 10−2 10−3 10−4 0

Figure 6.3

10

20

30

40 50 60 Age in years

70

80

90

100

(A) Proportion surviving as a function of age for human populations that vary greatly in mortality compared to a synthetic survivorship curve for wild chimpanzees. The horizontal dotted line intersects each survival curve at the median life expectancy. The vertical dotted lines indicate the exact median life expectancies for the wild chimpanzee, hunter-gatherers and Sweden 2010. (B) Annual probability of death by age for the same populations. The horizontal dotted lines indicate the illustrative actuarial definitions for ‘old’ (qx = 10−2) and ‘very old’ (qx = 10−1).

modest access to Western goods and a less mobile lifestyle (Gurven & Kaplan 2007). Well into the nineteenth century the survival experiences of even the world’s healthiest and wealthiest nations were not out of the range of what humans experienced during the Pleistocene. In other words, if we could obtain a truly random sample of foraging populations from across the Pleistocene (with, say, a time machine), the error bars around the estimated mean mortality profile would likely be very large. Only recently did any human population leave the range of these hypothetical error bars in terms of life expectancy or risk of death at any particular age (Burger et al. 2012; Finch 2012). Figure 6.3B shows the same populations in terms of the annual probability of death qx. The dotted lines in this case illustrate two illustrative thresholds in order to make a point about the rapid decrease in age-specific risk of death. Using the annual probability of death, let ‘old age’ be defined as an annual probability of death greater than 1 per cent and ‘very old age’ as a probability of death greater than 10 per cent (Table 6.2). For these thresholds, the age of becoming ‘old’ has increased from the mid-fifties to the

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:31:53, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.006

Mortality Reduction and Changes to the Force of Natural Selection

113

mid-sixties in the last fifty years, and the age of becoming ‘very old’ has increased from the mid-seventies to the mid-eighties. For contemporary industrial societies, let us further assume that these are the ages generally associated with old and very old individuals; we retire when old, around sixty-five, and twenty years later we are very old at eighty-five. Both of these figures imply great progress for those who live in the wealthy and healthy countries. However, as before, these figures obscure the evolutionary relevance of mortality reduction, failing to capture the magnitude of progress from the typical human mortality profile, as estimated by the mean hunter-gatherer survival curve (Burger et al. 2012; Gurven & Kaplan 2007). While the thresholds for ‘old’ and ‘very old’ might agree with contemporary experience, they do not work at all for characterising the survival patterns of typical humans, where this illustrative definition of old is an annual risk of death lower than what the average person would experience in their entire lifetime (Table 6.2 and Figure 6.3B). While the age at which a population reaches a probability of death of one in ten has changed a lot, the remaining life span at that age is strikingly constant at about six additional years. A Swede surviving to seventy-one in 1800, the age when ‘very old’, could expect about six more years, on average, just as a ‘very old’ Swede living to age eighty-six in 2010 could expect about six more years. Of particular note in both plots is the distance in each between hunter-gatherers, an approximation for the evolutionarily ‘normal’ survival pattern for humans, and chimpanzees, the nearest living taxonomic relative to humans. The distance between chimpanzees and humans is smaller than the distance between hunter-gatherers and contemporary industrial societies like Sweden, again showing a seemingly specieslevel jump in the observed values for this very flexible phenotypic trait (Burger et al. 2012). Most of the known phenotypic variation for human mortality has only been observable since the industrial revolution. Thus, mortality patterns for the humans we might study in biogerontology, evolutionary demography and the like are highly dependent on the time, prevailing conditions and location. This raises the question, examined in the next section, of whether such contingencies have a strong influence on the pattern of ageing we predict for that species. Classical theory provides indicators seemingly designed for addressing such a question.

Mortality Reduction and Changes to the Force of Natural Selection Classical explanations for senescence stem from the calculation that selection becomes less effective with age, resulting in organisms that also become progressively less effective with age (Charlesworth 2000). This decline in the effectiveness of selection, in terms of removing alleles with negative impacts on fitness, is partly influenced by changes in mortality that occur with age (Abrams 1991; Hamilton 1966). As such, intuition might suggest that the drastic changes in mortality and environment described earlier should lead to major shifts in this calculated effectiveness of natural selection. In fact, many of the empirical tests of Williams’ hypothesis assume that change in Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:31:53, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.006

114

Evolutionary Demography of the Human Mortality Profile

mortality should alter the rate of ageing without actually calculating the indicators designed by Hamilton to compute this effectiveness of selection (Gurven & Fenelon 2009; Hawkes et al. 2012). For the classical indicators of the force of selection, the age-specific sensitivity to a change in mortality is calculated using age schedules of survival and fertility as the derivative of fitness with respect to a change in the force of mortality (where fitness is the intrinsic population growth rate or Malthusian parameter r) (Hamilton 1966). Such a sensitivity measure was Hamilton’s main result in his effort to quantify Williams’ (1957) theory of senescence (Caswell 2007). While most studies focus only on one, Hamilton derived two measures of sensitivity (Abrams 1991). Williams et al. (2006) suggested that the second, lesser-used measure is the most useful for studies of senescence because the mortality-altering change occurs at age x but has consequences for all later ages, whereas the first indicator considers only the current age x. That said, both indicators are examined here. Following Williams et al (2006), they are indicator one s1 (x), here called ‘sensitivity to instantaneous mortality change’, and indicator two s2ðXÞ, or ‘sensitivity to prolonged or senescent mortality change’: s 1 ð xÞ ¼

s 2 ð xÞ ¼

dr ¼ dμx

dr ¼ dμx

∞ X

era la ma =T;

ð6:1aÞ

ða  xÞera la ma =T:

ð6:1bÞ

a¼xþ1

X ∞

a¼xþ1

In each case, r is the intrinsic growth rate (Malthusian parameter), la is the probability of P survival to age a, ma is age-specific fertility, and T is generation length (T ¼ ∞a¼1 aera . Each indicator is in terms of dr/dμx, the derivative of the fitness measure r with respect to the force of mortality μx. The only difference is the ða  xÞ term in s2 ð xÞ, which ensures that fitness effect is prolonged rather than instantaneous (Abrams 1991), because a change at age a influences the sensitivity a − x years later. The two indicators are plotted in Figure 6.4 for a range of human populations that vary in mortality and fertility, along with hunger-gatherers and wild chimpanzees for comparison. For both indicators, the strength of selection is high at birth and declines with age, but the sensitivity to instantaneous mortality change implies that selection is indifferent to age between birth and first reproduction, but the sensitivity to prolonged mortality declines monotonically from birth (Williams et al. 2006). Thus, senescence starts only with reproduction for the common indicator but right from birth for the second. Early-life selection pressure is much greater among chimps than among any human population for the instantaneous indicator (Figure 6.4A). For the prolonged, or senescent, survival sensitivity, most of the variation among the populations at each age is due to the shape of the maternity function, particularly the location of the age at peak fertility. Across both indicators, the sensitivity of fitness at each age is strikingly similar for this sample of human populations that differs greatly in mortality (and fertility). This is

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:31:53, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.006

Mortality Reduction and Changes to the Force of Natural Selection

115

Survival sensitivities

S(x)1,Instantaneous

(a) 0.05 0.04 0.03 0.02 0.01 0.00 0

S(x)2,Prolonged

(b)

10

20

30

1.0

40

50

Hunter–Gatherer Sweden 1900 Sweden 1950 Sweden 2000 Wild Chimps

0.8 0.6 0.4 0.2 0.0 0

Figure 6.4

10

20 30 Age in years

40

50

(A) Hamilton’s indicator for the sensitivity of fitness to a change in mortality where senescence begins at reproductive maturity. (B) Hamilton’s indicator for the sensitivity of fitness to a change in mortality where all future ages are affected due to a senescent change, causing senescence to begin at birth (Abrams 1991).

contrary to popular articulations of Williams’ hypothesis, where reductions in mortality are thought to decrease the rate of ageing due to greater survivorship to older ages (Hawkes 2010). In sum, the two measures of the force of selection demonstrate that very large changes in mortality have only modest changes in selection pressure (because fertility and r are changing too). Such observations present a layer cake of difficulties. The first layer: as Caswell (2007) has shown, Williams’ hypothesis should not be taken as a prediction because it ignores the indicators behind Hamilton’s results. A second layer is that Hamilton’s result itself leads to some predictions that are difficult to reconcile with actual observations on mortality increase with age. These include the ‘Wall of Death’, or the prediction that the risk of death from intrinsic ageing should shoot to infinity as selection pressure goes to zero (Charlesworth & Partridge 1997; Levitis et al. 2013; Tuljapurkar et al. 2007). The fact that the classical indicators, as necessary simplifications of a more complicated reality, make unsupported empirical predictions can be remedied by using sex-linked models of selection pressure, which show that selection pressure in both sexes is linked to the age-specific mating patterns (Tuljapurkar et al. 2007). Likewise, more appropriate

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:31:53, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.006

116

Evolutionary Demography of the Human Mortality Profile

measures might include sociality and intergenerational transfers (Chu & Lee 2006; Gurven et al. 2012). Third, one could argue that the Hamiltonian indicators are only useful for cross-species comparisons, where they have been more successful. Species with higher extrinsic mortality may indeed age faster (Ricklefs 2010). However, note that (1) the magnitude of mortality change across humans is greater than across several large-bodied species of primate, so there may be a lack of sufficient evolutionary time, but there is not a lack of sufficient variation in mortality for the result to be plausibly visible, and (2) if the indicators are not well suited for the shorter time-scale changes of within-species variation seen for humans in Figure 6.3, then certainly studying just mortality change in isolation is not a sufficient way to test the theory, yet mortalitymortality correlations are commonly employed, and (3) closely related species do not seem to follow the Williams’ prediction (when tested with Gompertz models) (Bronikowski et al. 2011). The question is, where does that leave us with respect to Williams’ hypothesis, Hamilton’s indicators and a means of predicting how mortality affects optimal ageing patterns? In defence of any particular study that has chosen one of the indicators or has focused on mortality-mortality correlations or correlations of parameters in the Gompertz model, theoretical models should be taken for their intended purpose of providing necessarily over-simplified abstractions of complex evolutionary processes. That said, there is some need to seek further modifications to the indicators for the strength of selection, perhaps along the lines of Tuljapurkar et al. (2007).

Discussion The phenomenal reduction in human mortality of the last century is accompanied by the socio-economic developments of the industrial revolution (Fogel & Costa 1997), but despite these great changes in mortality, the force of selection changes relatively little. This partly explains why researchers have had difficulty verifying Williams’ hypothesis using historical data from countries passing through the demographic transition (Finch et al. 1990; Gurven & Fenelon 2009; Hawkes 2010). When the rate of mortality is measured as the mortality-rate doubling time, it tends to change very little through the mortality transition (Gurven & Fenelon 2009). Williams’ hypothesis, for example, has engaged many novel and important studies but also has many exceptions (Hawkes 2010; Reznick et al. 2004; Walsh et al. 2014). This is partly due to a discipline-wide tendency to frame the prediction in terms of extrinsic mortality rather than in terms of the indicators of the force of selection, although the indicators may need modification as well. The mortality reductions due to economic growth are so large that we should (or would like to) be able to make predictions about actuarial ageing based on the sensitivity curves. However, in terms of the indicators developed to capture the fitness consequence of a change in mortality, there is little observed difference across populations, and hence no clear expectation linking mortality reduction to a change in the rate of ageing. That is, ageing could be argued to increase, decrease or remain constant in response to a decrease in mortality (Abrams 1993; Caswell 2007; Gurven & Fenelon 2009). This is shown by the similarity of the indicators Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:31:53, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.006

A Need for Derived Theories of Ageing

117

in Figure 6.4. Caswell (2007) explains this result quite succinctly: a change in truly extrinsic (age-independent) mortality changes the numerator and denominator of Hamilton’s indictors (Equations (6.1a) and (6.1b)) in ways that exactly offset each other. In this case, the indicators of the strength of selection by age are essentially silent with respect to predicting changes in the rate of ageing in response to reduced extrinsic mortality and completely so when the tests are based on Gompertz fits to human populations of the last 200 years or so. The recency of human longevity progress and the strong likelihood that it is driven by environment and technology are widely appreciated (Carey & Judge 2001; Fogel & Costa 1997; Wrigley 2010). Less widely appreciated, however, is that the economic role indicates that the drastic lowering of mortality, the conscious and unconscious removing of hazards by institutions and technologies, is in part the result of increasing investments made possible by economic growth (Table 6.1). Some sources of mortality that become attenuated during the mortality reduction likely have a strong age pattern (Carnes et al. 2006), and in aggregate, these various sources interact with age in ways that shift through time, making the link between source of mortality and age especially complicated. This, in turn, makes it difficult to compute an expected response of the ageing rate to the shift in mortality pressure. At this point researchers on human ageing lack a general framework for linking changes in ecology to feedbacks with selection on the repair mechanisms that resist senescence (Reznick et al. 2004; Williams et al. 2006). This is partly due to a lack of adequate attention to the ecology of the focal species, recalling the earlier quote from Walsh et al. (2014), but is also symptomatic of problems with classical theories of ageing (Jones et al. 2014; Vaupel et al. 2004).

A Need for Derived Theories of Ageing To find a way forward, Reznick et al. (2004) and others (Chen & Maklakov 2012; Walsh et al. 2014; Williams et al. 2006) have suggested the need to develop ecologically oriented ‘derived’ theories of ageing that include density- and condition-dependent interactions with senescent mortality. Both of these dependencies, one from density and the other from condition, apply to humans in many ways. If mortality is either density or condition dependent, then selection may favour faster, slower or unchanging rates of senescence depending on the nature of the interaction with age (Walsh et al. 2014). Before proceeding, a brief digression into the differences between intrinsic and extrinsic mortality is needed. Change in mortality with age can be thought of as a balance between an intrinsic susceptibility and extrinsic risk. At this point it is an open question how fixed intrinsic mortality is across environmental conditions (Figures 6.1 through 6.3), and we strongly suspect that intrinsic mortality is influenced by many cues that are not genetically fixed. Many statistical and evolutionary studies of ageing use the original actuarial definition of extrinsic mortality as age independent (Makeham 1867), even though actual sources of mortality probably very rarely lack an age pattern (Carnes et al. 2006; Gurven & Fenelon 2009). This has led to a common failure to explicitly address the important distinction between a parameter in a model Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:31:53, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.006

118

Evolutionary Demography of the Human Mortality Profile

that captures an age-independent quantity of mortality and actual age independency of mortality sources encountered in the organism’s ecology. As mortality that is simply external to the intrinsic decline in cell and tissue quality of the organism, extrinsic mortality clearly can vary by orders of magnitude in a wide range of species from humans to fruit flies. Examining the mortality changes in Figure 6.3, is this reduction due only to extrinsic mortality being removed? Or have the significant changes in population composition, condition, changed the age profile of the intrinsic risk as well? For many applications, extrinsic mortality probably should be thought of in less restrictive terms than simply age-independent or due to singular processes such as predation. Having to work harder at a given temperature because of changes in the density of resources or predators, for example, is an extrinsic risk, even though the cellular damage incurred by the metabolites generated might seem otherwise intrinsic in nature. In this way, the environmentally influenced condition of the individual affects his or her intrinsic susceptibility; these are not mutually exclusive. Empirical tensions emerge where the classical theory fails to address a balance between extrinsic factors that offset the observed mortality risk and the properties of the intrinsic risk itself (or perhaps the issue is only in our ability to measure the intrinsic risk). If the bulk of observed mortality is extrinsic, then one may have to dig quite deep, in terms of empirically removing all extrinsic causes, before the intrinsic mortality profile can be identified. If this is true, then the intrinsic profile is rarely observed. Condition- and density-dependent mortality are well suited for filling this ‘gap’ between the statistical nicety of viewing extrinsic mortality as strictly age-independent and the need for realising that many sources of mortality have distinct age signatures (Carnes et al. 2006).

Density Dependence and Ageing Density dependence affects ageing because it interacts with the age profile of mortality and can lead to changes in resource availability (Reznick et al. 2004). If a source of mortality is removed, this may alter the age profile of the population. As the population grows, resources will eventually become limiting, and different age groups can feel the effects of resource competition differently. If competitive ability is age-dependent, then the mortality consequences of resource competition will be age-dependent as well. If mortality is increased, then the survivors may experience increased access to resources by reduced competition (Reznick et al. 2004; Williams et al. 2006), which is one of the complicating factors for making predictions about ageing rate from mortality change (Abrams 1993). Such density-dependent feedbacks can change the pressure of selection against senescence. This density-dependent dynamic is central to the exceptional human demographic change over the last two centuries. The post-industrial age has resulted in a special kind of mammal population; humans can experience the benefit of reduced mortality while simultaneously escaping density-dependent feedbacks on food availability. In most mammal populations, there is an approximate balance between these forces, or at least a ‘push back’ from density dependence following reductions in mortality (Charlesworth & Partridge 1997). Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:31:53, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.006

Condition Dependence and Ageing

119

The GDPpc relationships earlier exhibit a pathway by which humans gain from increased access to resources without suffering the feedback of density dependence (at least in terms of food limitation from density), which would otherwise cause mortality to rise via reduced access to resources as populations grow (Gause 1934; Sibly et al. 2005). Likewise, humans lower extrinsic mortality with technology, culture and institutions (Fogel 1997; Lee 1987), but the economic growth ensures that reducing mortality does not create increased resource pressure because access to energy increases as mortality drops, as opposed to increasing, and this novel situation underlies some of the particularities of human demographic traits (Burger et al. 2011; DeLong et al. 2010). Using England as an example, both economic and ecological measures of density dependence can be identified from the 1500s to the late 1700s (Wrigley 1990, 2010) but largely disappear after the industrial revolution, or right about the time that GDPpc and life expectancy become monotonically increasing. The degree to which this simple dynamic is critical for understanding vital rate variation or plasticity in humans is under-appreciated.

Condition Dependence and Ageing The many (economic, institutional, etc.) changes responsible for lower mortality also alter the condition of people in terms of health, nutrition and exposure to risk. Like density dependence, condition dependence can interact with extrinsic mortality and the rate of ageing. As a simplistic example, consider that running speed is important for predator avoidance. If so, then the costs of losing foot speed with age are greater in a predator-rich environment than in a predator-poor environment. Such a change might alter the strength of selection on the physiological systems that underlie running speed (Reznick et al. 2004). In terms of evolutionary demographic processes, if technology or sociality removes the risk of predation, or some other environmental mortality risk, the selection gradient on predator-avoidance traits is fundamentally altered, as is the pressure on the social or cognitive traits that helped to remove the mortality risk. One might think that some of the traits that underlie human sociality and cognitive development interact with mortality and selection pressure in this way. Condition-dependent mortality has been shown to select for life span in ways that differ greatly from condition-independent mortality. For instance, laboratory studies on nematode worms (Caenorhabditis elegans) show that randomly induced mortality selects for shortened life span, but condition-dependent mortality selects for longer life span (Chen & Maklakov 2012; Walsh et al. 2014). Condition dependence may underlie both the evolution of the long human life span and the drastic increase of the last 200 years (Figures 6.1 through 6.3). While the experimental conditions afforded by studies of nematodes are not present for studies of human ageing, the rich historical record of some population goes a long way towards providing nearly comparable insights. For instance, Temby and Smith (2014) use historical data for the US state of Utah to demonstrate gene–environment interactions, whereby individuals in high-status positions (who experience better ecologies) had a more pronounced benefit of a family history of long life spans. Thus, the protective effects of genetic inheritance were Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:31:53, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.006

120

Evolutionary Demography of the Human Mortality Profile

enhanced by especially benign environments. As such, condition dependence should be viewed as a major part of the explanation for the reduction in mortality that leads to a doubling in median life span over evolutionary change from chimpanzees to humans as well as the quadrupling that occurred due to economic improvement within postindustrial humans (Tables 6.1 and 6.2). As organisms age, mortality from some sources may decline due to increased ability to avoid risk by somatic growth or by learning (Williams et al. 2006). Learning how to avoid danger and increased reliance on sociality through evolutionary time (across populations) and with age (within populations) were hallmark developments for the behavioural evolution of humans (Hill et al. 2009). If the mortality reduced by learning and social cooperation is greater than the increase in intrinsic susceptibility with age, then actuarial mortality could appear flat or declining due to the balance of the two (Williams et al. 2006). Indeed, Gurven and Kaplan (2007) find nearly flat and even declining mortality rates with age from early to middle adulthood for some huntergatherer populations, and Finch’s phase 2 of the human mortality profile, ‘basal mortality’, is noted for being especially low in average risk of death and notably nonincreasing as the Gompertz phase begins only after age forty. Carey and Judge (2001) and others (Fogel 1997; Fogel & Costa 1997) have argued that technological development and tool use extended human longevity pre-historically and recently. This comes from lowering mortality in general but also from improving diet that leads to stature increase and other physiological changes that change the distribution of condition by age in a population. Thus, change in technology via the human propensity for culture begets changes in condition that beget long life spans. The changes to condition that lead to lower mortality are evident in many of the changes in population composition made possible by post-industrial technologies and institutions (Carnes et al. 2006). Improved nutrition has lowered rates of organ failure, improved physiological status across whole populations and contributed to a 50 per cent increase in height in the last 200 to 300 years (Fogel & Costa 1997). Clearly, the removal of infectious diseases and causes of inflammation are among the most important for reducing multiple sources of mortality with differing age profiles. Likewise, increased height is often associated with lower mortality and morbidity, and height increases along with many of the other developments that lower mortality. By increasing nutrition levels and caloric availability, the average output per worker increased, and the requisite economic growth behind longevity increase was under way (Fogel & Costa 1997), and this benefitted from (or interacted with) changes in cause of death and the removal of many life-threatening events (Carnes et al. 2006; Finch 1994). These compositional changes indicate changing distributions of condition that are ‘biological but not genetic’ but have direct consequences for how selection may alter ageing and longevity (Fogel 1997). Work in classical life history theory and recent studies in epigenetics and maternal effects suggest ways in which condition dependence affects ageing. The positive co-variances observed between early-life and late-life fitness, or fitnessrelated traits, is especially noteworthy for proponents of positive pleoitropy (Maklakov et al. 2015), as noted earlier. In life history we know the simple rule that organisms that take longer to build themselves tend to be longer lasting, as seen in the slow growth, late Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:31:53, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.006

Methods and Measurement

121

age at maturity and longer life span of the great apes compared to mammals of similar size (Charnov & Berrigan 1993; Promislow & Harvey 1990). Likewise, there are phenotypically adaptive cues about the environment that organisms, including humans, obtain from their mothers by a variety of mechanisms that may trigger a similar growslow and live-longer correlation under certain ecological conditions (Wells 2007, 2011).

Methods and Measurement The tendency here has been to follow other recent treatments by interpreting the ‘rising tide’ of tension between theory and data as indicators that the evolutionary theory of ageing has problems and should be modified (Jones et al. 2014; Maklakov et al. 2015; Vaupel et al. 2004; Williams et al. 2006;). There are, however, also more mundane issues of definition, method and measurement that pose great challenges. For instance, while Williams’ defined the onset of ageing as beginning near maturity, it may be that it occurs at later ages when mortality accelerates. A reduction in extrinsic mortality could lead to a delay in onset with an unchanged or even faster rate of ageing. That is, the onset of ageing and the rate of ageing may both be subject to variation. If, hypothetically speaking, there were great variation in the age of onset but relatively little in the rate, but we measured ageing with Gompertz fits from set ages of thirty to eighty, we could find variation in the rate of ageing as an artefact of fitting the wrong model. Studies using simulations find that a wide range of empirically observed mortality profiles can be obtained, including accelerations or plateaus, if individuals are allowed to vary in the onset and rate of ageing (Le Cunff et al. 2013). Likewise, exceptions to Williams’ hypothesis are often based on fits of the Gompertz model to mortality data (e.g. Reznick et al. (2004) for guppies and Hawkes et al. (2009) for humans), which may be a flawed approach to testing the hypothesis in the first place, especially because of the dubious biological meaning behind the correlation between Gompertz parameters (Lenart & Missov 2014; Burger and Missov 2016). The problematic or challenging approach of measuring the rate of ageing with the Gompertz model is seen in the fact that Hawkes et al. (2012) estimate more variance in the rate of ageing than Gurven and Fenelon (2009), even though both fit Gompertz models to a similar sample of populations. Further, in the absence of very detailed data on cause of death, reliance on the Gompertz model for testing predictions from the evolutionary theory of ageing implicitly force the convention of defining extrinsic mortality as age independent, thus excluding the influence of density and condition. Some ageing researchers have suggested that the rate of ageing at the individual level may be roughly constant (Vaupel 2010). If so, then the difficulty of separating the onset of ageing from its rate interacts with the difficulty of satisfactorily distinguishing extrinsic from intrinsic mortality, as tests of predictions regarding the rate of ageing could be biased by variation in the onset or by a failure to capture intrinsic mortality. Likewise, there is the issue of inferring individual ageing patterns from population-level data, which is fraught with difficulty (Vaupel & Yashin 1985; Yashin et al. 2002). Mortality change by orders of magnitude can lead to minimal change to the force of selection, which would be expected Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:31:53, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.006

122

Evolutionary Demography of the Human Mortality Profile

if the theory applied to age-independent mortality. The force of selection may need to be measured with other indicators, perhaps those that account for condition dependence and inter-generational transfers (Carey & Judge 2001; Lee 2003).

Conclusion In sum, evolutionary theories of ageing provide limited predictive guidance for how the human life span and ageing pattern should evolve. Hamilton (1966) is often seen as a mathematical formalisation of Williams (1957), yet neither the verbal model from Williams nor the indicators of selection intensity from Hamilton is sufficient to explain or predict how ageing should change in response to the major reductions in mortality experienced by humans. Nor can they provide insight into other pressing questions, such as how much life span can increase or how much age-specific plasticity in mortality we should observe across populations. The rapid reductions in mortality of the last century are effectively ‘bought’ in that the ability to remove stresses and sources of extrinsic mortality requires developments, crudely indexed here with GDPpc. As mortality drops, the composition of the population changes, and this calls for an understanding of condition-dependent mortality, which interacts with selection on maintenance functions in ways that may influence the onset and rate of ageing. This means that ageing may depend on condition, while condition may depend on certain features of the human built environment. While part of the failure to support the Williams hypothesis, as classically framed, is certainly due to the short time span involved, there are pressing issues at work requiring new or modified indicators of selection pressure by age. It may be that there is a real but difficult-to-measure distinction between intrinsic and extrinsic mortality but that the two interact in such a way that labelling extrinsic mortality as ‘age independent’ makes testing the classical predictions with real-world data unrealistic. The canonical theory simply does not capture the complex properties of actuarial mortality, which are what most scientists actually observe (at least for humans). In the future, derived theories of ageing may help to pave the way to novel hypotheses and models for explaining human ageing patterns. Human ageing patterns can be used to inform general theory as much as general theory can be developed to understand humans as one particularly relevant instance of ageing across all those in the Tree of Life.

Acknowledgements This chapter was improved immensely thanks to comments from Sarah Myers, Daniel Levitis, Samuel Pavard and an anonymous reviewer. However, the mistakes that remain are clearly my own fault. I thank Owen Jones for excellent editorial advice. The staff of the Max Planck Institute for Demographic Research were supportive and inspiring in numerous ways, and I’m especially indebted to Trifon Missov, Jim Oeppen and Jim Vaupel. Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:31:53, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.006

References

123

References Abrams, P. A. (1993). Does increased mortality favor the evolution of more rapid senescence? Evolution, 47(3), 877–87. Abrams, P. A. (1991). The fitness costs of senescence: the evolutionary importance of events in early adult life. Evolutionary Ecology, 5, 343–60. Borgerhoff Mulder, M. (1998). The demographic transition: are we any closer to an evolutionary explanation? Trends in Ecology and Evolution, 13, 266–70. Bronikowski, A. M., Altmann, J., Brockman, D. K., et al. (2011). Aging in the natural world: comparative data reveal similar mortality patterns across primates. Science, 331, 1325–8. Burger, O., Baudisch, A. & Vaupel, J. W. (2012). Human mortality improvement in evolutionary context. Proceedings of the National Academy of Sciences of the United States of America, 109, 18210–14. Burger, O., DeLong, J. P. & Hamilton, M. J. (2011). Industrial energy use and the human life history. Scientific Reports, 1, 56. Burger, O. & Missov, T. I. (2016). Evolutionary Theory and the Problem of Correlated Gompertz Parameters. Journal of Theoretical Biology, 408, 34–41. http://dx.doi.org/ 10.1016/j.jtbi.2016.08.002 Carey, J. R. & Judge, D. S. (2001). Life span extension in humans is self-reinforcing: a general theory of longevity. Population and Development Review, 27, 411–36. Carnes, B. A., Holden, L. R., Olshansky, S. J., et al. (2006). Mortality partitions and their relevance to research on senescence. Biogerontology, 7, 183–98. Caswell, H. (2007). Extrinsic mortality and the evolution of senescence. Trends in Ecology and Evolution, 22, 173–4. Charlesworth, B. (2000). Fisher, Medawar, Hamilton and the evolution of aging. Genetics, 156, 927–31. Charlesworth, B. & Partridge, L. (1997). Ageing: levelling of the grim reaper. Current Biology, 7, R440–2. Charnov, E. L. & Berrigan, D. (1993). Why do female primates have such long lifespans and so few babies? Or life in the slow lane. Evolutionary Anthropology, 1, 191–4. Chen, H. & Maklakov A. A. (2012). Longer life span evolves under high rates of condition-dependent mortality. Current Biology, 22, 2140–3. Chu, C. Y. C. & Lee, R. D. (2006). The co-evolution of intergenerational transfers and longevity: An optimal life history approach. Theoretical Population Biology, 69, 193–201. DeLong, J. P., Burger, O. & Hamilton, M. J. (2010). Current demographics suggest future energy supplies will be inadequate to slow human population growth. PLoS ONE, 5, e13206. Finch, C. E. (2012). Evolution of the human lifespan, past, present, and future: phases in the evolution of human life expectancy in relation to the inflammatory load. Proceedings of the American Philosophical Society, 156, 9–44. Finch, C. E. (1994). Longevity, Senescence, and the Genome (University of Chicago Press). Finch, C. E., Pike, M. C. & Witten, M. (1990). Slow mortality rate accelerations during aging in some animals approximate that of humans. Science, 249, 902–5. Fogel, R. W. (1997). Economic and social structure for an ageing population. Philosophical Transactions of the Royal Society of London Series B: Biological Sciences, 352, 1905–17.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:31:53, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.006

124

Evolutionary Demography of the Human Mortality Profile

Fogel, R. W. & Costa, D. L. (1997). A theory of technophysioevolution, with some implications for forecasting population, health care costs, and pension costs. Demography, 34, 49–66. Gause, G. F. (1934). The Struggle for Existence (Baltimore: Williams & Wilkins). Gurven, M. & Fenelon, A. (2009). Has actuarial aging ‘slowed’ over the past 250 years? A comparison of small-scale subsistence populations and European cohorts. Evolution: International Journal of Organic Evolution, 63, 1017–35. Gurven, M. & Kaplan, H. (2007). Longevity among hunter-gatherers: a cross-cultural examination. Population and Development Review, 33, 321–65. Gurven, M., Stieglitz, J., Hooper, P. L., et al. (2012). From the womb to the tomb: the role of transfers in shaping the evolved human life history. Experimental Gerontology 47(10), 807–13. Hamilton, W. D. (1966). The moulding of senescence by natural selection. Journal of Theoretical Biology, 12, 12–45. Hawkes, K. (2010). How grandmother effects plus individual variation in frailty shape fertility and mortality: guidance from human–chimpanzee comparisons. Proceedings of the National Academy of Sciences of the United States of America 107(2), 8977–84. Hawkes, K., Smith, K. R. & Blevins, J. K. (2012). Human actuarial aging increases faster when background death rates are lower: a consequence of differential heterogeneity? Evolution, 66, 103–14. Hawkes, K., Smith, K. R. & Robson, S. L. (2009). Mortality and fertility rates in humans and chimpanzees: how within-species variation complicates cross-species comparisons. American Journal of Human Biology, 21, 578–86. Hill, K., Barton, M. & Hurtado, A. M. (2009). The emergence of human uniqueness: characters underlying behavioral modernity. Evolutionary Anthropology: Issues, News, and Reviews, 18, 187–200. Jones, O. R., Scheuerlein, A., Salguero-Gómez, R., et al. (2014). Diversity of ageing across the tree of life. Nature, 505, 169–73. Kirk, D. (1996). Demographic transition theory. Population Studies, 50, 361–87. Le Cunff, Y., Baudisch, A. & Pakdaman, K. (2013). How evolving heterogeneity distributions of resource allocation strategies shape mortality patterns. PLoS Comput Biol, 9, e1002825. Lee, R. D. (1987). Population dynamics of humans and other animals. Demography, 24, 443–65. Lee, R. D. (2003). Rethinking the evolutionary theory of aging: transfers, not births, shape senescence in social species. Proceedings of the National Academy of Sciences of the United States of America, 100, 9637–42. Lenart, A. & Missov, T. I. (2014). Goodness-of-fit tests for the Gompertz distribution. Communications in Statistics: Theory and Methods, 2014, 1–38. Levitis, D. A., Burger, O. & Lackey, L. B. (2013). The human post-fertile lifespan in comparative evolutionary context. Evolutionary Anthropology, 22, 66–79. Makeham, W. M. (1867). On the law of mortality. Journal of the Institute of Actuaries, 1866, 325–58. Maklakov, A. A., Rowe, L. & Friberg, U. (2015). Why organisms age: evolution of senescence under positive pleiotropy? BioEssays 37, 802–7. Oeppen, J. & Vaupel, J. W. (2002). Broken limits to life expectancy. Science, 296, 1029.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:31:53, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.006

References

125

Preston, S. H. (1975). The changing relation between mortality and level of economic development. Population Studies, 29, 231–48. Preston, S. H. (2007). The changing relation between mortality and level of economic development. International Journal of Epidemiology, 36, 484–90. Promislow, D. & Harvey, P. (1990). Living fast and dying young: a comparative analysis of life-history variation among mammals. Journal of Zoology, 220, 417–37. Reznick, D. N., Bryant, M. J., Roff, D. et al. (2004). Effect of extrinsic mortality on the evolution of senescence in guppies. Nature, 431, 1095–9. Ricklefs, R. E. (2010). Life-history connections to rates of aging in terrestrial vertebrates. Proceedings of the National Academy of Sciences of the United States of America, 107, 10314–19. Rose, M. R. (1991). Evolutionary Biology of Aging (New York: Oxford University Press). Rossi, I. A., Rousson, V. & Paccaud, F. (2012). The contribution of rectangularization to the secular increase of life expectancy: an empirical study. International Journal of Epidemiology 42, 250–8. Sibly, R. M., Barker, D., Denham, M. C., et al. (2005). On the regulation of populations of mammals, birds, fish, and insects. Science, 309, 607–10. Temby, O. F. & Smith, K. R. (2014). The association between adult mortality risk and family history of longevity: the moderating effects of socioeconomic status. Journal of Biosocial Science, 46, 703–16. Tuljapurkar, S. D., Puleston, C. O. & Gurven, M. D. (2007). Why men matter: mating patterns drive evolution of human lifespan. PLoS One, 2, e785. Vaupel, J. W. (2010). Biodemography of human ageing. Nature, 464, 536–42. Vaupel, J. W., Baudisch, A., Dölling, M., et al. (2004). The case for negative senescence. Theoretical Population Biology, 65, 339–51. Vaupel, J. W. & Yashin, A. I. (1985). Heterogeneity’s ruses: some surprising effects of selection on population dynamics. American Statistician, 39, 176–85. Walsh, M. R., Whittington, D. & Walsh, M. J. (2014). Does variation in the intensity and duration of predation drive evolutionary changes in senescence? Journal of Animal Ecology 83, 1279–88. Wells, J. C. (2007). The programming effects of early growth. Early Human Development, 83, 743–8. Wells, J. C. (2011). The thrifty phenotype: an adaptation in growth or metabolism? American Journal of Human Biology, 23, 65–75. Williams, G. C. (1957). Pleiotropy, natural selection, and the evolution of senescence. Evolution, 1957, 398–411. Williams, P. D., Day, T., Fletcher, Q. & Rowe, L. (2006). The shaping of senescence in the wild. Trends in Ecology and Evolution, 21, 458–63. Wrigley, E. A. (1990). Continuity, Chance and Change: The Character of the Industrial Revolution in England (Cambridge University Press). Wrigley, E. A. (2010). Energy and the English Industrial Revolution (Cambridge University Press). Yashin, A. I., Ukraintseva, S. V., Boiko, S. I. & Arbeev, K. G. (2002). Individual aging and mortality rate: how are they related? Proceedings of the Royal Society of London Series B: Biological Research, 49, 206–17.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:31:53, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.006

7

Senescence in Mammalian Life History Traits Jean-Michel Gaillard, Michael Garratt and Jean-François Lemaître

Short Summary For several decades, senescence was considered as non-existing in free-ranging mammalian populations simply because most animals were expected to die from environmentally driven causes of mortality before the age at which senescence starts. Thanks to an increasing number of long-term individually based monitoring schemes of knownage animals in the wild, evidence of senescence in most life history traits has now been reported in a large number of species across all mammalian orders. From a review of these studies, we found that actuarial senescence is the rule rather than the exception in most mammalian species studied so far. We also found clear evidence for reproductive senescence, especially in long-lived primates and ungulates, in which the oldest individuals show the existence of post-reproductive life. Senescence in life history traits is thus pervasive among mammalian species, and the key topics are now to assess the causes and consequences of senescence in mammalian population dynamics and to understand variation in the magnitude and targets of senescence among and within mammalian species. Published analyses have identified the intensity of sexual selection, the environmental context and some physiological and genetic mechanisms as major structuring factors shaping mammalian senescence within and between species. Although not yet quantified in most studies, the fitness costs of senescence do not appear as negligible as generally assumed and warrant further investigation.

Introduction From an evolutionary perspective, senescence occurs due to the decrease of the forces of natural selection with increasing age (Hamilton 1966; Medawar 1952; Williams 1957). In his classic paper, Hamilton (1966: 12) concluded that senescence is ‘an inevitable outcome of evolution’ in age-structured populations. Because mammalian species display strongly age-structured demographic parameters (Caughley (1966) for survival, Emlen (1970) for reproduction), we therefore expect, under our current evolutionary theory, that senescence should be universal in mammals. However, the picture is not that clear. Firstly, the theory formulated in the fifties through the seventies focused on senescence in survival (hereafter called ‘actuarial senescence’). Hamilton (1966) himself noticed that the problem of reproductive senescence is more complex than actuarial Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:32:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.007

Assessing Actuarial Senescence in Wild Mammals

127

senescence and did not address the question of senescence in life history traits other than survival and reproduction such as body mass or body condition, which describe qualitatively the individual phenotype and indirectly shape variation in individual fitness by their influence on demographic parameters. Secondly, mammals in the wild face a high environmentally driven mortality, which allows only a few individuals to reach the old ages at which the negative effects of senescence can be detected (Comfort 1979; Medawar 1952). Lastly, mortality at a given age within a cohort is not a random process, and individuals that survive are often heavier or in better body condition than individuals that die. Such viability selection (sensu Fisher 1930) generates an increase in the average individual quality with age, which can prevent senescence from being detected in studies performed at the population level (van de Pol & Verhulst 2006; Vaupel & Yashin 1985). Independently of the mere existence of senescence in mammalian populations, we can thus wonder about our ability to detect actuarial senescence in the wild. In this chapter we review our current knowledge of mammalian senescence in the wild. We first present recent advances in our ability to detect actuarial senescence in the wild and show how the increased abundance of age-specific mortality data is providing an insight into previously unsolved questions related to the proximate and ultimate causes of senescence. Because long-term individual-based monitoring of populations is required to assess reliably actuarial senescence in the wild, our current knowledge is biased towards the most intensively studied groups, such as large herbivores and primates. Then we discuss new findings in the study of mammalian reproductive senescence, and we show how patterns of reproductive senescence can be complex and uncoupled from patterns of actuarial senescence. Finally, we review the few studies that have investigated senescence in traits other than survival and reproduction. We discuss in each of these sections the main questions that further research on mammalian senescence should address to fill gaps in our current knowledge.

Assessing Actuarial Senescence in Wild Mammals Survival is ‘in general the most difficult estimation problem facing the field ecologist’ (Hibly & Mullen 1980). Following the pioneering work by Deevey (1947), age-specific survival has been mostly obtained from transversal life tables based on dead recoveries (e.g. Spinage 1972) or sampling of live individuals (e.g. Pielowski 1984). Such information has been repeatedly used to analyse survival patterns in mammals (Lynch & Fagan 2009; Promislow 1991; Sibly et al. 1997) and usually leads to detection of a marked decrease in survival at old ages (Figure 7.1). However, the validity of senescence patterns obtained from transversal life tables is highly questionable. There are problems of estimation of age (Vincent et al. 1994), strong and unrealistic assumptions of stability of demographic parameters and stationarity of population size (McCullough 1979; Menkens & Boyce 1993) and the assumption of a constant detection probability of live or dead animals across ages (Gimenez et al. 2008). This has led to weak accuracy and low precision of survival estimates, which has prevented a reliable assessment of actuarial senescence. Fortunately, since the eighties, there has been an ongoing marked Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:32:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.007

Senescence in Mammalian Life History Traits

Annual survival rate

128

1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0

Rangifer tarandus Ovis dalli Muscardinus avellanarius

0 Figure 7.1

2

4

6 8 Age (years)

10

12

14

Empirical evidence of actuarial senescence across mammalian species based on analyses of transversal life tables. No statistical support for actuarial senescence is available, but most transversal life tables indicate a marked decrease in annual survival rate at old ages. Age-specific variation in survival is displayed for female reindeer (Rangifer tarandus) in south Georgia (Leader-Williams 1988), in Dall sheep (Ovis dalli) in McKinley Park, Alaska (Murie 1944) and in the common dormouse (Muscardinus avellanarius) in Lithuania (Juskaitis 2008).

increase in the availability of age-specific estimates of demographic parameters from intensive monitoring of individually marked animals from birth to death in mammalian populations (Clutton-Brock & Sheldon 2010), which has provided the required empirical information to measure the magnitude of actuarial senescence, often using a capturerecapture sampling design (Lebreton et al. 1992). By accounting for imperfect detection, capture-recapture methods indeed allow accurate age-specific estimates to be obtained (Nichols 1992). From a literature survey (Nussey et al. 2013), there is clear evidence of actuarial senescence in large mammalian species (i.e. ungulates, marine mammals and primates) (Table 7.1). In small mammals, actuarial senescence has been much less studied (but see Millar 1994) and even suggested to be negligible (Slade 1995). However, these species also deserve thorough investigation. In scurids, for example, intensive monitoring has indicated that actuarial senescence might also be pervasive (e.g. Broussard et al. 2003, 2005; Descamps et al. 2008) (Table 7.1). To measure actuarial senescence, different metrics have been used. In some studies aiming only to provide evidence of senescence, the recorded survival decrease beyond some threshold age has been used. When an exact model of actuarial senescence is looked for, the rate of mortality increase with age is estimated (Jones et al. 2008), most often based on the Gompertz model (Gaillard et al. 2003). Indeed, following Gompertz’s (1825) classical work, the increase in mortality with age is estimated from the following equation: logðμÞ ¼ α þ β  age

ð7:1Þ

where μ is the annual mortality, α is the baseline mortality and β is the senescence rate. According to the Hamilton’s (1966) model, the baseline mortality should be minimal at Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:32:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.007

Assessing Actuarial Senescence in Wild Mammals

129

Table 7.1 Evidence of Actuarial Senescence in Female Mammals from Long-Term Studies Species

Reference

Moose (Alces alces) Pronghorn (Antilocapra americana) Subantarctic fur seal (Arctocephalus tropicalis) American bison (Bison bison) Muriqui (Brachyteles hypoxanthus) Northern fur seal (Callorhinus ursinus) Ibex (Capra ibex) Roe deer (Capreolus capreolus) White-headed capuchin (Cebus capucinus) Blue monkey (Cercopithecus mitis) Elk (Cervus elaphus canadensis) Red deer (Cervus elaphus) Persian fallow deer (Dama mesopotamica) Little red kaluta (Dasykaluta rosamondae) Virginia opossum (Delphinus virginiana) Asian elephant (Elephas maximus) Sea otter (Enhydra lutris) Horse (Equus caballus) Short-finned pilot whale (Globicephala macrorhynchus) Long-finned pilot whale (Globicephala melaena) Mountain gorilla (Gorilla beringei) Weddell seal (Leptonychotes weddellii) African elephant (Loxodonta africana) Alpine marmot (Marmota marmota) Badger (Meles meles) Mediterranean monk seal (Monachus schauinslandii) Dusky-footed woodrat (Neotoma fuscipes) Mule deer (Odocoileus hemionus) White-tailed deer (Odocoileus virginianus) Killer whale (Orcinus orca) Mountain goat (Oreamnos americana) Soay sheep (Ovis aries) Bighorn sheep (Ovis canadensis) Chimpanzee (Pan troglodytes) Lion (Panthera leo) Yellow baboon (Papio cynocephalus) Hamadryas baboon (Papio hamadryas) White-footed mouse (Peromyscus leucopus) Allied rock-wallaby (Petrogale assimilis) Red-tailed phascogale (Phascogale calura) New Zealand sea lion (Phocarctos hookeri) Harbour porpoise (Phocoena phocoena) Sperm whale (Physeter catodon) Eastern pipistrelle (Pipistrellus subflavus) Verreau’s sifaka (Propithecus verreauxi) Reindeer (Rangifer tarandus) Chamois (Rupicapra rupicapra)

Ericsson & Wallin (2001) Byers (1997) Beauplet et al. (2006) Pyne et al. (2010) Bronikowski et al. (2011) Lander (1981) Toïgo et al. (2007) Gaillard et al. (2004) Bronikowski et al. (2011) Bronikowski et al. (2011) Garrott et al. (2003) Clutton-Brock et al. (1988) Saltz (1996) Woolley (1991) Austad (1993) Robinson et al. (2012) Tinker et al. (2006) Garrott & Taylor (1990) Foote (2008) Foote (2008) Bronikowski et al. (2011) Rotella et al. (2012) Whitehouse & Hall-Martin (2000) Berger et al. (2016) McDonald et al. (2014) Baker & Thompson (2007) Lee & Tietje (2005) Bishop et al. (2009) DelGiudice et al. (2006) Foote (2008) Festa-Bianchet et al. (2003) Catchpole et al. (2000) Jorgenson et al. (1997) Nishida et al. (2003) Packer et al. (1998) Bronikowski et al. (2011) Packer et al. (1998) Millar (1994) Delean (2007) Bradley (1997) Chilvers et al. (2010) Moore & Read (2008) Evans & Hindell (2004) Davis (1966) Richard et al. (2002) Albon et al. (2002) Loison et al. (1999)

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:32:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.007

130

Senescence in Mammalian Life History Traits

Table 7.1 (cont.) Species

Reference

Columbian ground squirrel (Spermophilus columbianus) Richardson’s ground squirrel (Spermophilus richardsoni) Greater kudu (Tragelaphus strepsiceros) African buffalo (Syncerus caffer) Red squirrel (Tamiasciurus hudsonicus) Common brushtail possum (Trichosurus vulpecula) California sea lion (Zalophus californicus)

Neuhaus & Pelletier (2001) Broussard et al. (2005) Owen-Smith (1990) Sinclair (1977) Descamps et al. (2007) Isaac & Johnson (2005) Hernandez-Camacho et al. (2008b)

Note: The focal mammalian species and one illustrative reference are provided (from Nussey et al. 2013, modified).

the age of first reproduction, so the senescence rate is generally estimated from the age at first reproduction onwards. In the context of capture-recapture modelling (which allows accounting for imperfect detection) (Nichols 1992), the Gompertz model can be rewritten as (Gaillard et al. 2004) log½logðΦÞ ¼ ð1  αÞ þ β  age

ð7:2Þ

where Φ is the annual survival, α is the baseline mortality and β is the senescence rate. Fitting this model to survival data of twelve species of large herbivores has revealed strong evidence of actuarial senescence in both sexes (Gaillard et al. 2003). However, detailed analyses of case studies have identified several problems with this approach. Firstly, the theoretical assumption of a minimal mortality at the age of first reproduction does not seem to hold in most cases. For instance, while most ibex (Capra ibex) females usually give birth for the first time at two or three years of age (Toïgo et al. 2002), their survival does not show any sign of decline before seven years of age (Toïgo et al. 2007). Such clear evidence for delayed actuarial senescence compared to Hamilton’s prediction seems to be the rule rather than the exception in the most intensively monitored populations of mammals (Jones et al. 2008). Secondly, the two parameters of the Gompertz model, the baseline mortality and the rate of senescence, have a strong negative co-variation, meaning that any overestimation of the baseline mortality will lead to underestimation of the rate of senescence. Lastly, because mammals generally display a J- or U-shaped mortality curve involving a decreasing mortality with increasing age in early life, a minimal mortality at some age during adulthood and finally an increasing mortality with age late in life (Caughley 1966), the Gompertz model does not allow the full range of age-specific changes in survival to be modelled. The Siler model (Siler 1979) explicitly accounts for different age-specific changes in survival among early-life, prime-age and late-life stages by involving five parameters to fit three different functions of survival variation. More flexible bathtub models of senescence, which were initially developed to model age-specific changes in human mortality (see e.g. Bebbington et al. 2007), nowadays provide a promising way to get a reliable picture of the full age dependence in survival. These models can be easily implemented in free

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:32:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.007

131

Trait value

Assessing Actuarial Senescence in Wild Mammals

Age Figure 7.2

Age

Age

Assessing patterns of actuarial senescence in mammals: two metrics instead of one. While most analyses of senescence have focused on the strength of senescence (measured by the rate of senescence), the timing of senescence (measured by the onset of senescence) should also be investigated.

software such as E-Surge (Choquet and Nogué 2011) or BaSTA (Colchero et al. 2012). Bathtub models can be parameterised so as to give an estimate of the onset of senescence, providing information about the timing of senescence in addition to its strength measured by the actuarial senescence rate (Choquet et al. 2011). However, all these models require high-quality data on age-specific changes in survival to provide reliable estimates of senescence rates. While previous analyses of actuarial senescence have focused almost entirely on the rate of senescence (e.g. Ricklefs 2010), the age at onset of senescence provides a complementary metric for studying senescence that should also be considered (Figure 7.2), especially in the context of comparative analyses of senescence across species. For instance, a recent analysis in wild boar (Sus scrofa), a polygynous ungulate species that exhibits a unusual life history combining high fecundity and early age at primiparity with a potentially long life span (Focardi et al. 2008), has shown that the onset of actuarial senescence, rather than the rate of ageing, differs from that of other ungulate species (Gamelon et al. 2014). Comparative analyses of actuarial senescence across mammals in the wild are still very scarce. Since the fifties, most previous comparative inter-specific analyses have focused on longevity (de Magalhaes et al. 2007; Sacher 1959; Stearns 1983; Wilder et al. 2013), likely because of the high availability of these types of data (Carey & Judge 2000; de Magalhaes & Costa 2009). Some studies have been conducted using adult life expectancy (Gaillard et al. 1989) or survival rates (Promislow et al. 1990; Toïgo & Gaillard 2003), and the only few broad-scale studies that analysed the rate of senescence were either based on poor-quality data issued from life tables without accounting for data quality (Promislow 1991) or on heterogeneous data including a mixture of wild and captive populations and/or males and females (Ricklefs 2010). Nowadays, the increasing availability of high-quality estimates of both onset and rate of actuarial senescence should allow comparative analyses that account for phylogenetic inertia to be performed while also controlling for potentially confounding factors such as sex, body mass and diet. For instance, a comparative analysis across twenty-two species of large herbivores showed that captive populations have a lower rate of senescence than their wild counterparts (Lemaître et al. 2013) once correcting for

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:32:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.007

132

Senescence in Mammalian Life History Traits

body mass (to account for decreased rate of senescence with increasing mass) (Jones et al. 2008), diet (measured as the proportion of grass in the natural diet to account for the larger difficulty of maintaining browsers than grazers in optimal conditions in captivity) (Müller et al. 2010) and the quality of the data (to account for higher reliability of capture-recapture estimates compared to simple enumeration) (Péron et al. 2010). While empirical evidence recently accumulated from long-term detailed studies clearly demonstrates that actuarial senescence is pervasive in mammals, the factors shaping the observed variation in the intensity of senescence, in terms of either strength (rate of senescence) or timing (onset of senescence), remain to be identified. Besides the general influence of body size (with a decreasing intensity of senescence with increasing size) (Ricklefs 2010) and of the pace of life (with an increasing intensity of senescence with a faster pace of life) (Garratt et al. 2013; Jones et al. 2008; Lemaître & Gaillard 2013a; Péron et al. 2010; Ricklefs 2010), one knows very little about the influence of environmental factors (Do mammals living at high elevations have lower intensity of actuarial senescence than those living at low elevations? Do mammals living in the tropics have lower intensity of actuarial senescence than those living in temperate or arctic areas?) or social factors (Do mammals with complex social bonds have lower intensity of actuarial senescence than solitary mammals?) for shaping among-species variation in senescence patterns of mammals. So far, only a few recent studies have started to evaluate the influence of sexual selection on the diversity of senescence patterns currently observed in mammals (e.g. Tidière et al. 2014). For instance, a comparative study based on twenty-four species of large herbivores failed to detect any relationship between measures of male allocation to sexual competition (i.e. sexual size dimorphism, weapon size and relative testes mass) and actuarial senescence rates (Lemaître & Gaillard 2013b). However, this sample of large herbivores included species with relatively similar reproductive tactics, and evidence that male life expectancy is shorter than female life expectancy in polygynous mammals but is not in monogamous mammals (Clutton-Brock & Isvaran 2007) suggests that the role of sexual selection in shaping species-specific patterns of senescence might be detectable only at a very broad scale. There is thus now a great need to target this question using a wide range of contrasted mammals before drawing any definitive conclusion on a possible link between intensity of sexual selection and actuarial senescence patterns across species. Within a given species, one knows even less about the factors driving variation in senescence among populations. The ecological context could influence the intensity of actuarial senescence. Most captive populations of ruminants enjoy smaller rates of senescence than their wild counterparts (Lemaître et al. 2013), and senescence seems to be slower in island populations than in mainland ones (Austad 1993). Under classical evolutionary theories of senescence, any environmental factor causing increased fecundity associated with increased mortality in early adulthood should lead to accelerated senescence (Hamilton 1966; Williams 1957). Thus, mammals in populations under strong predation pressure or facing especially harsh climatic conditions should senesce faster than their counterparts facing more favourable conditions. There is, to our knowledge, no comparative analyses that have tested this Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:32:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.007

Assessing Actuarial Senescence in Wild Mammals

133

prediction in mammals, even across species, but recent inter-specific studies of bird longevity have provided supporting evidence (Valcu et al. 2014). However, we still lack a full understanding of the link between the amount of environmentally driven mortality and the intensity of senescence. This relationship might be complex because the type of mortality (i.e. whether mortality is selective or affects equally all individuals within a population) also shapes senescence patterns (Abrams 1993; Chen & Maklakov 2012; Williams et al. 2006). Variation in senescence patterns among individuals within a given population has been neglected for a long time. Detailed analyses of red deer (Cervus elaphus) on Rum Island have revealed high individual variation in the intensity of senescence. Females that were born at high population density suffered from earlier and greater actuarial senescence than their conspecifics that were born at low density (Nussey et al. 2007). From the few studies that have looked for among-individual variation in actuarial senescence, it appears that individual mammalian females that allocate more to reproduction early in their adulthood pay the cost at old age in terms of decreased survival, providing support for the disposable soma theory initially proposed by Kirkwood (1977), as recently reviewed by Lemaître et al. (2015). Such delayed reproductive costs in terms of survival have been so far reported in Asian elephants (Elephas maximus) (Hayward et al. 2014), red deer (Nussey et al. 2007), Weddell seals (Leptonychotes weddellii) (Hadley et al. 2007), Rhesus macaques (Macaca mulatta) (Blomquist 2009) and reindeer (Rangifer tarandus) (Weladji et al. 2008). The large amount of evidence for actuarial senescence in mammals has accumulated without accounting for individual heterogeneity in quality. The process of viability selection is expected to remove disproportionately individuals with a selective disadvantage, often characterised by a combination of low reproductive output, low survivorship and low body size. These individuals correspond to poor-quality individuals (sensu Wilson & Nussey 2010). Such individual differences are likely to mask senescence and have thus to be accounted for. Since mortality is a unique event in an individual’s life, correcting for individual heterogeneity cannot be based on repeated measurements at the individual level. Models including a mixture of high- and low-quality individuals like those used initially by Vaupel and Yashin (1985) offer a promising way to account for individual differences in survivorship. We thus expect the rate of actuarial senescence in mammals to be higher than reported up to now, especially in small species for which a higher magnitude of individual differences in survivorship is expected to occur (Péron et al. 2016). Another way to look for individual heterogeneity in survivorship might involve considering mortality causes. Taking advantage of technological improvements in monitoring individual fates (e.g. radio-telemetry or GPS monitoring) (Cagnacci et al. 2010), we should accumulate knowledge about individual mortality causes and include such heterogeneity to assess actuarial senescence. For instance, a preliminary analysis of age-specific survival in females of an intensively monitored roe deer population has shown that disentangling human-related mortalities and mortalities caused by natural factors led to increases in the rate of actuarial senescence, whereas no age variation occurred in human-related mortality (Koons et al. 2014).

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:32:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.007

134

Senescence in Mammalian Life History Traits

Assessing Reproductive Senescence in Wild Mammals Compared to actuarial senescence, reproductive senescence has been often overlooked. To the notable exception of humans and a handful of very long-lived species like killer whales, elephants and anthropoid primates, reproductive cessation (sensu Packer et al. 1998) is expected to be weak in mammals, and the process of reproductive senescence has not retained much attention. For instance, Caughley (1976) claimed that there is ‘no hint of the sharp decrease of fecundity at old age’ in large herbivores. Assessing reproductive senescence in mammals is much easier than assessing actuarial senescence, at least in females. Female reproductive status can be measured at different stages of the reproductive cycle (i.e. at mating, during pregnancy, at birth and at weaning) from captures, hunting or observations. Empirical evidence of reproductive senescence has been reported in two types of studies. Firstly, the number of female offspring produced per female at different ages has been tabulated (as mx values) (Seber 1973) in life tables built on data collected through hunting (game species) or trapping (small mammals, small to medium-sized carnivores). Contrary to actuarial senescence, evidence of reproductive senescence from life tables is far from being the rule. In more than half of these case studies, reproductive senescence has simply not been looked for, and the female reproductive output has been assumed to be constant from the mid-adulthood. Only the increase of reproductive output from age of primiparity to prime ages has been reported in such cases (e.g. Slade & Balph (1974) on Spermophilus armatus; Rodgers (1984) on Phacochoerus aethiopicus or Dinerstein (1991) on Rhinoceros unicornis). Reproductive senescence has also been assessed from detailed individual monitoring of known-age animals from birth to death in a large range of mammals (Table 7.2). From these studies, reproductive senescence displays different patterns according to the reproductive component analysed (Table 7.2). For instance, implantation failure in roe deer increases from a probability of 0.2 in females aged between two and seven years to 0.5 in females older than eight years of age (Hewison & Gaillard 2001), whereas pregnancy rates remain high and constant (>0.95) up to eleven years of age, after which they strongly decrease to below 0.5 from fourteen years onwards (J. M. Gaillard and collaborators, unpublished data). Similar patterns of variation have been reported in reindeer (Milner et al. 2003). Reproductive senescence thus corresponds to a complex process that involves agerelated declines in the efficiency of one or more functions or structures involved in the reproduction process. Thus, different reproductive traits do not consistently show similar senescence patterns across related mammalian species. For example, offspring performance (in terms of birth mass or survival) decreases with increasing female age in Soay sheep (Ovis aries) (Hayward et al. 2009), reindeer (Weladji et al. 2010) and whitetailed deer (Odocoileus virginianus) (DelGiudice 2007), whereas it does not change with age in bighorn sheep (Ovis canadensis) (Bérubé et al. 1999); pregnancy rate decreases with age in baboon (Papio anubis) (Packer et al. 1998), red deer (Nussey et al. 2009a), roe deer (J. M. Gaillard and collaborators, unpublished data) and reindeer (Rangifer

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:32:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.007

Assessing Reproductive Senescence in Wild Mammals

135

Table 7.2 Evidence of Reproductive Senescence in Female Mammals from Long-Term Studies Species

Reference

Moose (Alces alces)

Heard et al. (1997): fertility; Ericsson et al. (2001): litter size/offspring survival Lima & Paez (1997): reproductive rate Lunn et al. (1994): reproductive rate Beauplet et al. (2006): reproductive rate Green (1990): breeding proportion; Wilson et al. (2002): reproductive rate Lander (1981): pregnancy rate Toïgo et al. (2002): reproductive rate Hewison & Gaillard (2001): implantation failure; Gaillard et al. (2003): pregnancy rate/litter size Flook (1970): pregnancy rate Clutton-Brock et al. (1982): reproductive rate Saltz (1996): reproductive rate Austad (1993): fertility Robinson et al. (2012): reproductive rate Garrott et al. (1991): breeding proportion Holmes et al. (2007): reproductive rate Foote (2008): fertility

South American fur seal (Arctocephalus australis) Antarctic fur seal (Arctocephalus gazella) Sub-Antarctic fur seal (Arctocephalus tropicalis) Bison (Bison bison) Northern fur seal (Callorhinus ursinus) Ibex (Capra ibex) Roe deer (Capreolus capreolus) Elk (Cervus elaphus canadensis) Red deer (Cervus elaphus) Persian fallow deer (Dama mesopotamica) Virginia opossum (Delphinus virginiana) Asian elephant (Elephas maximus) Horse (Equus caballus) Steller sea lion (Eumatopias jubatus) Short-finned pilot whale (Globicephala macrorhynchus) Long-finned pilot whale (Globicephala melaena) Mountain gorilla (Gorilla beringei) Gray seal (Halichoerus grypus) Ring-tailed lemur (Lemur catta) African elephant (Loxodonta africana) Long-tailed macaque (Macaca fascicularis) Japanese macaque (Macaca fuscata) Rhesus macaque (Macaca mulatta)

Foote (2008): fertility Robbins et al. (2006): reproductive rate Bowen et al. (2006): birth rate Sussman (1991): birth rate/infant mortality Hanks (1972): reproductive rate Koyama et al. (1992): reproductive rate Wolfe & Noyes (1981): birth rate Hoffman et al. (2010): interbirth interval/offspring survival Barbary macaque (Macaca sylvanus) Paul et al. (1993): reproductive rate Mandrill (Mandrillus sphinx) Leigh et al. (2008): reproductive rate Alpine marmot (Marmota marmota) Tafani et al. (2013): litter size Badger (Meles meles) Dugdale et al. (2011): annual fecundity Northern elephant seal (Mirounga angustirostris) Sydeman et al. (1991): reproductive rate Mediterranean monk seal (Monachus schauinslandii) Harting et al. (2007): reproductive rate Mule deer (Odocoileus hemionus) Lawrence et al. (2004): pregnancy rate Killer whale (Orcinus orca) Olesiuk et al. (1990): reproductive rate Mountain goat (Oreamnos americana) Bailey (1991): reproductive rate Soay sheep (Ovis aries) Robinson et al. (2006): reproductive rate Bighorn sheep (Ovis canadensis) Bérubé et al. (1999): birth rate Chimpanzee (Pan troglodytes) Sugiyama (1994): birth rate Lion (Panthera leo) Packer et al. (1998): birth rate/offspring survival Hamadryas baboon (Papio hamadryas) Packer et al. (1998): birth rate White-footed mouse (Peromyscus leucopus) Morris (1996): litter size New Zealand sea lion (Phocarctos hookeri) Childerhouse et al. (2010): reproductive rate Verreau’s sifaka (Propithecus verreauxi) Richard et al. (2002): birth rate/offspring survival Reindeer (Rangifer tarandus) Adams & Dale (1998): fertility/reproductive pause Pyrenean chamois (Rupicapra pyrenaica) Crampe et al. (2006): reproductive rate

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:32:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.007

136

Senescence in Mammalian Life History Traits

Table 7.2 (cont.) Species

Reference

Chamois (Rupicapra rupicapra) Columbian ground squirrel (Spermophilus columbianus) Meerkat (Suricata suricatta)

Tettamanti et al. (2015): reproductive rate Broussard et al. (2003): litter success

Wild boar (Sus scrofa) African buffalo (Syncerus caffer) Red squirrel (Tamiasciurus hudsonicus) Gelada (Theropithecus gelada) Brown bear (Ursus arctos) Polar bear (Ursus maritimus) California sea lion (Zalophus californicus)

Sharp & Clutton-Brock (2010): litter size/offspring survival Aumaître et al. (1982): fertility Sinclair (1977): pregnancy rate McAdam et al. (2007): reproductive rate Dunbar (1980): birth rate Schwartz et al. (2003): reproductive rate Derocher & Stirling (1994): litter size Hernandez-Camacho et al. (2008a)

Note: The focal mammalian species, the focal reproductive trait and one illustrative reference are provided (from Nussey et al. 2013, modified).

tarandus) (Milner et al. 2003) but not in white-tailed deer (DelGiudice et al. 2006); and litter size decreases with increasing age from four years onwards in meerkats (Suricata suricatta) (Sharp & Clutton-Brock 2010), from six years onwards in roe deer (J. M. Gaillard and collaborators, unpublished data) and from ten years onwards in Alpine marmots (Marmota marmot) (Berger et al. 2015) but remains constant with age in white-tailed deer (DelGiudice et al. 2006), mule deer (Odocoileus hemionus) (Monteith et al. 2014), pronghorn (Antilocapra americana) (Byers 1997) and Soay sheep (Hayward et al. 2009). Why do such different patterns of age-specific variation occur among reproductive traits? The reproductive output at each stage of the reproductive cycle is tightly linked with the energy allocated to reproduction, and trade-offs are likely to occur. For instance, the offspring size-number trade-off is one of the most intensively studied trade-offs (Smith & Fretwell 1974; Stearns 1992). According to the Y model proposed by van Noordwijk & de Jong (1986), the direction of the trade-off should depend on the balance between variability in resource acquisition, which includes both resource availability and individual ability to exploit resources, and variability in resource allocation. When allocation is more variable among individuals than acquisition, a trade-off occurs. Finally, we regret that most published studies so far have focused on females, with only a few investigations of reproductive senescence occurring in mammalian males. Such bias probably arises from the fact that genetic data are required to determine reproductive success in males, while this is not automatically the case for females. However, males also suffer from reproductive senescence, as nicely revealed by a thorough study conducted on male red deer, in which breeding success declines from the age of ten years onwards (Nussey et al. 2009a). Similarly to females, male reproductive senescence can be complex. For instance, senescence in harem size is faster in male red deer that control the largest harems and spend the most time rutting during early adulthood (Lemaître et al. 2014). Male reproductive senescence can also involve an age-dependent decline in secondary Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:32:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.007

Assessing Reproductive Senescence in Wild Mammals

137

sexual traits, as reported in roe deer, for which antler size (a trait positively associated with reproductive success) (Vanpé et al. 2010) decreases after males reach eight years of age (Vanpé et al. 2007). Note that for red deer, whether antler size decreases with increasing age is not so clear (Mysterud et al. 2005 versus Nussey et al. 2009a). Alternatively, a decrease in male fertilisation efficiency at old ages might also account for reproductive senescence. Until now, studies that have documented senescence in ejaculate quality in mammals have focused on captive populations (Crosier et al. 2007; Thongtip et al. 2008) because it is particularly challenging to collect longitudinal data on ejaculate quality in the wild. The few published data are thus cross sectional (see Curren et al. (2013) for a study case in spotted hyena, Crocuta crocuta), which might explain why senescence in sperm-related traits has not been documented yet in free-ranging mammalian populations. As virtually all female mammals are iteroparous, reproductive traits can be measured repeatedly over the lifetime of a given individual, which makes it easy to correct for individual heterogeneity using generalised linear mixed models (van de Pol and Verhulst 2006; van de Pol and Wright 2009). However, accounting for confounding effects of selective disappearance requires repeated measures of the same individuals, which are available only from longitudinal studies. Age-specific reproduction assessed from transversal data often collected from reproductive tractus of dead females are likely to underestimate severely the intensity of reproductive senescence and should thus be taken with great caution. Observed changes in reproductive output between prime-age and old individuals are likely to be shaped by different processes. Senescence is obviously one process, but prime-age and old reproducers might have different strategies of energy allocation to reproduction (Berger et al. 2015). Terminal investment (Clutton-Brock 1984) and terminal allocation (Weladji et al. 2010) have been shown to increase reproductive allocation at old ages and counter-balance thereby the influence of reproductive senescence, even though the two processes co-occur (Hamel et al. 2012; Weladji et al. 2010). Thus, we can expect trade-offs between resource acquisition and allocation to reproduction to vary among ages. Are trade-offs age dependent? Surprisingly, very few empirical studies have tackled this question in mammals or in any other taxonomic group, and further investigation is necessary to understand what is the relative contribution of senescence in age-specific variation of reproductive traits at old ages. Both the onset and the rate of reproductive senescence isometrically scale with physiological time (sensu Linstedt and Calder 1981). In support, available data indicate that both the timing and the strength of senescence strongly depend on generation time (Garratt et al. 2013; Jones et al. 2008), a reliable metric of variation in the pace of life across mammals (Gaillard et al. 2005). For a given pace of life, we thus expect to find only little variation among mammalian orders in either the onset or the rate of senescence. The rapid increase in age-specific survival and reproductive data is providing the material to test the hypothesis that actuarial and reproductive senescence patterns should be synchronous (Williams 1957). In his thorough review of post-reproductive life span, Cohen (2004) proposed different scenarios of age-specific patterns in reproduction and survival to assess when post-reproductive life span might evolve (Figure 7.2). However, Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:32:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.007

138

Senescence in Mammalian Life History Traits

16 14 Age (years)

12 10 8 6

Reproduction

4

Survival

2 se oo M

ou ar ib C

gh

er Bi

ed R

oe

de

ee R

nd ei R

de

er

r

t oa t.G M Figure 7.3

or n

0

Onsets of actuarial and reproductive senescence estimated in a selected set of large herbivore species for which an intensive individual monitoring was available.

Mt goat (Oreamnos americana): Festa-Bianchet and Côté (2008) Reindeer (Rangifer tarandus): Albon et al. (2002) Roe deer (Capreolus capreolus): Gaillard et al. (1998) Red deer (Cervus elaphus): Clutton-Brock et al. (1982) Bighorn sheep (Ovis canadensis): Jorgenson et al. (1997) and Bérubé et al. (1999) Caribou (Rangifer tarandus): Adams and Dale (1998) and Thomas and Barry (1990) Moose (Alces alces): Ericsson and Wallin (2001) and Ericsson et al. (2001) the link between post-reproductive life span and age-specific variation in reproduction is more complex than expected. Indeed, an earlier age of reproductive cessation could involve either earlier or stronger reproductive senescence. As most studies (including Cohen’s (2004) scenarios displayed in Figure 7.2) focused on the rate of senescence and have neglected the onset of senescence, the interplay between the onset and the rate of reproductive senescence is still poorly understood. Although evidence from laboratory studies suggests that patterns of senescence across life history traits are uncoupled (Nussey et al. 2013; Promislow et al. 2006; Walker & Herndon 2010), few studies have yet directly targeted this question in the wild (but see Hayward et al. (2015) for a notable recent exception in Soay sheep). Nonetheless, results from separate studies but published on the same species might provide a few elements to tackle this question, although it is noteworthy that data sets used in these studies are evidently not strictly identical. So far, preliminary results in large herbivores suggest that the onset of reproductive senescence occurs later than the onset of actuarial senescence (Figure 7.3).

Evidence of Senescence in Other Mammalian Life History Traits Body Mass in Wild Mammals When looking for the underlying factors potentially explaining actuarial and reproductive senescence in wild mammals, the most obvious trait is probably the loss of body Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:32:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.007

Evidence of Senescence in Other Mammalian Life History Traits

139

mass. The strength of the relationship between body mass and body condition varies a lot across species. For instance, while body condition is almost perfectly measured using body mass in roe deer (Toïgo et al. 2006), a larger variation in mass at a given size occurs in the Columbian ground squirrel, Spermophilus columbianus (Dobson 1992). However, body mass and body condition are positively correlated in all mammals studied so far, and as a rule, heavier mammalian females survive better, give birth at an earlier age and produce more and larger offspring than lighter females (Clutton-Brock 1988; Gaillard et al. 2000; Sadleir 1969). Therefore, patterns of actuarial and reproductive senescence reported in the wild (Nussey et al. 2013) might be, in mammals, explained by a senescence in body condition. So far, senescence in body mass has already been documented in several species, principally large herbivores (e.g. Mysterud et al. (2001) in red deer; Weladji et al. (2010) in reindeer; and Nussey et al. (2011) in Soay sheep, roe deer and bighorn sheep) or seals (e.g. Proffitt et al. 2007). However, most of the published studies did not disentangle between-individual selective disappearance and within-individual changes in body mass in their analyses and thus potentially underestimated the magnitude of the body-mass decline with age. Indeed, if light individuals are more likely to die during early adulthood, old age classes will be constituted by the heaviest individuals of the population, which can make senescence more difficult to detect. Interestingly, studies controlling for selective disappearance processes have revealed contrasted patterns of senescence in body mass. For example, Nussey and colleagues found that the decrease in body mass accelerates with increasing age in roe deer, while body mass declines steadily only two years before individuals die in Soay sheep (Nussey et al. 2011). If the exact nature of these differences remains unknown, the authors suggested that possible interactions between phenotypic attributes and environmental conditions might play an important role. To go one step further, to understand the origin of body-mass senescence requires comparing patterns between males and females from the same species. Indeed, selective pressures are different between the sexes, which might ultimately lead to divergent patterns of senescence. For instance, in polygynous mammals, males allocate substantially to sexual competition through the defence of females or of territories and in some species through the growth of conspicuous weapons. It is thus possible to predict that in these species males should suffer from a much steeper decline in body mass than females, while in monogamous species, senescence patterns are expected to be similar between the sexes. To date, only a few detailed studies of sex-specific senescence in mass have been performed. Senescence patterns in mass of the grey mouse lemur (Microcebus murinus), a sexually monomorphic primate, did not differ between the sexes in either captive or wild populations (Hämäläinen et al. 2015), in support of the expectation. On the contrary, in the socially monogamous Alpine marmots (Marmota marmota), senescence in body mass was only observed in males. Male body mass declined from the age of eight years in association with an acceleration of the mass loss the year before death, but female body mass remained constant across the whole life (Tafani et al. 2013). In the European badger (Meles meles), males also showed stronger senescence in mass than females, likely in relation to intra-sexual competition during early life (Beirne et al. 2015). These results Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:32:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.007

140

Senescence in Mammalian Life History Traits

suggest that the factors underpinning senescence in body condition are likely to be complex and are yet to be understood. Obtaining longitudinal measures of body mass in the wild is difficult, especially for large species. However, such data are badly required to investigate at a broad scale the relationship between senescence in body condition and senescence in fitness-related traits.

Physiological Parameters in Wild Mammals Studies of senescence in physiological parameters in wild mammals are beginning to increase and offer an insight into mechanistic causes for increased mortality rates with age and further validation of the presence of senescence in the wild. Examination of physiological traits in wild animals is not easy, however, and is often restricted to biological markers that can be assayed in blood (Selman et al. 2012). Much of this work to date has been preliminary and has involved cross-sectional comparisons of individuals of different ages in a particular population. Limited information can be gained from studies of animals of only two age groups, such as between young and adults or young and old individuals, because it is not possible to determine those changes that occur as a consequence of senescence and those that change simply as a side effect of animals getting older. Perhaps more importantly, correlations between age and physiological trait expression can be generated by differential survival rates of animals of different genotypes and phenotypes in the wild, as was the case for a positive correlation between the concentration of anti-nuclear antibodies and age in wild Soay sheep (Graham et al. 2010). Target parameters for physiological senescence in wild mammals are usually based on those that have been implicated in ageing in laboratory and domesticated populations of animals, particularly mice and rats, the model mammalian organisms for biomedical science. Assessments of muscle physiology in wild animals of various age groups has confirmed that animals survive to an age where senescence can be detectable in aspects of muscle physiology, and these changes are largely consistent with muscle senescence reported in non-wild animals. In a range of wild cetacean species, animals judged to be approaching senescence via visual morphology show a change to slower muscle phenotype, which is largely consistent with the physiological muscle alterations observed in elderly humans – although the authors also point out that such alterations might be adaptive if they facilitate more prolonged underwater diving (Sierra et al. 2013). A more detailed analysis of muscle samples across forty-seven Weddell seals classified as ‘adult’ or ‘old’ further documented a range of changes in muscle physiology with age, including increased collagen build-ups, which the authors speculated might affect sprint capacity and contractile efficiency of these old individuals (Hindle et al. 2009a). Young-adult and prime-aged wild-caught individuals of two shrew species (Blarina brevicauda and Sorex palustris) confirmed age-related increases in collagen content in wild small mammals, highlighting that similar age-related changes are observed in shorter-lived terrestrial species (Hindle et al. 2009b). Changes in aspects of the immune system are known to occur with senescence in laboratory rodents, domesticated mammals and also humans. In the wild, where Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:32:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.007

Evidence of Senescence in Other Mammalian Life History Traits

141

parasites and pathogens are prevalent, senescence in the immune system could conceivably have a substantial impact on health and mortality (Maizels & Nussey 2013). Cross-sectional analysis of young, adult and old-aged wild Soay sheep have supported the notion that the immune system is impaired in senescent individuals in the wild, with alterations in T-cell subsets in old-age sheep largely mirroring the changes observed in laboratory rodents (Nussey et al. 2012). These age-related changes in the immune system might be a contributing factor to the elevation in rates of parasitism that occurs in sheep of this population as they age, demonstrated through a rare longitudinal analysis (Hayward et al. 2009). Ageing of the immune system and the potential costs of this in terms of parasite and pathogen burden might be one substantial factor contributing to increased rates of mortality in old-aged mammals living in the wild. Oxidative stress and an impaired ability to regulate defence mechanisms that protect against this are a central feature of senescence in laboratory rodents (Fraga et al. 1990; Sohal et al. 1995; Stadtman 1992), and there has been much investigation and discussion of the potential role of oxidative stress as a public cause of senescence across animals (Kirkwood & Austad 2000). Studies of oxidative stress and its relation to senescence in the wild are beginning to emerge, although the methodological issues in collection samples and their interpretation are numerous (Selman et al. 2012). Studies of wild Soay sheep and Eastern chipmunks (Tamias striatus) have explored the cross-sectional relationship between age and a marker of oxidative stress in blood. Surprisingly, while both studies noted an elevation in serum oxidative stress in young animals, old-aged individuals had similar levels of a serum marker of oxidative stress to middle-aged adults (Bergeron et al. 2011; Nussey et al. 2009b). In a more detailed analysis of oxidative stress in the muscles of wild-caught shrews, no consistent change in oxidative stress was observed: one marker increased with age, while another decreased (Hindle et al. 2010). Despite this, the activity of several antioxidants was consistently elevated in old individuals, which also occurs in the muscles of old laboratory rodents and is thought to be related to impaired transcriptional regulation and altered physiological responses to bouts of muscle exercise (Vasilaki et al. 2006). Thus, although the very limited assessment of oxidative stress to date has not revealed large increases in oxidative stress with old age, a more detailed analysis of redox function across a greater variety of tissues may prove enlightening. Changes in physiological parameters with age are detectable in wild populations of mammals (e.g. see Jégo et al. (2014) for a recent case study in roe deer), although research in this area is still in its infancy. In contrast to research in wild birds, which have been studied more intensively, some key aspects of physiology linked to senescence have not yet been assessed in wild mammals, including telomere attrition and production of inflammatory cytokines. Moving forward, it will be important to understand whether senescence-related changes in these physiological traits also occur in wild mammals. It is notable, however, that there is a growing appreciation that assessment of individual markers of physiological traits does not capture the complexity of physiological decline and loss of condition in wild animals (Cohen 2015; Milot et al. 2014), particularly since different markers of traits such as oxidative stress and inflammation can be uncorrelated (see Christensen et al Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:32:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.007

142

Senescence in Mammalian Life History Traits

(2015) for a recent example in Soay sheep). A more comprehensive approach of assessing multiple different markers of multiple aspects of physiology may provide a more robust and repeatable means to assess physiological dysregulation and identify senescent individuals in the wild and could include those aspects of physiology yet to be assessed. Once a broader understanding of the patterns of physiological deterioration in wild animals has been made, we may be able to move to the more complicated issue of which of these changes is causally involved in generating decreased reproduction and increased mortality with age.

Conclusion Empirical evidence clearly demonstrates that senescence is pervasive in most mammalian species studied to date. Patterns of senescence are trait dependent but seem to be shaped by the same major sources of variation. The intensity of sexual selection, the environmental context and some physiological and genetic mechanisms are the most structuring factors shaping mammalian senescence within and between species. Senescence is expected to have strong implications for conservation and management of mammalian populations. In particular, the occurrence of actuarial and reproductive senescence is expected to lead to a decrease in individual fitness and population growth rates, which has to be accounted for when modelling populations. Thus, thirty years ago, Lee Eberhardt (1985) stressed the need of including senescence when assessing population dynamics. However, to date, the fitness loss due to actuarial and reproductive senescence, and thus the costs of senescence in terms of population dynamics, remains poorly understood. A pioneering study on thirteen populations of twelve species of birds and mammals indicates that the senescence costs could be far from negligible (Bouwhuis et al. 2012). Future studies will be required to measure the impact of senescence on population dynamics in a wide range of mammals and to identify potential drivers of such senescence costs.

References Abrams, P. A. (1993). Does increased mortality favor the evolution of more rapid senescence? Evolution, 47, 877–87. Adams, L. G. & Dale, B. W. (1998). Reproductive performance of female Alaskan caribou. Journal of Wildlife Management, 62, 1184–95. Albon, S. D., Stien, A., Irvine, R. J., et al. (2002). The role of parasites in the dynamics of a reindeer population. Proceedings of the Royal Society of London Series B: Biological Sciences, 269, 1625–32. Aumaître, A., Morvan, C., Quéré, J. P., et al. (1982). Productivité potentielle et reproduction hivernale chez la laie (Sus scrofa scrofa) en milieu sauvage. Journée de la Recherche Porcine, 14, 109–24.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:32:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.007

References

143

Austad, S. N. (1993). Retarded senescence in an insular population of Virginia opossums (Didelphis virginiana). Journal of Zoology, 229, 695–708. Baker, J. D. & Thompson, P. M. (2007). Temporal and spatial variation in age-specific survival rates of a long-lived mammal, the Hawaiian monk seal. Proceedings of the Royal Society of London Series B: Biological Sciences, 274, 407–15. Bailey, J. A. (1991). Reproductive success in female mountain goats. Canadian Journal of Zoology, 69, 2956–61. Beauplet, G., Barbraud, C., Dabin, W., Kussener, C. & Guinet, C. (2006). Age-specific survival and reproductive performances in fur seals: evidence of senescence and individual quality. Oikos, 112, 430–41. Bebbington, M., Lai, C. & Zitikis, R. (2007). Modeling human mortality using mixtures of bathtub shaped failure distributions. Journal of Theoretical Biology, 245, 528–38. Beirne, C., Delahay, R. & Young, A. (2015). Sex differences in senescence: the role of intra-sexual competition in early adulthood. Proceedings of the Royal Society of London Series B: Biological Sciences, 282, 20151086. Berger, V., Lemaître, J. F., Gaillard, J.-M. & Cohas, A. (2015). How do animals optimize the size-number trade-off when aging? Insights from reproductive senescence patterns in marmots. Ecology, 96, 46–53. Berger, V., Lemaître, J. F., Gaillard, J.-M. & Cohas, A. (2016). Age-specific survival in the socially monogamous alpine marmot (Marmota marmota): evidence of senescence. Journal of Mammalogy 97, 992–1000. Bergeron, P., Careau, V., Humphries, M. M., et al. (2011). The energetic and oxidative costs of reproduction in a free-ranging rodent. Functional Ecology, 25, 1063–71. Bérubé, C. H., Festa-Bianchet, M. & Jorgenson, J. T. (1999). Individual differences, longevity, and reproductive senescence in bighorn ewes. Ecology, 80, 2555–65. Bishop, C. J., White, G. C., Freddy, D. J., et al. (2009). Effect of enhanced nutrition on mule deer population rate of change. Wildlife Monographs, 172, 1–28. Blomquist, G. E. (2009). Trade-off between age of first reproduction and survival in a female primate. Biology Letters, 5, 339–42. Bouwhuis, S., Choquet, R., Sheldon, B. C. & Verhulst, S. (2012). The forms and fitness cost of senescence: age-specific recapture, survival, reproduction, and reproductive value in a wild bird population. American Naturalist, 179, E15–27. Bowen, W. D., Iverson, S. J., McMillan, J. I. & Boness, D. J. (2006). Reproductive performance in grey seals: age-related improvement and senescence in a capital breeder. Journal of Animal Ecology, 75, 1340–51. Bradley, A. J. (1997). Reproduction and life history in the red-tailed phascogale, Phascogale calura (Marsupialia: Dasyuridae): the adaptive-stress senescence hypothesis. Journal of Zoology, 241, 739–55. Bronikowski, A. M., Altmann, J., Brockman, D. K., et al. (2011). Aging in the natural world: comparative data reveal similar mortality patterns across primates. Science, 331, 1325–8. Broussard, D. R., Michener, G. R., Risch, T. S. & Dobson, F. S. (2005). Somatic senescence: evidence from female Richardson’s ground squirrels. Oikos, 108, 591–601. Broussard, D. R., Risch, T. S., Dobson, F. S. & Murie, J. O. (2003). Senescence and age-related reproduction of female Columbian ground squirrels. Journal of Animal Ecology, 72, 212–19.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:32:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.007

144

Senescence in Mammalian Life History Traits

Byers, J. A. (1997). American Pronghorn: Social Adaptations and the Ghosts of Predators Past (University of Chicago Press). Cagnacci, F., Boitani, L., Powell, R. A. & Boyce, M. S. (2010). Animal ecology meets GPS-based radiotelemetry: a perfect storm of opportunities and challenges. Philosophical Transactions of the Royal Society of London Series B: Biological Sciences, 365, 2157–62. Carey, J. R. & Judge, D. S. (2000). Longevity Records: Life Spans of Mammals, Birds, Reptiles, Amphibians and Fish (Odense Monographs on Population Aging) (Odense University Press). Catchpole, E. A., Morgan, B. J. T., Coulson, T. N., et al. (2000). Factors influencing Soay sheep survival. Journal of the Royal Statistical Society Series C: Applied Statistics, 49, 453–72. Caughley, G. (1966). Mortality patterns in mammals. Ecology, 47, 906–18. Caughley, G. (1976). Wildlife management and the dynamics of ungulate populations. In Applied Biology, Vol. 1 (pp. 183–246), ed. T. H. Coaker (London: Academic Press). Chen H. Y. & Maklakov A. A. (2012). Longer life span evolves under high rates of condition-dependent mortality. Current Biology, 22, 2140–3. Childerhouse, S. J., Dawson, S. M., Fletcher, D. J., et al. (2010). Growth and reproduction of female New Zealand sea lions. Journal of Mammalogy, 91, 165–76. Chilvers, B. L., Wilkinson, I. S. & Mackenzie, D. I. (2010). Predicting life-history traits for female New Zealand sea lions, Phocarctos hookeri: integrating short-term mark-recapture data and population modeling. Journal of Agricultural Biological and Environmental Statistics, 15, 259–78. Choquet, R. & Nogué, E. (2011). E-SURGE 1–8 User’s Manual (CEFE, UMR 5175, Montpellier, France), available at http://ftp.cefe.cnrs.fr/biom/soft-cr/. Choquet, R., Viallefont, A., Rouan, L., et al. (2011). A semi-Markov model to assess reliably survival patterns from birth to death in free-ranging populations. Methods in Ecology and Evolution, 2, 383–9. Christensen, L. L., Selman, C., Blount, J. D., et al. (2015). Plasma markers of oxidative stress are uncorrelated in a wild mammal. Ecology and Evolution, 5, 5096–5108. Clutton-Brock, T. H., Guinness, F. E. & Albon, S. D. (1982). Red Deer: Ecology and Behaviour of Both Sexes (University of Chicago Press). Clutton-Brock, T. H. (1984). Reproductive effort and terminal investment in iteroparous animals. American Naturalist, 123, 212–29. Clutton-Brock, T. H. (1988). Reproductive Success: Studies of Individual Variation in Contrasting Breeding Systems (University of Chicago Press). Clutton-Brock, T. H., Albon, S. D. & Guinness, F. E. (1988). Reproductive success in male and female red deer. In Reproductive Success: Studies of Individual Variation in Contrasting Breeding Systems (pp. 325–43), ed. T. H. Clutton-Brock (University of Chicago Press). Clutton-Brock, T. H. & Isvaran, K. (2007). Sex differences in ageing in natural populations of vertebrates. Proceedings of the Royal Society of London Series B: Biological Sciences, 274, 3097–3310. Clutton-Brock, T. H. & Sheldon, B. C. (2010). Individuals and populations: the role of long-term, individual-based studies of animals in ecology and evolutionary biology. Trends in Ecology and Evolution, 25, 562–73.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:32:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.007

References

145

Cohen, A. A. (2004). Female post-reproductive lifespan: a general mammalian trait. Biological Reviews, 79, 733–50. Cohen, A. A. (2015). Complex systems dynamics in aging: new evidence, continuing questions. Biogerontology, 17, 1–16. Colchero, F., Jones, O. R. & Rebke, M. (2012). BaSTA: an R package for Bayesian estimation of age-specific survival from incomplete mark-recapture/recovery data with covariates. Methods in Ecology and Evolution, 3, 466–70. Comfort, A. (1979). The Biology of Senescence (London: Churchill Livingston). Crampe, J. P., Loison, A., Gaillard, J.-M., et al. (2006). Monitoring of the reproduction in isard females (Rupicapra pyrenaica pyrenaica) in a non-hunted population and demographic consequences. Canadian Journal of Zoology, 84, 1263–8. Crosier, A. E., Marker, L., Howard, J., et al. (2007). Ejaculate traits in the Namibian cheetah (Acinomyx jubatus): influence of age, season and captivity. Reproduction Fertility and Development, 19, 370–82. Curren, L. J., Weldele, M. L. & Holekamp, K. E. (2013). Ejaculate quality in spotted hyenas: intraspecific variation in relation to life-history traits. Journal of Mammalogy, 94, 90–9. Davis, W. H. (1966). Population dynamics of bat Pipistrellus subflavus. Journal of Mammalogy, 47, 383. Deevey, E. S. (1947). Life tables for natural populations of animals. Quarterly Review of Biology, 22, 283–314. Delean, J. S. C. (2007). Longitudinal Population Demography of the Allied Rock Wallaby, Petrogale assimilis. Unpublished PhD dissertation, James Cook University, Australia. de Magalhaes, J. P. & Costa, J. (2009). A database of vertebrate longevity records and their relation to other life-history traits. Journal of Evolutionary Biology, 22, 1770–4. de Magalhaes, J. P., Costa, J. & Church, G. M. (2007). Analysis of the relationship between metabolism, developmental schedules, and longevity using phylogenetic independent contrasts. Journals of Gerontology Series A, 62, 149–60. Delgiudice, G. D., Fieberg, J., Riggs, M. R., et al. (2006). A long-term age-specific survival analysis of female white-tailed deer. Journal of Wildlife Management, 70, 1556–68. DelGiudice, G. D., Lenarz, M. S. & Powell, M. C. (2007). Age-specific fertility and fecundity in northern free-ranging white-tailed deer: evidence for reproductive senescence? Journal of Mammalogy, 88, 427–35. Derocher, A. E. & Stirling, I. (1994). Age-specific reproductive performance of female polar bears (Ursus maritimus). Journal of Zoology, 234, 527–36. Descamps, S., Boutin, S., Berteaux, D. & Gaillard, J.-M. (2007). Female red squirrels fit Williams’ hypothesis of increasing reproductive effort with increasing age. Journal of Animal Ecology, 76, 1192–1201. Dinerstein, E. (1991). Demography and habitat use by greater one-horned rhinoceros in Nepal. Journal of Wildlife Management, 55, 401–11. Dobson, F. S. (1992). Body mass, structural size, and life-history patterns of the Columbian ground squirrel. American Naturalist, 140, 109–25. Dugdale, H. L., Pope, L. C., Newman, C., et al. (2011). Age-specific breeding success in a wild mammalian population: selection, constraint, restraint and senescence. Molecular Ecology, 20, 3261–74.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:32:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.007

146

Senescence in Mammalian Life History Traits

Dunbar, R. I. M. (1980). Demographic and life-history variables of a population of Gelada baboons (Theropithecus, gelada). Journal of Animal Ecology, 49, 485–506. Eberhardt, L. L. (1985). Assessing the dynamics of wild populations. Journal of Wildlife Management, 49, 997–1012. Emlen, J. M. (1970). Age specificity and ecological theory. Ecology, 51, 588–601. Ericsson, G. & Wallin, K. (2001). Age-specific moose (Alces alces) mortality in a predator-free environment: evidence for senescence in females. Ecoscience, 8, 157–63. Ericsson, G., Wallin, K., Ball, J. P. & Broberg, M. (2001). Age-related reproductive effort and senescence in free-ranging moose, Alces alces. Ecology, 82, 1613–20. Evans, K. & Hindell, M. A. (2004). The age structure and growth of female sperm whales (Physeter macrocephalus) in southern Australian waters. Journal of Zoology, 263, 237–50. Festa-Bianchet, M., Gaillard, J.-M. & Côté, S. D. (2003). Variable age structure and apparent density dependence in survival of adult ungulates. Journal of Animal Ecology, 72, 640. Festa-Bianchet, M. & Côté, S. D. (2008). Mountain Goats: Ecology, Behavior, and Conservation of an Alpine Ungulate (Washington, DC: Island Press). Fisher, R. A. (1930). The Genetical Theory of Natural Selection (Oxford: Clarendon Press). Flook, D. R. (1970). Causes and implications of an observed sex differential in the survival of wapiti. Canadian Wildlife Service Report, 11, 1–70. Focardi, S., Gaillard, J.-M., Ronchi, F. & Rossi, S. (2008). Survival of wild boars in variable environments: unexpected life-history variation in an unusual ungulate. Journal of Mammalogy, 89, 1113–23. Foote, A. D. (2008). Mortality rate acceleration and post-reproductive lifespan in matrilineal whale species. Biology Letters, 4, 189–91. Fraga, C. G., Shigenaga, M. K., Park, J. W., et al. (1990). Oxidative damage to DNA during aging: 8-hydroxy-2’-deoxyguanosine in rat organ DNA and urine. Proceedings of the National Academy of Sciences of the United States of America, 87, 4533–7. Gaillard, J.-M., Pontier, D., Allainé, D., et al. (1989). An analysis of demographic tactics in birds and mammals. Oikos, 56, 59–76. Gaillard, J.-M., Liberg, O., Andersen, R., et al. (1998). Population dynamics of roe deer. In European Roe Deer: The Biology of Success (pp. 309–35), ed. R. Andersen, P. Duncan & J. D. C. Linnell (Oslo: Scandinavian University Press). Gaillard, J.-M., Festa-Bianchet, M., Yoccoz, N. G., et al. (2000). Temporal variation in fitness components and population dynamics of large herbivores. Annual Review of Ecology and Systematics, 31, 367–93. Gaillard, J.-M., Duncan, P., Delorme, D., et al. (2003). Effects of hurricane Lothar on the population dynamics of European roe deer. Journal of Wildlife Management, 67, 767–73. Gaillard, J.-M., Loison, A., Festa-Bianchet, M., et al. (2003). Ecological correlates of life span in populations of large herbivorous mammals. Population & Development Review, 29, 39–56.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:32:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.007

References

147

Gaillard, J.-M., Viallefont, A., Loison, A. & Festa-Bianchet, M. (2004). Assessing senescence patterns in populations of large mammals. Animal Biodiversity and Conservation, 27, 47–58. Gaillard, J.-M., Yoccoz, N. G., Lebreton, J.-D., et al. (2005). Generation time: a reliable metric to measure life-history variation among mammalian populations. American Naturalist, 166, 119–23. Gaillard, J.-M., Hewison, A. J. M., Klein, F., et al. (2013). How does climate change influence demographic processes of widespread species? Lessons from the comparative analysis of contrasted populations of roe deer. Ecology Letters, 16, 48–57. Gamelon, M., Focardi, S., Gaillard, J.-M., et al. (2014). Do age-specific survival patterns of wild boar fit current evolutionary theories of senescence? Evolution, 68, 3636–43. Garratt, M., Gaillard, J.-M., Brooks, R. C. & Lemaître, J. F. (2013). Diversification of the eutherian placenta is associated with changes in the pace of life. Proceedings of the National Academy of Sciences of the United States, 110, 7760–5. Garrott, R. A. & Taylor, L. (1990). Dynamics of a feral horse population in Montana. Journal of Wildlife Management, 54, 603–12. Garrott, R. A., Eagle, T. C. & Plotka, E. D. (1991). Age-specific reproduction in feral horses. Canadian Journal of Zoology, 69, 738–43. Garrott, R. A., Eberhardt, L. L., White, P. J. & Rotella, J. (2003). Climate-induced variation in vital rates of an unharvested large-herbivore population. Canadian Journal of Zoology, 81, 33–45. Gimenez, O., Viallefont, A., Charmantier, A., et al. (2008). The risk of flawed inference in evolutionary studies when detectability is less than one. American Naturalist, 172, 441–8. Gompertz, B. (1825). On the nature of the function expressive of the law of human mortality and on a new model of determining life contingencies. Philosophical Transactions of the Royal Society of London Series B: Biological Sciences, 115, 513–85. Graham, A. L., Hayward, A. D., Watt, K. A., et al. (2010). Fitness correlates of heritable variation in antibody responsiveness in a wild mammal. Science, 330, 662–5. Green, W. C. H. (1990). Reproductive effort and associated costs in bison (Bison bison): do older mothers try harder? Behavioral Ecology, 1, 148–60. Hadley, G. L., Rotella, J. J. & Garrott, R. A. (2007). Evaluation of reproductive costs for Weddell seals in Erebus Bay, Antarctica. Journal of Animal Ecology, 76, 448–58. Hämäläinen, A., Dammhahn, M., Aujard, F., et al. (2015). Senescence or selective disappearance? Age trajectories of body mass in wild and captive populations of a small-bodied primate. Proceedings of the Royal Society of London Series B: Biological Sciences, 281, 20140830. Hamel, S., Craine, J. M. & Towne, E. G. (2012). Maternal allocation in bison: co-occurrence of senescence, cost of reproduction, and individual quality. Ecological Applications, 22, 1628–39. Hamilton, W. D. (1966). Moulding of senescence by natural selection. Journal of Theoretical Biology, 12, 12–45. Hanks, J. (1972). Reproduction of elephant, Loxodonta africana, in Luangwa Valley, Zambia. Journal of Reproduction and Fertility, 30, 13–26. Harting, A. L., Baker, J. D. & Johanos, T. C. (2007). Reproductive patterns of the Hawaiian monk seal. Marine Mammal Science, 23, 553–73.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:32:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.007

148

Senescence in Mammalian Life History Traits

Hayward, A. D., Wilson, A. J., Pilkington, J. G., et al. (2009). Ageing in a variable habitat: environmental stress affects senescence in parasite resistance in St Kilda Soay sheep. Proceedings of the Royal Society of London Series B: Biological Sciences, 276, 3477–85. Hayward, A. D., Wilson, A. J., Pilkington, J. G., et al. (2013). Reproductive senescence in female Soay sheep: variation across traits and contributions of individual ageing and selective disappearance. Functional Ecology, 27, 184–95. Hayward, A. D., Mar, K. U., Lahdenperä, M. & Lummaa, V. (2014). Early reproductive investment, senescence and lifetime reproductive success in female Asian elephants. Journal of Evolutionary Biology, 27, 772–83. Hayward, A. D., Moorad, J., Regan, C. E., et al. (2015). Asynchrony of senescence among phenotypic traits in a wild mammal population. Experimental Gerontology, 71, 56–68. Heard, D., Barry, S., Watts, G. & Child, K. (1997). Fertility of female moose (Alces alces) in relation to age and body composition. Alces, 33, 165–76. Hernandez-Camacho, C. J., Aurioles-Gamboa, D. & Gerber, L. R. (2008a). Age-specific birth rates of California sea lions (Zalophus californianus) in the Gulf of California, Mexico. Marine Mammal Science, 24, 664–76. Hernandez-Camacho, C. J., Aurioles-Gamboa, D., Laake, J. & Gerber, L. R. (2008b). Survival rates of the California sea lion, Zalophus californianus, in Mexico. Journal of Mammalogy, 89, 1059–66. Hewison, A. J. M. & Gaillard, J.-M. (2001). Phenotypic quality and senescence affect different components of reproductive output in roe deer. Journal of Animal Ecology, 70, 600–8. Hibly, A. R. & Mullen, A. J. (1980). Simultaneous determination of fluctuating age structure and mortality from field data. Theoretical Population Biology, 18, 192–203. Hindle, A. G., Horning, M., Mellish, J.-A. E. & Lawler, J. M. (2009a). Diving into old age: muscular senescence in a large-bodied, long-lived mammal, the Weddell seal (Leptonychotes weddellii). Journal of Experimental Biology, 212, 790–6. Hindle, A. G., Lawler, J. M., Campbell, K. L. & Horning, M. (2009b). Muscle senescence in short-lived wild mammals, the soricine shrews Blarina brevicauda and Sorex palustris. Journal of Experimental Zoology Part A: Ecological Genetics and Physiology, 311A, 358–67. Hindle, A. G., Lawler, J. M., Campbell, K. L. & Horning, M. (2010). Muscle aging and oxidative stress in wild-caught shrews. Comparative Biochemistry and Physiology Part B: Biochemistry and Molecular Biology, 155, 427–34. Hoffman, C. L., Higham, J. P., Mas-Rivera, A., et al. (2010). Terminal investment and senescence in rhesus macaques (Macaca mulatta) on Cayo Santiago. Behavioral Ecology, 21, 972–8. Holmes, E. E., Fritz, L. W., York, A. E. & Sweeney, K. (2007). Age-structured modeling reveals long-term declines in the natality of western Steller sea lions. Ecological Applications, 17, 2214–32. Isaac, J. L. & Johnson, C. N. (2005). Terminal reproductive effort in a marsupial. Biology Letters, 1, 271–5. Jégo, M., Lemaître, J. F., Bourgoin, G., et al. (2014). Haematological parameters do senesce in the wild: evidence from different populations of a long-lived mammal. Journal of Evolutionary Biology, 27, 2745–52.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:32:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.007

References

149

Jones, O. R., Gaillard, J.-M., Tuljapurkar, S., et al. (2008). Senescence rates are determined by ranking on the fast-slow life-history continuum. Ecology Letters, 11, 664–73. Jorgenson, J. T., Festa-Bianchet, M., Gaillard, J.-M. & Wishart, W. D. (1997). Effects of age, sex, disease, and density on survival of bighorn sheep. Ecology, 78, 1019–32. Juškaitis, R. (2008). The Common Dormouse Muscardinus avellanarius: Ecology, Population Structure and Dynamics (Institute of Ecology of Vilnius University). Kirkwood, T. B. L. (1977). Evolution of ageing. Nature, 270, 301–4. Kirkwood, T. B. L. & Austad, S. N. (2000). Why do we age? Nature, 408, 233–8. Koons, D. N., Gamelon, M., Gaillard, J.-M., et al. (2014). Methods for studying cause-specific senescence in the wild. Methods in Ecology and Evolution, 5, 924–33. Koyama, N., Takahata, Y., Huffman, M. A., et al. (1992). Reproductive parameters of female Japanese macaques: 30 years data from the Arashyiama troops, Japan. Primates, 33, 33–47. Lander, R. H. (1981). A life table and biomass estimate for Alaskan fur seals. Fisheries Research, 1, 55–70. Lawrence, R. K., Demarais, S., Relyea, R. A., et al. (2004). Desert mule deer survival in southwest Texas. Journal of Wildlife Management, 68, 561–9. Leader-Williams, N. (1988). Reindeer on South Georgia (Cambridge University Press). Lebreton, J.-D., Burnham, K. P., Clobert, J. & Anderson, D. R. (1992). Modeling survival and testing biological hypotheses using marked animals: a unified approach with case studies. Ecological Monographs, 62, 67–118. Lee, D. E. & Tietje, W. D. (2005). Dusky-footed woodrat demography and prescribed fire in a California oak woodland. Journal of Wildlife Management, 69, 1211–20. Leigh, S. R., Setchell, J.-M., Charpentier, M., et al. (2008). Canine tooth size and fitness in male mandrills (Mandrillus sphinx). Journal of Human Evolution, 55, 75–85. Lemaître, J. F. & Gaillard, J.-M. (2013a). Polyandry has no detectable mortality cost in female mammals. PloS ONE, 8, e66670. Lemaître, J. F. & Gaillard, J.-M (2013b). Male survival patterns do not depend on male allocation to sexual competition in large herbivores. Behavioral Ecology, 24, 421–8. Lemaître, J. F., Gaillard, J.-M., Lackey, L. B., et al. (2013). Comparing free-ranging and captive populations reveals intra-specific variation in aging rates in large herbivores. Experimental Gerontology, 48, 162–7. Lemaître, J. F., Gaillard, J.-M., Pemberton, J. M., et al. (2014). Early life expenditure in sexual competition is associated with increased reproductive senescence in male red deer. Proceedings of the Royal Society of London B: Biological Sciences, 281, 20140792. Lemaître, J. F., Berger, V., Bonenfant, C., Douhard, M., Gamelon, M., et al. (2015). Early-late life trade-offs and the evolution of ageing in the wild. Proceedings of the Royal Society of London Series B: Biological Sciences, 282, 20150209. Lima, M. & Paez, E. (1997). Demography and population dynamics of South American fur seals. Journal of Mammalogy, 78, 914–20. Lindstedt, S. L. & Calder, W. A. (1981). Body size, physiological time, and longevity of homeothermic animals. Quarterly Review of Biology, 56, 1–16. Loison, A., Festa-Bianchet, M., Gaillard, J.-M., et al. (1999). Age-specific survival in five populations of ungulates: evidence of senescence. Ecology, 80, 2539–54.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:32:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.007

150

Senescence in Mammalian Life History Traits

Lunn, N. J., Boyd, I. L. & Croxall, J. P. (1994). Reproductive performance of female Antarctic fur seals: the influence of age, breeding experience, environmental variation and individual quality. Journal of Animal Ecology, 63, 827–40. Lynch, H. J. & Fagan, W. F. (2009). Survivorship curves and their impact on the estimation of maximum population growth rates. Ecology, 90, 1116–24. Maizels, R. M. & Nussey, D. H. (2013). Into the wild: digging at immunology’s evolutionary roots. Nature Immunology, 14, 879–83. McAdam, A. G., Boutin, S., Sykes, A. K. & Humphries, M. M. (2007). Life histories of female red squirrels and their contributions to population growth and lifetime fitness. Ecoscience, 14, 362–9. McCullough, D. R. (1979). The George Reserve Deer Herd: Population Ecology of a K-Selected Species (Ann Arbor: University of Michigan Press). McDonald, J. L., Smith, G. C., McDonald, R. A., et al. (2014). Mortality trajectory analysis reveals the drivers of sex-specific epidemiology in natural wildlife-disease interactions. Proceedings of the Royal Society of London Series B: Biological Sciences, 281, 20140526. Medawar, P. B. (1952). An Unsolved Problem of Biology (London: Lewis). Menkens, G. E. & Boyce, M. S. (1993). Comments on the use of time-specific and cohort life tables. Ecology, 74, 2164–8. Millar, J. S. (1994). Senescense in a population of small mammals? Ecoscience, 1, 317–21. Milner, J. M., Stien, A., Irvine, R. J., et al. (2003). Body condition in Svalbard reindeer and the use of blood parameters as indicators of condition and fitness. Canadian Journal of Zoology, 81, 1566–78. Milot, E., Cohen, A. A., Vézina, F., et al. (2014). A novel integrative method for measuring body condition in ecological studies based on physiological dysregulation. Methods in Ecology and Evolution, 5, 146–55. Monteith, K. L., Bleich, V. C., Stephenson, T. R., et al. (2014). Life-history characteristics of mule deer: effects of nutrition in a variable environment. Wildlife Monographs, 186, 1–62. Moore, J. E. & Read, A. J. (2008). A Bayesian uncertainty analysis of cetacean demography and bycatch mortality using age-at-death data. Ecological Applications, 18, 1914–31. Morris, D. W. (1996). State-dependent life history and senescence of white-footed mice. Ecoscience, 3, 1–6. Müller, D. W. H., Gaillard, J.-M., Lackey, L. B., et al. (2010). Comparing life expectancy of three deer species between captive and wild populations. European Journal of Wildlife Research, 56, 205–8. Murie, A. (1944). The Wolves of Mount McKinley (Washington, DC: US Department Interior, National Park Service). Mysterud, A., Yoccoz, N. G., Stenseth, N. C. & Langvatn, R. (2001). Effects of age, sex and density on body weight of Norwegian red deer: evidence of density-dependence. Proceedings of the Royal Society of London Series B: Biological Sciences, 268, 911–19. Mysterud, A., Meisingset, E., Langvatn, R., et al. (2005). Climate-dependent allocation of resources to secondary sexual traits in red deer. Oikos, 111, 245–52.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:32:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.007

References

151

Neuhaus, P. & Pelletier, N. (2001). Mortality in relation to season, age, sex, and reproduction in Columbian ground squirrels (Spermophilus columbianus). Canadian Journal of Zoology, 79, 465–70. Nichols, J. D. (1992). Capture-recapture models. Bioscience, 42, 94–102. Nishida, T., Corp, N., Hamai, M., et al. (2003). Demography, female life history, and reproductive profiles among the chimpanzees of Mahale. American Journal of Primatology, 59, 99–121. Nussey, D. H., Kruuk, L. E. B., Morris, A. & Clutton-Brock, T. H. (2007). Environmental conditions in early life influence ageing rates in a wild population of red deer. Current Biology, 17, R1000–1. Nussey, D. H., Kruuk, L. E. B., Morris, A., et al. (2009a). Inter- and intrasexual variation in aging patterns across reproductive traits in a wild red deer population. American Naturalist, 174, 342–57. Nussey, D. H., Pemberton, J. M., Pilkington, J. G. & Blount, J. D. (2009b). Life history correlates of oxidative damage in a free-living mammal population. Functional Ecology, 23, 809–17. Nussey, D. H., Coulson, T., Delorme, D., et al. (2011). Patterns of body mass senescence and selective disappearance differ among three species of free-living ungulates. Ecology, 92, 1936–47. Nussey, D. H., Watt, K., Pilkington, J. G., et al. (2012). Age-related variation in immunity in a wild mammal population. Aging Cell, 11, 178–80. Nussey, D. H., Froy, H., Lemaitre, J. F., et al. (2013). Senescence in natural populations of animals: widespread evidence and its implications for bio-gerontology. Ageing Research Reviews, 12, 214–25. Oleziuk, P. F., Bigg, M. A. & Ellis, G. M. (1990). Life history and population dynamics of resident killer whales (Orcinus orca) in the coastal waters of British Columbia and Washington State. Report of the International Whaling Commission, 12, 209–43. Owen-Smith, N. (1990). Demography of a large herbivore, the greater kudu Tragelaphus strepsiceros, in relation to rainfall. Journal of Animal Ecology, 59, 893–913. Packer, C., Tatar, M. & Collins, A. (1998). Reproductive cessation in female mammals. Nature, 392, 807–11. Paul, A., Kuester, J. & Podzuweit, D. (1993). Reproductive senescence and terminal investment in female Barbary macaques (Macaca sylvanus) at Salem. International Journal of Primatology, 14, 105–24. Péron, G., Gimenez, O., Charmantier, A., et al. (2010). Age at the onset of senescence in birds and mammals is predicted by early-life performance. Proceedings of the Royal Society of London Series B: Biological Sciences, 277, 2849–3285. Péron, G., Gaillard, J.-M., Barbraud, C., Bonenfant, C., Charmantier, A., et al. (2016). Evidence of reduced heterogeneity in adult survival of long-lived species. Evolution 70, 2909–2914. Pielowski, Z. (1984). Some aspects of population structure and longevity of field roe deer. Acta Theriologica, 29, 17–33. Proffitt, K. M., Garrott, R. A., Rotella, J. J. & Wheatley, K. E. (2007). Environmental and senescent related variations in Weddell seal body mass: implications for age-specific reproductive performance. Oikos, 116, 1683–90. Promislow, D. E. L. (1991). Senescence in natural populations of mammals: a comparative study. Evolution, 45, 1869–87.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:32:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.007

152

Senescence in Mammalian Life History Traits

Promislow, D. E. L. & Harvey, P. H. (1990). Living fast and dying young: a comparative analysis of life-history variation among mammals. Journal of Zoology, 220, 417–37. Promislow, D. E. L., Fedorka, K. M. & Burger, J. M. S. (2006). Evolutionary biology of aging: future directions. In Handbook of the Biology of Aging, eds. E. J. Masoro & S. N Austad (pp. 217–42). (Burlington, MA: Academic Press). Pyne, M. I., Byrne, K. M., Holfelder, K. A., et al. (2010). Survival and breeding transitions for a reintroduced bison population: a multistate approach. Journal of Wildlife Management, 74, 1463–71. Richard, A. F., Dewar, R. E., Schwartz, M. & Ratsirarson, J. (2002). Life in the slow lane? Demography and life histories of male and female sifaka (Propithecus verreauxi verreauxi). Journal of Zoology, 256, 421–36. Ricklefs, R. E. (2010). Life-history connections to rates of aging in terrestrial vertebrates. Proceedings of the National Academy of Sciences of the United States of America, 107, 10314–19. Robbins, A. M., Robbins, M. M., Gerald-Steklis, N. & Steklis, H.D. (2006). Age-related patterns of reproductive success among female mountain gorillas. American Journal of Physical Anthropology, 131, 511–21. Robinson, M. R., Pilkington, J. G., Clutton-Brock, T. H., et al. (2006). Live fast, die young: trade-offs between fitness components and sexually antagonistic selection on weaponry in Soay sheep. Evolution, 60, 2168–81. Robinson, M. R., Mar, K. U. & Lummaa, V. (2012). Senescence and age-specific trade-offs between reproduction and survival in female Asian elephants. Ecology Letters, 15, 260–6. Rotella, J. J., Link, W. A., Chambert, T., et al. (2012). Evaluating the demographic buffering hypothesis with vital rates estimated for Weddell seals from 30 years of mark-recapture data. Journal of Animal Ecology, 81, 162–73. Rodgers, W. A. (1984). Warthog ecology in South East Tanzania. Mammalia, 48, 327–50. Sacher, G. A. (1959). Relation of life span to brain weight and body weight in mammals. In The Lifespan of Animals (CIBA Foundation: Colloquia on Aging, Vol. 5, pp. 115–33) (Hoboken, NJ: Wiley). Sadleir, R. M. F. S. (1969). The Ecology of Reproduction in Wild and Domestic Animals (London: Methuen). Saltz, D. (1996). Minimizing extinction probability due to demographic stochasticity in a reintroduced herd of Persian fallow deer Dama dama mesopotamica. Biological Conservation, 75, 27–33. Schwartz, C. C., Keating, K. A., Reynolds, H. V., et al. (2003). Reproductive maturation and senescence in the female brown bear. Ursus, 14, 109–19. Seber, G. A. F. (1973). The Estimation of Animal Abundance and Related Parameters (London: Griffin). Selman, C., Blount, J. D., Nussey, D. H. & Speakman, J. R. (2012). Oxidative damage, ageing, and life-history evolution: where now? Trends in Ecology and Evolution, 27, 570–7. Sharp, S. P. & Clutton-Brock, T. H. (2010). Reproductive senescence in a cooperatively breeding mammal. Journal of Animal Ecology, 79, 176–83. Sibly, R. M., Collett, D., Promislow, D. E. L., et al. (1997). Mortality rates of mammals. Journal of Zoology, 243, 1–12.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:32:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.007

References

153

Sierra, E., Fernandez, A., de los Monteros, A. E., et al. (2013). Muscular senescence in cetaceans: adaptation towards a slow muscle fibre phenotype. Scientific Reports, 3, 1795. Siler, W. (1979). Competing-risk model for animal mortality. Ecology, 60, 750–7. Sinclair, A. R. E. (1977). The Africa Buffalo: A Study of Resource Limitation of Populations (University of Chicago Press). Slade, N. A. (1995). Failure to detect senescence in persistence of some grassland rodents. Ecology, 76, 863–70. Slade, N. A. & Balph, D. F. (1974). Population ecology of Uinta ground squirrels. Ecology, 55, 989–1003. Smith, C. C. & Fretwell, S. D. (1974). Optimal balance between size and number of offspring. American Naturalist, 108, 499–506. Sohal, R. S., Agarwal, S. & Sohal, B. H. (1995). Oxidative stress and aging in the Mongolian gerbil (Meriones unguiculatus). Mechanisms of Ageing and Development, 81, 15–25. Spinage, C. A. (1972). African ungulates life tables. Ecology, 53, 645–52. Stadtman, E. R. (1992). Protein oxidation and aging. Science, 257, 1220–4. Stearns, S. C. (1983). The influence of size and phylogeny on patterns of covariation among life-history traits in the mammals. Oikos, 41, 173–87. Stearns, S. C. (1992). The Evolution of Life Histories (Oxford University Press). Sugiyama, Y. (1994). Age-specific birth rate and lifetime reproductive success of chimpanzees at Bossou, Guinea. American Journal of Primatology, 32, 311–18. Sussman, R. W. (1991). Demography and social organization of free-ranging Lemur catta in the Beza-Mahafaly Reserve, Madagascar. American Journal of Physical Anthropology, 84, 43–58. Sydeman, W. J., Huber, H. R., Emslie, S. D., et al. (1991). Age-specific weaning success of Northern elephant seals in relation to previous breeding experience. Ecology, 72, 2204–17. Tafani, M., Cohas, A., Bonenfant, C., et al. (2013). Decreasing litter size of marmots over time: a life history response to climate change? Ecology, 94, 580–6. Tettamanti, F., Grignolio, S., Filli, F., et al. (2015). Senescence in breeding success of Alpine chamois (Rupicapra rupicapra): the role of female quality and age. Oecologia, 178, 187–95. Thomas, D. C. & Barry, S. J. (1990). A life table for female barren-ground caribou in North Central Canada. Rangifer, 3, 177–84. Thongtip, N., Saikhun, J., Mahasawangkul, S., et al. (2008). Potential factors affecting semen quality in the Asian elephant (Elephas maximus). Reproductive Biology and Endocrinology, 6, 9. Tidière, M., Gaillard, J.-M., Muller, D. W. H., et al. (2014). Males do not senesce faster in large herbivores with highly seasonal rut. Experimental Gerontology, 60, 167–72. Tinker, M. T., Doak, D. F., Estes, J. A., et al. (2006). Incorporating diverse data and realistic complexity into demographic estimation procedures for sea otters. Ecological Applications, 16, 2293–2312. Toïgo, C., Gaillard, J.-M., Gauthier, D., et al. (2002). Female reproductive success and costs in a temperate capital breeder: the effect of contrasting environmental conditions. Ecoscience, 9, 427–33.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:32:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.007

154

Senescence in Mammalian Life History Traits

Toïgo, C. & Gaillard, J.-M. (2003). Causes of sex-biased adult survival in ungulates: sexual size dimorphism, mating tactic or environmental harshness? Oikos, 101, 376–84. Toïgo, C., Gaillard, J.-M., van Laere, G., et al. (2006). How does environmental variation influence body mass, body size, and body condition? Roe deer as a case study. Ecography, 29, 301–8. Toïgo, C., Gaillard, J.-M., Festa-Bianchet, M., et al. (2007). Sex- and age-specific survival of the highly dimorphic alpine ibex: evidence for a conservative life-history tactic. Journal of Animal Ecology, 76, 679–86. Valcu, M., Dale, J., Griesser, M., Nakagawa, S. & Kempenaers, B. (2014). Global gradients of avian longevity support the classic evolutionary theory of ageing. Ecography, 37, 930–8. Van Noordwijk, A. J. & de Jong, G. (1986). Acquisition and allocation of resources: their influence on variation in life history tactics. American Naturalist, 128, 137–42. van de Pol, M. & Verhulst, S. (2006). Age-dependent traits: a new statistical model to separate within- and between-individual effects. American Naturalist, 167, 766–73. van de Pol, M. & Wright, J. (2009). A simple method for distinguishing within- versus between-subject effects using mixed models. Animal Behaviour, 77, 753–8. Vanpé, C., Gaillard, J.-M., Kjellander, P., et al. (2007). Antler size provides an honest signal of male phenotypic quality in roe deer. American Naturalist, 169, 481–93. Vanpé, C., Gaillard, J.-M., Kjellander, P., et al. (2010). Assessing the intensity of sexual selection on male body mass and antler length in roe deer Capreolus capreolus: is bigger better in a weakly dimorphic species? Oikos, 119, 1484–92. Vasilaki, A., McArdle, F., Iwanejko, L. M. & McArdle, A. (2006). Adaptive responses of mouse skeletal muscle to contractile activity: the effect of age. Mechanisms of Ageing and Development, 127, 830–9. Vaupel, J. W. & Yashin, A. I. (1985). Heterogeneity’s ruses: some surprising effects of selection on population dynamics. American Statistician, 39, 176–85. Vincent, J. P., Angibault, J. M., Bideau, E. & Gaillard, J.-M. (1994). Problem of age determination: overlooked source of error in vertical life table calculations [in French]. Mammalia, 58, 293–9. Walker, M. L., & Herndon, J. G. (2010). Mosaic aging. Medical Hypotheses, 74, 1048–51. Weladji, R. B., Loison, A., Gaillard, J.-M., et al. (2008). Heterogeneity in individual quality overrides costs of reproduction in female reindeer. Oecologia, 156, 237–47. Weladji, R. B., Holand, O., Gaillard, J.-M., et al. (2010). Age-specific changes in different components of reproductive output in female reindeer: terminal allocation or senescence? Oecologia, 162, 261–71. Whitehouse, A. M. & Hall-Martin, A. J. (2000). Elephants in Addo Elephant National Park, South Africa: reconstruction of the population’s history. Oryx, 34, 46–55. Wilder, S. M., Le Couteur, D. G. & Simpson, S. J. (2013). Diet mediates the relationship between longevity and reproduction in mammals. Age, 35, 921–7. Williams, G. C. (1957). Pleiotropy, natural selection and the evolution of senescence. Evolution, 11, 398–411. Williams, P. D., Day, T., Fletcher, Q. & Rowe, L. (2006). The shaping of senescence in the wild. Trends in Ecology and Evolution, 21, 458–63.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:32:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.007

References

155

Wilson, A. J. & Nussey, D. N. (2010). What is individual quality? An evolutionary perspective. Trends in Ecology and Evolution, 25, 207–14. Wilson, G. A., Olson, W. & Strobeck, C. (2002). Reproductive success in wood bison (Bison bison athabascae) established using molecular techniques. Canadian Journal of Zoology, 80, 1537–48. Wolfe, L. D. & Noyes, M. J. S. (1981). Reproductive senescence among female Japanese macaques (Macaca fuscata fuscata). Journal of Mammalogy, 62, 698–705. Woolley, P. A. (1991). Reproduction in Dasykaluta rosamondae (Marsupiala, Dasyuridae): field and laboratory observations. Australian Journal of Zoology, 39, 549–68.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:32:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.007

8

Avian Escape Artists? Patterns, Processes and Costs of Senescence in Wild Birds Sandra Bouwhuis and Oscar Vedder

Short Summary Birds, in comparison to mammals, are relatively long lived for their size and were once thought to largely escape senescence. As longitudinal studies have accumulated, it has, however, become clear that many avian traits show signs of late-life deterioration. The question of whether birds senesce has therefore been replaced by questions regarding (1) variation in rates of senescence among traits or individuals, (2) the way trait deterioration translates to fitness declines, (3) fitness costs of senescence and (4) processes underlying the observed patterns. While more studies are needed, the first results are intriguing. Within-individual age trajectories often vary among traits, but how varying trait declines translate to fitness declines is still largely unknown. With respect to among-individual variation in rates of senescence, results largely comply with predictions from life history theory: increased investment in early-life reproduction accelerates senescence. Population-level fitness costs of senescence are low and independent of peak survival rate, such that long-lived species do not experience higher costs of senescence than short-lived species. Whether there is intra-specific variation in fitness costs of senescence, however, remains an open question. It is thus clear that ornithologists are now more than ever in a position to pursue promising investigations into the extent, causes and consequences of individual variation in avian senescence.

Introduction ‘Senescence’ is a decrease in whole-organism performance with age thought to have evolved because unavoidable extrinsic mortality reduces the strength of selection against poor performance with age (Fisher 1930; Hamilton 1966; Medawar 1952; Williams 1957), thereby allowing three mechanisms to operate (also see Chapter 5). Firstly, deleterious mutations causing pathologies late in life could escape purging and accumulate over evolutionary time (‘mutation accumulation’) (Medawar 1952). Secondly, antagonistically pleiotropic alleles with beneficial effects early but detrimental effects late in life could be selected for (‘antagonistic pleiotropy’) (Williams 1957). Thirdly, on the phenotypic level, Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:33:29, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.008

Development of the Study of Senescence in Wild Birds

157

favouring the allocation of limited resources to growth or early reproduction over somatic maintenance could promote physiological ageing and reduced late-life performance (‘disposable soma’) (Kirkwood & Rose 1991; also see Chapter 2). Understanding biological variation in the processes contributing to senescence requires both field and laboratory research on a broad taxonomic range (Monaghan 2008; see also Chapter 1). In this light, birds are especially interesting models for two reasons. Firstly, several natural populations of bird species have now been monitored for multiple decades and have started to provide sufficient data to facilitate investigation of demographic and physiological patterns of senescence, especially in interaction with ecological conditions. Secondly, birds live longer than similar-sized mammals, both in the wild and in captivity (e.g. Lindstedt & Calder 1976), implying that avian senescence is slow or delayed, and its study may provide information on how maintenance of somatic integrity into old age can be achieved (Holmes & Austad 1995a). Two decades ago it was for these two reasons that Holmes and Austad (1995b) stated that ornithologists had come into a position to ‘make invaluable contributions to understanding the ultimate, or evolutionary, as well as proximate, or physiological, mechanisms underlying the variety of life spans and patterns of senescence in animals’. With this statement, Holmes and Austad posed questions to be addressed using avian models, which included whether rates of mortality and reproductive and physiological senescence are correlated, whether senescence patterns relate to life span within and among species, whether senescence patterns relate to lifetime reproductive success and come with a fitness cost and whether there is evidence for trade-offs between early- and late-life performance. After providing some historical and methodological background and reviewing evidence for avian senescence, these are the questions we address in this chapter.

Development of the Study of Senescence in Wild Birds In the early days of the development of evolutionary theories of senescence, it was hypothesised that birds should display little senescence because flight allows threedimensional movement, which increases chances of escape from predators and accidents and protects birds from the age-independent mortality that would lead to limited selection against senescence (Williams 1957). Interestingly, and paradoxically in hindsight, the absence of senescence in birds was also sometimes explained by the hypothesised ultimate cause of the evolution of senescence, that is, by the fact that extrinsic mortality caused few birds to reach old age. Lack (1943), for example, stated for robins, Erithacus rubecula, that ‘old age [as a likely cause of mortality] has been ruled out, since extremely few robins reach old age’. Nevertheless, little over a decade later, Richdale published a pioneering eighteen-year population study on yellow-eyed penguins, Megadyptes antipodes, and reported declines in aspects of reproductive performance with age (Richdale 1957). Similarly, declines in annual survival probability with age were inferred from a comparison of expected and observed age distributions within several wild bird populations (Botkin & Miller 1974) and directly observed in another early long-term population study on great tits, Parus major (Perrins 1979). Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:33:29, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.008

158

Avian Escape Artists? Patterns, Processes and Costs of Senescence in Wild Birds

When considering whether one can infer avian senescence from these early studies, there are two important points to raise. The first is that since senescence is a decrease in whole-organism performance with age, it, by definition, is a within-individual process and one that can arise in many ways. From an evolutionary perspective, a withinindividual performance decline in a single trait does not necessarily provide evidence for senescence because this decline could be part of an adaptive state-dependent reallocation of resources among traits rather than result in a fitness decline (e.g. Partridge & Barton 1996). A within-individual decline in clutch size with age, for example, would not represent a senescent decline if the total number of offspring produced did not decrease with age accordingly. Similarly, a within-individual decline in the number of offspring produced with age would not represent a senescent decline if it would be compensated for by an improved survival probability for the parent. To determine whether senescence occurs, one would therefore ideally quantify the within-individual change in both survival and offspring production with age. Most studies of avian senescence to date, however, report the age specificity of one or few traits, making it hard to distinguish between whole-organism senescence and what we could call ‘trait senescence’. We will therefore often use the term ‘senescence’ to refer to both phenomena. The second point is that because senescence is a within-individual process, it cannot be inferred from population-level data (e.g. Forslund & Pärt 1995; Vaupel & Yashin 1985). Population-level patterns of trait changes with age result from a combination of within-individual trait change and changes in a population’s phenotypic composition with age, and among-individual correlations between survival and reproductive performance may mask, or falsely suggest, senescence in reproductive traits in cross-sectional analyses. A positive correlation between reproductive performance and survival probability would, for example, result in individuals with poor reproductive performance being over-represented in early-age classes, and population-level patterns of reproductive performance will underestimate senescence. Alternatively, if increased reproduction comes with survival costs, individuals with high reproductive performance are over-represented in early-age classes, and population-level patterns of reproductive performance will overestimate within-individual deterioration. For an accurate description of senescence patterns, it is thus essential to use longitudinal data and to correct for among-individual heterogeneity in life span. The first study on birds that tested whether population-level patterns of age-specific reproductive performance reflected within-individual age-related changes in reproductive success came from Newton and Rothery (1998), who compared the population-level change in reproductive success of female Eurasian sparrowhawks, Accipiter nisus, between series of consecutive ages. They confirmed that individuals improved offspring production to the age of five, after which offspring production deteriorated. Using a similar analytical approach, Møller and de Lope (1999) found morphology, migratory performance and reproductive success to deteriorate in old age in barn swallows, Hirundo rustica, and the intensity of ectoparasite infestation to increase. Individual female red-billed choughs, Pyrrhocorax pyrrhocorax, were found to lay smaller clutches, be more likely to fail to reproduce at all and to fledge fewer offspring when Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:33:29, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.008

Statistical Considerations

159

old than during middle age (Reid et al. 2003). While the sparrowhawk population-level pattern could be entirely attributed to within-individual processes, in the choughs the population-level pattern was driven by a combination of within- and among-individual processes, reiterating the importance of distinguishing between the two.

Statistical Considerations The method used in the early longitudinal studies of avian age-specific performance described earlier was recently developed into a more extensive decomposition method based on the Price equation (Price 1970), in which the population-level change in trait expression between two age classes is compared to the same pattern in a data set restricted to individuals that were alive in both age classes, while the difference is attributed to selective (dis)appearance of individuals with certain phenotypes (Rebke et al. 2010). This method has the advantage of providing an accurate assessment of the proportional contribution of within-individual change and selective (dis)appearance to the population-level change with age across life spans but does not allow significance testing of causes of age-dependent trait variation or assessment of among-individual variation in senescence rate. Moreover, care must be taken in cases where age and life span interact to determine trait expression. If longer-lived individuals, for example, change less at each age interval than shorter-lived individuals (e.g. Kervinen et al. 2015), the proportion of the total age-dependent variation attributed to within-individual change will be underestimated, because long-lived individuals contribute more observations to the calculations than short-lived individuals (Zhang et al. 2015b). Two simple methods to statistically test the significance of contributions of withinand between-individual changes to age-dependent trait expression have been published recently (van de Pol & Verhulst 2006; van de Pol & Wright 2009) to bring existing statistical concepts (Kreft et al. 1995) to the attention of evolutionary ecologists and ornithologists. The first method consists of simultaneously testing for effects of age and age at first and last reproduction on performance measures such that the age at first and last reproduction terms capture the trait variation that is caused by individuals of consistently good or poor performance being non-randomly distributed over age classes (‘selective (dis)appearance’), leaving the age term to reflect the average withinindividual change in the trait with age (van de Pol & Verhulst 2006). The second method, based on the same concept but additionally useful in cases where individuals cannot be followed over their entire life span, uses a within-subject centring approach. This involves subtracting an individual’s mean trait value from each of its observed values such that, in senescence studies, the mean age values capture between-individual effects of selective (dis)appearance on a performance measure, while the age deviations capture within-individual performance changes with age (van de Pol & Wright 2009). Combined with the accumulation of high-quality individual-based data from longterm studies on wild vertebrates (Clutton-Brock & Sheldon 2010), necessary for sufficient statistical power, these methods have sparked an enormous increase in the annual number of studies published on senescence in wild animals. Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:33:29, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.008

160

Avian Escape Artists? Patterns, Processes and Costs of Senescence in Wild Birds

A Review The longitudinal analysis approaches described earlier have rapidly become the most widely used approaches in avian senescence studies. In order to complement reviews by Bennett and Owens (2002), Brunet-Rossinni and Austad (2006) and Nussey et al. (2013), in which cross-sectional as well as longitudinal studies of senescence and agespecific trait values were collated, we therefore summarise avian studies that use one of these approaches (found by citation of the methodological papers) and list all studies that describe how performance measures change with individual age. Our collation includes thirty-five studies of thirty populations of twenty-four species and seventy-nine trait–species combinations (Table 8.1), in which traits comprise various measures of reproductive performance (e.g. phenology, ejaculate characteristics, clutch size and breeding success), physiology (e.g. body condition, telomere length, resistance to oxidative stress and metabolic rate) and mate attraction or signalling (e.g. behaviour and morphology). With respect to performance declines with age, fifty-two out of seventy-nine trait–species combinations (65.8 per cent) provide evidence for within-individual deterioration, while significant selective appearance and disappearance effects are found for fourteen out of twenty-eight (50.0 per cent) and twenty-eight out of sixty-one (45.9 per cent) trait–species combinations, respectively. While formal meta-analysis is required to test whether certain traits or species are especially prone to senescence and selective effects, the numbers illustrate that senescence is widespread, but not ubiquitous, and that studies investigating the extent and sources of variation in senescence rates between traits, populations and species are needed to facilitate a more complete understanding of the evolutionary ecology of senescence (also see Monaghan et al. 2008; Nussey et al. 2013). Our summary of longitudinal studies of senescence accounting for among-individual heterogeneity does not include studies of age-specific mortality or actuarial senescence, in which breeding dispersal and imperfect detection probability pose problems and in which among- and within-individual processes affecting mortality risk are harder to disentangle because death is often unobserved and occurs only once for each individual. Methodological advances are, however, also made in this area of research. Agedependent breeding dispersal is often hard to establish (but see Breton et al. (2014) for an example), but studies of age-specific survival probabilities now generally take into account the age specificity of recapture probability. This is important because rates of actuarial senescence can be over- or underestimated depending on whether birds become more ‘trap happy’ or less likely to be observed with advancing age, respectively (e.g. Bouwhuis et al. 2012). Moreover, the concept of heterogeneity in individual ‘frailty’ (Vaupel et al. 1979) is now often incorporated in survival models (e.g. McDonald et al. 1996). Avian studies using these techniques have yielded improved estimates for survival senescence (e.g. Aubry et al. 2011; Cam et al. 2002; Marzolin et al. 2011; Péron et al. 2010) and confirmed findings from studies on reproductive senescence (e.g. Bouwhuis et al. 2009) that the onset (Marzolin et al. 2011) or rate (Cam et al. 2002; Péron et al. 2010) of senescence is often underestimated when performing population-

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:33:29, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.008

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:33:29, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.008

Table 8.1 Taxonomically Ordered Studies that Quantify the Contributions of Within- and Among-Individual Changes to Avian Age-Dependent Trait Expression Following Methods Described by van de Pol and Verhulst (2006) and/or van de Pol and Wright (2009) Species

Trait

N

Age SA

SD

(Reference)

Mute swan (Cygnus olor)

Laydate Clutch size Recruit production Copulation propensity Semen transfer failure Sperm number Sperm motility Comb size Gamete quality Gamete quantity Breeding success Laydate Brood size Breeding success Recruitment success Fledgling production Body mass Lyre length Blue chroma Eye comb size Tarsus length Lek attendance Fighting rate Territory distance Ejaculate size Ejaculate motility Sperm quality Breeding success

459; 753 459; 842 652 41 41 41 38 99 125 116 983 468 468 468 358 264 164 163 161 162 164 164 164 164 1792 1792 1792 2124

Yes Yes No Noa Noa Noa Noa Yesa Yesa Yesa Yes Yes Yes Yes Yes No Yes Yes Yes Yes No Yes Yes Yes Yes Yes No Yes

— — — — — — — — — — No No Yes Yes — No — — — — — — — — No Yes Yes Yes

Yes Yes Yes — — — — — — — No No No No Yes Yes Yes Yes Yes Yes No Yes Yes No Yes Yes Yes Yes

(McCleery et al. 2008; Auld et al. 2013) (McCleery et al. 2008; Auld et al. 2013) (Auld et al. 2013) (Dean et al. 2010) (Dean et al. 2010) (Dean et al. 2010) (Dean et al. 2010) (Cornwallis et al. 2014) (Cornwallis et al. 2014) (Cornwallis et al. 2014) (Froy et al. 2013) (Kim et al. 2011) (Kim et al. 2011) (Kim et al. 2011) (Torres et al. 2011) (Blas et al. 2009) (Kervinen et al. 2015) (Kervinen et al. 2015) (Kervinen et al. 2015) (Kervinen et al. 2015) (Kervinen et al. 2015) (Kervinen et al. 2015) (Kervinen et al. 2015) (Kervinen et al. 2015) (Preston et al. 2011) (Preston et al. 2011) (Preston et al. 2011) (Aubry et al. 2009)

Domestic fowl (Gallus gallus domesticus)

Wandering albatross (Diomedea exulans) Blue-footed booby (Sula nebouxii)

Black kite (Milvus migrans) Black grouse (Tetrao tetrix)

Houbara bustard (Chlamydotis undulata)

Black-legged kittiwake (Rissa tridactyla)

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:33:29, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.008

Table 8.1 (cont.) Species Black-tailed gull (Larus crassirostris) Lesser black-backed gull (Larus fuscus) Common tern (Sterna hirundo)

Trait

Telomere length Arrival date Arrival date Breeding probability Laydate Egg volume Clutch size Brood size Fledgling production Tawny owl (Strix aluco) Fledgling production Alpine swift (Apus melba) Resistance to oxidative stress Lance-tailed manakin (Chiropxiphia lanceolata) Siring success Barn swallow (Hirundo rustica) Arrival date Laydate Clutch size Fledgling production Tail length Body condition Seychelles warbler (Acrocephalus sechellensis) Telomere length Fledgling production Pied flycatcher (Ficedula hypoleuca) Ornamentation Clutch size Fledging success Recruitment success Collared flycatcher (Ficedula albicollis) Male wing-patch size Female wing-patch size Forehead-patch height Forehead-patch width Laydate Clutch size Fledgling production

N 25 66 477 499 470 470 470 470 470 223 229 43 639 488; 1622 1101 472; 1841 4419 729 204 283 466 466 466 466 1654 1577 1663 1663 1688 1701 1313

Age SA a

No Yes No Yes No No No No No Yes Yes Yes Yes Yes Yes Yes Yes Yesa Yesa Yes No No Yes No No No No No Yes Yes Yes

— Yes Yes Yes Yes Yes No No No No — Yes — — — — — — — No No No No Yes — — — — — — —

SD

(Reference)

— No No Yes Yes No No No No No Yes No No No No Yesb Yes — — No No No No Yes Yes No No Yes Yes Yes Yes

(Mizutani et al. 2013) (Bosman et al. 2013) (Zhang et al. 2015b) (Zhang et al. 2015b) (Zhang et al. 2015b) (Zhang et al. 2015b) (Zhang et al. 2015b) (Zhang et al. 2015b) (Zhang et al. 2015b) (Millon et al. 2011) (Bize et al. 2014) (DuVal 2012) (Balbontín et al. 2007) (Balbontín et al. 2007; Balbontín et al. 2012a) (Balbontín et al. 2012a) (Balbontín et al. 2007; Balbontín et al. 2012a) (Balbontín et al. 2011) (Balbontín et al. 2012b) (Barrett et al. 2013) (Hammers et al. 2012) (Potti et al. 2013) (Potti et al. 2013) (Potti et al. 2013) (Potti et al. 2013) (Evans et al. 2011) (Evans et al. 2011) (Evans et al. 2011) (Evans et al. 2011) (Evans et al. 2011) (Evans et al. 2011) (Evans et al. 2011)

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:33:29, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.008

Great tit (Parus major)

Blue tit (Cyanistes caeruleus)

Wood thrush (Hylocichla mustelina) Jackdaw (Corvus monedula) House sparrow (Passer domesticus)

Zebra finch (Taeniopygia guttata)

Clutch size Brood size Fledgling production Recruit production Plumage colouration Basal metabolic rate Resistance to oxidative stress Laydate Clutch size Fledgling production Recruit production Fledgling production Telomere length Social dominance Number of broods Clutch size Brood size Fledgling production Recruit production Basal metabolic rate

4935 4935 4935 4935; 760 1500 48 33 1143 1143 814 814 540 46 69 339 339 339 339 339 46

No Yes Yes Yes Noa Yesa Yes Yes Yes Yes No Yes Yesa Noa Yes Yes Yes No Yes Yes

— — — — — — — Yes Yes No No — — — — — — — — —

No No No Yes — — No — — — — Yes — — No No No No No No

(Bouwhuis et al. 2009) (Bouwhuis et al. 2009) (Bouwhuis et al. 2009) (Bouwhuis et al. 2009; Bouwhuis et al. 2010c) (Evans & Sheldon 2013) (Broggi et al. 2010) (Bize et al. 2014) (Auld & Charmantier 2011) (Auld & Charmantier 2011) (Auld & Charmantier 2011) (Auld & Charmantier 2011) (Brown & Roth 2009) (Salomons et al. 2009) (Verhulst et al. 2014) (Schroeder et al. 2011) (Schroeder et al. 2011) (Schroeder et al. 2011) (Schroeder et al. 2011) (Schroeder et al. 2011) (Moe et al. 2009)

Note: Reported for each trait–species combination are sample size (number of individuals N) and whether studies found evidence for senescence (age), selective appearance (SA) and disappearance (SD), where – means that the effect was not tested. Note that studies reported data from a single population of each species, except in the case of barn swallows and great tits, where studies reported data from four and six populations, respectively. a Tested with a within-subject centring approach (van de Pol & Wright 2009). b Inconsistent results across studies.

164

Avian Escape Artists? Patterns, Processes and Costs of Senescence in Wild Birds

level analyses. We therefore hope that these methods will increasingly be employed to facilitate identification of general patterns and species- or population-specific processes underlying age-specific survival and initiate investigation of between- and withinindividual trait variation on age-specific mortality risk (e.g. Zhang et al. 2015c).

Do All Traits Senesce at an Equal Rate? With performance declines proving to be detectable in wild bird populations and to be widespread (Table 8.1), it is now possible to initiate more detailed investigation of causes and consequences of variation in senescence. One question is whether different traits senesce in synchrony, as was predicted when an interest in the evolutionary ecology of ageing first arose, based on the expectation that an age-specific decline in general physiological state would affect all traits equally (Williams 1957). Since senescence of specific traits has often only been investigated in few or a single species, it is too early to use formal meta-analysis to answer this question. If we, however, split Table 8.1 to roughly quantify the occurrence of senescence in three trait types, we find that performance declines are most often reported for physiological trait–species combinations (8 of 9: 88.9 per cent), followed by trait–species combinations that capture reproductive performance (36 of 54: 66.7 per cent) and traits associated with sexual selection (8 of 16: 50.0 per cent). The first avian study adopting a longitudinal analysis approach and reporting the within-individual age specificity of more than a single trait mentioned that, in mute swans, Cygnus olor, the decline in clutch size with old age appears weaker than the increasing delay in laying date, although the difference was not formally tested (McCleery et al. 2008). A second longitudinal study, on great tits, compared the rate of senescence among four reproductive traits and showed that the onset of reproductive senescence advances and its rate becomes steeper over the course of successive stages within a reproductive attempt (Bouwhuis et al. 2009). Specifically, while clutch size shows no sign of senescence, senescence does affect brood size and the number of fledglings produced, and senescence in the number of fledglings produced contributes most strongly to senescence in the number of local recruits produced. These results were interpreted as suggesting that the most energetically expensive traits are subject to the strongest senescence effects, but a need for additional studies to test for generality of this pattern was noted and needs to be re-emphasised here because senescence rates of the same reproductive traits were not found to differ in a subsequent study on house sparrows, Passer domesticus (Schroeder et al. 2011). In the same population of great tits, Bouwhuis et al. (2012) found the correlation between the age-specific change in recruit production and survival probability to be strongly positive (Pearson’s correlation coefficient ρ = 0.844), suggesting that these two traits senesce at a similar rate and that, at the population level, a trade-off between the two is not observed. Among six bird species, however, the correlation between the age-specific change in recruit production and survival probability ranges from ρ = 0.316 to 0.987 and declines with the species’ maximum age-specific survival (Bouwhuis et al. 2012). Perhaps the mechanisms Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:33:29, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.008

Do All Individuals Senesce at an Equal Rate?

165

underlying reproductive and survival senescence are therefore more likely to be uncoupled in long-lived species, which are predicted to canalise survival. Four more studies have reported within-individual trajectories of various traits with age. The first, on collared flycatchers, Ficedula albicollis, provided evidence for a within-individual increase in ornamental traits with age, while reproductive performance was found to senesce (Evans et al. 2011). The second found that, in houbara bustards, Chlamydotis undulata, the predicted relationship between behavioural sexual display and ejaculate quality is uncoupled in old age because ejaculate quality shows strong senescence, while display behaviour remains intense (Preston et al. 2011). The third, on domestic chickens, Gallus gallus, reported senescent declines in male gamete quantity and quality, while male comb size is unaffected by age (Cornwallis et al. 2014). Female comb size and gamete quantity both senesce, while senescence in gamete quality depends on a female’s average gamete quality (Cornwallis et al. 2014). Finally, a fourth study showed little variation in rates of change in trait expression between seven sexually selected morphological and behavioural traits in black grouse, Tetrao tetrix, with only the change in testosterone-dependent eye comb size exceeding that of other traits (Kervinen et al. 2015). The balance of evidence thus is that traits often do not senesce in parallel, raising the question as to why this is so. Besides the hypothesis that the energetic ceiling lowers with age such that the most energetically expensive traits senesce fastest (Bouwhuis et al. 2009), hypotheses regarding optimal age-dependent life history trade-offs can be derived from observations that different trait types are subject to different selection pressures (e.g. Kingsolver et al. 2012) and that traits may be involved in cross-trait trade-offs (e.g. McGlothlin et al. 2007), which may themselves change with age (e.g. Cornwallis et al. 2014). We hope that future studies will address these hypotheses and report on agespecific selection pressures and trade-offs, as assessed in longitudinal multi-trait studies.

Do All Individuals Senesce at an Equal Rate? Individuals differ in traits associated with reproduction and survival, and life history theory predicts that individuals also vary in rates of senescence in relation to factors influencing their ability to prevent or repair somatic damage, such as investment in early-life reproduction (‘disposable soma’) (Kirkwood & Rose 1991; also see Chapter 2) or the quality of natal conditions (Metcalfe & Monaghan 2003). According to theory, senescence should start earlier or be accelerated in individuals that produce a large quantity of offspring in early life, suffer from adverse environmental conditions during early-life reproduction or experience poor natal conditions. Current results largely comply with predictions from life history theory. In both great tits and willow tits, Poecile montanus, females who skip breeding early in life enjoy higher survival probabilities in the senescent phase of life (McCleery et al. 1996; Orell & Belda 2002), while collared flycatchers rearing experimentally enlarged broods in their first year of life lay smaller clutches later in life (Gustafsson & Pärt 1990). Similarly, jackdaws, Corvus monedula, rearing experimentally enlarged broods throughout life Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:33:29, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.008

166

Avian Escape Artists? Patterns, Processes and Costs of Senescence in Wild Birds

suffer from a threefold higher age-specific increase in mortality rate (Boonekamp et al. 2014), and reproductive senescence is accelerated in common guillemots, Uria aalge, that reproduce successfully or experience poor climatic conditions during early-life reproduction (Reed et al. 2008). In great tits, rates of reproductive senescence increase alongside rates of early-life reproduction (Bouwhuis et al. 2010a). In the same population of great tits, immigrant females suffer from faster rates of senescence than locally hatched females (Bouwhuis et al. 2010a), which could be interpreted as increased costs of early-life reproduction being associated with increased senescence rate if breeding in an unfamiliar environment bears larger costs than breeding in the familiar rearing environment or if immigrants are poor-quality individuals evicted from their natal populations. Finally, among-individual variation in senescence has been linked to the age at which individuals start to reproduce, since young recruits, which invest more heavily in early-life reproduction, have a faster rate, and an earlier onset, of senescence in blue-footed boobies, Sula nebouxii, and Seychelles warblers, Acrocephalus sechellensis, respectively (Hammers et al. 2013; Kim et al. 2011). Although inter-individual variation in senescence rates has been related to variation in investment in early-life reproduction, evidence for an effect of natal conditions on senescence rate is relatively scarce. In great tits, there is evidence neither for populationlevel effects of natal breeding density, breeding success or food availability nor for nestof-origin-specific effects of hatching date or number of siblings (Bouwhuis et al. 2010a). Female offspring of old mothers, however, experience faster senescence than those of younger mothers (Bouwhuis et al. 2010b), and such trans-generational effects of parental age on offspring performance seem general (e.g. Bouwhuis et al. 2015; Schroeder et al. 2015). The mechanism that could link age effects across generations is currently unknown, but it is known that fitness traits of offspring can be affected epigenetically by changes in the parental environment (e.g. Hager et al. 2009). Indeed, since offspring during their development depend on their parents for information about future resource availability, it may be beneficial for parents to adjust their offspring’s resource allocation to life history components as a form of adaptive trans-generational phenotypic plasticity. Birds may be an especially useful taxon to study such processes because they lay eggs, the constituents of which can relatively easily be measured and manipulated. Maternally derived levels of prenatal androgens (e.g. testosterone and androstenedione), in particular, have been studied intensively and have been found to promote pre- and post-natal growth and competitiveness while also varying with a range of environmental conditions (Groothuis et al. 2005). Moreover, experimentally increased yolk androgen levels have been found to shorten life span in black-headed gulls, Larus ridibundus (Eising 2004), increase adult basal metabolic rate in zebra finches, Taeniopygia guttata (Nilsson et al. 2011) and pied flycatchers, Ficedula hypoleuca (Ruuskanen et al. 2013), and reduce DNA damage repair efficiency in domestic chickens (Treidel et al. 2013). The possibility that mothers, via androgen allocation to their eggs, modulate offspring senescence rates through antagonistic effects on early- and late-life performance thus constitutes a promising future research avenue.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:33:29, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.008

Fitness Costs of Avian Senescence

167

Fitness cost senescence (%)

100

80

60

40

20

0 0.0

Figure 8.1

0.2

0.4 0.6 0.8 Peak survival rate

1.0

Fitness costs of senescence in relation to peak survival rate, assessed for eight bird species (white circles, dashed line) and seven populations of five mammal species (black circles, solid line). (Modified from Bouwhuis et al. (2012) by adding the long-lived common tern (peak survival 0.92) (Zhang et al. 2015a) and wandering albatross (peak survival 0.96) (Pardo et al. 2013)).

Fitness Costs of Avian Senescence With studies on inter-individual variation in senescence rates within populations and species only just emerging, current work on fitness costs of avian senescence is mostly limited to inter-specific comparisons. In this work, fitness is quantified as the ‘reproductive value’, which is defined as the number of offspring a certain-aged individual can expect to produce during its remaining life span, assuming the prevailing age-specific survival probabilities and reproductive success (Fisher 1930). Specifically, the ‘fitness cost of senescence’ is calculated as the proportional difference between the observed reproductive value at the age of onset of reproduction and the hypothetical reproductive value if actuarial and/or reproductive senescence were not to occur (Bonduriansky & Brassil 2002; Bouwhuis et al. 2012). For eight bird species, the population-level fitness cost of senescence ranges from 1.4 to 18.6 per cent, averaging 8.9 per cent (Bouwhuis et al. 2012; updated with data from Pardo et al. 2013; Zhang et al. 2015a). This is markedly lower than for seven populations of five species of mammals, for which the fitness cost of senescence ranges from 9.7 to 62.5 per cent, averaging 42.0 per cent (Bouwhuis et al. 2012). In mammals, the fitness cost of senescence is strongly positively correlated with peak survival rate such that longlived mammals experience a much higher fitness cost of senescence than short-lived mammals (Bouwhuis et al. 2012). The slope of this relationship within birds is much weaker and not statistically significant (estimate ± SE: 3.9 ± 12.5, χ² = 0.098, p = 0.754; Figure 8.1). Understanding the taxonomic difference in this relationship poses an interesting challenge, especially since species of both birds and mammals with Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:33:29, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.008

168

Avian Escape Artists? Patterns, Processes and Costs of Senescence in Wild Birds

a longer generation time (i.e. a higher weighted average age of reproductive females) were found previously to have a later onset and slower rate of senescence (Jones et al. 2008; also see Chapter 7). Decomposing the fitness cost of senescence to fledgling production and survival has revealed that, on average, reproductive senescence contributes less to the cost of senescence than survival senescence, with the average ratio being 0.344 (Bouwhuis et al. 2012; updated with data from Pardo et al. 2013; Zhang et al. 2015a). The ratio of reproductive to survival senescence, however, ranges from 0.021 to 0.738, and this variation cannot be explained by inter-specific variation in peak survival rate (Bouwhuis et al. 2012; updated with data from Pardo et al. 2013; Zhang et al. 2015a). Further investigation of which traits govern fitness costs of senescence and in relation to which trade-offs and components of life history therefore is another requirement for advancing our understanding of the evolutionary ecology of senescence. A first attempt at quantifying fitness costs of senescence at the intra-specific level has recently been made in a study on common terns. In this species, the age of recruitment to the breeding population is highly variable, and recruitment age is known to influence processes of age-specific reproductive performance (Limmer & Becker 2010). Indeed, recruitment age affects age-specific breeding probability and fledgling production (Zhang et al. 2015a), but the effects are not large enough to translate to significant effects on age-specific reproductive values, such that fitness costs of senescence range between 17 and 21 per cent and do not differ between birds of different recruitment ages (Zhang et al. 2015a). Whether intra-specific variation in fitness costs of senescence is generally small, and how such variation in costs affects lifetime reproductive success, remains to be shown.

Conclusion Two decades ago, Holmes and Austad (1995b) stated that ornithologists had come into a position to ‘make invaluable contributions to the understanding of the ultimate, or evolutionary, as well as proximate, or physiological, mechanisms underlying the variety of life spans and patterns of senescence in animals’. Since then, cross-sectional studies of avian senescence have been supplemented with an increasing number of invaluable longitudinal studies. With the improved statistical tools for studying within-individual age trajectories of traits, the ever-extending long-term individual-based data sets and the accumulating longitudinal avian studies showing that senescence is abundant, but not ubiquitous, we fully agree. More than ever, ornithologists are in a position to pursue promising investigations into the extent, causes and consequences of individual variation in senescence. The field lies open for research on proximate and ultimate causes of differences in senescence between traits, regulation of among-individual variation in senescence and the establishment of general patterns in fitness costs associated with senescence, for all of which birds are a wonderful study taxon.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:33:29, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.008

Acknowledgements

169

Acknowledgements We are grateful to Owen Jones, Jon Brommer, Simon Verhulst, Ben Sheldon, Tobias Uller, Sinead English and an anonymous reviewer for constructive criticism on this work; to Deborah Pardo for providing us with data on fitness costs of senescence in the wandering albatross; and to the Alexander von Humboldt Foundation for supporting OV with a research fellowship.

References Aubry, L. M., Cam, E., Koons, D. N., et al. (2011). Drivers of age-specific survival in a long-lived seabird: contributions of observed and hidden sources of heterogeneity. Journal of Animal Ecology, 80, 375–83. Aubry, L. M., Koons, D. N., Monnat, J. Y. & Cam, E. (2009). Consequences of recruitment decisions and heterogeneity on age-specific breeding success in a long-lived seabird. Ecology, 90, 2491–2502. Auld, J. R. & Charmantier, A. (2011). Life history of breeding partners alters age-related changes of reproductive traits in a natural population of blue tits. Oikos, 120, 1129–38. Auld, J. R., Perrins, C. M. & Charmantier, A. (2013). Who wears the pants in a mute swan pair? Deciphering the effects of male and female age and identity on breeding success. Journal of Animal Ecology, 82, 826–35. Balbontín, J., de Lope, F., Hermosell, I. G., et al. (2011). Determinants of age-dependent change in a secondary sexual character. Journal of Evolutionary Biology, 24, 440–8. Balbontín, J., Hermosell, I. G., Marzal, A., et al. (2007). Age-related change in breeding performance in early life is associated with an increase in competence in the migratory barn swallow Hirundo rustica. Journal of Animal Ecology, 76, 915–25. Balbontín, J., Møller, A. P., Hermosell, I. G., et al. (2012a). Geographical variation in reproductive ageing patterns and life-history strategy of a short-lived passerine bird. Journal of Evolutionary Biology, 25, 2298–2309. Balbontín, J., Møller, A. P., Hermosell, I. G., et al. (2012b). Lifetime individual plasticity in body condition of a migratory bird. Biological Journal of the Linnean Society, 105, 420–34. Barrett, E. L. B., Burke, T. A., Hammers, M., et al. (2013). Telomere length and dynamics predict mortality in a wild longitudinal study. Molecular Ecology, 22, 249–59. Bennett, P. M. & Owens, I. P. F. (2002). Evolutionary Ecology of Birds (Oxford University Press). Bize, P., Cotting, S., Devevey, G., et al. (2014). Senescence in cell oxidative status in two bird species with contrasting life expectancy. Oecologia, 174, 1097–1105. Blas, J., Sergio, F. & Hiraldo, F. (2009). Age-related improvement in reproductive performance in a long-lived raptor: a cross-sectional and longitudinal study. Ecography, 32, 647–57. Bonduriansky, R. & Brassil, C. E. (2002). Rapid and costly ageing in wild male flies. Nature, 420, 377–9.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:33:29, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.008

170

Avian Escape Artists? Patterns, Processes and Costs of Senescence in Wild Birds

Boonekamp, J. J., Salomons, M., Bouwhuis, S., et al. (2014). Reproductive effort accelerates actuarial senescence in wild birds: an experimental study. Ecology Letters, 17, 599–605. Bosman, D. S., Vercruijsse, H. J. P., Stienen, E. W. M., et al. (2013). Age of first breeding interacts with pre- and post-recruitment experience in shaping breeding phenology in a long-lived gull. PLoS ONE, 8(12), e82093. Botkin, D. B. & Miller, R. S. (1974). Mortality rates and survival of birds. American Naturalist, 108, 181–92. Bouwhuis, S., Charmantier, A., Verhulst, S. & Sheldon, B.C. (2010a). Individual variation in rates of senescence: natal origin effects and disposable soma in a wild bird population. Journal of Animal Ecology, 79, 1251–61. Bouwhuis, S., Charmantier, A., Verhulst, S. & Sheldon, B. C. (2010b). Transgenerational effects on ageing in a wild bird population. Journal of Evolutionary Biology, 23, 636–42. Bouwhuis, S., Choquet, R., Sheldon, B. C. & Verhulst, S. (2012). The forms and fitness cost of senescence: age-specific recapture, survival, reproduction, and reproductive value in a wild bird population. American Naturalist, 179, E15–27. Bouwhuis, S., Sheldon, B. C., Verhulst, S. & Charmantier, A. (2009). Great tits growing old: selective disappearance and the partitioning of senescence to stages within the breeding cycle. Proceedings of the Royal Society of London Series B: Biological Sciences, 276, 2769–77. Bouwhuis, S., van Noordwijk, A. J., Sheldon, B. C., et al. (2010c). Similar patterns of age-specific reproduction in an island and mainland population of great tits Parus major. Journal of Avian Biology, 41, 615–20. Bouwhuis, S., Vedder, O. & Becker, P. H. (2015). Sex-specific pathways of parental age effects on offspring lifetime reproductive success in a long-lived seabird. Evolution, 69, 1760–71. Breton, A. R., Nisbet, I. C. T., Mostello, C. S. & Hatch, J. J. (2014). Age-dependent breeding dispersal and adult survival within a metapopulation of common terns Sterna hirundo. Ibis, 156, 534–47. Broggi, J., Hohtola, E., Koivula, K., et al. (2010). Idle slow as you grow old: longitudinal age-related metabolic decline in a wild passerine. Evolutionary Ecology, 24, 177–84. Brown, W. P. & Roth, R. R. (2009). Age-specific reproduction and survival of individually marked wood thrushes, Hylocichla mustelina. Ecology, 90, 218–29. Brunet-Rossinni, A. K. & Austad, S. N. (2006). Senescence in wild populations of mammals and birds. In Handbook of the Biology of Aging, ed. E. J. Masoro & S. N. Austad pp. 243–66). (Burlington MA: Academic Press). Cam, E., Link, W. A., Cooch, E. G., et al. (2002). Individual covariation in life-history traits: seeing the trees despite the forest. American Naturalist, 159, 96–105. Clutton-Brock, T. H. & Sheldon, B. C. (2010). Individuals and populations: the role of long-term, individual-based studies of animals in ecology and evolutionary biology. Trends in Ecology and Evolution, 25, 562–73. Cornwallis, C. K., Dean, R. & Pizzari, T. (2014). Sex-specific patterns of aging in sexual ornaments and gametes. American Naturalist, 184, E66–78. Dean, R., Cornwallis, C. K., Løvlie, H., et al. (2010). Male reproductive senescence causes potential for sexual conflict over mating. Current Biology, 20, 1192–6.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:33:29, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.008

References

171

DuVal, E. H. (2012). Variation in annual and lifetime reproductive success of lance-tailed manakins: alpha experience mitigates effects of senescence on siring success. Proceedings of the Royal Society of London Series B: Biological Sciences, 279, 1551–9. Eising, C. (2004). Mother knows best? Costs and benefits of differential maternal hormone allocation in birds. PhD thesis, University of Groningen. Evans, S. R., Gustafsson, L. & Sheldon, B. C. (2011). Divergent patterns of age-dependence in ornamental and reproductive traits in the collared flycatcher. Evolution, 65, 1623–36. Evans, S. R. & Sheldon, B. C. (2013). Pigments versus structure: examining the mechanism of age-dependent change in a carotenoid-based colour. Journal of Animal Ecology, 82, 418–28. Fisher, R. A. (1930). The Genetical Theory of Natural Selection (Oxford: Clarendon Press). Forslund, P. & Pärt, T. (1995). Age and reproduction in birds: hypotheses and tests. Trends in Ecology and Evolution, 10, 374–8. Froy, H., Phillips, R. A., Wood, A. G., et al. (2013). Age-related variation in reproductive traits in the wandering albatross: evidence for terminal improvement following senescence. Ecology Letters, 16, 642–9. Groothuis, T. G. G., Müller, W., von Engelhardt, N., et al. (2005). Maternal hormones as a tool to adjust offspring phenotype in avian species. Neuroscience and Biobehavioral Reviews, 29, 329–52. Gustafsson, L. & Pärt, T. (1990). Acceleration of senescence in the collared flycatcher Ficedula albicollis by reproductive costs. Nature, 347, 279–81. Hager, R., Cheverud, J. M. & Wolf, J. B. (2009). Change in maternal environment induced by cross-fostering alters genetic and epigenetic effects on complex traits in mice. Proceedings of the Royal Society of London Series B: Biological Sciences, 276, 2949–54. Hamilton, W. D. (1966). Moulding of senescence by natural selection. Journal of Theoretical Biology, 12, 12–45. Hammers, M., Richardson, D. S., Burke, T. & Komdeur, J. (2012). Age-dependent terminal declines in reproductive output in a wild bird. PLoS ONE, 7(7), e40413. Hammers, M., Richardson, D. S., Burke, T. & Komdeur, J. (2013). The impact of reproductive investment and early-life environmental conditions on senescence: support for the disposable soma hypothesis. Journal of Evolutionary Biology, 26, 1999–2007. Holmes, D. J. & Austad, S. N. (1995a). Birds as animal-models for the comparative biology of aging: a prospectus. Journals of Gerontology Series A: Biological Sciences and Medical Sciences, 50, B59–66. Holmes, D. J. & Austad, S. N. (1995b). The evolution of avian senescence patterns: implications for understanding primary aging processes. American Zoologist, 35, 307–17. Jones, O. R., Gaillard, J.-M., Tuljapurkar, S., et al. (2008). Senescence rates are determined by ranking on the fast-slow life-history continuum. Ecology Letters, 11, 664–73.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:33:29, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.008

172

Avian Escape Artists? Patterns, Processes and Costs of Senescence in Wild Birds

Kervinen, M., Lebigre, C., Alatalo, R. V., et al. (2015). Life-history differences in age-dependent expressions of multiple ornaments and behaviors in a lekking bird. American Naturalist, 185, 13–27. Kim, S. Y., Velando, A., Torres, R. & Drummond, H. (2011). Effects of recruiting age on senescence, lifespan and lifetime reproductive success in a long-lived seabird. Oecologia, 166, 615–26. Kingsolver, J. G., Diamond, S. E., Siepielski, A. M. & Carlson, S. M. (2012). Synthetic analyses of phenotypic selection in natural populations: lessons, limitations and future directions. Evolutionary Ecology, 26, 1101–18. Kirkwood, T. B. L. & Rose, M. R. (1991). Evolution of senescence: late survival sacrificed for reproduction. Philosophical Transactions of the Royal Society of London Series B: Biological Sciences, 332, 15–24. Kreft, I. G. G., de Leeuw, J. & Aiken, L. S. (1995). The effect of different forms of centering in hierarchical linear models. Multivariate Behavioural Research, 30, 1–21. Lack, D. (1943). The Life of the Robin (London: Witherby). Limmer, B. & Becker, P. H. (2010). Improvement of reproductive performance with age and breeding experience depends on recruitment age in a long-lived seabird. Oikos, 119, 500–7. Lindstedt, S. L. & Calder, W. A. (1976). Body size and longevity in birds. Condor, 78, 91–4. Marzolin, G. A., Charmantier, A. & Gimenez, O. (2011). Frailty in state-space models: application to actuarial senescence in the dipper. Ecology, 92(3), 562–7. McCleery, R. H., Clobert, J., Julliard, R. & Perrins, C. M. (1996). Nest predation and delayed cost of reproduction in the great tit. Journal of Animal Ecology, 65, 96–104. McCleery, R. H., Perrins, C. M., Sheldon, B. C. & Charmantier, A. (2008). Age-specific reproduction in a long-lived species: the combined effects of senescence and individual quality. Proceedings of the Royal Society of London Series B: Biological Sciences, 275, 963–70. McDonald, D. B., Fitzpatrick, J. W. & Woolfenden, G. E. (1996). Actuarial senescence and demographic heterogeneity in the Florida scrub jay. Ecology, 77, 2373–81. McGlothlin, J. W., Jawor, J. M. & Ketterson, E. D. (2007). Natural variation in a testosterone-mediated trade-off between mating effort and parental effort. American Naturalist, 170, 864–75. Medawar, P. B. (1952). An Unsolved Problem in Biology (London: Lewis). Metcalfe, N. B. & Monaghan, P. (2003). Growth versus lifespan: perspectives from evolutionary ecology. Experimental Gerontology, 38, 935–40. Millon, A., Petty, S. J., Little, B. & Lambin, X. (2011). Natal conditions alter age-specific reproduction but not survival or senescence in a long-lived bird of prey. Journal of Animal Ecology, 80, 968–75. Mizutani, Y., Tomita, N., Niizuma, Y. & Yoda, K. (2013). Environmental perturbations influence telomere dynamics in long-lived birds in their natural habitat. Biology Letters, 9, 20130511. Moe, B., Rønning, B., Verhulst, S. & Bech, C. (2009). Metabolic ageing in individual zebra finches. Biology Letters, 5, 86–9. Møller, A. P. & de Lope, F. (1999). Senescence in a short-lived migratory bird: age-dependent morphology, migration, reproduction and parasitism. Journal of Animal Ecology, 68, 163–71.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:33:29, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.008

References

173

Monaghan, P. (2008). Early growth conditions, phenotypic development and environmental change. Philosophical Transactions of the Royal Society of London Series B: Biological Sciences, 363, 1635–45. Monaghan, P., Charmantier, A., Nussey, D. H. & Ricklefs, R. E. (2008). The evolutionary ecology of senescence. Functional Ecology, 22, 371–8. Newton, I. & Rothery, P. (1998). Age-related trends in the breeding success of individual female sparrowhawks Accipiter nisus. Ardea, 86, 21–31. Nilsson, J. F., Tobler, M., Nilsson, J. A. & Sandell, M. I. (2011). Long-lasting consequences of elevated yolk testosterone for metabolism in the zebra finch. Physiological and Biochemical Zoology, 84, 287–91. Nussey, D. H., Froy, H., Lemaitre, J. F., et al. (2013). Senescence in natural populations of animals: widespread evidence and its implications for bio-gerontology. Ageing Research Reviews, 12, 214–25. Orell, M. & Belda, E. J. (2002). Delayed cost of reproduction and senescence in the willow tit Parus montanus. Journal of Animal Ecology, 71, 55–64. Pardo, D., Barbraud, C. & Weimerskirch, H. (2013). Females better face senescence in the wandering albatross. Oecologia, 173, 1283–94. Partridge, L. & Barton, N.H. (1996). On measuring the rate of ageing. Proceedings of the Royal Society of London Series B: Biological Sciences, 263, 1365–71. Péron, G., Crochet, P. A., Choquet, R., et al. (2010). Capture-recapture models with heterogeneity to study survival senescence in the wild. Oikos, 119, 524–32. Perrins, C. M. (1979). British Tits (London: Collins). Potti, J., Canal, D. & Serrano, D. (2013). Lifetime fitness and age-related female ornament signalling: evidence for survival and fecundity selection in the pied flycatcher. Journal of Evolutionary Biology, 26, 1445–57. Preston, B. T., Saint Jalme, M., Hingrat, Y., et al. (2011). Sexually extravagant males age more rapidly. Ecology Letters, 14, 1017–24. Price, G. R. (1970). Selection and covariance. Nature, 227, 520. Rebke, M., Coulson, T., Becker, P. H. & Vaupel, J. W. (2010). Reproductive improvement and senescence in a long-lived bird. Proceedings of the National Academy of Sciences of the United States of America, 107, 7841–6. Reed, T. E., Kruuk, L. E. B., Wanless, S., et al. (2008). Reproductive senescence in a long-lived seabird: rates of decline in late-life performance are associated with varying costs of early reproduction. American Naturalist, 171, E89–101. Reid, J. M., Bignal, E. M., Bignal, S., et al. (2003). Age-specific reproductive performance in red-billed choughs Pyrrhocorax pyrrhocorax: patterns and processes in a natural population. Journal of Animal Ecology, 72, 765–76. Richdale, L. E. (1957). A Population Study of Penguins (Oxford: Clarendon Press). Ruuskanen, S., Lehikoinen, E., Nikinmaa, M., et al. (2013). Long-lasting effects of yolk androgen on phenotype in the pied flycatcher (Ficedula hypoleuca). Behavioral Ecology and Sociobiology, 67, 361–72. Salomons, H. M., Mulder, G. A., van de Zande, L., et al. (2009). Telomere shortening and survival in free-living corvids. Proceedings of the Royal Society of London Series B: Biological Sciences, 276, 3157–65. Schroeder, J., Burke, T., Mannarelli, M. E., et al. (2011). Maternal effects and heritability of annual productivity. Journal of Evolutionary Biology, 25, 149–56.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:33:29, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.008

174

Avian Escape Artists? Patterns, Processes and Costs of Senescence in Wild Birds

Schroeder, J., Nakagawa, S., Rees, M., et al. (2015). Reduced fitness in progeny from old parents in a natural population. Proceedings of the National Academy of Sciences of the United States of America, 112, 4021–5. Torres, R., Drummond, H. & Velando, A. (2011). Parental age and lifespan influence offspring recruitment: a long-term study in a seabird. PLoS ONE, 6(11), e27245. Treidel, L. A., Whitley, B. N., Benowitz-Fredericks, Z. M. & Haussmann, M. F. (2013). Prenatal exposure to testosterone impairs oxidative damage repair efficiency in the domestic chicken (Gallus gallus). Biology Letters, 9, 20130684. van de Pol, M. & Verhulst, S. (2006). Age-dependent traits: a new statistical model to separate within- and between-individual effects. American Naturalist, 167, 766–73. van de Pol, M. & Wright, J. (2009). A simple method for distinguishing within- versus between-subject effects using mixed models. Animal Behaviour, 77, 753–8. Vaupel, J. W., Manton, K. G. & Stallard, E. (1979). The impact of heterogeneity in individual frailty on the dynamics of mortality. Demography, 16, 439–54. Vaupel, J. W. & Yashin, A. I. (1985). Heterogeneity’s ruses: some surprising effects of selection on population-dynamics. American Statistician, 39, 176–85. Verhulst, S., Geerdink, M., Salomons, H. M. & Boonekamp, J. J. (2014). Social life histories: jackdaw dominance increases with age, terminally declines and shortens lifespan. Proceedings of the Royal Society of London Series B: Biological Sciences, 281, 20141045. Williams, G. C. (1957). Pleiotropy, natural selection, and the evolution of senescence. Evolution, 11, 398–411. Zhang, H., Rebke, M., Becker, P. H. & Bouwhuis, S. (2015a). Fitness prospects: effects of age, sex and recruitment age on reproductive value in a long-lived seabird. Journal of Animal Ecology, 84, 199–207. Zhang, H., Vedder, O., Becker, P. H. & Bouwhuis, S. (2015b). Age-dependent trait variation: the relative contribution of within-individual change, selective appearance and disappearance in a long-lived seabird. Journal of Animal Ecology, 84(3), 797–807. Zhang, H., Vedder, O., Becker, P.H. & Bouwhuis, S. (2015c). Contrasting between- and within-individual trait effects on mortality risk in a long-lived seabird. Ecology, 96, 71–9.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:33:29, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.008

9

The Evolution of Senescence in Nature Andrew I. Furness and David N. Reznick

Short Summary Williams (1957) generated interest in senescence from a comparative evolutionary perspective by developing a variety of testable predictions. Perhaps the most wellknown is that high (extrinsic) mortality risk will cause an earlier onset and more rapid increase in intrinsic deterioration associated with age. This prediction has been amenable to a diversity of empirical tests, many of which found support, some of which have not. Later authors (e.g. Abrams 1993; Caswell 2007; Charlesworth 1980; Williams & Day 2003) modified the simplicity of the Williams’ prediction by incorporating ecological complexity. Risk of dying may not be a sufficient descriptor of the factors that shape the evolution of senescence in the wild. A grasshopper facing the onset of winter and a water flea facing the desiccation of its forest pool are certain to die when their environment deteriorates, but other organisms do not face such temporal boundaries. A guppy escaping a predator may or may not survive depending on its inherent speed and condition. These and other modifiers of an individual’s probability of dying can, according to this body of theory, modify how we expect senescence to evolve. Here we review the empirical study of senescence with the goal of evaluating whether or not these modifications offer additional explanatory power. We conclude that progress has been sufficient to confirm that they do, and then we offer suggestions for new directions of research.

Introduction The ultimate causes of senescence, intrinsic physiological decline in fitness components, have been speculated upon since the time of Aristotle and called an unsolved problem in biology (Medawar 1952). Senescence reduces individual fitness, so why hasn’t selection eliminated it – producing potentially immortal, fitness-maximising organisms? Applying evolutionary thinking to explain the phenomenon began in earnest with the classical work of Peter Medawar (1952) and George C. Williams (1957). Both authors concluded that the ultimate cause of senescence is the declining of force of natural selection with age. Even in an idyllic world free of intrinsic deterioration or senescence, there will be constant risk of accidental death due to external or extrinsic sources of mortality that are independent of age and condition. An inevitable consequence of this Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:34:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.009

176

The Evolution of Senescence in Nature

cumulative mortality risk is that few organisms will survive to very late ages. Thus, the early life history becomes of greater importance in determining fitness, and the late life history becomes a dustbin shielded from the purifying effects of selection. Medawar proposed that mutations in genes that have late-acting effects are able to accumulate and spread to fixation because most organisms fail to reach very old age. Reproduction happens before the deleterious effects of these genes are manifested, so they can persist and contribute to deterioration in advanced ages. Williams proposed that alleles that cause vigour early in life can be favoured even if they have detrimental consequences later in life. These two population genetic mechanisms, mutation accumulation and antagonistic pleiotropy, thought to underlie the evolution of senescence, are not mutually exclusive; research has focused on distinguishing the relative importance of both mechanisms, and each has some support (Charlesworth & Hughes 1996; Partridge & Barton 1993; Partridge & Gems 2002; Snoke & Promislow 2003; but see Moorad and Promislow 2009). The end result is that the expression of genes with late-acting mildly deleterious effects is expected in nearly every tissue, causing a near-universal and coordinated senescence (Williams 1957). This intuitively simple yet far-reaching verbal theory for how senescence evolves was given a sound mathematical foundation (Hamilton 1966; see also Chapter 3) and has gone on to serve as a framework to explain many comparative aspects of ageing and as a source of testable predictions. Theories are ultimately supported, discarded or modified on the basis of whether empirical evidence matches the predictions derived from them. Starting from simple and well-defined assumptions grounded in evolutionary theory, Williams (1957) derived nine predictions that went a long way towards shaping subsequent empirical research on senescence. Perhaps the most widely cited prediction, and one that has largely become synonymous with the evolutionary theory of senescence, is that high levels of (extrinsic) mortality are expected to result in the evolution of higher rates of intrinsic mortality due to senescence. In an environment with high extrinsic mortality, the probability of reaching old age is reduced relative to that in a low-mortality environment. Given limited resources, this tips the optimal investment strategy towards early reproduction and away from long-term bodily maintenance, resulting in more rapid senescence or intrinsic physiological deterioration when sources of extrinsic mortality are removed (e.g. in a laboratory environment). In contrast, in an environment with low extrinsic mortality, there is a high probability of reaching old age, favouring a more balanced investment in reproduction and somatic maintenance and resulting in delayed senescence. We hereafter refer to this as the ‘Williams prediction’. Support for this prediction has been sought using a variety of approaches including (1) laboratory selection experiments in which mortality or reproductive schedules are directly manipulated for several generations and the subsequent rate of intrinsic mortality is compared to that of control lines, (2) laboratory experimental evolution studies in which conditions of high and low mortality risk are created and lines are allowed to evolve under such regimes for several generations, after which the rate of senescence is compared, (3) comparisons among species that presumably differ in mortality risk owing to key traits or adaptations and (4) comparisons among populations of the same species that differ in mortality risk owing to an environmental factor. We first review the Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:34:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.009

Evidence from Laboratory Artificial Selection and Experimental Evolution Studies

177

evidence for and against the Williams prediction that has accrued from these different approaches. This prediction has, for the most part, received empirical support from wellcontrolled laboratory studies (selection experiments and experimental evolution) and broad-scale comparisons among species. However, comparative population studies of senescence in the field have produced mixed results, highlighting the fact that additional complicating factors are present. A requirement for the Williams prediction to hold is age heterogeneity in the form of either mortality risk or effect of density. An increase in mortality that affects all age classes equally is predicted to have no effect on the evolution of senescence (Caswell 2007; see also Chapter 4). Furthermore, theoretical and recent experimental work has demonstrated how density regulation and condition-dependent mortality can lead to a greater diversity of ageing responses, some opposite that of the Williams prediction. We review, discuss and interpret comparative studies of senescence in the wild with reference to these predictions and attempt to come to some general conclusions about when we might expect support for the classical (Williams) versus derived predictions (developed below). We find that this derived theory matches quite well with results from both the field and laboratory and largely subsumes the Williams prediction. In a sense, the classic theory represents a subset of the ageing response likely to be observed when a specific set of conditions is met. Such conditions are more readily met in a laboratory setting, thus potentially explaining the consistent match between laboratory results and classical theory. Under field conditions, which we argue are often characterised by an interaction between multiple mortality sources, density dependence and other complicating factors, a greater diversity of evolved ageing responses is both predicted and observed. Future empirical work aimed at elucidating these responses will be quite informative.

Evidence from Laboratory Artificial Selection and Experimental Evolution Studies Artificial selection experiments in Drosophila melanogaster have demonstrated that delaying the onset of reproduction in evolving laboratory populations can lead to the evolution of postponed senescence (Luckinbill et al. 1984, 1985; Rose 1984, 1991), as predicted by theory (Hamilton 1966). One experiment stands out in the way it was explicitly designed to test the Williams prediction. Stearns et al. (2000) created conditions in which replicate lines of D. melanogaster experienced either high or low adult mortality over the course of many generations. Mortality was imposed twice weekly. In the high adult-mortality treatments the probability of adult flies surviving for one week was only 1 per cent, while in the low adult-mortality treatments the weekly adult probability of survival was 64 per cent for the first 13.5 months of the experiment and 81 per cent thereafter. Population densities were kept constant for both adult flies and larvae, ensuring that treatment differences were due solely to adult mortality rate. Traits including fecundity, development time and body weight were assayed in separate vials in parallel with the main experiment at various time points. ‘Senescence’, defined as Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:34:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.009

178

The Evolution of Senescence in Nature

intrinsic mortality rate, was measured after the high adult-mortality flies had experienced approximately ninety generations and the low adult-mortality treatment group fifty generations of their respective selection regime. Five thousand flies of each sex from each line were reared in replicate cages for their life span, with dead flies sexed and counted every three days and replaced with white-eyed mutants such that a constant density was achieved over the course of the intrinsic mortality assay. Following an increase in larval rearing density, the experimental conditions of imposed high and low adult mortality generated significant divergence in the early life history in a manner consistent with theory (Charlesworth 1980; Gadgil & Bossert 1970; Schaffer 1974). Flies in the high adult-mortality treatment groups rapidly evolved smaller body size, earlier time to eclosion and higher early fecundity relative to flies in the low adultmortality treatment groups. Intrinsic mortality was significantly higher in the high adultmortality lines, directly confirming Williams’ prediction that higher extrinsic mortality rates lead to the evolution of higher intrinsic mortality rates (i.e. more rapid senescence).

Evidence from Inter-Specific Comparisons Inter-specific comparisons of senescence have generally taken two different forms. One approach involves fitting models (often Gompertz and Weibull) to demographic data collected in the field such that measures of age-specific mortality can be extracted and compared (Finch et al. 1990; Pletcher et al. 2000; Promislow 1991; Ricklefs 1998; Ricklefs & Scheuerlein 2002; Williams et al. 2006). A second approach has been to test whether species that have presumably been somewhat insulated from extrinsic mortality owing to a key adaptation, or habitat shift, have evolved a longer life span relative to related congeners that lack this attribute. This is the Williams prediction re-worked to be testable with patterns among species. Adaptations that reduce extrinsic mortality (e.g. flight, protective shell, large brain and large body size) tend to be associated with species that live a long time (e.g. birds, turtles, humans and elephants). When the effect of body size is removed, bats have a life span that is three times as long as terrestrial mammals (Austad & Fischer 1991). Presumably, having evolved flight removed bats from a major mortality source – predation – thus paving the way for the evolution of increased longevity. Similarly, birds have much longer life spans than mammals of comparable sizes (Holmes and Austad 1995). Eusocial heavily protected insect queens can live over 100 times longer than solitary insects (Keller & Genoud 1997), and the naked mole rat, a semi-eusocial mammal that is well protected due to its subterranean lifestyle, has an unusually long life span for a rodent (Sherman & Jarvis 2002). After correcting for body size, poisonous fish, reptiles and amphibians were found to have longer life spans than non-poisonous species (Blanco & Sherman 2005). In short, adaptations that apparently reduce mortality are often correlated with longer life span (at the species level). These broad-scale comparisons are consistent with the Williams prediction and valuable in that they highlight its potential explanatory power at a coarse level. However, such comparisons assume that a key adaptation or habitat shift has served as a reliable proxy of extrinsic Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:34:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.009

Evidence from Intra-Specific Comparative Studies in the Wild

179

mortality hazard, that maximum life span appropriately encapsulates differences in rate of senescence, and they do not necessarily exclude other interpretations (Williams et al. 2006). In particular, mean or maximum life span is a very coarse surrogate for patterns of senescence, and it is conceivable that the two are decoupled.

Evidence from Intra-Specific Comparative Studies in the Wild There have been several comparative studies of senescence that employ a field assessment of environment and mortality risk for multiple populations combined with a laboratory examination of senescence (summarised in Table 9.1). This approach complements well-controlled laboratory selection and experimental evolution studies and allows one to assess whether such results are applicable to the more complicated situation in the field. We first detail the generalised experimental approach before summarising several of these studies in more depth. The general pattern has been to identify populations that differ in regards to an environmental factor predicted to cause differences in mortality risk and thus likely to select for opposing life history strategies (Reznick 1993). Inter-population heterogeneity in mortality caused by predation (Austad 1993; Reznick et al. 1996), season length (Dudycha & Tessier 1999; Tatar et al. 1997; Tozzini et al. 2013) and a combination of predation and fluctuating resource levels (Bronikowski & Arnold 1999; Sparkman et al. 2007) have been used. In each case, the a priori predictions were the same – populations having historically experienced a higher level of extrinsic mortality were expected to have evolved a faster rate of senescence (intrinsic deterioration). An important step is demonstrating differences in extrinsic mortality between populations by characterising the environment or, ideally, measuring age- or size-specific mortality in the different environments. If feasible, intrinsic mortality (or other measures of intrinsic decline in fitness components such as reproduction or aspects of physiology) can be quantified by rearing individuals of each population in a common-garden environment free of extrinsic mortality sources present in the wild. In such instances, it is important to match environmental conditions in the laboratory to those in the field (as is reasonable) or to test over a range of conditions, absent extrinsic mortality sources, as environment has been shown to have a large influence on the expression of senescence (Kawasaki et al. 2008). Such an experimental approach is only feasible for relatively short-lived organisms that can be reared successfully in captivity. For larger, longer-lived organisms, such a comparative approach has been modified such that repeated measurements have been taken from the same individuals so that various physiological indices can be obtained as organisms age (Austad 1993). Alternatively, long-lived organisms have been reared in the laboratory for short periods such that various measures of senescence can be assayed before the organisms are released back into the wild (Bronikowski & Arnold 1999; Robert & Bronikowski 2010; Sparkman et al. 2007). One of the earliest intra-specific comparative tests was performed on Virginia opossums from the mainland United States (South Carolina) versus those that had evolved on a barrier island separated from the Georgia coast by five miles of open Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:34:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.009

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:34:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.009

Table 9.1 Intra-Specific Comparative Studies of Senescence in the Wild

Species

Number of populations studied

Agent of selection causing mortality differences among populations

Water flea (Daphnia ambigua)

6

Predation (high/ intermediate/low)

Water flea (Daphnia pulex- pulicaria)

8

Ephemeral vs. permanent water bodies

Virgina opossum (Didelphis virginiana)

2

Predation (high/low)

Grasshopper (Melanoplus sanguinipes/ devastator)

5

Season length at different altitudes

10 Annual killifish (populations from closely related sympatric species – Nothobranchius furzeri, N. kunthae, N. pienaari and N. rachovii) Guppy (Poecilia 4 reticulata)

Western terrestrial garter snake (Thamnophis elegans)

5

Duration of seasonal water bodies fish inhabit (dry vs. humid regions)

Measure(s) of senescence

Study details

Third-generation offspring Lifetime rate derived from wild-caught of offsrping populations were reared in production, intrinsic common-garden laboratory mortality setting Clonal lines of different Life span, intrinsic populations were mortality risk, established and reared in fecundity a common-garden laboratory setting Age-specific mortality In the field, young female rate acceleration, opossums were radioage-related collared and tracked such reproductive output, that senescence could be tail tendon collagen assayed denaturation rate Mortality rate Male grasshoppers that were (survivorship) derived from field-caught females from five different elevations were reared in a common laboratory setting at two different temperatures Mortality rate, accumulation of lipofuscin

Result (in reference to the Williams prediction)

Reference

Inconsistent

Walsh et al. (2014)

Consistent

Dudycha and Tessier (1999)

Consistent

Austad (1993)

Consistent

Tatar et al. (1997)

Consistent Captive-bred individuals derived from wild-caught fish were reared in common-garden laboratory setting

Tozzini et al. (2013); Tezibasi et al. (2008)

Second-generation offspring Inconsistent, except Reznick et al. Mortality rate, for swimming (2004) derived from wild-caught reproductive output, performance females were reared in swimming a common-garden performance laboratory setting Sparkman et al. Pregnant female snakes were Inconsistent for Maximum life span, Lakeshore (warm (2007); maximum life collected from the wild and median life span, temperatures, Robert and span and raised in the lab until reproductive effort, continuous food Bronikowski reproductive parturition; age of the incidence of supply, fast growth, (2010); success, consistent females was estimated reproductive failure, low adult survival, Sparkman for median life through mark/recapture several increased predation et al. (2013) span and records, skeletonphysiological pressure) vs. physiological chronology, or the measures mountain (cooler measures relationship between size temperatures, slow and age growth, variable food supply, high adult survival, lower predation pressure) habitat Predation (high/low)

Note: Williams (1957) predicted that populations subjected to higher levels of extrinsic mortality will exhibit more rapid senescence.

182

The Evolution of Senescence in Nature

ocean (Austad 1993). The island population had a 4,000- to 5,000-year history of reduced mortality risk relative to the mainland population owing to the complete absence of mammalian predators (bobcats and feral canines) on the island. Senescence (age-specific mortality rate acceleration, age-related reproductive output and tail tendon collagen denaturation rate) was assayed in radio-collared wild opossums. Female opossums from the island population showed higher survivorship, slower signs of reproductive senescence in their second year of life and delayed age-specific physiological decline compared to the mainland population (Austad 1993), thus supporting the Williams prediction that populations subjected to historically low rates of extrinsic mortality will evolve delayed senescence. Reznick et al. (2004) performed a comparative study of senescence in populations of guppies derived from habitats that differ in mortality risk. The island of Trinidad has a series of rivers running in parallel as they drain from the Northern Range Mountains. In the lower reaches of such rivers guppies co-occur with several piscivorous species (high predation sites). As one moves further upstream, discrete barrier waterfalls often truncate the upstream distribution of predatory species, creating a situation where guppies co-occur with only the relatively benign killifish – Rivulus hartii (low predation sites). Due to the close geographical proximity of such guppy populations, the habitat can be nearly identical, yet individual-based mark-recapture analysis indicates that the probability of survival for six months is twenty to thirty times greater for lowpredation guppies than for their high-predation counterparts (Reznick & Bryant 2007). To establish stock for the senescence experiment, Reznick et al. (2004) collected twenty-five wild-caught females from each of two paired high- and low-predation sites within two independent river drainages (Oropuche and Yarra). Females store sperm, so each wild-caught female produced litters of young that became numbered lines for future crosses. Each female contributed equally to the first and second laboratory generations, ensuring that the experimental population was free of adaptation to a laboratory setting. Crosses that produced subsequent laboratory generations were always between lines, so there was no inbreeding. The senescence assay was performed on the second generation of laboratory-reared fish, which minimises the contribution of latent environmental and trans-generational effects to any differences in senescence observed among populations. Guppies were reared on high or low food levels, designed to mimic naturally occurring resource levels in their respective environments and induce growth rates and asymptotic sizes comparable to that seen in the field. Experimental fish were reared individually from an age of twenty-five to thirty days until death. Data on age and size at all reproductive events, number of offspring, and age at death were collected. An index of physiological senescence, the fast-start escape response, used to evade predator strikes, was assayed on a subset of fish from each population at an age of twelve and twenty-six months. The quantification of fitness components (survival, reproduction, and physiological performance) over the duration of the life span allowed for the measurement of senescence as the age-specific decline in these traits. The laboratory rearing environment was designed to closely replicate all aspects of the natural environment (water quality, temperature, photo period, two food level that resulted in growth rates Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:34:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.009

Evidence from Intra-Specific Comparative Studies in the Wild

183

observed in high and low predation communities) but without extrinsic mortality – thus isolating the effect of intrinsic deterioration or senescence. In direct contrast to the Williams prediction, high predation guppies are longer lived, have longer reproductive life span, and higher fecundities over the duration of their reproductive lives than do low-predation guppies. They also continue to reproduce into more advanced ages than guppies from low-predation environments. The pattern of senescence in regards to physiological performance was somewhat different. At an age of twelve months, guppies from high-predation environments had higher maximum acceleration during the escape response than guppies from low-predation environments. This physiological response declined more rapidly with age in high-predation guppies such that at an age of twenty-six months, survivors from both populations exhibited equal performance levels. The early life history of these fish was consistent with earlier studies on other populations of guppies from high- and low-predation environments (Reznick 1982; Reznick et al. 1996). These differences were as predicted by life history theory that models the evolutionary response to higher mortality rate (e.g. Charlesworth 1980; Gadgil & Bossert 1970; Law 1979; Michod 1979). Specifically, guppies from high-predation sites matured at an earlier age, had higher fecundity and had a higher rate of investment in reproduction. The fact that the late life history (pattern of senescence) is largely the opposite of the predicted pattern is an anomalous result that provides an important counter-example to the Williams prediction. Tozzini et al. (2013) compared captive life span and physiological senescence (lipofuscin accumulation in tissues) in replicated populations of annual killifish derived from geographical regions that differ in habitat duration. Annual killifish of the genus Nothobranchius inhabit temporary pools that completely dry on a seasonal basis. In the wild, the maximum life span of fish is dictated by the duration of the water bodies (aquatic pools) they inhabit. The species under study are found across a rainfall gradient (humid to dry) that has led to consistent differences in habitat duration between populations taken from the ends of the species’ distribution. Specifically, populations from a high-rainfall region close to the Indian Ocean experience consistently longer pool duration than those that inhabit inland low-rainfall regions. Since pool duration imposes an upper limit on fish life span in the wild, it is expected to be a strong selective factor shaping patterns of senescence. In particular, populations from dry regions, which have historically experienced short-duration aquatic environments and thus a consistently early truncation of life span, are expected to have relatively more rapid senescence when reared in captivity – relative to populations native to high-rainfall and longer-duration aquatic environments. An assumption is that pool desiccation on an annual basis imposes a bout of complete and condition-independent mortality on the entire population, but levels of extrinsic mortality during the time when pools are filled do not significantly differ between populations from these different geographical regions. Consistent with the Williams prediction, Tozzini et al. (2013) found that populations inhabiting dry regions have significantly shorter life spans and more rapid accumulation of lipofuscin in the brain and liver when reared in a common-garden laboratory environment. The accumulation of lipofuscin in tissues has been used as a biomarker of physiological ageing in studies of other fishes (Ding et al. 2010; Strauss 1999). Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:34:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.009

184

The Evolution of Senescence in Nature

These and other comparative ageing studies are summarised in Table 9.1. In short, some studies have found results that are inconsistent with the Williams prediction for some or all measures of senescence (Reznick et al. 2004; Sparkman et al. 2007; Walsh et al. 2014; see also Gray & Cade 2000; Miller et al. 2002; Morbey et al. 2005), while others have found support (Austad 1993; Dudycha & Tessier 1999; Robert & Bronikowski 2010; Tatar et al. 1997; Terzibasi et al. 2008; Tozzini et al. 2013).

Summary of Evidence The classic prediction, as originally stated by Williams (1957), has been readily adopted because it makes intuitive sense, accounts for many broad-scale comparative patterns of senescence and has garnered support from laboratory study. Yet, theoretical work suggests that there are situations in nature in which this prediction is unlikely to hold. Comparative studies of wild populations reveal cases that are inconsistent with the Williams prediction, although unambiguous empirical tests are relatively rare. Rather than being dismissed as anomalies, such cases warrant further examination. In this regard, experimental evolution study in Drosophila is important because it has directly tested the Williams prediction in a well-controlled laboratory setting in which mortality was applied randomly with respect to age (after eclosion) and condition and densities were tightly controlled (Stearns et al. 2000). The resulting match between theory and empirical work is reassuring. However, the question arises as to the applicability of these conditions in the wild and whether the relaxation of such assumptions fundamentally changes the predicted ageing response. In the sections that follow we discuss theory that explores these questions.

The Derived Theories: Increased Scope of Response Condition Dependence Recent theory argues for the importance of condition-dependent mortality in determining how patterns of senescence evolve in the wild (Abrams 1993; Williams & Day 2003). The central premise behind such theory is that if susceptibility to death by some environmental hazard (be it predation, temperature, starvation or disease) is dependent upon an organism’s physical condition or phenotype (i.e. is not completely extrinsic or haphazard), then selection can favour delayed senescence in the physiological trait(s) that affect susceptibility to this hazard (Abrams 1993, 2004; Bronikowski & Promislow 2005; Williams & Day 2003; Williams et al. 2006). If predators select the most senescent individuals in a population, an increase in predation will simultaneously increase mortality and select for slower rates of senescence over the most affected age classes (Abrams 1993). The primary prediction of Williams and Day (2003) is that an increase in condition-dependent mortality will favour decreased agerelated deterioration early in life and a steeper increase in age-related physiological

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:34:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.009

The Derived Theories: Increased Scope of Response

185

decline later on. This represents a positive-feedback loop – because the least robust individuals are most susceptible to death, selection favours individuals that have robust phenotypes, which, given various demographic assumptions, can lead to a variety of different responses (Williams & Day 2003; Williams et al. 2006). The most counter-intuitive prediction is that given high levels of conditiondependent mortality, delayed senescence can evolve, which runs counter to the Williams prediction. This theory could potentially go a long way towards explaining patterns observed in comparative studies of senescence in the wild that do not match the Williams prediction (Abrams 2004; Reznick et al. 2004; Williams et al. 2006). An ingenious experimental evolution study using the nematode worm Caenorhabditis remanei is particularly valuable in beginning to bridge the gap between this theory and comparative empirical work in the field (Chen & Maklakov 2012). By manipulating both mortality level (high and low) and mortality source (random versus condition dependent) in a crossed design, Chen and Maklakov (2012) were able to disentangle their effects. Chen and Maklakov (2012) quantified the evolutionary response in longevity in replicate populations of worms exposed to one of four selection regimes: high mortality applied haphazardly (i.e. condition independent), low mortality applied haphazardly, high mortality applied in a condition-dependent manner, and low mortality applied in a condition-dependent manner (Figure 9.1). Condition-dependent mortality was attained by exposing the populations to heat stress and transferring the most vigorous individuals to the next generation. Haphazard mortality was induced by randomly removing some proportion of the population so that mortality was independent of genotype. Importantly, mortality level in the two low-mortality treatments and the two high-mortality treatments (28.6 and 85.5 per cent per generation, respectively) were kept constant such that the only difference was whether the mortality was random or condition dependent. The crossed experimental design derives its power from built-in controls. Comparison of the high- and low-random mortality treatments allowed for testing of the Williams prediction (as in Stearns et al. 2000). These two treatments simultaneously served as controls when testing the effect of condition dependence. That is, if condition-dependent mortality had no effect on the evolution of senescence, then no differences are expected between high- and low-mortality treatments. When selection was applied haphazardly (random), high mortality led to the evolution of increased intrinsic mortality rates, corroborating the Williams prediction. When selection was condition dependent, high mortality led to the evolution of decreased intrinsic mortality rates, providing direct support for the importance of condition-dependent mortality in fully reversing the direction of the evolved response. Chen and Maklakov (2012) thus neatly reconciled these contrasting predictions. This experiment will surely become a classic and provides a potential explanation for the discordant results of empirical studies of senescence in natural populations. Condition-dependent mortality apparently leads to delayed senescence due to a positive genetic correlation (pleiotropy) between the trait(s) that are under selection and longevity or, more generally, between robustness or condition early and late in life (Chen & Maklakov 2012; Dowling 2012; Kimber & Chippindale 2013). There are Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:34:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.009

186

The Evolution of Senescence in Nature

Cumulative survivorship

Time

Mortality (condition independent) falls equally on all age classes

High, selective (condition Low, selective (condition High, random (condition Low, random (condition dependent) adult mortality dependent) adult mortality independent) adult mortality independent) adult mortality

A

B

D

E

Cumulative survivorship

C

Time

Figure 9.1

Time

Time

Time

Time

Graphical representation of potential evolutionary responses in longevity to given selection regimes. The initial population can be thought of as ancestral or baseline and having experienced a history of low adult mortality. Each replicate population can be thought of as undergoing a given selection regime for x number of generations, after which the evolutionary response would be measured as cumulative survivorship of a cohort in the absence of extrinsic mortality, thus isolating the effect of intrinsic deterioration (senescence) on survival. In A, condition-independent mortality falls equally upon all age classes. If population regulation is density independent or density dependent but not age specific, there is no predicted evolutionary change in patterns of senescence regardless of mortality level (Abrams 1993; Caswell 2007; Williams et al. 2006; see also Chapter 4). In B–E, either high or low mortality is applied to the adult stage in either a random (condition-independent) or selective (condition-dependent) manner (following the experimental design of Chen and Maklakov 2012). High adult mortality applied in a condition-dependent manner can (somewhat counter-intuitively) select for delayed senescence (B: theory: Abrams 1993; Williams & Day 2003; empirical: Chen & Maklakov 2012), while high adult mortality applied in a random (condition-independent) manner selects for more rapid senescence (D: Chen & Maklakov 2012; Stearns et al. 2000).

several biological reasons why heat stress as the condition-dependent mortality source was a particularly appropriate choice for nematode worms (Chen & Maklakov 2012). Here positive pleiotropy between ability to survive heat stress and robustness late in life is perhaps mediated by heat-shock proteins (Dowling 2012). If condition-dependent mortality is the norm in wild populations, then we might expect populations having experienced a history of high condition-dependent mortality to be superior in an array of phenotypes that, depending upon underlying genetic correlations, are only indirectly related to the phenotype under selection.

Density Dependence Abrams (1993) analysed how an increase in extrinsic mortality affects the evolution of senescence in populations that are exposed to density regulation, which may represent

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:34:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.009

Comparative Field Studies of Ageing and Condition-Dependent Mortality

187

a more realistic representation of life in a natural environment. Abrams found that the Williams prediction is upheld only under a specific set of assumptions regarding the nature of population growth. Specifically, when population growth is density independent and the intrinsic rate of increase r is the appropriate fitness measure (Charlesworth 1980), a uniform increase in extrinsic mortality, by definition, affects all age classes equally, does not change the age structure of a population and therefore does not change the optimal values for any life history traits (Abrams 1993; Bassar et al. 2010; Charlesworth 1980; Reznick et al. 2002; Taylor et al. 1974). When population growth is density dependent and relative fitness can be measured by net reproductive rate, increased extrinsic mortality can favour either an increase or decrease in the rate of senescence depending upon the age classes most affected by density regulation. Although the math can get complicated, the underlying logic is quite simple. A constant environment and stable population are assumed. An increase in extrinsic mortality, by definition age and condition independent, decreases population density. Under density-dependent population regulation, survivorship and fertility of remaining individuals are adjusted to maintain a stable population. The way senescence evolves in such circumstances is a function of how different age classes respond to changes in density. Consider the case where density dependence affects survivorship of older individuals to a greater extent than younger individuals. An increase in extrinsic mortality reduces population density. Older age classes are the prime beneficiaries of reduced densities, which afford them an increase in survivorship. This increase in survivorship in late age classes in turn favours the evolution of delayed senescence. In contrast, if density affects the survival of young age classes or the fertility of all age classes, then increased intrinsic mortality favours more rapid senescence. Lastly, if density equally affects survival of all age classes, a situation perhaps unlikely to occur in nature, then an increase in extrinsic mortality is predicted to affect no change in the rate of senescence, as in the densityindependent case.

Comparative Field Studies of Ageing and Condition-Dependent Mortality The likelihood of condition-dependent mortality in the wild depends upon the agent of selection. Fish or Daphnia in pools that dry out on a seasonal basis will all die, regardless of their condition. However, if death is instead caused by predators, disease or environmental stress, then mortality will likely be selective. Predation frequently acts as a potent agent of selection driving the evolution of prey phenotype (Endler 1986; Reznick et al. 1990). Furthermore, it is a prevailing idea in ecology that predators disproportionately target weak, injured, old or otherwise poorcondition prey (Curio 1976; Slobodkin 1968; Temple 1987) because they are easier to catch and subdue. This phenomenon is perhaps best documented in large mammalian carnivores. Mountain lions in the northern Front Range of Colorado selectively prey upon prion-infected mule deer (Krumm et al. 2010); wolves and coyotes disproportionately select prey impaired by malnutrition, age or disease (Crisler 1956; Gese & Goth Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:34:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.009

188

The Evolution of Senescence in Nature

1995; Lingle & Wilson 2001; Mech 1970); and lions on the Serengeti prey upon old wildebeest, zebra and buffalo (as judged by tooth wear) to a far greater extent than expected by their actual numbers (Schaller 1972). Predation has been demonstrated to impose selection upon prey escape performance (Ghalambor et al. 2004; Irschick et al. 2007; Janzen et al. 2007; Strobbe et al. 2010). In a semi-natural field enclosure experiment, Strobbe et al. (2010) demonstrated that dragonfly predators select for both swimming speed and underlying physiological performance (higher activity levels of the enzyme arginine kinase) in damselfly prey. In general, when predation represents a significant cause of mortality, it is likely to be applied in a condition-dependent manner because condition is likely to affect an individual’s ability to escape. Age is perhaps the main factor that reliably affects condition. It is under such circumstances that an increase in mortality can theoretically lead to an increase, decrease or unchanged rate of senescence depending upon density dependence (Williams et al. 2006) and what measures of ageing are used (Bronikowski & Promislow 2005). In five studies in which differences in mortality among populations are purported to be due to predation, one found results consistent with the Williams prediction (Austad 1993), two found results consistent for some measures of ageing and inconsistent for others (Robert & Bronikowski 2010; Sparkman et al. 2007), and two found results largely inconsistent with the Williams prediction (Reznick et al. 2004; Walsh et al. 2014). In all of these studies, mortality differences among populations were confirmed, often through mark-recapture, although the strength of evidence regarding predation as the cause of these mortality differences, and its age specificity, differs among studies. Issues that cannot be addressed by mark-recapture alone are the extent to which mortality differences among populations are due to extrinsic factors, like predation; intrinsic factors, like the consequences of high or low growth rates early in life; or an interaction between the two. In the mainland versus island population of opossums studied by Austad (1993), predation as cause of mortality differences was inferred due to a lack of predators on the island but not the mainland. Likewise, in a comparative study of senescence in Daphnia ambigua, the presence, absence or seasonal presence of a fish predator (the alewife) in different lakes was used as a reliable proxy of predation pressure. Avian predators are more abundant in lakeshore (versus mountain) habitats of the Western terrestrial garter snake (Sparkman et al. 2013), although snakes from lakeshore populations also have a more abundant and consistent food supply, experience warmer temperatures and are infected with different species of parasites (Bronikowski & Arnold 1999; Bronikowski & Vleck 2010) which likely all contribute to variation along a fast-slow life history continuum. The higher mortality rates sustained by guppies in high-predation localities, as revealed with mark-recapture studies, are accompanied by other evidence of predator-induced mortality. There are direct observations of predators attacking guppies in the field (Endler 1978) or preying on guppies in designed laboratory experiments (Mattingly & Butler 1994). We are not aware of any comparative ageing studies in which disease-induced mortality has been identified as a factor that shapes the evolution of senescence. However, it is known that immune function generally declines with age, particularly acquired immunity, making older age classes more susceptible to disease. Palacios Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:34:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.009

Conclusion and Suggestions for Future Work

189

et al. (2007) provide a fairly comprehensive examination of in vitro and in vivo senescence in components of the immune system in free-living tree swallows and found that some components show age-related declines, while other aspects of immune function remain unchanged. Mortality owing to disease is likely to be age and condition dependent. When mortality is caused by abiotic factors, it may or may not be condition dependent depending upon the nature of the environment. The ability to survive cold winter temperatures has been shown to be condition or size dependent (Cote & FestaBianchet 2001; Korslund & Steen 2006; Rodel 2004; Unsworth et al. 1999). Bumpus (1899), for example, found that the English sparrows killed by a winter storm were not a random subset of the population: ‘We found that there are fundamental differences between the surviving birds and those eliminated, and we conclude that the birds which survived because they possessed certain structural characters.’ There are, however, certain instances in which mortality caused by environmental conditions is not dependent upon an individual’s condition or phenotype. These cases can be rather informative because it is here that we would expect support for the Williams prediction to be strongest. Organisms that have an annual life cycle and live in seasonal environments that are only hospitable by the adult phase for a portion of the year are representative of this scenario; the end of the growth season, marked by the first hard freeze or the drying of a vernal pool, will kill all individuals, regardless of condition. If there are differences among populations in the duration of season length, then populations are expected to exhibit local adaptation. In particular, the Williams hypothesis predicts that populations from shorter-duration habitats would evolve more rapid senescence because, after the mean date of catastrophe, the late-life history is shielded from selection. Furthermore, density regulation might be less likely in environments where the interval favourable for growth is short and/or unpredictable in duration. Three comparative studies of ageing in the residents of seasonal environments – annual killifish adapted to seasonally ephemeral aquatic pools across a rainfall gradient (Tozzini et al. 2013), grasshoppers that inhabit regions with different growing season lengths across an elevation gradient (Tatar et al. 1997) and Daphnia found in seasonally ephemeral versus permanent aquatic environments (Dudycha & Tessier 1999) – all unambiguously support the classic prediction for all the various measures of senescence used (Table 9.1).

Conclusion and Suggestions for Future Work Theoreticians have shown that adding ecological complexity such as density regulation and condition dependence can modify how extrinsic risk of mortality is predicted to shape the evolution of senescence. Intra-specific comparative studies in the wild conform with the view that extrinsic risk of mortality alone is often an insufficient descriptor of the selection imposed on the organism. For instance, the onset of winter or seasonal habitat drying in annual organisms represents non-selective mortality risk, whereas predation, disease and environmental stress are likely to be selective. Furthermore, mortality risk is often confounded with other features of the Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:34:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.009

190

The Evolution of Senescence in Nature

environment, such as density and resource levels. In nature, it is quite unlikely for populations to be identical with respect to all environmental factors save the level of extrinsic mortality. One reason for this is that differences among populations in mortality risk will often affect population density and hence per capita resource availability (Walsh 2013). If such effects do not fall equally upon all age classes, then the range of possible evolutionary outcomes increases (Abrams 1993; Charlesworth 1980). Such density dependence may be common in natural populations (Bassar et al. 2010). Thus, we are left with several discordant empirical results that do not fit with classic theory and with a body of theory that takes into account the realistic assumptions of condition- and density-dependent mortality in the wild. Chen and Maklakov (2012) began to experimentally bridge this gap by demonstrating that mortality level and whether or not mortality was condition dependent played interactive roles in shaping the evolution of senescence. Increased mortality applied over thirteen generations lead to the evolution of either increased or decreased longevity depending upon whether such mortality is applied in a random or condition-dependent manner. This experiment provides a potential explanation for the discordant results of the empirical studies of senescence in natural populations. If our goal is to understand how senescence evolves in nature, then what remains to be done is to wed theory, comparative studies of natural populations and designed experiments. We suggest an approach similar to what was developed in the discipline of ecological genetics in the 1960s – an integration of field studies that characterise how selection works in nature with designed experiments that address the candidate agents of selection defined by the field programme. The field component of such a research programme must include more than an assessment of differences among populations/ species in risk of mortality. We must also know the causes of mortality and assess whether or not mortality is dependent upon condition, age or size or some combination of these factors (Chen & Maklakov 2012; Williams & Day 2003). Recent quantitative genetic studies of senescence in wild populations based upon longitudinal markrecapture data combined with pedigrees provide this sort of detailed information (reviewed in Charmantier et al. 2014; Wilson et al. 2008). Studies have established that measurable senescence occurs in natural populations (Nussey et al. 2013; see also Chapters 7 and 8), there are inter-individual differences in senescence manifested as genotype-by-age interactions (Hayward et al. 2013; Wilson et al. 2007), there are often trade-offs between early- and late-life performance (Bouwhuis et al. 2010; Descamps et al. 2006; Reed et al. 2008) and differences in rate of senescence among different measured traits is relatively common (Charmantier et al. 2014; Hayward et al. 2013). Longitudinal mark/recapture studies (Nussey et al. 2008) may be well suited to test whether mortality is condition dependent because each time an organism is captured or re-sighted, indices of condition can be obtained (e.g. parasite counts, blood samples or body weight). Working with species that have short generation times and are readily cultured in the laboratory would facilitate the integration of field studies with laboratory experiments. Experimental evolution studies are particularly powerful in that they can be factorial and tailored to test the effect of mortality level and different types of selection/mortality Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:34:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.009

Acknowledgements

191

characterised in the field setting and incorporate differences in population density or food availability (Mueller 1987). Both direct mortality and the indirect effects of mortality on density and resource availability could potentially impose selection on development rate and ageing, and these factors might interact. Prior studies also suggest the importance of including multiple measures of senescence – not only intrinsic mortality but also measures of physiological deterioration and deterioration in reproductive performance because the measure used can often dictate interpretation (Bronikowski and Promislow 2005; Nussey et al. 2013). If mortality is conditiondependent, then the particular traits that most determine susceptibility to mortality – escape performance, immune function, resistance to environmental stress (and those phenotypes genetically correlated with such traits) – are presumably under strong selection and might exhibit delayed senescence early in life relative to other traits that do not affect mortality risk. Lastly, a fuller understanding of the evolution of senescence will require a better understanding of underlying life history trade-offs. For example, if resources are limiting, any aspect of maintenance that requires resources could contribute to the trade-offs that shape senescence. We traditionally think of life history trade-offs, such as growth or reproductive investment, but immunity and disease resistance also could come into play. For instance, in low-risk environments where there is a high probability of reaching old age, selection may favour significant investment in immune function. In high-risk environments where organisms are unlikely to live very long, selection in favour of high reproductive investment early in life might be made at the expense of investment in immune function. A standard laboratory in which senescence of populations adapted to such high- and low-risk environments are being compared will be free of extrinsic mortality risk and exclude risk of disease and parasitism. Individuals from high-risk environments could reap the benefit of not investing heavily in immune function without paying the cost of increased susceptibility to disease in such a ‘common garden’. If there are significant population (genotype)-byenvironment interactions, then a common-garden laboratory environment free of extrinsic mortality may not provide a fair representation of the relative rates of senescence. The fundamental premises underlying the evolutionary theory of senescence are strong (Rose et al. 2007). Yet, models made more realistic by including biological and ecological complexity (e.g. condition-dependent mortality and density dependence) are capable of generating predictions regarding the evolution of senescence that sometimes run counter to the Williams prediction. There is now sufficient empirical research to argue that these added complexities matter and that the scope of possible evolutionary outcomes is greater than originally anticipated.

Acknowledgements We thank Anne Bronikowski, Daniel Promislow and Owen Jones for helpful comments.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:34:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.009

192

The Evolution of Senescence in Nature

References Abrams, P. A. (1993). Does increased mortality favor the evolution of more rapid senescence? Evolution, 47(3), 877–87. Abrams, P. A. (2004). Mortality and lifespan. Nature, 431, 1048–9. Austad, S. N. (1993). Retarded senescence in an insular population of Virginia opossums (Didelphis virginiana). Journal of Zoology, 229(4), 695–708. Austad, S. N. & Fischer, K. E. (1991). Mammalian aging, metabolism, and ecology: evidence from the bats and marsupials Journal of Gerontology, 46(2), B47–53. Bassar, R. D., López Sepulcre, A., Walsh, M. R., et al. (2010). Bridging the gap between ecology and evolution: integrating density regulation and life history evolution. Annals of the New York Academy of Sciences, 1206(1), 17–34. Blanco, M. A. & Sherman, P. W. (2005). Maximum longevities of chemically protected and non-protected fishes, reptiles, and amphibians support evolutionary hypotheses of aging. Mechanisms of Ageing and Development, 126(6–7), 794–803. Bouwhuis, S., Charmantier, A., Verhulst, S., et al. (2010) Individual variation in rates of senescence: natal origin effects and disposable soma in a wild bird population. Journal of Animal Ecology, 79(6), 1251–61. Bronikowski, A. & Vleck, D. (2010). Metabolism, body size and life span: a case study in evolutionarily divergent populations of the garter snake (Thamnophis elegans). Integrative and Comparative Biology, 50(5), 880–7. Bronikowski, A. M. & Arnold, S. J. (1999). The evolutionary ecology of life history variation in the garter snake Thamnophis elegans. Ecology, 80(7), 2314–25. Bronikowski, A. M. & Promislow, D. E. L. (2005). Testing evolutionary theories of aging in wild populations. Trends in Ecology and Evolution, 20(6), 271–3. Bumpus, H. C. (1899). The elimination of the unfit as illustrated by the introduced sparrow, Passer domesticus. Biological Lectures, Woods Hole Marine Biological Station, 6, 209–26. Caswell, H. (2007). Extrinsic mortality and the evolution of senescence. Trends in Ecology and Evolution, 22(4), 173–4. Charlesworth, B. (1980). Evolution in Age-Structured Populations (Cambridge University Press). Charlesworth, B. & Hughes, K. A. (1996). Age-specific inbreeding depression and components of genetic variance in relation to the evolution of senescence. Proceedings of the National Academy of Sciences of the United States of America, 93(12), 6140–5. Charmantier, A., Brommer, J. E. & Nussey, D. H. (2014). The quantitative genetics of senescence in wild animals. In Quantitative Genetics in the Wild, ed. A. Charmantier, D. Garant & L. E. B. Kruuk (pp. 68–83 ) (Oxford University Press). Chen, H. & Maklakov, A. (2012). Longer life span evolves under high rates of condition-dependent mortality. Current Biology, 22, 2140–3. Cote, S. D. & Festa-Bianchet, M. (2001). Birthdate, mass and survival in mountain goat kids: effects of maternal characteristics and forage quality. Oecologia, 127(2), 230–8. Crisler, L. (1956). Observations of wolves hunting caribou. Journal of Mammalogy, 37, 337–46. Curio, E. (1976). Ethology of Predation (Berlin: Springer-Verlag).

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:34:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.009

References

193

Descamps, S., Boutin, S., Berteaux, D., et al. (2006). Best squirrels trade a long life for an early reproduction. Proceedings of the Royal Society of London Series B: Biological Sciences, 273, 2369–74. Ding, L., Kuhne, W. W., Hinton, D. E., et al. (2010). Quantifiable biomarkers of normal aging in the Japanese medaka fish (Oryzias latipes). PLoS ONE, 5(10), 1–11. Dowling, D. K. (2012). Aging: evolution of life span revisited. Current Biology, 22(22), R947–9. Dudycha, J. L. & Tessier, A. J. (1999). Natural genetic variation of life span, reproduction, and juvenile growth in Daphnia. Evolution, 53(6), 1744–56. Endler, J. A. (1978). A predator’s view of animal color patterns. In Evolutionary Biology, ed. M. K. Hecht, W. C. Steere & B. Wallace (pp. 319–64 ) (New York: Plenum Press). Endler, J. A. (1986). Natural Selection in the Wild (Princetion, NJ: Princeton University Press). Finch, C. E. (1990). Longevity, Senescence, and the Genome (University of Chicago Press). Gadgil, M. & Bossert, W. H. (1970). Life historical consequences of natural selection. American Naturalist, 104(935), 1–24. Gese, E. M. & Grothe, S. (1995). Analysis of coyote predation on deer and elk during winter in Yellowstone National park, Wyoming. American Midland Naturalist, 133, 36–43. Ghalambor, C. K., Reznick, D. N. & Walker, J. A. (2004). Constraints on adaptive evolution: the functional trade-off between reproduction and fast-start swimming performance in the Trinidadian guppy (Poecilia reticulata). American Naturalist, 164(1), 38–50. Gray, D. A. & Cade, W. H. (2000). Senescence in field crickets (Orthoptera; Gryllidae): examining the effects of sex and a sex-biased parasitoid. Canadian Journal of Zoology, 78(1), 140–3. Hamilton, W. D. (1966). Moulding of senescence by natural selection. Journal of Theoretical Biology, 12(1), 12–45. Hayward, A. D., Wilson, A. J., Pilkington, J. G., et al. (2013). Reproductive senescence in female Soay sheep: variation across traits and contributions of individual ageing and selective disappearance. Functional Ecology, 27, 184–95. Holmes, D. J. & Austad, S. N. (1995) The evolution of avian senescence patterns: implications for understanding primary aging processes. American Zoologist, 35(4), 307–17. Irschick, D. J., Bailey, J. K., Schweitzer, J., et al. (2007). New directions for studying selection in nature: studies of performance and communities. Physiological and Biochemical Zoology, 80, 567–77. Janzen, F. J., Tucker, J. K. & Paukstis, G. L. (2007). Experimental analysis of an early life-history stage: direct or indirect selection on body size of hatchling turtles? Functional Ecology, 21, 162–70. Kawasaki, N., Brassil, C., Brooks, R., et al. (2008). Environmental effects on the expression of life span and aging: an extreme contrast between wild and captive cohorts of Telostylinus angusticollis (Diptera: Neriidae). American Naturalist, 172 (3), 346–57.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:34:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.009

194

The Evolution of Senescence in Nature

Keller, L. & Genoud, M. (1997). Extraordinary lifespans in ants: a test of evolutionary theories of ageing. Nature, 389(6654), 958–60. Kimber, C. M. & Chippindale, A. K. (2013). Mutation, condition, and the maintenance of extended lifespan in Drosophila. Current Biology, 23, 2283–7. Korslund, L. & Steen, H. (2006) Small rodent winter survival: snow conditions limit access to food resources. Journal of Animal Ecology, 75(1), 156–66. Krumm, C. E., Conner, M. M., Hobbs, N. T., et al. (2010). Mountain lions prey selectively on prion-infected mule deer. Biology Letters, 6, 209–11. Law, R. (1979). Optimal life histories under age-specific predation. American Naturalist, 114(3), 399–417. Lingle, S. & Wilson, W. F. (2001). Detection and avoidance of predators in white-talied deer (Odocoileus virgianianus) and mule deer (O. hemionus). Ethology, 107, 125–47. Luckinbill, L. S., Arking, R., Clare, M.J., et al. (1984). Selection for delayed senescence in Drosophila melanogaster. Evolution, 38(5), 996–1003. Luckinbill, L. S. & Clare, M. J. (1985). Selection for life span in Drosophila melanogaster. Heredity, 55(1), 9–18. Mattingly, H. T. & Butler, M. J., IV (1994). Laboratory predation on the Trinidadian guppy: implications for the size-selective predation hypothesis and guppy life history evolution. Oikos, 69(1), 54–64. Mech, L. D. (1970). The Wolf: Ecology and Behavior of an Endangered Species (Garden City, NJ: American Museum of Natural History Press). Medawar, P. B. (1952). An Unsolved Problem of Biology (London: Lewis). Michod, R. E. (1979). Evolution of life histories in response to age-specific mortality factors. American Naturalist, 113(4), 531–50. Miller, R. A., Harper, J. M., Dysko, R. C., et al. (2002). Longer life spans and delayed maturation in wild-derived mice. Experimental Biology and Medicine, 227(7), 500–8. Moorad, J. A. & Promislow, D. E. L. (2009). What can genetic variation tell us about the evolution of senescence? Proceedings of the Royal Society of London Series B: Biological Sciences, 276, 2271–8. Morbey, Y. E., Brassil, C. E. & Hendry, A. P. (2005). Rapid senescence in pacific salmon. American Naturalist, 166(5), 556–68. Mueller, L. D. (1987). Evolution of accelerated senescence in laboratory populations of Drosophila. Proceedings of the National Academy of Sciences of the United States of America, 84(7), 1974–7. Nussey, D. H., Coulson, T., Festa-Bianchet, M., et al. (2008). Measuring senescence in wild animal populations: towards a longitudinal approach. Functional Ecology, 22, 393–406. Nussey, D. H., Froy, H., Lemaître, J.-F., et al. (2013). Senescence in natural populations of animals: widespread evidence and its implications for bio-gerontology. Ageing Research Reviews, 12, 214–25. Palacios, M. G., Cunnick, J. E., Winkler, D. W., et al. (2007). Immunosenescence in some but not all immune components in a free-living vertebrate, the tree swallow. Proceedings of the Royal Society of London Series B: Biological Sciences, 274, 951–7. Partridge, L. & Barton, N. H. (1993). Optimality, mutation and the evolution of aging. Nature, 362(6418), 305–11.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:34:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.009

References

195

Partridge, L. & Gems, D. (2002). Mechanisms of ageing: public or private? Nature Reviews Genetics, 3(3), 165–75. Pletcher, S. D., Khazaeli, A. A. & Curtsinger, J. W. (2000). Why do life spans differ? Partitioning mean longevity differences in terms of age-specific mortality parameters. Journals of Gerontology Series A: Biological Sciences and Medical Sciences, 55(8), B381–9. Promislow, D. E. L. (1991). Senescence in natural populations of mammals: a comparative study. Evolution, 45(8), 1869–87. Reed, T. E., Kruuk, L. E. B., Wanless, S., et al. (2008). Reproductive senescence in a long-lived seabird: rates of decline in late-life performance are associated with varying costs of early reproduction. American Naturalist, 171(2), E89–101. Reznick, D. (1993). New model systems for studying the evolutionary biology of aging: crustacea. Genetica, 91(1), 79–88. Reznick, D. & Bryant, M. (2007). Comparative long-term mark-recapture studies of guppies (Poecilia reticulata): differences among high and low predation localities in growth and survival. Annales Zoologici Fennici, 44, 152–60. Reznick, D., Bryant, M. J. & Bashey, F. (2002). r-and K-selection revisited: the role of population regulation in life-history evolution. Ecology, 83(6), 1509–20. Reznick, D., Bryga, H. & Endler, J. A. (1990). Experimentally induced life-history evolution in a natural population. Nature, 346, 357–9. Reznick, D. & Endler, J. A. (1982). The impact of predation on life-history evolution in Trinidadian guppies (Poecilia reticulata). Evolution, 36(1), 160–77. Reznick, D. N., Bryant, M. J., Roff, D., et al. (2004). Effect of extrinsic mortality on the evolution of senescence in guppies. Nature, 431(7012), 1095–9. Reznick, D. N., Butler, M. J., Rodd, F. H., et al. (1996). Life-history evolution in guppies (Poecilia reticulata): 6. Differential mortality as a mechanism for natural selection. Evolution, 50(4), 1651–60. Ricklefs, R. E. (1998) Evolutionary theories of aging: confirmation of a fundamental prediction, with implications for the genetic basis and evolution of life span. American Naturalist, 152(1), 24–44. Ricklefs, R. E. & Scheuerlein, A. (2002). Biological implications of the Weibull and Gompertz models of aging. Journals of Gerontology Series A: Biological Sciences and Medical Sciences, 57(2), B69–76. Robert, K. A. & Bronikowski, A. M. (2010). Evolution of senescence in nature: physiological evolution in populations of garter snake with divergent life histories. American Naturalist, 175(2), 147–59. Rodel, H. G., Bora, A., Kaetzke, P., et al. (2004). Over-winter survival in subadult European rabbits: weather effects, density dependence, and the impact of individual characteristics. Oecologia, 140(4), 566–76. Rose, M. R. (1984). Laboratory evolution of postponed senescence in Drosophila melanogaster. Evolution, 38(5), 1004–10. Rose, M. R. (1991). Evolutionary Biology of Aging (New York: Oxford University Press). Rose, M. R., Rauser, C. L., Benford, G., et al. (2007). Hamilton’s forces of natural selection after forty years. Evolution, 61(6), 1265–76. Schaffer, W. M. (1974). Optimal reproductive effort in fluctuating environments. American Naturalist, 108(964), 783–90.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:34:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.009

196

The Evolution of Senescence in Nature

Schaller, G. B. (1972). The Serengeti Lion: A Study in Predator-Prey Relations (University of Chicago Press). Sherman, P. W. & Jarvis, J. U. M. (2002). Extraordinary life spans of naked mole-rats (Heterocephalus glaber). Journal of Zoology, 258(3), 307–11. Slobodkin, L. B. (1968). How to be a predator. American Zoologist, 8(1), 43–51. Snoke, M. S. & Promislow, D. E. L. (2003). Quantitative genetic tests of recent senescence theory: age-specific mortality and male fertility in Drosophila melanogaster. Heredity, 91(6), 546–56. Sparkman, A. M., Arnold, S. J. and Bronikowski, A. M. (2007). An empirical test of evolutionary theories for reproductive senescence and reproductive effort in the garter snake Thamnophis elegans. Proceedings of the Royal Society of London Series B: Biological Sciences, 274(1612), 943–50. Sparkman, A. M., Bronikowski, A. M., Billings, J. G., et al. (2013). Avian predation and the evolution of life histories in the garter snake Thamnophis elegans. American Midland Naturalist, 170(1), 66–85. Stearns, S. C., Ackermann, M., Doebeli, M., et al. (2000). Experimental evolution of aging, growth, and reproduction in fruitflies. Proceedings of the National Academy of Sciences of the United States of America, 97(7), 3309–13. Strauss, R. E. (1999). Brain-tissue accumulation of fluorescent age pigments in four Poeciliid fishes (Cyprinodontiformes) and the estimation of ‘biological age’. Growth Development and Aging, 63(4), 151–70. Strobbe, F., McPeek, M. A., De Block, M., et al. (2010). Survival selection imposed by predation on a physiological trait underlying escape speed. Functional Ecology, 24, 1306–12. Tatar, M., Gray, D. W. & Carey, J. R. (1997). Altitudinal variation for senescence in Melanoplus grasshoppers. Oecologia, 111(3), 357–64. Taylor, H. M., Gourley, R. S., Lawrence, C. E., et al. (1974). Natural-selection of life-history attributes: analytical approach. Theoretical Population Biology, 5(1), 104–22. Temple, S. A. (1987). Do predators always capture substandard individuals disproportionately from prey populations? Ecology, 68(3), 669–74. Terzibasi, E., Valenzano, D. R., Benedetti, M., et al. (2008). Large differences in aging phenotype between strains of the short-lived annual fish Nothobranchius furzeri. PLoS ONE, 3(12), 1–13. Tozzini, E. T., Dorn, A., Ng’oma, E., et al. (2013). Parallel evolution of senescence in annual fishes in response to extrinsic mortality. BMC Evolutionary Biology, 13(77), 1–12. Unsworth, J. W., Pac, D. F., White, G. C., et al. (1999). Mule deer survival in Colorado, Idaho, and Montana. Journal of Wildlife Management, 63(1), 315–26. Walsh, M. R. (2013). The evolutionary consequences of indirect effects. Trends in Ecology and Evolution, 28(1), 23–9. Walsh, M. R., Whittington, D. & Walsh, M. J. (2014) Does variation in the intensity and duration of predation drive evolutionary changes in senescence? Journal of Animal Ecology, 83, 1279–88. Williams, G. C. (1957). Pleiotropy, natural selection, and the evolution of senescence. Evolution, 11(4), 398–411.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:34:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.009

References

197

Williams, P. D. & Day, T. (2003). Antagonistic pleiotropy, mortality source interactions, and the evolutionary theory of senescence. Evolution, 57(7), 1478–88. Williams, P. D., Day, T., Fletcher, Q., et al. (2006). The shaping of senescence in the wild. Trends in Ecology and Evolution, 21(8), 458–63. Wilson, A. J., Charmantier, A. & Hadfield, J. D. (2008). Evolutionary genetics of ageing in the wild: empirical patterns and future perspectives. Functional Ecology, 22, 431–42. Wilson, A. J., Nussey, D. H., Pemberton, J. M., et al. (2007). Evidence for a genetic basis of aging in two wild vertebrate populations. Current Biology, 17, 2136–42.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:34:36, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.009

10 Explaining Extraordinary Life Spans The Proximate and Ultimate Causes of Differential Life Span in Social Insects Eric Lucas and Laurent Keller

Short Summary The striking differences in life span observed among some social insect castes offer unique opportunities to study ageing and have therefore attracted increasing attention. While evolutionary theories of ageing can explain the long life span of social insect queens, experimental evidence to support them is lacking or contradictory. Furthermore, how social insects age is still poorly understood. Senescence patterns vary between behavioural worker castes, and senescence in honeybee workers can even be reversed by inducing a caste transition; but explicit comparisons between queens and workers are needed to understand how queen longevity is linked to senescence. The ability of queens in advanced insect societies to combine long life spans with high investment in reproduction presents a physiological puzzle that may be solved in honeybees by the unique relationship between juvenile hormone and vitellogenin. How this is achieved in other species remains unclear. We finish with a consideration of the challenges facing research into social insect ageing and discuss how these can best be met.

Introduction: What Is So Particular about Social Insects? Social insects, particularly those of advanced insect societies such as ants and honeybees, have attracted increasing interest in recent years as a focus in the study of ageing. Given the existence of model organisms that have been extensively studied and for which a wealth of data and manipulative techniques have been developed (Dietzl et al. 2007; Duffy 2002; Hamilton et al. 2005), why should we dedicate our time to the study of social insects? One of the main draws of social insects for ageing research is the extreme life spans found in many species. The longest-lived social insects can live for well over twenty years and are, to our knowledge, the longest-lived of all adult insects. Investigating how they achieve this remarkable longevity may elucidate processes involved in extremely long life that cannot be detected in organisms that live a few weeks at most (Jemielity et al. 2005). Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:35:23, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.010

Introduction: What Is So Particular about Social Insects?

199

The second interesting characteristic is the huge variation in life span that exists in some species between individuals bearing the same genome but different phenotypes. So-called eusocial insect societies are defined by their reproductive division of labour between behavioural castes (Wilson 1971). A colony of eusocial insects will typically be composed of one or several reproductive queens who dominate reproduction and a group of workers who conduct non-reproductive tasks such as caring for the brood and queens, building and maintaining the colony and foraging (Wilson 1971). In the most advanced eusocial insects, such as honeybees and most species of ants, queens and workers are morphologically and irreversibly differentiated. In these species, queens have highly developed ovaries, allowing them to lay eggs at high rates (well over 1,000 eggs per day in honeybee queens) (Tanaka & Hartfelder 2004; Wilson 1971). Workers are smaller than queens, typically cannot mate and generally do not have developed ovaries. While the way in which morphological caste is determined can in some cases have a large genetic component, eggs and young larvae of most species are totipotent with respect to caste (Schwander et al. 2010). The same genotype is therefore capable of creating individuals of different castes. Division of labour can also exist among workers. In many species there is morphological variation between workers that suits them to specific tasks, ranging from simple variation in size to highly specific castes (Wilson 1971). Even in species in which workers are morphologically monomorphic, they may nevertheless display strong behavioural polymorphism as they specialise in a different subset of the tasks required for colony growth and reproduction. This is often a result of individuals progressing through tasks as they age (Mersch et al. 2013; Robinson 1992; Seeley 1982), a process known as ‘age polyethism’. The fundamental differences in behaviour and body shape between eusocial insect castes are accompanied by an equally varied physiology and sometimes huge differences in adult life span (Kramer & Schaible 2013). In the black garden ant Lasius niger, for example, the oldest recorded queen lived twenty-nine years (Hölldobler & Wilson 1990), whereas workers, even in laboratory conditions, live no more than one or two years. Among workers, differences in life span have also been recorded between different castes (Chapuisat & Keller 2002). It is therefore possible to study not only the processes at work in extremely long-lived insects but also the differences in these processes between conspecifics whose life spans are, in some cases, an order of magnitude smaller. Another area in which research on social insects is highly rewarding is the study of the effect that the social environment, that is, the conspecifics in an individual’s presence, has on life span (Amdam et al. 2009; Carey 2001; Hartmann & Heinze 2003; Rueppell et al. 2008, 2015; Schrempf et al. 2011; Smedal et al. 2009). The evolution of sociality in insects is associated with a marked increase in the life span of reproductive individuals (Keller & Genoud 1997), and the role that an individual plays in the society has a strong effect on life span, probably more so than anything else (Remolina & Hughes 2008; Rueppell 2009). As humans also live in cooperative societies, the idea that social interactions can significantly affect life span is of inherent interest. Researchers have also been interested by the fact that social insect queens have both long life spans and high fecundity. In most organisms that have been studied, increasing Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:35:23, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.010

200

Explaining Extraordinary Life Spans

fertility comes at a cost to life span, and long-lived phenotypes are often associated with low reproductive output (Flatt 2011). Queens, however, lay eggs at very high rates (Tanaka & Hartfelder 2004; Wilson 1971) and yet have the highest longevity of any individual in the colony. This has prompted research into the mechanisms of how this apparent exception to the rule is achieved (Corona et al. 2007). In this review we examine, from an evolutionary and physiological perspective, what research in recent years has taught us about ageing in social insects.

Extraordinary Life Spans The heterogeneity and extremes in life span found in social insects appear to be directly linked to a social mode of life with caste differentiation (Keller & Genoud 1997; Kramer & Schaible 2013). The challenge is to understand the basis of this longevity from both an evolutionary point of view and a physiological one.

Selective Pressure Affecting Senescence Tightly linked to life span is the process of senescence, that is, the increasing frailty and degeneration that typically accompanies ageing. To the extent that senescence leads to increased mortality, extraordinary life spans must be accompanied by extraordinarily slow senescence, without which mortality will eventually reach high levels and become a limiting factor to longevity. Why we grow weaker as we reach old age is a question that has been the focus of much theoretical attention. Evolutionary explanations of senescence are typically inspired by three fundamental theories: mutation accumulation, antagonistic pleiotropy and somatic maintenance. These all rely on the same underlying fact that the strength of selection on a trait decreases with the age at which it applies, mainly because the survival curve of a given organism is always decreasing: the probability of an individual surviving to a given age t1 will always be smaller than the probability of it surviving to some later age t2 because of mortality due to external factors such as predation or disease. While this underlying basis is the same, the theories nevertheless are different in that mutation accumulation is essentially an argument about genetic drift, while the other two provide adaptive explanations (Partridge & Barton 1993). The theory of mutation accumulation posits that deleterious mutations expressed at older ages will accumulate more rapidly than those that affect younger ages (Medawar 1952). Deleterious mutations that appear in the genome can be removed by natural selection, but these random mutations are continuously appearing and, by chance, occasionally spread to fixation. The accumulation of these mutations is directly linked to the power of natural selection to remove them. Because selection is weaker at older ages, a greater load of deleterious mutations will be found in genes that are relevant at age t2 than in those relevant at age t1. Antagonistic pleiotropy (Williams 1957) again relies on selection at older ages being weaker than that at younger ages but considers the fate of mutations with pleiotropic Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:35:23, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.010

Extraordinary Life Spans

201

effects across ages (i.e. mutations that have one effect at one age and a different effect at another). Imagine a mutation that confers a selective advantage in young individuals (age t1) but carries an equal and opposite cost in older individuals (age t2). Such a mutation would spread because of the weaker strength of selection at age t2. If we assume that this pleiotropy is inevitable (i.e. that it is impossible to achieve the benefit at age t1 without incurring the cost at age t2), then this cost at older ages will become fixed in the population and lead to senescence. In this case, in contrast to mutation accumulation, selection is driving the evolution of senescence rather than being unable to prevent it. The disposable soma theory (Kirkwood 1977) approaches the question of senescence somewhat differently. Like all complex entities, organisms slowly wear down with time. Investing in repairing this wear and tear presumably diverts energy away from other processes, such as reproduction, leading to a trade-off between reproduction and maintenance. The theory of disposable soma proposes that it is usually not optimal to invest sufficiently in the maintenance process to completely avoid senescence (Kirkwood & Holliday 1979). This is because the benefit of investing into reproduction will be obtained in the present, whereas the cost of failing to invest in maintenance will be borne in the future in the form of deteriorating condition. In effect, a mutation that changes the level of somatic maintenance is antagonistically pleiotropic because decreasing investment into maintenance gives an immediate increase in fecundity and a delayed increase in mortality. In certain cases, such as when extrinsic mortality is very low, the delayed cost may nevertheless be greater than the immediate benefit. In these cases, organisms should invest sufficiently in maintenance to experience negligible senescence (Baudisch & Vaupel 2010). The scarcity of organisms that are known to escape senescence, however, implies that such cases may be rare. In summary, mutation accumulation suggests that senescence is due to age-specific deleterious mutations spreading by drift, while antagonistic pleiotropy and disposable soma propose that selection drives the evolution of optimal strategies at young ages despite future negative repercussions. All these theories rely on the premise that the strength of selection declines with age. This decline is itself directly affected by the rate of extrinsic mortality (i.e. mortality due to extrinsic factors such as predation or disease): the stronger the rate of extrinsic mortality, the fewer individuals will survive to older ages, the weaker selection will be at those ages, and the faster the organism will senesce due to mutation accumulation, antagonistic pleiotropy and lack of somatic maintenance. These broad ideas are consistent with observed longevities in insect societies. Social insect queens, once they have become established, lead very sheltered lives in which they for the most part remain inside their nest and are cared for and fed by their worker force (Wilson 1971). In this way, they enjoy a level of protection from predation that is far higher than those experienced by solitary insects or workers, which must leave their nests to find food. The evolution of sociality is therefore associated with a reduction in the extrinsic mortality of queens and should lead to reduced senescence and increased life span in this caste (Baudisch & Vaupel 2010; Keller & Genoud 1997). Similarly, it has been shown that in weaver ants the worker caste that forages and thus experiences higher levels of Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:35:23, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.010

202

Explaining Extraordinary Life Spans

extrinsic mortality has a shorter life span than other castes even in laboratory conditions, where extrinsic mortality is removed (Chapuisat & Keller 2002). This again follows the predictions of evolutionary theories of ageing. In the light of these theories, it is interesting to consider the fate of a mutation affecting life span or senescence in social insects. Because queens and workers do not differ genetically, one might consider that any mutation that affects the life span of queens must also affect that of workers. However, queens and workers successfully differ in all manner of characteristics despite their lack of genetic differentiation. This is reflected by substantial levels of caste-biased gene expression (Ferreira et al. 2013; Grozinger et al. 2007; Judice et al. 2006; Ometto et al. 2011). A mutation might therefore be caste specific by affecting a gene that is tightly linked to one caste and not the other. In the case of mutation accumulation, the fate of a mutation that is both age and caste specific will differ depending on whether it is specific to queens or to workers. In the case of somatic maintenance, a mutation might affect, for example, the extent of DNA repair in queens but not workers. For such caste-specific genes, all the same arguments as presented earlier apply. The fate of a mutation that does affect life span similarly across castes is also of interest. Such a mutation may spread if it is beneficial in one caste, even if it is deleterious in another, leading to a sub-optimal life history. Alternatively, a stable polymorphism may be reached (Hall et al. 2013) and will remain until a mutation appears that can caste-specifically reverse this change. Mutations with caste-specific effects therefore make existing theories of ageing applicable to the differences in life span found within social insect species. Furthermore, while the classical theories deal with organisms that increase their fitness through only direct reproduction, Lee (2003) extended these ideas to include the possibility of transferring resources between members of a population. This framework has been verbally applied to non-reproductive worker castes in honeybees (Amdam & Page 2005) to link the high extrinsic mortality of foragers to their rapid senescence. From a different perspective, Amdam and Omholt, (2002) argued that workers can be considered to be a honeybee colony’s soma, while the queen represents the germ line. Since resources can be shifted between workers in a colony, one can then consider somatic maintenance from the point of view of the whole colony. In this framework, resources necessary for maintenance of the individual worker are drawn away from foragers because of their high extrinsic mortality in order to avoid the cost of losing those resources if the worker should die.

Physiological Basis of Life Span Much of the research into social insect ageing, particularly in honeybees, concerns the physiological differences that underlie life span discrepancies between castes. Understanding the physiology of ageing not only sheds light on how senescence proceeds but also offers tests of evolutionary theories. For example, if mutation accumulation is the primary driver of senescence, then differences that are relevant to senescence between castes should only be found between old individuals of each caste, because at younger ages selection will be strong in all castes (while in some Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:35:23, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.010

Extraordinary Life Spans

203

species selection may be weaker on workers than queens at all ages (Hall & Goodisman 2012), the difference will still be smaller at younger ages). We might also expect that different species would experience different senescence processes, as different mutations would have appeared in their genomes (Partridge & Gems 2002). However, if antagonistic pleiotropy or disposable soma were the main drivers of senescence, then we should find physiological differences, relevant to ageing, early in the lives of queens and workers. This is because, for antagonistically pleiotropic traits, different life history strategies will be optimal for each of the castes: a given phenotype that incurs an immediate cost at young ages in exchange for future rewards might be optimal for queens but not for workers. The theory of disposable soma, in particular, makes the broad physiological prediction that long-lived phenotypes should be the ones that invest most in systems that help to prevent or repair accumulating damage to the organism. However, it is not clear a priori which such systems will be most relevant to senescence and therefore which ones will be up-regulated in queens. Study of the physiology of social insects is in a position to reveal whether a difference in somatic maintenance underlies differences in life span between castes and, if so, which systems in particular are involved.

Oxidative Stress Linked to the theory of somatic maintenance is the free-radical theory of ageing (Harman 1992; Shringarpure & Davies 2009). Reactive oxygen species (ROS) are oxidative compounds produced primarily as a by-product of respiration but also by cytosolic processes, and they are responsible for causing oxidative damage to macromolecules such as DNA, proteins and lipids (Finkel & Holbrook 2000). Because ROS cause damage to the fundamental building blocks of the cell and can be dealt with by costly processes such as producing antioxidant enzymes, differential investment in antioxidant systems may underlie variation in life span. For this reason, several studies have compared the expression of antioxidant genes between long- and short-lived castes. Results have, however, been either ambiguous or contrary to predictions. Parker et al. (2004) investigated the expression and activity of Cu-Zn superoxide dismutase (SOD) in workers, queens and males of the ant L. niger. Contrary to predictions, the long-lived caste (queens) showed lower Cu-Zn SOD activity than males in the head and abdomen than workers in the thorax. It should be noted that, in this study, individual age was not controlled and may therefore have been a confounding factor in the results. In honeybees, Corona et al. (2005) investigated the expression of eight genes coding for antioxidant enzymes in age-controlled samples of queen and worker brain, thorax and abdomen. While they found that young queens tended to have higher levels of antioxidant gene expression than young workers (in brain and thorax but not abdomen), the opposite was true at older ages. This was because the expression of these genes increased with age in workers and decreased in queens. Two recent studies have investigated the activity of antioxidant enzymes in the abdominal fat cells and trophocytes of queens (Hsieh & Hsu 2013) and workers (Hsu & Hsieh 2014). They found no consistent effect of age among the different antioxidants. Unfortunately, the queen and Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:35:23, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.010

204

Explaining Extraordinary Life Spans

worker profiles cannot be compared because the ages chosen to represent ‘young’ and ‘old’ individuals of each caste were very different. While the available data are therefore limited, it does not appear that the expression of genes that serve to reduce levels of ROS is higher in queens than in workers. There may be several explanations for this. Firstly, we cannot exclude the possibility that the disposable soma theory is wrong or that its relevance is insignificant in social insects. Secondly, the levels of ROS found in healthy organisms may not in fact reduce life span. In model organisms, over-expression of antioxidants has been found to extend life span in certain cases, but counter-examples abound (Doonan et al. 2008; Gems & Doonan 2009; Pérez et al. 2009). There is also evidence that ROS are in fact necessary for some functions, such as growth or as a signal molecule in the reaction to hypoxia (Gapper & Dolan 2006; Hamanaka & Chandel 2010). Longer-lived phenotypes may well have higher levels of investment into somatic maintenance, but it may come at a different point in the process of avoiding or repairing oxidative damage or may be in a different process entirely (Gems & Doonan 2009). For example, a study of gene expression in the fat body among honeybee worker castes found that the longest-lived caste, the winter bees (Page & Peng 2001), showed increased expression of several DNA repair genes (Seehuus et al. 2013). However, Aamodt (2009) compared the expression of nine genes from DNA repair pathways between the flight muscles of honeybee queens and workers of various ages and found no difference between queens and workers in six of them. Of the other three genes, one was up-regulated in workers, while the other two were upregulated in queens relative to older workers only. Long-lived castes may also have physiologies that are more resistant to oxidative stress. For example, there is evidence that the fatty acid composition of honeybee queens is richer in peroxidation-resistant lipids than that of workers (Haddad et al. 2007). Based on numerous studies in other taxa that found no effect of molecular damage on life span, it has even been suggested that molecular damage, whether due to ROS or other factors, is irrelevant to senescence (Blagosklonny 2007; discussed in Gems & Partridge 2013). A third possibility is that workers may simply express higher levels of antioxidant genes because they generate larger quantities of ROS that need to be removed (Corona et al. 2005). Workers may have a lower efficiency of energy production during respiration, leading to increased ROS production. Furthermore, workers and queens engage in different tasks. If the workers’ tasks are more energetically demanding than those of queens, they may generate more ROS than queens and thus need to produce more antioxidants. This explanation fits the findings in honeybees (Corona et al. 2005), in which workers expressed more antioxidant genes than queens at older ages. At these ages, the workers are no longer nurse bees that remain in the nest and care for the brood but are foragers that must expend a great deal of energy in flight and have low systemic levels of a protein thought to confer resistance to oxidative stress: vitellogenin (Seehuus et al. 2006b).

Vitellogenin Another major focus of research into the physiology of ageing in social insects, particularly in honeybees, is vitellogenin (Vg) and its link to nutrition, the Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:35:23, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.010

Extraordinary Life Spans

205

insulin/insulin-like growth factor signalling (IIS) pathway and juvenile hormone (JH) (Amdam & Omholt 2003; Münch & Amdam 2010; Münch et al. 2008). Vitellogenins are yolk proteins found in a wide range of vertebrate and invertebrate taxa (Byrne et al. 1989) and are linked to fertility (Bownes et al. 1991). One of the most intriguing findings to emerge from social insect ageing research is the unique nature of the interaction between Vg and JH in honeybees. Whereas in other organisms, such as Drosophila, the production of Vg is stimulated by JH (Postlethwait & Shirk 1981; Postlethwait & Weiser 1973; Sheng et al. 2011), the two molecules are negatively correlated in honeybees due to their mutually repressive effect on each other (Corona et al. 2007; Guidugli et al. 2005; Pinto et al. 2000) (this is not the case across the social insects, Bell 1973; Libbrecht et al. 2013; Röseler, 1977). Furthermore, while Vg is typically associated with short life spans (Murphy et al. 2003), the opposite appears true in honeybees: not only are levels high in queens, but they are also higher in nurse bees than foragers (Amdam et al. 2005; Corona et al. 2007). High levels of Vg are therefore associated with low mortality. These findings suggest a potentially crucial role of Vg in regulating honeybee life history and longevity. Studies have investigated why Vg might be associated with increased life span in honeybees and whether life span can be affected by manipulating Vg. It appears that Vg plays a role in increasing worker immunocompetence (Amdam et al. 2004, 2005) and that knocking down the expression of Vg reduces resistance to oxidative stress (Seehuus et al. 2006b) and shortens life span (Nelson et al. 2007; but see Ihle et al. 2015). It has therefore been argued that Vg may be associated with longevity due to its antioxidant and immune properties, but then the fact that it is not associated with longevity in other organisms needs to be explained. Either Vg lacks these properties in these organisms (but there is evidence for antioxidant activity of some Vgs in other species, Nakamura et al. 1999; Zheng et al. 2012), or these properties are not fundamental to determining life span. Vg in honeybees interacts with JH (Guidugli et al. 2005) and therefore forms part of the IIS pathway, a network of interacting molecules that leads to global changes in physiology and that is frequently linked to longevity (Corona et al. 2007; Kenyon 2001; Sheng et al. 2011). It is therefore possible that the most important factors responsible for modulating life span are affected by the downstream effects of Vg, while its antioxidant and immunocompetence properties have only a limited direct effect on life span. Given the abundance of Vg in the honeybee haemolymph and the many roles that have been identified for it (Amdam et al. 2003a, 2004; Seehuus et al. 2006b), it is unlikely that it is solely produced as a dedicated signal molecule (D. Münch, personal communication). Rather, it may act as an intermediate cue that serves to link downstream life-span-affecting processes to characteristics such as physiological state or caste. The major physiological shift in the life history of honeybee workers occurs when they switch from nursing to foraging. This change is associated with shifts in the titers of Vg and JH (Amdam et al. 2005; Corona et al. 2007; Robinson 1992). Interestingly, this shift in caste can be controlled and even reversed (Robinson 1992). By manipulating the proportion of foragers in the nest, young bees can be forced to become precocious foragers, or foragers can be coaxed into returning to the nurse caste, thereby decoupling Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:35:23, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.010

206

Explaining Extraordinary Life Spans

age and caste. Such manipulations have revealed that oxidative damage is more prevalent in foragers than in nurses, independent of age (Seehuus et al. 2006a). Similarly, foragers that revert back to the nurse caste retrieve their previous levels of Vg, JH and immunocompetence (Amdam et al. 2005). This provides further evidence that these molecules are more strongly linked to caste than age and gives correlative support for the case of an effect of Vg on stress resistance and immunity. A weakness of these techniques, however, is that only a subset of individuals will change caste, and the experimenter cannot control which individuals these will be. This means that a confounding factor to the conclusions will be the potential bias in the individuals that choose to revert to nurses (i.e. it is possible that the foragers that are most nurse-like will be the ones to revert and will therefore be over-represented in the pool of reverted foragers).

Detecting Senescence It is important to distinguish life span from senescence. The concept of senescence implies a deterioration that progresses with age, typically associated with increasing mortality, decreasing fertility and a decline in performance at many behavioural activities. A constant, high level of mortality throughout life will lead to short life span, but it cannot be considered to be senescence. Workers of many social insects typically start their life as nurses and progress to foraging later in life (Wilson 1971), but foragers suffer much greater mortality than nurses in part at least due to predation and exhaustion in the field. Several studies, however, have disentangled the effects of age and caste and shown that within caste it is possible to find evidence of senescence.

Senescence in Workers One of the typical signs of senescence in organisms is an age-specific increase in mortality. Dukas (2008) measured mortality of marked honeybee foragers in the field and found that mortality increased greatly after about ten days of foraging. Increases in mortality with age have also been recorded in nurses and foragers even when foraging was restricted to a flight room (Rueppell et al. 2007a). Rueppell et al. (2007b) found an increase in mortality with age in laboratory conditions but did not find any other signs of deterioration with age. For example, there was no evidence that walking speed or responsiveness to light or sucrose decreased with age. Another potential manifestation of senescence is a decline in resistance to various forms of stress and disease. By manipulating the social environment to prevent transitions to the forager state, Remolina et al. (2007) were able to obtain honeybee nurses that varied greatly in age from ten to fifty days. This experiment showed that older nurses have reduced resistance to starvation, heat stress and oxidative stress. In bumblebees, where the marked transition from nurse to forager does not exist and workers may forage at any age, the ability of workers to mount an immune response decreases with age (Doums et al. 2002). The physiological basis of this decline is unclear; there is some evidence that haemocyte counts decrease with age in both bumblebees and honeybees Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:35:23, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.010

Detecting Senescence

207

(Moret & Schmid-Hempel 2009; Schmid et al. 2008; but see Doums et al. 2002), while the activity of phenoloxidase (an enzyme important for immune response) decreases with age in bumblebees, increases with age in honeybee nurses and did not change with age in honeybee foragers (Moret & Schmid-Hempel 2009; Schmid et al. 2008). Recently, and particularly in honeybees, there has been a surge in research investigating patterns of senescence in the form of learning ability. This is because learning is a cognitive function that has been shown to deteriorate with age in many species (Grotewiel et al. 2005) and because it can be relatively easily measured using a standardised process in laboratory conditions, thus avoiding confounding factors such as the influence of predation on estimates of mortality or exposure to parasites on immune reaction. It has therefore been used to investigate senescence in workers and to compare the extent of senescence between nurses, foragers and winter bees. Workers can be trained to associate certain stimuli (most often odours, but tactile stimuli have also been used) with a food reward, and aspects of learning and memory consolidation can be measured by how quickly individuals learn these associations, how long they retain this in their memory and how effectively they can use what they have learned to discriminate between stimuli (Behrends & Scheiner 2010). In nurse and winter bees, there is no evidence of a senescent decline in learning (Behrends et al. 2007; Münch et al., 2013). In foragers, studies have been conducted on the effects of chronological age and foraging age (foraging age represents the time since a worker switched from nursing to foraging) in both naturally age-structured colonies and single-cohort colonies in which all workers are of the same age. Studies of chronological age in age-structured colonies have produced contrasting results. One study found an effect of chronological age on forager learning ability (Behrends et al. 2007), but as chronological and foraging age were not independent, the two cannot be disentangled. Another study did not find evidence of a significant decline in learning with chronological age (Rueppell et al. 2007b), but this study confounded the effects of age and caste. Studies of foraging age, with one exception, indicate that foragers do display learning senescence. In naturally age-structured colonies, two studies found an effect of foraging age on learning ability (Baker et al. 2012; Münch et al. 2010), while another did not (Hystad et al. 2014). The studies of Hystad et al. (2014) and Baker et al. (2012) used foragers of similar ages obtained in similar ways, but both only estimated foraging age rather than measuring it exactly. One possibility is therefore that this estimation reduces the power of the analysis, leading to variable results. In singlecohort colonies, learning ability has consistently been found to decline with foraging age (Baker et al. 2012; Behrends et al. 2007; Scheiner & Amdam 2009). Behrends et al. (2007) pointed out that the division of labour in single-cohort colonies cannot be determined by age and may therefore reinforce biases in other characters. If the individuals that are most prone to senescence are the ones that become foragers, then the high levels of learning senescence in foragers could be an artefact of the setup. Indeed, there is evidence that individuals with artificially lowered life span tend to transition to foraging earlier in life (Kuszewska & Woyciechowski 2013; Woyciechowski & Moroń 2009). However, the fact that learning senescence has been detected in two studies from an agestructured colony does not support this. Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:35:23, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.010

208

Explaining Extraordinary Life Spans

Whether the higher rate of senescence observed in foragers is due only to the weakest subset of individuals becoming foragers can be tested by causing a reversion to nursing behaviour. Studies applying such manipulations have found that reverted foragers show higher learning ability than individuals that remain as foragers (Baker et al. 2012; Behrends et al. 2007), implying that foragers can recover from their senescent decline in learning if they are made to revert to nursing behaviour. Furthermore, evidence that learning senescence in foragers can be found in the absence of a single-cohort setup comes from a study showing that winter bees that transition to the forager caste show significantly lower learning ability than those that remain as winter bees (Münch et al. 2013). The picture that emerges from these studies is that senescence in learning ability occurs only in foragers and can even be reversed if the individual ceases to be a forager. An interesting question is whether senescence represents an inevitable process associated with the physiology of foragers or is instead simply a by-product of the act of foraging and the energetic demands that it entails. By restricting opportunities for flight, Tolfsen et al. (2011) were able to investigate the effect of foraging age while controlling for foraging activity. They found no evidence that foraging age per se affected learning ability because foragers with limited foraging opportunities did not show signs of senescence. In contrast, foragers that were free to forage showed a marked decline in learning ability with age. The learning senescence observed in forager honeybees is therefore a direct result of foraging. This may explain why reversion to nursing allows a recovery from this decline. It would be interesting to investigate whether old, senescent foragers that are prevented from further foraging using the technique in the study by Tolfsen et al. (2011) also show a reversal of their decline in learning. These results on learning refer to the trait of acquisition, that is, the speed at which an association can be learned. It should be pointed out that studies on the ability to remember an association some time after the learning experience, or on the ability to discriminate between a learned stimulus and a related one, found little or no effect of age (Scheiner & Amdam 2009; Behrends & Scheiner 2010) or even that older foragers are better at discrimination (Behrends et al. 2007). Senescence in worker honeybees is therefore a heterogeneous process, with different functions showing different patterns of decline (summarised by Münch 2013).

Senescence in Queens Senescence is hard to measure in social insect queens because their long life spans make obtaining old queens of known age very difficult. Heinze and Schrempf (2012) showed that the rate of egg laying increases throughout the life of queens of the ant Cardiocondyla obscurior, even when the number of workers is kept constant to remove the effect of colony growth. It should be noted that queens in this species are relatively short lived, with a mean life span of about twenty-six weeks (Heinze & Schrempf 2012); this made the study possible but also means that this result may not be representative of longer-lived species. In the wild, queen fertility and mortality must be inferred by studying whole colonies. Long-term studies of this kind in the ant Pogonomyrmex barbatus revealed that the mortality of colonies increases with age (Gordon & Kulig 1998), but the picture Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:35:23, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.010

Reproduction and Longevity: Is the Trade-Off Really Broken?

209

regarding fertility is less clear. It appears that older colonies produce greater total numbers of sexuals but that the dry mass of these sexuals is lower than in younger colonies (Wagner & Gordon 1999). Recently Ingram et al. (2013) used genotyping to allocate colony queens to their parent colonies in this population. Because the ages of the colonies were known, it was then possible to estimate the age-specific rate at which colonies produced successful foundress queens. This study did not reveal a significant increase or decrease in successful queen production over the twenty-five- to thirty-year range in colony age. It therefore appears that in P. barbatus mortality increases with age, while reproductive success is more or less constant in mature colonies. What is still lacking, to our knowledge, is an explicit comparison of senescence between queens and workers. This may be because their physiologies are so different that it makes comparison difficult. Such data would make a valuable contribution to our understanding of the differences in ageing between these castes.

Reproduction and Longevity: Is the Trade-Off Really Broken? One feature of social insects that is frequently remarked upon is the apparent overcoming of the trade-off between longevity and reproduction (Corona et al. 2007; Hartmann & Heinze 2003; Parker 2010; Remolina & Hughes 2008). It is usually assumed that resources invested into avoiding senescence and mortality must come at the cost of reproduction. It may therefore seem puzzling that queens of social insects have both a higher reproductive output and greater longevity than their workers or than females of solitary species. In comparing social insect queens with solitary species, an important point to note is that the queen has access to vastly more resources than a solitary female (Remolina & Hughes 2008). She is fed by her workers and therefore has access to large reserves of energy that should allow her to invest heavily in both egg laying and somatic maintenance. The trade-off that solitary individuals face is therefore on an entirely different scale to that of a social insect queen. This also applies to the comparison between queens and workers. Moreover, it should be pointed out that while workers do not lay eggs, they invest in other activities such as foraging, brood care and nest maintenance. Their energetic expenditure is therefore considerable, and it is not at all clear that this is proportionally lower than the queen’s investment into reproduction. Effectively, the workers are primarily trading life span for workload rather than for egg laying (Tsuji et al. 1996, 2012). The trade-off puzzle exists only if we consider the physiological link between egg laying and longevity and the observation that in many organisms physiological changes that increase one will tend to decrease the other (Partridge & Gems 2002), for example, through the IIS pathway (Partridge et al. 2005). It is therefore interesting to study the physiological regulation leading to queens investing more than workers into egg laying without sacrificing longevity (Parker 2010). In honeybees, it has been suggested that the unique relationship between Vg and JH (see earlier section) is part of the answer (Corona et al. 2007; Remolina & Hughes 2008). However, given that this relationship does not Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:35:23, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.010

210

Explaining Extraordinary Life Spans

exist in all social insects, it cannot be the only way in which fertility and life span can be positively correlated. Dietary restriction is one of the most widespread ways in which life span can be increased (Mair & Dillin 2008). How, then, can increased levels of nutrient supplies in social insect queens be presented as an argument to explain their increased life span? It has been argued that life span extension in cases of dietary restriction makes evolutionary sense because energy is diverted away from reproduction and towards survival until conditions improve and resources become more abundant (Shanley & Kirkwood 2000). Conversely, when resources are plentiful, energy can be diverted towards reproduction to take advantage of the situation while it lasts, at a cost to maintenance. If this is the case, why should high levels of nutrition in queens not lead to short life? Dietaryrestricted organisms are experiencing a shortage relative to that which they might encounter in the future. They therefore invest into survival because conditions may improve. If resources were consistently scarce, it would make no sense to delay reproduction. In the case of a social insect queen, the abundance of food that she receives compared to the workers, or compared to solitary species, is not a result of transient conditions in the environment. This is not to say that there will be no fluctuation in the resources available to the queen but simply that a large amount of resources represents a normal state of affairs rather than a temporary boon. The optimal strategies for investment into reproduction or survival therefore depend on whether the current availability of resources is likely to be stable. It would be interesting to compare life span and reproduction between queens that are nutrient limited and queens that are provided unlimited food. Unfortunately, such a study is difficult because the queen obtains all her food from the workers, making it impossible to experimentally manipulate food availability without affecting other processes in the colony.

Challenges and Opportunities in Social Insect Research Much research into social insect ageing has been performed on the honeybee, predominantly by comparing honeybee nurses and foragers. As mentioned earlier, the greatest life spans are, however, found in queens, and a comparison of queens and workers has the potential to be highly informative regarding the evolution of life span on the scale of years or decades. However, such studies also present several challenges. The first is directly linked to the long life of the queens. While this extended longevity creates a fascinating system of study, it invariably means that obtaining old queens in laboratory conditions, or determining queen life span, is difficult. To some extent, this problem is mitigated because to understand the difference in longevity between queens and workers, one does not need to look any further than the workers’ maximum life span. This is because after this age, there is no longer anything that the queen does ‘differently’ to workers. Nevertheless, in species such as L. niger, which has been a focus of ageing research due to the extreme longevity of its queens (Gräff et al. 2007; Jemielity et al. 2007; Parker et al. 2004), workers may still live for up to two years in laboratory Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:35:23, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.010

Challenges and Opportunities in Social Insect Research

211

conditions. In honeybees, this problem is less severe, as queens rarely live for more than a few years and workers typically die after a few weeks (Page & Peng 2001). A second challenge is that queens and workers are very different and therefore differ in many ways that are not necessarily linked to ageing. When comparing the two castes, therefore, one will inevitably find differences, and it is not trivial to determine which ones are relevant to ageing (Jemielity et al. 2005; Rueppell 2009). One approach to resolve this problem is to make a priori predictions based on previous findings or theoretical predictions of ageing models. Examples of this are studies that have investigated differences in the expression of genes linked to somatic maintenance or the IIS pathway (Aamodt 2009; Corona et al. 2005, 2007; Parker et al. 2004; Seehuus et al. 2013), in the length of telomeres (Jemielity et al. 2007) or in the extent of accumulated carbonylation damage (Seehuus et al. 2006a). However, this inherently limits discoveries to those that are linked to model predictions or to findings that have already been made in other organisms. The morphological difference between queens and workers also underlies a third difficulty, as it makes comparisons of features such as gene expression very tricky. A whole-body comparison of gene expression will be affected by the different proportions that each organ represents in each phenotype. Queens, for example, have extremely developed ovaries, and the expression patterns in these ovaries will therefore represent a significant proportion of the expression data. Therefore, when a difference is identified between queens and workers, this is not necessarily an indication that any particular tissue or cell type is behaving differently; it may simply represent a difference in the extent to which each tissue is represented in the sample. This problem can be reduced by performing tissue-specific analysis when feasible. The problem of morphological differences can also to a large extent be avoided by studying species in which workers and queens are morphologically identical. This is the case of many species in which workers are able to become queens (Wilson 1971). These species have received too little attention in the field of ageing. In the case of many such bees and wasps, laboratory colonies are difficult to maintain, and it is therefore difficult to show that the intrinsic mortality (i.e. independent of factors such as predation) of queens is smaller than that of workers. However, species of ants exist that also have such a social system and that can be kept in the laboratory. Workers that become reproductive are called ‘gamergates’, and it has been shown that these gamergates live substantially longer than workers (Hartmann & Heinze 2003; Tsuji et al. 1996). In the ant Harpegnathos saltator, the true queen can be replaced by a gamergate after her death. Recent work in this species has shown that gamergates show increased levels of resistance to disease and oxidative stress (Schneider et al. 2011) relative to normal workers. These species therefore offer a great opportunity for more accurately investigating the mechanisms that underlie caste-related differences in life span. A fourth challenge is to adequately control for age when comparing castes. If a sample of queens and a sample of workers are taken at random from colonies, then the average age of these two samples will be very different. Many physiological features, such as gene expression, change with age (Jin et al. 2001; Kim et al. 2005; Lee et al. 1999), and Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:35:23, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.010

212

Explaining Extraordinary Life Spans

a simple comparison of these two randomly selected samples is as likely to be linked to age as it is to caste. For example, some of the evidence that antioxidant expression is lower in queens than in shorter-lived castes (Parker et al. 2004) may be confounded by age effects. Even performing comparisons between individuals of equal chronological age may not completely remove the problem because the biological age of the castes may be very different. For example, a one-year-old queen of L. niger is still young, whereas a similar-aged worker is much further through its life. Is it fair to compare the two and conclude that a difference between them might be a causative factor of senescence rather than an effect of it? At what age, therefore, should queens and workers be compared? This problem may again be reduced by studying species in which caste is plastic because queens and workers can be compared soon after the workers have converted to queens. Finally, a challenge of research in social insects is that the molecular tools available are limited compared to those established for model organisms. For example, for most model organisms it is possible to obtain mutant lines in which genetic sequences are inserted or removed as desired, allowing fine-scale experimentation into the effects of specific mutations. Such possibilities do not currently exist for social insects. The use of interference RNA (RNAi) is in its infancy in social insects, having only been put to use for a few genes in a few species (Amdam et al. 2003b; Beye et al. 2003; Miyazaki et al. 2014; Zhou et al. 2006, 2008), but the use of this technique in honeybees has shown promising proliferation in recent years (Kamakura 2011; Li-Byarlay et al. 2013; Wang et al. 2010, 2012; Wolschin et al. 2011). Furthermore, there are up-and-coming, promising methods of genetic and transcriptomic manipulation using the RNA-guided doublestranded DNA (dsDNA)–binding protein Cas9 (Mali et al. 2013). Once established, these will provide unprecedented opportunities for investigating the genetic underpinnings of life span in social insects.

Conclusion Despite the many challenges associated with social life and long life span, social insects provide fertile ground for research into both the evolutionary and physiological underpinnings of longevity and senescence. From them we have learned about the plasticity, and even reversibility, of senescence, about the nature of the evolutionary trade-off between life span and reproduction, about the interactive effects of physiology and nutrition on ageing and about how cooperative and competitive social environments affect ageing strategies. However, experimental studies to test evolutionary theories have focused on investigating molecular damage and could be expanded to include alternative theories such as mutation accumulation. As always, further research will involve new technologies arising and becoming cheaper, but carefully considered experiments that overcome the limitations of these organisms will be the most important factor contributing to further understanding the ageing process through research into social insects.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:35:23, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.010

References

213

References Aamodt, R. M. (2009). Age-and caste-dependent decrease in expression of genes maintaining DNA and RNA quality and mitochondrial integrity in the honeybee wing muscle. Experimental Gerontology, 44(9), 586–93. Amdam, G. & Page, R. (2005). Intergenerational transfers may have decoupled physiological and chronological age in a eusocial insect. Ageing Research Reviews, 4(3), 398–408. Amdam, G. V., Aase, A. L. T. O., Seehuus, S. C., et al. (2005). Social reversal of immunosenescence in honey bee workers. Experimental Gerontology, 40(12), 939–47. Amdam, G. V., Norberg, K., Hagen, A., & Omholt, S. W. (2003a). Social exploitation of vitellogenin. Proceedings of the National Academy of Sciences of the United States of America, 100(4), 1799–1802. Amdam, G. V. & Omholt, S. W. (2002). The regulatory anatomy of honeybee lifespan. Journal of Theoretical Biology, 216, 209–28. Amdam, G. V. & Omholt, S. W. (2003). The hive bee to forager transition in honeybee colonies: the double repressor hypothesis. Journal of Theoretical Biology, 223(4), 451–64. Amdam, G. V., Rueppell, O., Fondrk, M. K., et al. (2009). The nurse’s load: early-life exposure to brood-rearing affects behavior and lifespan in honey bees (Apis mellifera). Experimental Gerontology, 44(6), 467–71. Amdam, G. V., Simões, Z. L. P., Hagen, A., et al. (2004). Hormonal control of the yolk precursor vitellogenin regulates immune function and longevity in honeybees. Experimental Gerontology, 39(5), 767–73. Amdam, G. V., Simões, Z. L. P., Guidugli, K. R., et al. (2003b). Disruption of vitellogenin gene function in adult honeybees by intra-abdominal injection of double-stranded RNA. BMC biotechnology, 3(1), 1. Baker, N., Wolschin, F. & Amdam, G. V. (2012). Age-related learning deficits can be reversible in honeybees Apis mellifera. Experimental Gerontology, 47(10), 764–72. Baudisch, A. & Vaupel, J. W. (2010). Senescence vs sustenance: evolutionary-demographic models of aging. Demographic Research, 23(23), 655–68. Behrends, A. & Scheiner, R. (2010). Learning at old age: a study on winter bees. Frontiers in Behavioral Neuroscience, 4, article 15. Behrends, A., Scheiner, R., Baker, N. & Amdam, G. V. (2007). Cognitive aging is linked to social role in honey bees (Apis mellifera). Experimental Gerontology, 42(12), 1146–53. Bell, W. J. (1973). Factors controlling initiation of vitellogenesis in a primitively social bee, Lasioglossum zephyrum (Hymenoptera: Halictidae). Insectes Sociaux, 20(3), 253–60. Beye, M., Hasselmann, M., Fondrk, M. K., et al. (2003). The gene csd is the primary signal for sexual development in the honeybee and encodes an SR-type protein. Cell, 114(4), 419–29. Blagosklonny, M. V. (2007). Paradoxes of aging. Cell Cycle, 6(24), 2997–3003.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:35:23, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.010

214

Explaining Extraordinary Life Spans

Bownes, M., Lineruth, K. & Mauchline, D. (1991). Egg production and fertility in Drosophila depend upon the number of yolk-protein gene copies. Molecular and General Genetics, 228(1–2), 324–7. Byrne, B. M., Gruber, M. & Ab, G. (1989). The evolution of egg yolk proteins. Progress in Biophysics and Molecular Biology, 53(1), 33–69. Carey, J. R. (2001). Demographic mechanisms for the evolution of long life in social insects. Experimental Gerontology, 36, 713–22. Chapuisat, M. & Keller, L. (2002). Division of labour influences the rate of ageing in weaver ant workers. Proceedings of the Royal Society of London Series B: Biological Sciences, 269(1494), 909–13. Corona, M., Hughes, K. A., Weaver, D. B., & Robinson, G. E. (2005). Gene expression patterns associated with queen honey bee longevity. Mechanisms of Ageing and Development, 126(11), 1230–8. Corona, M., Velarde, R. A., Remolina, S., et al. (2007). Vitellogenin, juvenile hormone, insulin signaling, and queen honey bee longevity. Proceedings of the National Academy of Sciences of the United States of America, 104(17), 7128–33. Dietzl, G., Chen, D., Schnorrer, F., et al. (2007). A genome-wide transgenic RNAi library for conditional gene inactivation in drosophila. Nature, 448(7150), 151–6. Doonan, R., McElwee, J. J., Matthijssens, F., et al. (2008). Against the oxidative damage theory of aging: superoxide dismutases protect against oxidative stress but have little or no effect on life span in Caenorhabditis elegans. Genes & Development, 22(23), 3236–41. Doums, C., Moret, Y., Benelli, E. & Schmid-Hempel, P. (2002). Senescence of immune defence in Bombus workers. Ecological Entomology, 27(2), 138–44. Duffy, J. B. (2002). GAL4 system in Drosophila: a fly geneticist’s Swiss army knife. Genesis, 34, 1–15. Dukas, R. (2008). Mortality rates of honey bees in the wild. Insectes Sociaux, 55(3), 252–5. Ferreira, P. G., Patalano, S., Chauhan, R., et al. (2013). Transcriptome analyses of primitively eusocial wasps reveal novel insights into the evolution of sociality and the origin of alternative phenotypes. Genome Biology, 14(2), R20. Finkel, T. & Holbrook, N. J. (2000). Oxidants, oxidative stress and the biology of ageing. Nature, 408(6809), 239–47. Flatt, T. (2011). Survival costs of reproduction in Drosophila. Experimental Gerontology, 46(5), 369–75. Gapper, C. & Dolan, L. (2006). Control of plant development by reactive oxygen species. Plant Physiology, 141(2), 341–5. Gems, D. & Doonan, R. (2009). Antioxidant defense and aging in C. elegans. Cell Cycle, 8(11), 1681–7. Gems, D. & Partridge, L. (2013). Genetics of longevity in model organisms: debates and paradigm shifts. Annual Review of Physiology, 75, 621–44. Gordon, D. M. & Kulig, A. (1998). The effect of neighbours on the mortality of harvester ant colonies. Journal of Animal Ecology, 67(1), 141–8. Gräff, J., Jemielity, S., Parker, J. D., et al. (2007). Differential gene expression between adult queens and workers in the ant Lasius niger. Molecular Ecology, 16(3), 675–83. Grotewiel, M. S., Martin, I., Bhandari, P. & Cook-Wiens, E. (2005). Functional senescence in Drosophila melanogaster. Ageing Research Reviews, 4(3), 372–97.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:35:23, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.010

References

215

Grozinger, C. M., Fan, Y., Hoover, S. E. R. & Winston, M. L. (2007). Genome-wide analysis reveals differences in brain gene expression patterns associated with caste and reproductive status in honey bees (Apis mellifera). Molecular Ecology, 16(22), 4837–48. Guidugli, K. R., Nascimento, A. M., Amdam, G. V., et al. (2005). Vitellogenin regulates hormonal dynamics in the worker caste of a eusocial insect. FEBS Letters, 579(22), 4961–5. Haddad, L. S., Kelbert, L. & Hulbert, A. J. (2007). Extended longevity of queen honey bees compared to workers is associated with peroxidation-resistant membranes. Experimental Gerontology, 42(7), 601–9. Hall, D. W. & Goodisman, M. A. D. (2012). The effects of kin selection on rates of molecular evolution in social insects. Evolution, 66(7), 2080–93. Hall, D. W., Soojin, V. Y. & Goodisman, M. A. D. (2013). Kin selection, genomics and caste-antagonistic pleiotropy. Biology Letters, 9(6), 20130309. Hamanaka, R. B. & Chandel, N. S. (2010). Mitochondrial reactive oxygen species regulate cellular signaling and dictate biological outcomes. Trends in Biochemical Sciences, 35(9), 505–13. Hamilton, B., Dong, Y., Shindo, M., et al. (2005). A systematic RNAi screen for longevity genes in C. elegans. Genes & Development, 19(13), 1544–55. Harman, D. (1992). Free-radical theory of aging. Mutation Research/DNAging, 275(3– 6), 257–66. Hartmann, A. & Heinze, J. (2003). Lay eggs, live longer: division of labor and life span in a clonal ant species. Evolution, 57(10), 2424–9. Heinze, J. & Schrempf, A. (2012). Terminal investment: individual reproduction of ant queens increases with age. PloS One, 7(4), e35201. Hölldobler, B. & Wilson, E. O. (1990). The Ants (Berlin: Springer-Verlag). Hsieh, Y. & Hsu, C. (2013). Oxidative stress and anti-oxidant enzyme activities in the trophocytes and fat cells of queen honeybees (Apis mellifera). Rejuvenation Research, 16(4), 295–303. Hsu, C. & Hsieh, Y. (2014). Oxidative stress decreases in the trophocytes and fat cells of worker honeybees during aging. Biogerontology, 15, 129–37. Hystad, E. M., Amdam, G. V. & Eide, L. (2014). Mitochondrial DNA integrity changes with age but does not correlate with learning performance in honey bees. Experimental Gerontology, 49, 12–18. Ihle, K. E., Fondrk, M. K., Page, R. E. & Amdam, G. V. (2015). Genotype effect on lifespan following vitellogenin knockdown. Experimental Gerontology, 61, 113–22. Ingram, K. K., Pilko, A., Heer, J. & Gordon, D. M. (2013). Colony life history and lifetime reproductive success of red harvester ant colonies. Journal of Animal Ecology, 82(3), 540–50. Jemielity, S., Chapuisat, M., Parker, J. D. & Keller, L. (2005). Long live the queen: studying aging in social insects. Age, 27(3), 241–8. Jemielity, S., Kimura, M., Parker, K. M., et al. (2007). Short telomeres in short-lived males: what are the molecular and evolutionary causes? Aging Cell, 6(2), 225–33. Jin, W., Riley, R. M., Wolfinger, R. D., et al. (2001). The contributions of sex, genotype and age to transcriptional variance in Drosophila melanogaster. Nature Genetics, 29 (4), 389–95.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:35:23, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.010

216

Explaining Extraordinary Life Spans

Judice, C. C., Carazzole, M. F., Festa, F., et al. (2006). Gene expression profiles underlying alternative caste phenotypes in a highly eusocial bee, Melipona quadrifasciata. Insect Molecular Biology, 15(1), 33–44. Kamakura, M. (2011). Royalactin induces queen differentiation in honeybees. Nature, 473(7348), 478–83. Keller, L. & Genoud, M. (1997). Extraordinary lifespans in ants: a test of evolutionary theories of ageing. Nature, 389(6654), 958–60. Kenyon, C. (2001). A conserved regulatory system for aging. Cell, 105(2), 165–8. Kim, S. N., Rhee, J., Song, Y., et al. (2005). Age-dependent changes of gene expression in the Drosophila head. Neurobiology of Aging, 26(7), 1083–91. Kirkwood, T. B. L. (1977). Evolution of aging. Nature, 270(5635), 301–4. Kirkwood, T. B. L. & Holliday, R. (1979). Evolution of ageing and longevity. Proceedings of the Royal Society of London Series B: Biological Sciences, 205(1161), 531–46. Kramer, B. H. & Schaible, R. (2013). Colony size explains the lifespan differences between queens and workers in eusocial Hymenoptera. Biological Journal of the Linnean Society, 109(3), 710–24. Kuszewska, K. & Woyciechowski, M. (2013). Reversion in honeybee, Apis mellifera, workers with different life expectancies. Animal Behaviour, 85(1), 247–53. Lee, C. K., Klopp, R. G., Weindruch, R. & Prolla, T. A. (1999). Gene expression profile of aging and its retardation by caloric restriction. Science, 285(5432), 1390–3. Lee, R. (2003). Rethinking the evolutionary theory of aging: transfers, not births, shape social species. Proceedings of the National Academy of Sciences of the United States of America, 100(16), 9637–42. Li-Byarlay, H., Li, Y., Stroud, H., et al. (2013). RNA interference knockdown of DNA methyl-transferase 3 affects gene alternative splicing in the honey bee. Proceedings of the National Academy of Sciences of the United States of America, 110(31), 12750–5. Libbrecht, R., Corona, M., Wende, F., et al. (2013). Interplay between insulin signaling, juvenile hormone, and vitellogenin regulates maternal effects on polyphenism in ants. Proceedings of the National Academy of Sciences of the United States of America, 110 (27), 11050–5. Mair, W. & Dillin, A. (2008). Aging and survival: the genetics of life span extension by dietary restriction. Annual Review of Biochemistry, 77, 727–54. Mali, P., Esvelt, K. M. & Church, G. M. (2013). Cas9 as a versatile tool for engineering biology. Nature Methods, 10(10), 957–63. Medawar, P. B. (1952). An Unsolved Problem of Biology: An Inaugural Lecture Delivered at University College, London, 6 December, 1951 (London: Lewis). Mersch, D. P., Crespi, A. & Keller, L. (2013). Tracking individuals shows spatial fidelity is a key regulator of ant social organization. Science, 340(6136), 1090–3. Miyazaki, S., Okada, Y., Miyakawa, H., et al. (2014). Sexually dimorphic body color is regulated by sex-specific expression of yellow gene in ponerine ant, Diacamma sp. PLoS ONE, 9(3), e92875. Moret, Y. & Schmid-Hempel, P. (2009). Immune responses of bumblebee workers as a function of individual and colony age: senescence versus plastic adjustment of the immune function. Oikos, 118(3), 371–8.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:35:23, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.010

References

217

Münch, D. (2013). Brain aging and performance plasticity in honeybees. In Invertebrate Learning and Memory, ed. R. Menzel & P. Benjamin (pp. 487–500). (New York: Academic Press). Münch, D. & Amdam, G. V. (2010). The curious case of aging plasticity in honey bees. FEBS Letters, 584(12), 2496–2503. Münch, D., Amdam, G. V. & Wolschin, F. (2008). Ageing in a eusocial insect: molecular and physiological characteristics of life span plasticity in the honey bee. Functional Ecology, 22(3), 407–21. Münch, D., Baker, N., Kreibich, C. D., et al. (2010). In the laboratory and during free-flight: old honey bees reveal learning and extinction deficits that mirror mammalian functional decline. PloS ONE, 5(10), e13504. Münch, D., Kreibich, C. D. & Amdam, G. V. (2013). Aging and its modulation in a long-lived worker caste of the honey bee. Journal of Experimental Biology, 216(9), 1638–49. Murphy, C. T., McCarroll, S. A., Bargmann, C. I., et al. (2003). Genes that act downstream of DAF-16 to influence the lifespan of Caenorhabditis elegans. Nature, 424 (6946), 277–83. Nakamura, A., Yasuda, K., Adachi, H., et al. (1999). Vitellogenin-6 is a major carbonylated protein in aged nematode, Caenorhabditis elegans. Biochemical and Biophysical Research Communications, 264(2), 580–3. Nelson, C. M., Ihle, K. E., Fondrk, M. K., et al. (2007). The gene vitellogenin has multiple coordinating effects on social organization. PLoS BIOLOGY, 5(3), 673–7. Ometto, L., Shoemaker, D., Ross, K. G. & Keller, L. (2011). Evolution of gene expression in fire ants: the effects of developmental stage, caste, and species. Molecular Biology and Evolution, 28(4), 1381–92. Page, R. E. & Peng, C. Y. S. (2001). Aging and development in social insects with emphasis on the honey bee, Apis mellifera l. Experimental Gerontology, 36, 695–711. Parker, J. D. (2010). What are social insects telling us about aging? Myrmecological News, 13, 103–10. Parker, J. D., Parker, K. M., Sohal, B. H., et al. (2004). Decreased expression of Cu-Zn superoxide dismutase 1 in ants with extreme lifespan. Proceedings of the National Academy of Sciences of the United States of America, 101(10), 3486–9. Partridge, L. & Barton, N. (1993). Optimality, mutation and the evolution of aging. Nature, 362(6418), 305–11. Partridge, L. & Gems, D. (2002). Mechanisms of aging: public or private? Nature Reviews Genetics, 3(3), 165–75. Partridge, L., Gems, D. & Withers, D. J. (2005). Sex and death: what is the connection? Cell, 120(4), 461–72. Pérez, V. I., Bokov, A., Van Remmen, H., et al. (2009). Is the oxidative stress theory of aging dead? Biochimica et Biophysica Acta (BBA): General Subjects, 1790(10), 1005–14. Pinto, L. Z., Bitondi, M. M. G. & Simoes, Z. L. P. (2000). Inhibition of vitellogenin synthesis in Apis mellifera workers by a juvenile hormone analogue, pyriproxyfen. Journal of Insect Physiology, 46(2), 153–60. Postlethwait, J. H. & Shirk, P. D. (1981). Genetic and endocrine regulation of vitellogenesis in Drosophila. American Zoologist, 21(3), 687–700.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:35:23, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.010

218

Explaining Extraordinary Life Spans

Postlethwait, J. H. & Weiser, K. (1973). Vitellogenesis induced by juvenile hormone in the female sterile mutant apterous-four in Drosophila melanogaster. Nature, 244 (139), 284–5. Remolina, S. C., Hafez, D. M., Robinson, G. E. & Hughes, K. A. (2007). Senescence in the worker honey bee Apis mellifera. Journal of Insect Physiology, 53(10), 1027–33. Remolina, S. C. & Hughes, K. A. (2008). Evolution and mechanisms of long life and high fertility in queen honey bees. Age, 30(2–3), 177–85. Robinson, G. E. (1992). Regulation of division of labor in insect societies. Annual Review of Entomology, 37(1), 637–65. Röseler, P. (1977). Juvenile hormone control of oögenesis in bumblebee workers, Bombus terrestris. Journal of Insect Physiology, 23(8), 985–92. Rueppell, O. (2009). Aging of social insects. In Organization of Insect Societies: From Genome to Sociocomplexity, ed. J. Gadau & J. Fewell (pp. 51–73) (Cambridge, MA: Harvard University Press). Rueppell, O., Bachelier, C., Fondrk, M. K. & Page Jr, R. E. (2007a). Regulation of life history determines lifespan of worker honey bees (Apis mellifera L.). Experimental Gerontology, 42(10), 1020–32. Rueppell, O., Christine, S., Mulcrone, C. & Groves, L. (2007b). Aging without functional senescence in honey bee workers. Current Biology, 17(8), R274–5. Rueppell, O., Königseder, F., Heinze, J. & Schrempf, A. (2015). Intrinsic survival advantage of social insect queens depends on reproductive activation. Journal of Evolutionary Biology, 28(12), 2349–54. Rueppell, O., Linford, R., Gardner, P., et al. (2008). Aging and demographic plasticity in response to experimental age structures in honeybees (Apis mellifera l). Behavioral Ecology and Sociobiology, 62(10), 1621–31. Scheiner, R. & Amdam, G. V. (2009). Impaired tactile learning is related to social role in honeybees. Journal of Experimental Biology, 212(7), 994–1002. Schmid, M. R., Brockmann, A., Pirk, C. W. W., et al. (2008). Adult honeybees (Apis mellifera l.) abandon hemocytic, but not phenoloxidase-based immunity. Journal of Insect Physiology, 54(2), 439–44. Schneider, S. A., Schrader, C., Wagner, A. E., et al. (2011). Stress resistance and longevity are not directly linked to levels of enzymatic antioxidants in the ponerine ant Harpegnathos saltator. PLoS ONE, 6(1), e14601. Schrempf, A., Cremer, S. & Heinze, J. (2011). Social influence on age and reproduction: reduced lifespan and fecundity in multi-queen ant colonies. Journal of Evolutionary Biology, 24, 1455–61. Schwander, T., Lo, N., Beekman, M., et al. (2010). Nature versus nurture in social insect caste differentiation. Trends in Ecology and Evolution, 25(5), 275–82. Seehuus, S., Krekling, T. & Amdam, G. V. (2006a). Cellular senescence in honey bee brain is largely independent of chronological age. Experimental Gerontology, 41(11), 1117–25. Seehuus, S., Norberg, K., Gimsa, U., et al. (2006b). Reproductive protein protects functionally sterile honey bee workers from oxidative stress. Proceedings of the National Academy of Sciences of the United States of America, 103(4), 962–7. Seehuus, S., Taylor, S., Petersen, K. & Aamodt, R. M. (2013). Somatic maintenance resources in the honeybee worker fat body are distributed to withstand the most life-threatening challenges at each life stage. PLoS ONE, 8(8), e69870.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:35:23, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.010

References

219

Seeley, T. D. (1982). Adaptive significance of the age polyethism schedule in honeybee colonies. Behavioral Ecology and Sociobiology, 11(4), 287–93. Shanley, D. P. & Kirkwood, T. B. L. (2000). Calorie restriction and aging: a life-history analysis. Evolution, 54(3), 740–50. Sheng, Z., Xu, J., Bai, H., et al. (2011). Juvenile hormone regulates vitellogenin gene expression through insulin-like peptide signaling pathway in the red flour beetle, Tribolium castaneum. Journal of Biological Chemistry, 286(49), 41924–36. Shringarpure, R. & Davies, K. J. A. (2009). Free radicals and oxidative stress in aging. In Handbook of Theories of Aging, ed. V. L. Bengston, D. Gans, N. M. Putney & M. Silverstein (chap. 13, pp. 229–43) (New York: Springer). Smedal, B., Brynem, M., Kreibich, C. D. & Amdam, G. V. (2009). Brood pheromone suppresses physiology of extreme longevity in honeybees (Apis mellifera). Journal of Experimental Biology, 212(23), 3795–801. Tanaka, E. D. & Hartfelder, K. (2004). The initial stages of oogenesis and their relation to differential fertility in the honey bee (Apis mellifera) castes. Arthropod Structure and Development, 33(4), 431–42. Tolfsen, C. C., Baker, N., Kreibich, C. & Amdam, G. V. (2011). Flight restriction prevents associative learning deficits but not changes in brain protein-adduct formation during honeybee ageing. Journal of Experimental Biology, 214(8), 1322–32. Tsuji, K., Kikuta, N. & Kikuchi, T. (2012). Determination of the cost of worker reproduction via diminished life span in the ant Diacamma sp. Evolution, 66(5), 1322–31. Tsuji, K., Nakata, K. & Heinze, J. (1996). Lifespan and reproduction in a queenless ant. Naturwissenschaften, 83(12), 577–8. Wagner, D. & Gordon, D. M. (1999). Colony age, neighborhood density and reproductive potential in harvester ants. Oecologia, 119(2), 175–82. Wang, Y., Brent, C. S., Fennern, E., & Amdam, G. V. (2012). Gustatory perception and fat body energy metabolism are jointly affected by vitellogenin and juvenile hormone in honey bees. PLoS GENETICS, 8(6), e1002779. Wang, Y., Mutti, N. S., Ihle, K. E., et al. (2010). Down-regulation of honey bee IRS gene biases behavior toward food rich in protein. PLoS GENETICS, 6(4), e1000896. Williams, G. C. (1957). Pleiotropy, natural selection, and the evolution of senescence. Evolution, 11(4), 398–411. Wilson, E. O. (1971). The Insect Societies (Cambridge, MA: Bellknap Press/Harvard University Press). Wolschin, F., Mutti, N. S. & Amdam, G. V. (2011). Insulin receptor substrate influences female caste development in honeybees. Biology Letters, 7(1), 112–15. Woyciechowski, M. & Moroń, D. (2009). Life expectancy and onset of foraging in the honeybee (Apis mellifera). Insectes Sociaux, 56(2), 193–201. Zheng, H., Zhang, Q., Liu, H., et al. (2012). Cloning and expression of vitellogenin (Vg) gene and its correlations with total carotenoids content and total antioxidant capacity in noble scallop Chlamys nobilis (Bivalve: Pectinidae). Aquaculture, 366, 46–53. Zhou, X., Oi, F. M. & Scharf, M. E. (2006). Social exploitation of hexamerin: RNAi reveals a major caste-regulatory factor in termites. Proceedings of the National Academy of Sciences of the United States of America, 103(12), 4499–504. Zhou, X., Wheeler, M. M., Oi, F. M. & Scharf, M. E. (2008). RNA interference in the termite Reticulitermes flavipes through ingestion of double-stranded RNA. Insect Biochemistry and Molecular Biology, 38(8), 805–15.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:35:23, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.010

11 Senescence in Modular Animals Botryllid Ascidians as a Unique Ageing System Baruch Rinkevich

Short Summary While multicellular organisms go through predictable ageing pathways, some defy progressive ageing by displaying continued growth and negligible senescence. These organisms share two archetypal life history traits: (1) sessile life mode and (2) colonial structures, conglomerates of repeated basic subunits, the modules (i.e. zooids, polyps, leaves). In many of these organisms, no boundaries exist between germ/somatic cell lines, and while the size and age of each individual basic module are usually constrained, the whole colony size/age may escape intrinsic restriction, revealing colonial entities with an unknown upper life span limit. Furthermore, colonial astogenic processes such as fission, fragmentation, fusion between ramets and partial mortality may dramatically alter actual sizes, blur predictions of age and also reveal the trait for negligible senescence with age. Model organisms, such as botryllid ascidians, are an indispensable tool in ageing research. In Botryllus schlosseri, ageing is marked by independent, sometimes contrasting types of ageing and senescence processes at the basic module (zooid) level, at the ramet level and at the genet level; they also exhibit novel rejuvenilisation processes (at the genet-ramet levels) following acute stress, where the stressed organism becomes younger, slowing down senescence. Colonial organisms may also present spatial and stochastic age-mosaic modules, postponement of senescence by a high regenerative power and replacement of basic modules that do not age according to classical criteria, indicating that ageing in modular organisms is possible but not obligatory. When senescence at the whole-genet level occurs, it may reflect a sharp contrast to senescence in unitary organisms. The offsetting of senescence at the basic module level may develop when vigorous totipotent stem cells exist, and there is no formal separation of soma from the germ line. The Botryllus system thus reveals, within the same colonial entity, constructed senescence/rejuvenilisation phenomena, such as semelparity versus iteroparity, programmed life span versus wear-and-tear senescence, weekly ageing of colonial modules versus whole-ramet-genet survivorship, rejuvenilisation versus extrinsic ageing and the immortality of germ/somatic cell lines. Though still in its infancy, studying ageing and senescence processes in sessile marine colonial organisms may lead to a better understanding of the evolutionary routes of senescence. Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:36:32, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.011

Introduction: To Get Old

221

Introduction: To Get Old Anyone can get old, all you have to do is live long enough. – Groucho Marx

‘Senescence’ is defined as the progressive decline in performance and the survival diminution that organisms experience with ageing, and exemplified as advanced losses of Darwinian fitness with time (Kirkwood 2005). In contrast, the term ‘ageing’ refers to the changes occurring in an organism during ontogeny, embracing the processes of embryogenesis, growth/differentiation, maturity, senescence and mortality. While ageing is one of the most prominent biological features of multicellular organisms, senescence is still an enigmatic discipline, as its evolutionary attributes are under intense debate, and various fundamental queries, such as the evolutionary connections between ageing and senescence, are still unanswered (Kirkwood 2005). A major unsolved query is why populations of multicellular organisms embraced the evolutionary trait of senescence, which causes each organism to accrue senescence along with age. In the ontogenetic continuum from birth to death, ageing is near the end pole of the sequence of development. In contrast, it has long been documented that several animal taxa, including cnidarians, sponges, bryozoans, urochordates and several chordates, do not fit with the general dogma of ageing but display continued growth and negligible senescence (Finch 1990, 1998, 2009). The same conclusion had been experimentally underscored, revealing that a wide range of environmental and biological stresses that could directly damage the body or increase the rate of wear and tear result in the paradoxical effect of rejuvenilisation and/or extended life span (Mitteldorf, 2010; Voskoboynik et al. 2002). It is interesting to note that the wide group of organisms that defy the progressive senescence mode, including plants and invertebrate animals, shares two life history traits: sessile life mode and colonial bodily structures. Evolutionary perspectives have posited several major incentives for senescence. Most accepted are those associated with the accumulation of mutations with time and the pleiotropy-related trade-offs between early and late reproduction (Kirkwood 2005; Medawar 1952). According to the evolutionary theory of mutation accumulation, mutations with a deleterious effect will pile up at late ontogenetic stages due to the decline in the strength of selection on age-specific characteristics, resulting in survival diminution and decreased fecundity. The idea that bodies wear out with age is deeply rooted in the scientific literature. The pleiotropic notion suggests negative genetic impacts on traits late in life, if those genes carry in parallel positive effects at earlier ontogenetic stages. A more elaborated evolutionary concept for senescence, which connects both theories, is the disposable soma tenet (Kirkwood & Holliday 1979). This theory advocates that when resources are limited (the most common scenario in nature), allocation of energy to early reproduction is favoured at the expense of somatic maintenance, altogether leading to enhanced senescence and death. However, there is no physical law that prohibits the life of an organism from continuing indefinitely, a claim further supported by documented cases in animals and plants. A good example is Turritopsis nutricula (Family: Oceanidae), an immortal hydrozoan, thought to

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:36:32, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.011

222

Senescence in Modular Animals

live forever without detectable signs of senescence. Through trans-differentiation (a cellular process characterised by the transformation of a somatic cell into another type of mature somatic cell without going through an intermediate pluripotent state), this species is capable of reverting completely from the medusa stage back into a sexually immature stage, after having reached sexual maturity as a solitary stage (Piraino et al. 2004). Another support for the tenet that there is no ubiquitous rule dictating senescence with age is the ‘negative senescence’ phenomenon (Finch 1990, 1998; Vaupel et al. 2004), where organisms (plants and animals alike) experience an extended period of life following the start of reproduction, during which mortality continues to decline. This extended period of life is expressed differently in various organisms. In Animalia, this phenomenon is specifically noted in colonial organisms (e.g. corals) (Babcook, 1991; Grigg, 1977) in which mortality and reproductive efforts are inversely related to colony size and age. However, this is not the general rule even within these taxa (e.g. Elahi & Edmunds 2007a; Rinkevich & Loya 1986).

Unitary versus Colonial Marine Animals Most studies on ageing processes in the Animalia deal with unitary aclonal species (and also aclonal/colonial species, such as some insects), individuals that present easy-totrack explicit ontogenetic pathways. Usually these organisms (e.g. vertebrates, insects and annelids) depict an ultimate meiotic reproductive pathway leading to easily distinctive individuals with genetically determined body sizes and predictable senescence pathways that conclude with the organism’s death. Colonial and modular animals (primarily of aquatic environments, many of which are clonal and sessile), however, do not follow this ‘unitary organismal’ pathway of development. They are typified by offspring individuals and/or sub-cloned ramets that emerge from somatic constituents without passing through regular meiotic cell cycles, thereby bypassing the sexual recombination of genetic constituent (Tuomi & Vuorisalo 1989). Colonial organisms (which are also clonal) represent structural individuals that consist of repeated subunits, the basic modules (Hughes 1989; Rinkevich 2002; Rosen 1986). Upon accomplishing ontogeny, the first established basic modules (i.e. polyp or zooid) commence astogeny, where similarly sized modules are continuously added – a process also known as ‘asexual reproduction’ – in various shapes and sizes, altogether forming simple to highly complex colonial pattern formation with basic modules that are arranged in a hierarchical subdivision. This modular assembly usually develops as a three-dimensional tessellation of its components, of which each level of modular organisation may contribute directly to the shape (Hughes 1989; Jackson & Coats 1986; Jackson & Hughes 1985; Rosen 1986). Structures may also reflect high plasticity. Within a specific species, different individuals can grow in various shapes, including arborescent, plating, massive, encrusting and linear forms (each may represents a typical disparate growth rate), changing pattern formations following signals from intrinsic and/ or extrinsic cues. Therefore, adaptive reiteration is an important trait at the ramet-genet levels. Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:36:32, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.011

Ageing in Modular Animals

223

Life histories of colonial/modular organisms may be characterised by limited motility capacities (when encrusting growth form is considered) but are usually referred to as sessile organisms that do not move on substrates after larval settlement and metamorphosis. Still, under field conditions, it is not always trivial to correctly identify genets. The outcomes of movement-like events, such as colonial fission, fusion and partial mortality (Hughes & Jackson 1980; Rosen 1986), further add abstractions to present knowledge, particularly when considering topics such as population structure, dynamics or colonial size and age of individual genets. In colonial forms, size is usually not a good predictor of whole-colony age (Hughes & Jackson 1980), nor is the separation between germ line and soma. Although the maximal size of each individual basic module within a colony is usually constrained (Rinkevich 2002), the whole-colony size in modular, and sessile animals may escape any intrinsic size restriction (Hughes 1989; Jackson & Coats 1986; Jackson & Hughes 1985; Rosen, 1986). Thus, individual colonies consist of a mosaic of modules representing different ages and different sizes. Moreover, whole-colony size-regulating mechanisms such as colonial fission, fragmentation, fusion between ramets (colonial fragments), injury, partial predation and partial mortality may dramatically alter the actual size of a specific colony, thus revealing that colony size and age are poorly allied. In some extreme cases, such as coral colonies (Elahi & Edmunds 2007b; Hughes & Jackson 1980; Jackson & Coats 1986; Rosen 1986), a small colony could be centuries old, whereas larger colonies could be just a few decades old. The same applies to physiological attributes in colonial animals (or plants) (Salguero-Gómez & Casper 2010), as no age-related decrease in physiological capacity is seen among most clonal animals. Modularity in sessile marine animals further confers a survival advantage because it allows the localisation and potential isolation of the damage to the basic module level (i.e. zooid or polyp) or to the ramet level, consequently avoiding failure in the entire genet and revealing the ‘partial mortality’ trait (Hughes 1989; Jackson & Coats 1986; Hughes & Jackson 1980). Similar is the ability to regenerate fast and most efficiently after major damage (e.g. Rinkevich et al. 1995, 2007), a phenomenon not characteristic of most individual entities. Long-lived ramets may also accumulate genetic heterogeneity through mutations, a mechanism that can be beneficial for survival and therefore contribute to longevity, as documented in plants (Borges 2009). Modularity facilitates additional biological attributes, such as decoupling body size from organ size constraints and partial relaxation of metabolic allometry at the colony level (Hughes 1989). Hence, the above-mentioned life history traits suggest that major biological attributes in colonial organisms are dynamic and ephemeral, where plastic and exponential indeterminate growths are major properties that differentiate colonial organisms from unitary entities.

Ageing in Modular Animals Ageing in modular animals holds some unique characteristics that distinguish it from ageing in unitary organisms. The asexual iteration of body plans instigates age-cohort Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:36:32, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.011

224

Senescence in Modular Animals

differences among modules of a specific genet, with the youngest modules usually positioned at the peripheral margins, concealing the chronological age of the genet, which dates from the point of settlement (in sessile colonial marine organisms). Regenerating colonial forms create further complex and stochastic age-related sets of basic modules. In contrast, the whole entity in unitary organisms is of a single age unit. In addition, the quantification of whole-colony versus genet age is challenging in many modular organisms, following constant fragmentations, colony fission, fusion and partial mortality that complicate any obvious relationship between age and size (Hughes & Jackson 1980; Jackson & Coats 1986; Jackson & Hughes 1985; Rosen 1986). Moreover, there are genotypes of assorted species that do not age according to the classical criteria, thus having the ability to persist (almost) indefinitely through asexual propagation (Jackson & Coates 1986). Part of this phenomenon is due to the postponement of senescence by employing processes such as a high regenerative power and the replacement of basic modules (Rosen 1986) that expand the longevity of whole genets. The challenge of defining the ecological/evolutionary ‘unit’ in colonial organisms (Rinkevich 2000) is another noticeable issue when elucidating ageing patterns. While the literature suggests that the size of a specific colony may increase indefinitely with age (Jackson & Hughes 1985), it may also lead to the conclusion that genets of some colonial marine animals do not necessarily senesce (e.g. some Hawaiian black corals were estimated to be over 4,200 years old) (Roark et al. 2009). The phenomena of budding, partial mortality, fusion and fission can be concurrently expressed at three hierarchical levels of colonial organisation: basic modules, ramets and genets. While either one of these phenomena can be executed more than once by a single genetic individual, senescence at the whole-genet level occurs only once, as a consequence of the ageing and death of all its individual ramets and basic modules. Therefore, in contrast to unitary animals, the ageing and death of a colonial basic module are not necessarily followed by colonial death but rather by the development of the colony into a fragmented colonial living surface area, depicting the concept of partial mortality and consent shrinkage (Palumbi & Jackson 1983). Claims were then raised (Heininger 2012) that modular organisms that have no segregated germ line may evade senescence. It is therefore of specific interest to note that when longevity mutants of unitary organisms are considered (e.g. Caenorhabditis elegans) (Curran et al. 2009), these entities exhibit a unique soma-to-germ-line transformation phenomenon that specifically contributes to their enhanced survival, recalling the somatic stem cell phenotype encountered in colonial animal taxa such as tunicates, sponges and corals. As the separation of soma from the germ line is invalid in many sessile colonial animals (separation of germ line and somatic cells occurs early in the ontogeny of most unitary animals; Buss, 1983), this exclusive situation allows, in different colonial marine organisms (e.g. corals, sponges, bryozoans and tunicates), to offset senescence of modules by either increasing the translocation of nutrients to newly developing parts of the colony, energetically sustaining vigorously growing distal colony areas, or by supporting neotenic generations of basic modules (Palumbi & Jackson 1983; Lauzon et al. 2000; Voskoboynik et al. 2002, 2004). It is also in a close functional relationship with the regenerative power manifested by so many modular organisms (Sköld & Obst 2011).

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:36:32, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.011

Facets for Ageing in B. schlosseri

225

Therefore, modular organisms do not necessarily undergo systemic senescence (Borges 2009). They also do not obey the inherent complexity of ageing (sensu Kirkwood 2005), where most organisms’ phenotypes undergo modification with time. Although longevity of each individual module within a colony can be constrained, the whole colony (at the genet level) may escape senescence, revealing colonial entities that are up to thousands of years old (e.g. corals) (Brown et al. 2009; Roark et al. 2009) with an unknown upper life span limit. Thus, senescence in modular organisms is possible (Rinkevich & Loya 1986) but not obligatory, where high rates of reproduction in colonial organisms could further slow down the process of senescence when developed (e.g. Tanner 2001). However, classical senescence attributes and universal physiological constraints such as growth and reproductive arrest were assigned to old colonies when compared with younger genotypes (Elahi & Edmunds 2007b; Hughes & Reynolds 2005; Kojis & Quinn 1985; Meesters & Bak 1995). Model organisms reflecting unique facets of senescence processes are an indispensable tool in ageing research, as they may hold answers to unsolved evolutionary attributes. One such model system is botryllid ascidians, a small group of encrusting colonial sessile urochordates (commonly associated with hard bottom biological assemblages) that present various types of senescence phenomena simultaneously on various morphological organisations. This includes the weekly senescence/rejuvenilisation cycles, part of a complex life history trait that may express a whole-genet programmed life span, ageing at the ramet level and rejuvenilisation after acute damage. Thus, to further elaborate a holistic view regarding ageing phenomena, in what follows I shall discuss senescence in B. schlosseri, a cosmopolitan shallow-waters encrusting colonial sea squirt that presents all the aforementioned ageing phenomena.

Facets for Ageing in B. schlosseri A Botryllus colony (Figure 11.1 (top)) is composed of several to thousands of similar units, zooids, each 1 to 3 mm in length. Zooids are embedded within the translucent gelatinous matrix, the tunic. Zooids within a colony commonly form star-shaped clusters (systems) around common cloacal apertures (siphons). All zooids within a system, as well as all systems within a single colony, are interconnected by an extensive network of blood vessels that ramify throughout the tunic. This common blood system bears sausage-like enlargements called ‘vascular ampullae’, which are found along the periphery of the colony and are scattered between the systems and near the upper surface of the colony. Colonies of this species exhibit several pathways for ageing processes on the basic module level and on the entire genet or ramet level.

Senescence at the Basic Module Level The life cycle of B. schlosseri, as in other botryllid ascidians, is characterised by a weekly unique developmental process in which all organismal zooids of a specific colony die and new zooids are reborn in a highly synchronised and cyclical development Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:36:32, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.011

226

Senescence in Modular Animals

Figure 11.1

B. schlosseri, a model colonial urochordate. (Top) Part of a colony with three colonial systems, each containing several zooids arranged around a common branchial siphon, which undergoes a weekly senescence process, blastogenesis (A–D). (A 0 –D 0 ) Schematic illustrations of the four major blastogenic cycles A–D. (A″–D″) The four major blastogenic cycles at the colony level. (A‴–D‴) Higher magnification for morphological status at the four major blastogenic cycles. Abbreviations: am = ampulla; as = atrial siphon; az = absorbed zooid; bd1 = primary bud; bd2 = secondary bud; bs = brachial siphon; bv = blood vessel; en = endostyle; ht = heart; st = stomach; zo = zooid. Bar = 500 micrometres.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:36:32, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.011

Facets for Ageing in B. schlosseri

227

route called ‘blastogenesis’, through a process of budding called ‘palleal budding’ (Berrill 1950a, 1950b, 1951). Each colony arises from a sexually produced chordate larva that attaches itself to a substratum and undergoes metamorphosis. The resulting first zooid (called ‘oozooid’) immediately begins to produce buds that start as outgrowths from its lateral wall and are genetically identical to the parent. This small colony continues to grow through the highly coordinated blastogenic cycles of palleal budding during which three successive generations of asexually derived zooids occur simultaneously side by side (mature zooids and first- and second-generation buds), the oldest generation succumbing to apoptotic death and morphological resorption. The whole process of blastogenesis consists of four major stages, marked by the letters A–D (Figure 11.1A–D) (Mukai & Watanabe 1976). Stages A–C, lasting four to five days under 20°C, begin with the formation of small vesicles at both sides of the zooids, which break off the parents’ (zooids) epidermis and peribranchial epithelium and segregate into blastula-like structures (Berrill 1941). As the cells proliferate, each vesicle undergoes a series of invaginations, differentiates and develops into an adult zooid within one week following initiation (Berrill 1941; Izzard 1973; Voskoboynik et al. 2007). At stage D, which lasts twenty-four to thirty-six hours, all adult zooidal tissues in the colony die, mainly by an apoptotic process, and are phagocytosed by specialised blood cells, the macrophages (Ballarin et al. 2008; Cima et al. 2010; Lauzon et al. 1992, 1993; Manni et al. 2007), which at this stage increase in frequency among circulating blood haemocytes (Ballarin et al. 1998; Manni et al. 2007). At the same time, the developing primary buds form their own vasculature and mature into a new generation of functional zooids (primary buds become zooids when their siphons are opened), which replaces the resorbed old generation of zooids. Zooid apoptosis in botryllid colonies is a wave-like process, beginning at the anterior end of each zooid and gradually advancing towards the posterior region (Lauzon et al. 1992). Blastogenesis is altered in response to various environmental conditions, such as temperature, physical insults and irradiation. With doses of 3,000 to 4,000 rads and above, irradiation arrests the formation of new buds and interrupts normal takeover, turning the colony into a chaotic bulk of vessels, buds and zooid segments. Death supervenes after a period of up to one month of poor conditions, characterised by loss of colonial organisation (Rinkevich & Weissman 1990). Zooidal life span during blastogenesis is further regulated by two independent signals: a bud-dependent signal and a bud-independent signal (Lauzon et al. 2007). Blocking phagocytic ingestion of senescent cells leads to an interruption of zooid degeneration (Voskoboynik et al. 2004), suggesting that the removal of signals on apoptotic cells, the removal of cell corpses by phagocytes and completion of the blastogenic cycle are closely linked. Thus, filtering zooids cannot support the development of a successive generation of buds to adulthood, except through their death (Manni et al. 2007). Bud maturation is therefore related to and conditioned by the regressing zooids: under conditions of stress, takeover occurs earlier, while maturation of primary buds is simultaneously mediated by circulating stem cells (Rinkevich et al. 2013). These birth–death blastogenesis cycles continue throughout the life span of the colony. Thus, unlike most species, where the body is long living and maintained by cellular replacements, B. schlosseri continuously replaces the colonial units on a weekly basis. It should Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:36:32, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.011

228

Senescence in Modular Animals

be noted that during the blastogenesis cycles, connective vasculature and tunic matrices remain intact, allowing constant signal transductions and pattern formation at the colony level. Thus, individual modules have a fixed short life span of approximately twenty-one days from first initiation to death (fourteen days spent developing into first and secondary buds and seven days as functional adult zooids at 20°C). Interestingly, developing buds and regressing zooids that extirpated from B. schlosseri colonies at different blastogenic stages revealed dramatic changes in life expectancies and ageing processes under in vitro conditions, exhibiting intrinsic phenomena that are probably masked by astogenic controls (Rabinowitz & Rinkevich 2004, 2011; Rabinowitz et al. 2009). The life expectancy of isolated buds in vitro, away from any discrete colonial regulatory cues, reached fifty to sixty, five times the life expectancy of intact, in vivo developing zooids, whereas the life expectancy of in vitro buds that remained unattached to the substrates lengthened to at least five months. The prevailing mode of death in vitro was necrotic, in contrast to the apoptotic mode of zooidal deterioration at blastogenesis. Extirpated buds and monolayers are also surprisingly very active on the molecular/biochemical levels (Pl10, P-MEK, MAP kinase and cadherin expressions) (Rabinowitz et al. 2009), also expressing molecular stemness flag (de novo expression of Piwi) (Rabinowitz & Rinkevich 2011). We therefore conclude (Rabinowitz & Rinkevich 2004) that not only are the underlying colonial growth mechanisms replaced by different developmental pathways under in vitro conditions but also that the internal colonial-level clocks programming death are replaced by a new biological mechanism with different timetables.

Senescence at the Ramet Level Many ramets, or genets where they are not sub-cloned during their life span to individual ramets, live for just a few weeks to several months in the wild (Brunetti and Copello 1978; Chadwick-Furman & Weissman 1995; Rinkevich et al. 1992; Yund & Stires, 2002). Towards the end of their life span, asexual reproduction begins to slow down so that colonies shrink over several weeks to months, and eventually blastogenesis is halted and is no longer able to replace tissues, leading to whole-colony death (ChadwickFurman & Weissman 1995). A similar phenomenon has been recorded in plants (Roach & Gampe 2004). This type of ageing typifies many iteroparous (repeatedly reproducing; see later) colonies. Ageing traits in these colonies are also subject to variation in environmental conditions (Boyd et al. 1986: Chadwick-Furman & Weissman 1995; Rinkevich et al. 1998; Yund & Stires 2002). Of special interest here is the very slow ageing process developing in laboratory-raised colonies of B. schlosseri, revealing dawdling growth compared to colonies from the field, which frequently fragment into small ramets (Brunetti 1974: Brunetti & Copello 1978: Chadwick-Furman & Weissman 1995). Some of these colonies were held in the laboratory six to thirteen years – compared to the several weeks to a few months’ life span of colonies in the field – and were fragmented into hundreds of ramets before they (the whole sum of ramets/genets) wore out and died (Rinkevich and Shapira 1998; Rinkevich unpublished data) (a representative case is depicted in Figure 11.2). Similar to the Botryllus laboratory longDownloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:36:32, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.011

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:36:32, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.011

Figure 11.2

A representative case for a long-lived B. schlosseri colony that was repeatedly sub-cloned to ramets. A founder ramet from a laboratory-raised colony (no record of the specific birthday) was taken (September 1994) and followed for about one year (until death was commenced), during which eleven ramets were sampled (nos. 1–6, 11, 12 and 17–19). Repeated sub-cloning of these ramets has resulted in a total of sixty daughter ramets from the chosen original ramet over the next three years from the starting day, at which time point all ramets disintegrated and died. The ‘life span’ of each ramet (from day of subcloning to death) ranged between two and twenty months (however, only three ramets survived more than sixteen months from the day of initiation). Therefore, the survival of the genet depended on repeated sub-cloning and the ‘production’ of daughter ramets. When several ramets were taken simultaneously, they usually revealed a ‘programmed life span’ phenomenon (sensu Rinkevich et al. 1992), succumbing to death within a short period. Asterisks denote sub-clones taken for experimentation (so their life span was not followed). Arrowheads denote failures in documenting the life spans of the ramets (last documented time point is presented).

230

Senescence in Modular Animals

lived cultured genets are the long-lived genets of the Japanese colonial tunicate Polyandrocarpa misakiensis (Family: Styelidae). In this species, colonies develop zooids that are four to five months old. Before dying by autolysis, zooids develop buds that mature into functional zooids of the next generation, a cyclical phenomenon that continues until the death of the genet. By ensuring this cyclical process, several genets were cultured in the laboratory for more than forty years following their field collection (Kawamura et al. 2012). The Botryllus and Polyandrocarpa systems thus further highlight our limited understanding of the evolution of senescence under natural conditions. Yet, the relaxed conditions in the aforementioned two laboratory-raised tunicates support evolutionary theories of ageing (e.g. Hamilton 1966), which consider senescence a result of an age-dependent decrease by selection forces. These theories envisage that populations experiencing relaxed mortality due to environmental conditions (e.g. controlled laboratory conditions) will show slower ageing. Moreover, while in unitary organisms the body is static and cells within organs are replaced by long-lived pools of stem cells with restricted lineage potential (such as hematopoietic stem cells) through homeostatic processes, in the Botryllus system, the basic modules are ephemeral, senescent and repeatedly replaced by new basic modules. Each of these modules is fully operational for only a single week, revealing different aspects for the disposable soma tenet (Kirkwood and Holliday 1979).

Senescence at the Genet Level The Botryllus system reveals two phenomena of senescence at the whole-genet level. The first deals with semelparous (reproducing once during their lifetime) and iteroparous (reproducing several times) sexually reproductive genets in wild Botryllus colonies (Grosberg 1988; Harvell & Grosberg 1988). Semelparous colonies are characterised by rapid, determinate growth that attains a single clutch emerging at early age, followed by the immediate death of the whole genet. Iteroparous colonies, however, grow much slower than semelparous genotypes; they postpone sexual reproduction until a later age for the consequent production of several to many clutches before senescing (Grosberg 1988). Iteroparous colonies also cease growing when entering into intra-specific interactions with conspecifics (Brunetti 1974). These genetically determined life history morphs appear to be regulated by genetics, developed as an adaptive response to environmental signals (Chadwick-Furman & Weissman 1995), since semelparous colonies revive the summer populations, while iteroparous colonies dominate the autumn populations. Therefore, neither size nor age can accurately predict the onset of senescence during the life cycle of Botryllus genets. The second type of senescence is termed ‘programmed life span’ (Rinkevich et al. 1992). This phenomenon appears when all respective ramets derived from a specific genet undergo regressive changes in concert, leading within one to two weeks to whole-genet death that is unlinked to a sexual reproductive effort or to any other external factor (Rinkevich et al. 1992). This pathway of senescence and death is distinct from the processes developed during blastogenesis. In laboratory settings, Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:36:32, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.011

Facets for Ageing in B. schlosseri

231

this non-random programmed life span has been developed simultaneously in all subclones of specific genets, months after their separation, thus strongly suggesting that it is mediated by heritable components (Lauzon et al. 2000; Rinkevich et al. 1992). Moreover, allogeneic fusion between a genotype exhibiting the programmed life span phenomenon and a young colony that does not exhibit this phenomenon has evoked a senescence signal in the whole new entity, causing the death of both partners within the chimera. An early induced senescence and death of the chimeric entity following parasitic cell lineage invasions may actually be advantageous because this response can alter the frequency of germ cell parasitism and reduce the potency of germ-line parasitic forms, as super-parasitic forms that suppress the development of the host’s germ cells may be eliminated by this induced programmed life span (Simon-Blecher et al. 2004). This non-random senescence proceeds according to a series of characteristic changes within the colony that include systemic constriction and congestion of the vasculature, accompanied by massive accumulation of pigment cells in the zooid body wall, blood vessels and ampullae; gradual shrinkage of individual zooids (bodily shrinkage, however, may serve for some species as a fundamental plastic trait) (Salguero-Gómez & Casper 2010); loss of colonial architecture and, ultimately, death. At the ultra-structure level, individual cells exhibit changes typical of ischemic cell death, culminating in necrotic cell lysis rather than apoptosis (Lauzon et al. 2000). Therefore, such a nonrandom senescence differs from the weekly blastogenic senescence on the morphological/cellular levels as well.

Rejuvenilisation following Acute Stress When ramets of long-lived B. schlosseri genets were acutely treated with lethal doses of the antioxidant butylated hydroxytoluene (BHT), blastogenesis was completely arrested, and colonies deteriorated to a morphologically chaotic state (Voskoboynik et al. 2002, 2004). Many of the treated colonies died. However, rescued ramets resorbed BHT-treated zooids, regenerated entirely new sets of zooids and then revealed young colony phenotypes, including enhanced growth rates, compared to control ramets of the same genets. More interestingly, they revealed up to a 4.6-fold of post-treatment life expectancy. In colonies rescued from BHT treatment, zooid resorption started immediately and was completed within a few days, concomitantly with the re-operation of circulating phagocytes, revealing important roles of the phagocytic cells in the coordination between death and clearance signals during blastogenesis (Voskoboynik et al. 2004). These results thus invalidate the two metabolic theories of senescence (Finch 2009; Finkel and Holbrook 2000) that interpret senescence either in terms of the rate of living, where a fixed total metabolic potential is consumed over an expected lifetime, after which the organism wears out and dies, or in terms of accumulative oxidative damage resulting in progressive and irreversible changes in metabolic pathways. The possible existence of an ageing clock that can be set by the environment has also been suggested (Voskoboynik et al. 2002). Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:36:32, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.011

232

Senescence in Modular Animals

Discussion on Coloniality/Individuality and Senescence We turn not older with years, but newer every day – Emily Dickinson

Entering the realm of ageing across broad phyletic samples of multicellular organisms, it emerges that senescence endorses a host of degenerative processes that are characterised by cumulative loss of tissue and cellular functions while not predicting any accepted unequivocal signature for ageing at the whole-organism level. While ageing processes are more homogeneous when expressed in unitary organisms, this is not the case when comparisons are made with colonial organisms. These organisms represent spatial and stochastic age-mosaic modules within each specific colony, postponement of senescence by high regeneration power, replacement of modules and genotypes/species that do not age according to the classical criteria. All of the preceding discussion reveals that ageing in modular organisms is possible (e.g. Meesters & Bak 1995; Rinkevich & Loya 1986) but not obligatory and that modularity often displays continued growth and reveals the quality of negligible senescence (Finch 1990, 1998, 2009) with age. When senescence occurs, it may reflect a sharp contrast to senescence in unitary organisms that show variation in ageing phenotypes (Kirkwood 2005) as exhibiting a clear pathway for programmed ageing (Rinkevich et al. 2002). Another difference between modular and unitary organisms is that some of the colonial morphs have the ability to become younger and slow down senescence, or, in contrast, they can express the induced programmed life span phenotype. While a wide range of colonial organisms exhibit one of the aforementioned characteristics, B. schlosseri exhibits all of them. In this species, much of the ageing phenomena are not defined by the accumulation of stochastic damages or by limited energy allocation to somatic maintenance. Accepting the holistic view regarding all ageing phenomena, it is appropriate to challenge the notion of biological senescence and what it means while also considering that senescence is associated with stem cell activity and vigour. Theoretically, stem cell entities are immortal, showing no senescence, and entities that are made solely of stem cells may survive indefinitely, without an apparent senescence. Two such natural phenomena exist. The first is the ‘agent’ causing canine transmissible venereal tumour (CTVT), a tumour that is naturally transmissible as an allograft in all dog breeds through coitus, licking, biting and sniffing tumour-affected areas (Das & Das 2000). CTVT is a globally distributed parasitic cell line forming chimera entities. Shared alleles between CTVT samples of dogs from various sites worldwide revealed that this cell line, which has appeared about 6,000 years ago (Rebbeck et al. 2009) when dogs were first domesticated, is still proliferating without any sign of senescence. The second example is the devil’s facial tumour disease (DFTD) (Murchison 2008), a clonally derived allograft transmitted by biting between natural populations of the marsupial carnivorous Tasmanian devil. While DFTD is a newcomer (first reported around mid-1990s), it is not showing any sign of senescence either, contagiously passing from one devil to another. As in many other sessile colonial animals, in B. schlosseri there is no formal separation of soma from the germ line, an exclusive situation that allows offsetting the senescence of

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:36:32, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.011

References

233

basic modules, especially when vigourous totipotent stem cells exist. It has been suggested that in botryllid ascidians, stem cells mediate a myriad of biological events, including aestivation, hibernation, blastogenesis, ‘colonial chimerism’, somatic and germ-line parasitism and diverse regeneration scenarios, serving as the evolutionary units of natural selection (reviewed in Rinkevich et al. 2013). Following the discovery that stem cells in Botryllus can contribute to a wide range of developing tissues (Voskoboynik et al. 2008), recent studies (Rinkevich et al. 2013) have documented the existence of a novel niche for putative germ and somatic stem cells in B. schlosseri (cell islands (CIs)). Cells within CIs express markers associated with somatic and germ stem cells in addition to expressing gene products of the BMP, TGF-β, Wnt, retinoic acid, FGF, JAK/STAT and Ras oncogene pathways, implicating CIs as signalling centres for stem cells (germ-line and somatic stem cells). These stem cells display self-renewal and pluripotency through cyclical weekly waves of migration from the old generation of zooids before their degradation, homing in on developing buds, and continuously migrating (on a weekly basis) from aged and damaged niches to newly formed niches as a means of facilitating their long-term fates. It is possible, therefore, that extended life span vigour in organisms such as B. schlosseri is at least partly based on the exuberance of their stem cell lineages that show no impairment throughout the animal’s life span. While evolutionary theories explain the widespread occurrence of ageing in multicellular organisms and significant attention has been given to studying senescence processes in unitary organisms, the time is ripe for capturing the interest in ageing within modular animals and the emergence of colonial organisms as the most suitable model cases for studying senescence. Unique cases such as the Botryllus system reveal within a single entity, in harmony and partially even in concert, constructing senescence and ageing phenomena, such as semelparity versus iteroparity, programmed life span versus wear-and-tear random senescence, weekly ageing of colonial basic modules versus whole-ramet-genet survivorship and senescence versus rejuvenilisation (caused by extrinsic hazards) of different ramets from the same genet all coherently associated with totipotent stem cells that continuously give rise to germ and somatic cell lineages.

Acknowledgements This study was supported by a grant from the Israel Academy of Sciences (68/10). I thank G. Paz for drawing the figures.

References Babcook, R. (1991). Comparative demography of three species of scleractinian corals using age- and size-dependent classifications. Ecological Monographs, 61, 225–44. Ballarin, L., Cima, F. & Sabbadin, A. (1998). Apoptosis during the takeover phase of the ascidian Botryllus schlosseri colonial life cycle. Animal Biology, 7, 86.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:36:32, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.011

234

Senescence in Modular Animals

Ballarin, L., Burighel, P. & Cima, F. (2008). A tale of death and life: natural apoptosis in the colonial ascidian Botryllus schlosseri (Urochordata, Ascidiacea). Current Pharmaceutical Design, 14, 138–47. Berrill, N. J. (1941). The development of the bud in Botryllus. Biology Bulletin, 80, 169–84. Berrill, N. J. (1950a). The Tunicata (London: Ray Society). Berrill, N. J. (1950b). The Tunicata with an Account of the British Species (London: Ray Society). Berrill, N. J. (1951). Regeneration and budding in tunicates. Biological Reviews, 26, 451–75. Borges, R. M. (2009). Phenotypic plasticity and longevity in plants and animals: cause and effect? Journal of Biosciences, 84, 605–11. Boyd, H. C., Brown, S. K., Harp, J. A. & Weissman, I. L. (1986). Growth and sexual maturation of laboratory-cultured Monterey Botryllus schlosseri. Biological Bulletin, 170, 91–109. Brown, D. P., Basch, L., Barshis, D., et al. (2009). American Samoa’s island of giants: massive Porites colonies at Ta’u island. Coral Reefs, 28, 735. Brunetti, R. (1974). Observations on the life cycle of Botryllus schlosseri (Pallas) (Ascidiacea) in the Venetian lagoon. Italian Journal of Zoology, 41, 225–51. Brunetti, R. & Copello, M. (1978). Growth and senescence in colonies of Botryllus schlosseri (Pallas) (Ascidiacea). Italian Journal of Zoology, 45, 359–64. Buss, L. W. (1983). Evolution, development and the units of selection. Proceedings of the National Academy of Science the United States of America, 80, 1387–91. Chadwick-Furman, N. E. & Weissman, I. L. (1995). Life histories and senescence of Botryllus schlosseri (Chordata, Ascidiacea) in Monterey Bay. Biological Bulletin, 89, 36–41. Cima, F., Manni, L., Basso, G., et al. (2010). Hovering between death and life: natural apoptosis and phagocytes in the blastogenetic cycle of the colonial ascidian Botryllus schlosseri. Developmental and Comparative Immunology, 34, 272–85. Curran, S. P., Wu, X. D., Riedel, C. G. & Ruvkun, G. (2009). A soma-to-germline transformation in long-lived Caenorhabditis elegans mutants. Nature 459, 1079–84. Das, U. & Das, A. K. (2000). Review of canine transmissible venereal sarcoma. Veterinary Research Communications, 24, 545–56. Elahi, R. & Edmunds, P. J. (2007a). Tissue age affects calcification in the scleractinian coral Madracis mirabilis. Biological Bulletin, 212, 20–8. Elahi, R. & Edmunds, P. J. (2007b). Consequences of fission in the coral Siderastrea siderea: growth rates of small colonies and clonal input to population structure. Coral Reefs, 26, 271–6. Finch, C. E. (1990). Longevity, Senescence, and the Genome (University of Chicago Press). Finch, C. E. (1998). Variations in senescence and longevity include the possibility of negligible senescence. The Journals of Gerontology Series B: Biological Sciences 53A, B235–9. Finch, C. E. (2009). Update on slow aging and negligible senescence: mini-review. Gerontology, 55, 307–13. Finkel, T. & Holbrook, N. J. (2000). Oxidants, oxidative stress and the biology of ageing. Nature, 408, 239–47.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:36:32, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.011

References

235

Grigg, R. (1977). Population dynamics of two gorgonian corals. Ecology, 58, 278–90. Grosberg, R. K. (1988). Life history within a population of the colonial ascidian Botryllus schlosseri: I. The genetic and environmental control of seasonal variation. Evolution, 42, 900–20. Hamilton, W. D. (1966). The moulding of senescence by natural selection. Journal of Theoretical Biology, 12, 12–45. Harvell, C. D. & Grosberg, R. K. (1988). The timing of sexual maturity in clonal animals. Ecology, 69, 1855–64. Heininger, K. (2012). The germ-soma conflict theory of aging and death: obituary to the ‘Evolutionary theories of aging’. WebmedCentral AGING, 3(4), WMC003275. doi: 10.9754/journal.wmc. 2012.003275 Hughes, K. A. & Reynolds, R. M. (2005). Evolutionary and mechanistic theories of aging. Annual Review of Entomology, 50, 421–45. Hughes, R. N. (1989). A Functional Biology of Clonal Animals (London: Chapman & Hall). Hughes, T. P. & Jackson, J. B. C. (1980). Do corals lie about their age? Some demographic consequences of partial mortality, fission, and fusion. Science, 209, 713–15. Izzard, C. S. (1973). Development of polarity and bilateral asymmetry in the palleal bud of Botryllus schlosseri (Pallas). Journal of Morphology, 139, 1–26. Jackson, J. B. C. & Coates, A. G. (1986). Life cycles and evolution of clonal modular animals. Philosophical Transactions of the Royal Society London Series B: Biological Sciences, 313, 7–22. Jackson, J. B. C. & Hughes, T. P. (1985). Adaptive strategies of coral reef invertebrates. American Scientist, 73, 265–73. Kawamura, K., Kitamura, S., Sekida, S., et al. (2012). Molecular anatomy of tunicate senescence: reversible function of mitochondrial and nuclear genes associated with budding cycles. Development, 139, 4083–93. Kirkwood, T. B. (2005). Understanding the odd science of aging. Cell, 120, 437–47. Kirkwood, T. B. L. & Holliday, F. R. S. (1979). The evolution of ageing and longevity. Proceedings of the Royal Society of London Series B: Biological Sciences, 205, 531–46. Kojis, B. L. & Quinn, N. J. (1985). Puberty in Goniastrea favulus age or size limited? In Proceedings of the 5th International Coral Reef Congress, Vol. 4 (pp. 289–93). Lauzon, R. J., Ishizuka, K. J. & Weissman, I. L. (1992). A cyclical, developmentally regulated death phenomenon in a colonial urochordate. Developmental Dynamics, 94, 71–83. Lauzon, R. J., Patton, C. W. & Weissman, I. L. (1993). A morphological and immunohistochemical study of programmed cell death in Botryllus schlosseri (Tunicata, Ascidiacea). Cell Tissue Research, 272, 115–27. Lauzon, R. J., Kidder, S. J. & Long, P. (2007). Suppression of programmed cell death regulates the cyclical degeneration of organs in a colonial urochordate. Developmental Biology, 301, 92–105. Lauzon, R. J., Rinkevich, B., Patton, C. W. & Weissman, I. L. (2000). A morphological study of non-random senescence in a colonial urochordate. Biological Bulletin, 198, 367–78. Manni, L., Zaniolo, G., Cima, F., et al. (2007). Botryllus schlosseri: a model ascidian for the study of asexual reproduction. Developmental Dynamics, 236, 335–52.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:36:32, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.011

236

Senescence in Modular Animals

Medawar, P. B. (1952). An Unsolved Problem in Biology (London: Lewis. Meesters, E. H. & Bak, R. P. M. (1995). Age-related deterioration of a physiological function in the branching coral Acropora palmata. Marine Ecology Progress Series, 121, 203–9. Mitteldorf, J. (2010). Aging is not a process of wear and tear. Rejuvenation Research, 13, 322–6. Mukai, H. & Watanabe, H. (1976). Studies on the formation of germ cells in a compound ascidian Botryllus primigenus Oka. Journal of Morphology, 148, 377–82. Murchison, E. P. (2008). Clonally transmissible cancers in dogs and Tasmanian devils. Oncogene, 27, S19–30. Sköld, H. N. & Obst, M. (2011). Potential for clonal animals in longevity and ageing studies. Biogerontology, 12, 387–96. Palumbi, S. R. & Jackson, J. B. C. (1983). Aging in modular organisms: ecology of zooid senescence in Steginoporella sp. (Bryozoa, Cheilostomata). Biological Bulletin, 164, 267–78. Piraino, S., De Vito, D., Schmich, J. et al. (2004). Reverse development in Cnidaria. Canadian Journal of Zoology, 82, 1748–54. Rabinowitz, C. & Rinkevich, B. (2004). In vitro delayed senescence of extirpated buds from zooids of the colonial tunicate, Botryllus schlosseri. Journal of Experimental Biology, 207, 1523–32. Rabinowitz, C. & Rinkevich, B. (2011). De novo emerged stemness signatures in epithelial monolayers developed from extirpated palleal buds. In Vitro Cellular and Developmental Biology: Animal, 47, 26–31. Rabinowitz, C., Alphasi, G. & Rinkevich, B. (2009). Further portrayal of epithelial monolayers, emergent de novo from extirpated ascidians’ palleal buds. In Vitro Cellular and Developmental Biology: Animal, 45, 334–42. Rebbeck, C. A., Thomas, R., Breen, M., et al. (2009). Origins and evolution of a transmissible cancer. Evolution, 63, 2340–9. Rinkevich, B. (2000). A critical approach to the definition of Darwinian units of selection. Biological Bulletin, 199, 231–40. Rinkevich, B. (2002). The branching coral Stylophora pistillata: the contribution of genetics in shaping colony landscape. Israel Journal of Zoology, 48, 71–82. Rinkevich, B. & Loya, Y. (1986). Senescence and dying signals in a reef building coral. Experientia, 42, 320–2. Rinkevich, B. & Shapira, M. (1998). An improved diet for inland broodstock and the establishment of an inbred line from Botryllus schlosseri, a colonial sea squirt (Ascidiacea). Aquatic Living Resources, 11, 163–71. Rinkevich, B. & Weissman, I. L. (1990). Botryllus schlosseri (Tunicata) whole colony irradiation: do senescent zooid resorption and immunological resorption involve similar recognition events? Journal of Experimental Zoology, 253, 189–201. Rinkevich, B., Shlemberg, Z. & Fishelson, L. (1995). Whole body protochordate regeneration from totipotent blood cells. Proceedings of the National Academy of Sciences of the United States of America, 92, 7695–9. Rinkevich, B., Porat, R. & Goren, M. (1998). On the development and reproduction of Botryllus schlossen (Tunicata) colonies from the eastern Mediterranean Sea: plasticity of life history traits. Invertebrate Reproduction and Development, 34, 207–18.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:36:32, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.011

References

237

Rinkevich, B., Lauzon, R. J., Brown, B. W. & Weissman, I. L. (1992). Evidence for a programmed lifespan in a colonial protochordate. Proceedings of the National Academy of Sciences the United States of America, 89, 3456–550. Rinkevich, Y., Paz, G., Rinkevich, B. & Reshef, R. (2007). Systemic bud induction and retinoic acid signaling underlie whole body regeneration in urochordate Botrylloides leachi. PLoS Biology, 5, 900–13. Rinkevich, Y., Voskoboynik, A., Rosner, A., et al. (2013). Repeated, long-term cycling of putative stem cells between niches in a basal chordate. Developmental Cell, 24, 76–88. Roark, E. B., Guilderson, T. P., Dunbar, R. B., et al. (2009). Extreme longevity in proteinaceous deep-sea corals. Proceedings of the National Academy of Sciences the United States of America, 106, 5204–8. Roach, D. A. & Gampe, J. (2004). Age-specific demography in Plantago: uncovering age-dependent mortality in a natural population. American Naturalist, 164, 60–9. Rosen, B. R. (1986). Modular growth and form of corals: a matter of metamers? Philosophical Transactions of the Royal Society of London Series B: Biological Sciences, 313, 115–42. Rosner, A., Moiseeva, E., Rinkevich, Y., et al. (2009). Vasa and the germ line lineage in colonial urochordate. Developmental Biology 331, 113–28. Salguero-Gómez, R. & Casper, B. B. (2010). Keeping plant shrinkage in the demographic loop. Journal of Ecology, 98, 312–23. Simon-Blecher, N., Achituv, Y., & Rinkevich, B. (2004). Protochordate concordant xeno-transplantation settings reveal outbreaks of donor cells and divergent lifespan traits. Developmental and Comparative Immunology, 28, 983–91. Tanner, J. E. (2001). The influence of clonality on demography: patterns in expected longevity and survivorship. Ecology, 82, 1971–81. Tuomi, J. & Vuorisalo, T. (1989). Hierarchical selection in modular organisms. Trends in Ecology & Evolution, 14, 209–13. Vaupel, J. W., Baudisch, A., Dölling, M., et al. (2004). The case for negative senescence. Theoretical Population Biology, 65, 339–51. Voskoboynik, A., Reznick, A. Z. & Rinkevich, B. (2002). Rejuvenescence and extension of a urochordate lifespan following a single, acute administration of an anti-oxidant, butylated hydroxytoluene. Mechanisms of Ageing and Development, 123, 1203–10. Voskoboynik, A., Rinkevich, B.A. Weiss, E. et al. (2004). Macrophage involvement for successful degeneration of apoptotic organs in the colonial urochordate Botryllus schlosseri. Journal of Experimental Biology 207, 2409–16. Voskoboynik, A., Simon-Blecher, N. Soen, Y., et al. (2007). Striving for normality: whole body regeneration through a series of abnormal generations. FASEB Journal 21, 1335–44. Voskoboynik, A., Soen, B., Rinkevich, Y., et al. (2008). Identification of the endostyle as a stem cell niche in a basal chordate. Cell Stem Cell, 3, 456–64. Yund, P. O., & Stires, A. (2002). Spatial variation in population dynamics in a colonial ascidian (Botryllus schlosseri). Marine Biology, 141, 955–63.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:36:32, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.011

12 Hydra Evolutionary and Biological Mechanisms for Non-Senescence Ralf Schaible, Felix Ringelhan, Boris H. Kramer and Alexander Scheuerlein

Short Summary Species of the genus Hydra show no senescence and have the potential for extremely long life spans. Previous studies have demonstrated both constant mortality under constant laboratory conditions and instances of high mortality under specific environmental conditions. Here we review these studies and argue that these unique life history traits can be understood as adaptations to a specific ecological niche, given Hydra’s specific anatomy and biology. We argue that these adaptations are facilitated by the genet-ramet organisational system of a clonal organism.

Introduction The members of the phylum Cnidarians (together with the Ctenophora (comb jellies) and the Porifera (sponges)) are diploblastic animals with tissue-grade organisation and a nervous system. Cnidaria are sister to all bilateral species (Schierwater et al. 2009) and consist of four classes: Anthozoa, Hydrozoa, Scyphozoa and Cubozoa. Most of the Cnidaria are marine (e.g. corals or jellyfish), live solely as a polyp, or alternate between a polyp and a medusa stage. The freshwater Hydra is a derived member of Hydrozoa and consists of a cylindrical body with hypostome and tentacles at the oral end and a basal disk on the aboral end. Members of the genus Hydra never produce a medusa stage. They reproduce asexually by forming buds or sexually by producing gonads (Figure 12.1). Campbell (1987) clustered all known Hydra species into four groups in a single genus: viridissima, braueri, oligactis and vulgaris. This taxonomy was based mainly on differences in the morphological structure of embryonic theca, nematocysts and the phylogeny of endosymbiotic algae. Recent phylogenetic analyses (Kawaida et al. 2010; Martínez et al. 2010) confirmed this taxonomy based on morphological characteristics. The body plan of Hydra consists of two histologically distinct tissue layers, a simple nervous system and stem cells. A Hydra (Cnidaria, Hydrozoae) individual consists of

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:37:38, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.012

Introduction

Testis

Sperms

239

Eggs Female

Male

Fertilisation

Fertilised egg separated from mother polyp Development of a bud

Asexual cycle

Separation of the bud

Sexual cycle

Embryo hatch to a miniature hydra

After growing, hydra looks like an adult polyp

Figure 12.1

A schematic life cycle of Hydra. Asexual reproduction by forming a bud (i.e. budding) is the main reproductive output. Hydra can also reproduce sexually by forming gonads, eggs and sperm. After successful fertilisation, the egg produces a theca and falls into the sediment. It then hatches and develops into a small Hydra – the new genet or clone lineage.

30,000 to 200,000 cells clustered into twenty to twenty-five different functional types (Bode 1996; Bode et al. 1973; Bosch et al. 2010). Cell turnover rates are high, and all Hydra cells are renewed after a period of twenty to thirty days (Campbell 1967). Stem cells are composed of three separated, morphologically different lineages: ectodermal epithelial stem cells, endodermal epithelial stem cells and interstitial stem cells (Hobmayer et al. 2012; Steele 2002). Despite their morphological differences, the stem cell lineages are part of a highly dynamic system with a high rate of cell proliferation and continuous replacement of epithelial and interstitial stem cells. Cells in the central part of the gastric body can move either to the upper or lower body columns, where they differentiate into head or foot somatic tissue. During the process of asexual reproduction, they migrate towards a growing bud. Unlike planarians, which regenerate from single isolated stem cells (Salo 2006), viable Hydra can only regenerate from cell aggregates with all three stem cell tissue lineages. If Hydra tissue is isolated without disaggregation of individual cells,

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:37:38, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.012

240

Hydra: Evolutionary and Biological Mechanisms for Non-Senescence

a fragment of the size of a few hundred cells suffices to successfully regenerate a polyp (Shimizu et al. 1993). In contrast, a disaggregated and re-aggregated individual of Hydra attenuata must contain a minimum number of 2 × 104 cells (about one-fifth of a mature Hydra) and a fair proportion of epithelial cells (15 per cent) (Bode et al. 1973; Gierer et al. 1972) to be able to fully regenerate. Hydra species are extremely successful in establishing populations in small freshwater reservoirs such as rivers, lakes and ponds (Holstein & Emschermann 1995) in a broad range of climates (Kawaida et al. 2010; Martínez et al. 2010). Hydra individuals attach via the foot region on hard substrates such as stones, plants (i.e. common reed) or macro-algae (i.e. charophyceae), and individuals are dispersed by water currents or by waterfowl. Hydra mainly reproduce by budding (Bell & Wolfe 1985; Cuker & Mozley 1981; Ribi et al. 1985; Welch & Loomis 1924). Sexual reproduction is scant in Hydra and occurs predominantly when population density is high (Bell & Wolfe, 1985) or when unfavourable environmental conditions prevail such as extreme temperatures or chemical stressors (Sugiyama & Fujisawa 1977; Yoshida et al. 2006). Under natural conditions, Hydra individuals are very vulnerable to environmental factors such as seasonal variation in ambient temperature and availability of food, as well as unpredictable water-level changes, inter-species competition for resources and predation mortality. All these influences lead to considerable seasonal fluctuation of Hydra abundance in their natural habitat, which is well documented (Cuker & Mozley 1981; Bell & Wolfe 1985; Ribi et al. 1985; Welch & Loomis 1924). Other than population density in their natural habitat, knowledge on Hydra’s biology under natural conditions is scant due to methodological issues such as the challenge of marking and following soft-bodied individuals of microscopic size in the wild. Most of the knowledge on Hydra comes from laboratory settings. There Hydra serves as a model organism in various biological disciplines such as developmental biology, toxicity research, stem cell biology, genetics and life history biology (Campbell 1967; Bosch 2012; Bosch et al. 2010; Gierer et al. 1972; Jones et al. 2014; Quinn et al. 2012; Schaible & Sussman 2013; Schaible et al. 2011, 2015). The unique body of knowledge on ageing in Hydra under laboratory conditions has generated novel insights into the mechanisms that enable regeneration and the avoidance of senescence. In addition, Hydra offers a perfect model to identify changes in their mortality pattern: nonsenescence, whereby age-specific mortality and fertility rates remain constant (Schaible et al. 2015), and a classical senescence-like pattern, in which the probability of death increases with age (Schaible et al. 2014; Yoshida et al. 2006). In this chapter we compile published research on ageing in Hydra and review the findings in the light of evolutionary theory. As with most of ageing research, investigations on Hydra are still in early stages of development, with many ‘unknowns’ and no definitive solutions. We feel that it is time to step back, review the evidence and elaborate hypotheses to explain different senescence patterns found in Hydra and to answer the question why some Hydra species seem to defy the senescence process completely. Firstly, we will illustrate the different ageing patterns that have been reported in Hydra. Secondly, we will argue that this phenomenon can be understood in the light of evolution Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:37:38, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.012

Mortality Patterns in Hydra

241

both at the genet and the ramet levels, complemented by the notion of how Hydra interacts with its immediate environment. Thirdly, we will highlight relevant aspects of Hydra’s physiology and anatomy that may help to understand Hydra’s unique position among other living organisms with respect to its simple body plan and its intriguing regeneration capabilities. Fourthly, we will argue that Hydra provides a unique opportunity to gain insights into the ageing process that will be relevant to understand the evolution of ageing across the Tree of Life.

Mortality Patterns in Hydra Genet-Ramet System in Hydra In order to better understand the different mortality patterns found in the genus Hydra, it is important to introduce the genet-ramet dual system of a clonal organismal organisation (Buss 1985). A ‘genet’ is the product of successful fertilisation of an egg by sperm, and it represents a novel and unique genetic composition. In Hydra, a sexually derived individual (genet clonal lineage) can continually produce new independent physiological and demographic units (ramets) by budding. Therefore, all individuals produced by budding, including the buds of a bud, are genetically identical ‘ramets’ that collectively represent a single genet. Since selection acts on genes alone (Orive 1995), evolutionary explanations of observed demographic rates should consider both the genet and the ramet levels, as suggested by Caswell (1985) and Orive (1995). However, due to methodological difficulties, the measurement and interpretation of demographic rates at the ramet and genet level have rarely been attempted (but see Damman and Cain 1998; Hartnett & Bazzaz 1985; Karlson 1986; Tanner 2001; de Witte et al. 2011). Here we describe the mortality pattern of Hydra along the life course of a genet, the actual clonal lineage with various implications and consequences caused by different senescence patterns on the ramet level.

Mortality of Eggs and Buds After sexual reproduction and successful external fertilisation, the Hydra egg produces a theca, a protective shell around the embryo, and a small polyp hatches and develops into a grown polyp – a new genet or clone lineage. This process is rather costly in Hydra, as only around 20 per cent of fertilised eggs hatch successfully (Moore & Campbell 1973; Sugiyama & Fujisawa 1977). Eggs derived from inbred strains (crosses of individuals of the same strain and pond) have even lower hatch rates. Sugiyama and Fujisawa (1977) suggested that inbreeding depression results in high frequencies of recessive deleterious alleles in Hydra strains that negatively affect hatching rate. Once hatched, the hatchling grows rapidly for the first approximately thirty days to become a miniature polyp. According to Grassi et al. (1995), most hatchlings die during this critical period of body growth. Three yet untested hypotheses have been put forward to explain the high mortality that ensues after a new individual has been produced sexually. Firstly, sexual reproduction might lead to novel genotypes that are therefore

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:37:38, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.012

242

Hydra: Evolutionary and Biological Mechanisms for Non-Senescence

‘untested’. Genetic constellations that have been formed anew might be incompatible, produce errors in development and growth and result in non-viable genets. Second, instabilities in the complicated and dynamic processes during growth may cause high mortality. Somatic growth is a complex process circling around the ‘what should be generated first’ question, form or function (Ricklefs 1979). Form is built easily but cannot assist in acquiring new resources, whereas functional tissue is nutritionally and energetically demanding to build (von Bertalanffy 1948). Thirdly, adverse environmental conditions may lead to stress and complications during the settlement, feeding, growth and development phases of hatchlings. Favourable substrate and feeding conditions differ between hatchlings and already established polyps. For example, food items must be much smaller for hatchlings, and hatchlings’ substrate requirements are more eclectic (Grassi et al. 1995). In contrast to sexual reproduction, budding, which is a form of asexual reproduction, does not lead to elevated early-life mortality. Budding is initiated by the growth of excess tissue, which bulges around the stalk of the Hydra and eventually forms a bud. Once the bud is large enough, it detaches from the parents (Otto & Campbell 1977). The lack of independent tissue differentiation and the fact that growth takes place while the bud is still attached to the parent apparently minimise the environmental effects and instabilities during development and growth. Although hard to measure due to the difficulty of identifying a bud in its early stages, budding mortality seems to be extremely low (R. Schaible, unpublished data). Hence, vegetative budding seems to be the safest and most effective way to propagate quickly.

Mortality of Adults Once the newly formed genet survives to the adult stage, it experiences extremely low mortality in the laboratory across all ages. In a pioneering study by Martínez (1998), H. vulgaris showed a low and constant mortality rate over a period of four years. This was confirmed by a more recent study by Schaible et al. (2015), who found in an experiment with a much higher number of individuals over a much longer observation period that individuals of both H. vulgaris and H. magnipapillata strain 105 showed an extremely low but constant rate of mortality. Furthermore, these authors demonstrated that fertility of Hydra stays constant throughout the life span of an individual. Additionally, they found that mortality and fertility rates are constant irrespective of ramet or genet age, which ranged from zero to forty-one years. It is important to note that mortality rate is extremely low at an average of only one annual death in 167 individuals and that it remains constant at that level even when sexual stages are produced (Schaible et al. 2015). Consequently, individual Hydra ramets may live for hundreds of years, although they are not immortal. The situation is somewhat different when sudden changes in the environment affect Hydra’s life history. We will illustrate this with two examples: 1. In H. oligactis, adverse environmental conditions (temperature decline) induced a sexual phase, which was followed by a period of elevated mortality (Brien 1953; Yoshida et al. 2006). In the study by Yoshida and colleagues, only a few individuals

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:37:38, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.012

Ecological Aspects of the Genet-Ramet System in Hydra

243

of both the female (n = 150) and the male (n = 148) cohorts were alive after 200 days. Once individuals entered the sexual reproduction mode, Yoshida et al. (2006) observed several important changes: (a) a strong decline in budding rate, (b) no differentiation of cells into tentacle cells or foot cells and (c) no new production of endodermal and ectodermal stem cells. It seems therefore that H. oligactis use all resources for terminal sexual reproduction, similar to semelparous organisms (Burghardt & Metcalf 2016). In contrast, the closely related H. magnipapilla and H. vulgaris do not undergo deterioration after the onset of sexual reproduction (Martínez 1998; Yoshida et al. 2006). Since it is unclear whether the increasing mortality rate observed solely in H. oligactis individuals is a consequence of intrinsic ageing processes or the result of the allocation of resources into reproduction at the cost of somatic maintenance, Schaible et al. (2014) termed this behaviour as ‘senescence-like’. 2. After Hydra are collected in the field and transferred to the laboratory, they show elevated mortality levels (e.g. H. vulgaris and H. oligactis in Hase 1909). Many authors have argued that this high mortality is caused by unregulated growth of commensals, parasites or other micro-organisms that were introduced along with wild Hydra (Augustin et al. 2010; Fraune & Bosch 2007; Warren & Robson 1998).

Ecological Aspects of the Genet-Ramet System in Hydra Maximal Resource Use by Clonal Reproduction The feature that a single individual can continuously produce new independent physiological units (Watson & Casper 1984) by budding without undergoing meiosis may be interpreted as growth (Harper 1977), as the somatic structure that bears the identical genome in each cell increases with time. From this perspective, the sum of all individual ramets determines the size of a genet, and ‘budding’ is equivalent to growth (i.e. indeterminate growth) (Caswell 1985; Orive 1995). In contrast, ramet size has been shown to be determinate and to be independent of food supply (Otto & Campbell 1977). If food is plentiful, excess resources are not channelled into the growth of a single ramet but into the production of buds (Schaible et al. 2011) and thus into genet growth. This ecological strategy is not unique to Hydra, as it has been found in other organisms such as plants (de Witte & Stoecklin 2010) and invertebrates (Nilsson Skold & Obst 2011). As in all clonal organisms, the probability of death for a genet is determined by the product of ramet mortality times the number of ramets comprising the genet. Consequently, the probability of extinction of a complete genet diminishes as the number of ramets within the genet increases. Genet size also determines genet fertility, which can be calculated as the product of the number of ramets and the probability of a ramet producing a female egg. Assuming that there is no density-dependent feedback of genet size on the probability of reproducing sexually, genet fertility should increase with genet size. Consequently, a Hydra genet may potentially persist for an unlimited amount of time once it has been Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:37:38, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.012

244

Hydra: Evolutionary and Biological Mechanisms for Non-Senescence

established. In other species such as plants, many studies have shown that individual genets may expand over large geographical areas, and their ages, given slow growth rates, may exceed thousands of years (Ally et al. 2010; de Witte & Stoecklin 2010; Ender 1997; Watkinson 1992). Since a genet is able to grow over time, it can be compared to organisms with indeterminate growth that show an increase in reproductive output with increasing size (age). Just as in other indeterminate growers such as mussels (Abele et al. 2009), turtles (Congdon et al. 2003), snakes (Robert and Bronikowski 2010; Sparkman et al. 2009) and crocodiles (Tumarkin-Deratzian et al. 2007), the increased reproductive potential at high age of a genet attenuates the decline of the force of selection with age and can even revert due to the higher fitness of old and large genets compared to young, unestablished genets. Consequently, a non-senescent ageing strategy should be able to outcompete other strategies (Orive 1995). An important consequence of the indeterminate growth capacity of Hydra genets is the practically unlimited persistence of a genet once it has become established.

Extrinsic Mortality and the Genet-Ramet System Catastrophic High Mortality and the Extinction of the Genet Sometimes environmental conditions can be so severe (e.g. desiccation or freezing events) that the extinction of a complete genet in a local area may seem immanent (Bell & Wolfe 1985). Hydra individuals exposed to such conditions should then switch to sexual reproduction because their eggs can survive extremely adverse conditions (Bell & Wolfe 1985; Yoshida et al. 2006). The duration and frequency of such hostile environmental deterioration events are likely to be higher in transient ponds than in permanent ones, and we predict that the frequency of adverse conditions determines both genetic distances among genets and the number of clonal lineages alive.

Extrinsic Mortality that Causes Ramet Mortality but May Not Cause Extinction of the Genet Evidence so far suggests that Hydra ramets in the wild are extremely vulnerable to stochastic factors such as environmental hazards and predation. Whether high extrinsic mortality actually favours senescence depends on how mortality risk interacts with the condition of individuals. If a high risk of extrinsic mortality leads to the evolution of individuals of better physical condition, then the onset of senescence will be delayed, as shown by Reznick et al. (2004). Here the fish from ponds with the highest risk of predation had longer life spans under laboratory conditions than fish from ponds with low predation. In contrast, if extrinsic mortality affects all individuals irrespective of their physical condition, shorter life spans will evolve (Kramer & Schaible 2013; Williams & Day 2003; Williams et al. 2006). In the case of Hydra, we suspect that unpredictable changes in their natural habitat will considerably affect Hydra ramet life histories. Which direction these changes will take, however, is impossible to predict (Schaible et al. 2011). We believe that a fruitful approach to understanding senescence patterns in Hydra would focus on the main sources of extrinsic mortality for Hydra by testing effects of predation, competition for resources and space, or diseases. Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:37:38, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.012

Ramet Anatomy and Cellular Processes Are Key to Low and Constant Mortality in Hydra

245

Ramet Anatomy and Cellular Processes Are Key to Low and Constant Mortality in Hydra Genet demography is determined by the production rate and death rate of ramets, with senescence at the ramet level of minor importance (de Witte et al. 2011; Orive 1995). As long as the production rate of ramets exceeds their death rate, the probability of death for the genet in natural conditions is minimal. This implies that long life span and low mortality of single ramets are not essential prerequisites for non-senescence on the genet level. Still, Schaible et al. (2015) showed that under optimal laboratory conditions without extrinsic mortality, Hydra ramets with various ages do not show senescence. Here we will review the anatomical and physiological properties that help Hydra ramets achieve this.

Lack of Germ-Line–Soma Separation and Multicellularity Current theories on the evolution of senescence focus on animals with a clear germ-line– soma segregation, where the germ line is potentially immortal, whereas the soma wears out and dies as in most metazoan species like humans, Drosophila and turtles (Kirkwood 1977, 1991; Kirkwood and Austad 2000; Hamilton 1966; Medawar 1952; see also Chapter 14). Theory is unclear, however, as to which patterns of senescence are expected in organisms without this clear distinction, such as bacteria, protozoans, basal metazoans and plants (Kirkwood and Austad 2000). Several studies suggest that senescence may be common in organisms without clear germ-line–soma-cell-line segregation. For example, senescence at the ramet level has been reported in bacteria (Stewart et al. 2005; Wang et al. 2010) and, entering the multicellular animal kingdom, for hydranths of hydrozoans (Brock 1974; Hughes 1987; Martínez 2002). It seems, therefore, that the phenomenon of senescence is deeply rooted in the Tree of Life itself, being an inseparable feature of life, where anything ‘old’ becomes replaced by something ‘new’. On the bacterial level, ‘old’ and ‘new’ units differ with respect to the concentration of damaged products. It is inevitable, therefore, that asymmetric cell division produces one damaged, enriched ‘old’ cell and one non-damaged ‘new’ cell (Stewart et al. 2005; Wang et al. 2010). This mechanism of rejuvenation is retained in multicellular organisms in the production of sexual gametes, where the germ line is separated from the soma and maintained infinitely, whereas the soma deteriorates after fulfilling its reproductive role (Kirkwood 1977). Why do Hydra ramets not senesce? One possible explanation is that multicellularity opens a way for Hydra polyps to overcome senescence (Martínez and Levinton 1992). A deeper insight on the cellular construction of a Hydra ramet reveals that each cell of the three stems cell lineages acts independently (Steele 2002) with respect to cell division, cell survival and the rate of mutation accumulation. Consequently, each cell may still be able to ‘wear out’ and senesce over time, as long as rejuvenation is ensured in the offspring cells, similar to the processes in bacteria (Stewart et al. 2005; Wang et al. 2010). We argue that this might be possible in Hydra by asymmetrical damage transmission during the division of stem cells. This means that each cell division results in one

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:37:38, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.012

246

Hydra: Evolutionary and Biological Mechanisms for Non-Senescence

damaged, enriched ‘old’ cell and one non-damaged ‘new’ cell. Given the well-known cell movement along the body column, we expect that ‘old’ cells loaded with damage should accumulate in the head and foot region of the polyp. Here they either differentiate into head or foot cells or are eliminated by apoptosis and sloughing (Chera et al. 2009). In contrast, the non-damaged cells remain in the central body column of the parental polyp. Since the multicellular body of Hydra polyps consists of a high proportion of stem cells with regular and continuous proliferation activity (see Schaible et al. 2014), segregation of damage-loaded cells into the edges of the Hydra body ensures that the body column of the parental generation remains free of damage. This process is analogous to cellular processes of organisms with sexual reproduction that protect the sequestered germ line from damage and mutations (Kirkwood 1977). However, whether there is asymmetric or symmetric damage accumulation at the cell or even at the individual level must be tested in future experiments.

Hydra Longevity as a By-Product of Selection for a Low Level of Cell Differentiation Bosch et al. (2010) described the whole body column excepting the head and foot region of Hydra as a stem cell niche containing a large fraction of cells with stem cell character. Hydra is able to build a multicellular body mainly of stem cells that retain the ability to take over the roles of differentiated cells (see Dańko et al. 2015). In this way, it is stem cells in Hydra that are responsible for several physiological or structural functions, such as the formation of an epithelial cell layer and the maintenance of body shape and structure (Technau and Steele 2011). We hypothesise that it is this property that allows Hydra to overcome the ‘Hayflick limits’ of cells, the cessation of reproductive capacity of somatic or differentiated cells (Hayflick 1965). This stem cell community provides each Hydra genet with an unlimited source of cell proliferation potential. In the light of Hydra’s cell turnover rate of three to four days (Bosch and David 1984), epithelial stem cells will have undergone 720 cell divisions over a period of eight years (see Schaible et al. 2015). Just like other cells in a body, stem cells must constantly contend with damage arising from both endogenous insults such as reactive oxygen species generated by cellular metabolism and exogenous insults generated by their surrounding environment (Blanpain et al. 2011; Sancar et al. 2004). Danko et al. (2015) demonstrated in their theoretical model on cellular processes in Hydra that damage accumulation can be avoided by the interplay of four well-known properties of Hydra: a high proportion and number of stem cells, a high and constant stem cell division rate, an effective cell selection mechanism and the ability of stem cells to take over roles of differentiated cells. However, how stem cells escape damage accumulation is still a matter of debate.

Successful Maintenance in Hydra and the Link to Organismal Complexity Complexity is a concept often employed when it comes to describe the evolution of a body plan and its development, but it is hard to quantify (Adami 2002). Still, organismal

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:37:38, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.012

Ramet Anatomy and Cellular Processes Are Key to Low and Constant Mortality in Hydra

247

complexity may provide an answer on how some organisms escape senescence, leading to an age-dependent increase in mortality. From a life history perspective, ‘maintenance’ can be defined as the processes that prevent organismal decline in function and ultimately the total failure of the organism (Stearns 1992). Organismal or somatic maintenance of the subunits of an organism can be achieved by three complementary processes, and the actual level of complexity of an organism determines which strategy is favourable. The first process prevents damage before it happens, such as cellular defence mechanisms that scavenge radical oxygen species (Asada 2006; Imlay 2013). In the second process, damage to parts of an organism are repaired, such as damaged DNA strands (Branzei and Foiani 2008) or broken bone tissue (Shapiro 2008). Third, damaged structures can be replaced, as it happens in damaged cells that undergo apoptosis (Roos and Kaina 2006). Note that these mechanisms act on different levels of organismal organisation (e.g. DNA, cell, tissue, organ) and that repair of one level might entail the replacement of the complete element of the level below, such as the replacement of proteins when cell surfaces are renewed or the replacement of cells to repair a tissue (Krafts 2010). Hydra has the extraordinary capacity to replace elements at each level, from cells to tissues, or even to complete tentacles. Hence, the question arises why some organisms are able to renew complete structure, whereas others must maintain or repair them in order to reach somatic maintenance, and how this is linked to the evolution of senescence. Organismal complexity may deliver an answer on how some organisms escape senescent processes that lead to an age-dependent increase in mortality. Increased diversification and functional specialisation (structural complexity) lead to higher ‘integrated-ness’, where higher-level functional units rely on lower-level functional units as a result of specialisation or division of labour (Adami 2002; Changizi et al. 2002). Thus, it is impossible to renew complete structures without compromising the functionality of other structures that rely on the function of the structure to be renewed (e.g. heart tissue in humans). Additionally, increasing costs for redundant units of increasingly complex structures will accrue. A strategy that prevents damage or repairs body parts thus may be the best way to avoid potentially life-threatening renewal in complex organisms. As a cost, however, debilitation, damage and mutation accumulation increase with age, and malfunction of functional units increases as the consequences of incompletely repaired structures in complex organisms, leading inevitably to senescence (Kirkwood 2005). In organisms with low complexity, a simple body plan, redundant functional units and low ‘integrated-ness’, such as Hydra, the dependency of different cells and tissues on one another is reduced, which is why renewal can be accomplished without incurring costs to other systems. Renewal of structure, rather than repair, completely prevents the accumulation of mutations and damage, as well as debilitation and malfunction, and is a key element for Hydra to achieve non-senescence.

Implications of Budding and High Cell Proliferation Rates on Life History Trade-Offs in Hydra Ramets Trade-offs play an essential role in shaping age-specific life histories (Stearns 1992). Resources that are used for a certain purpose cannot be used for another; for example, Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:37:38, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.012

248

Hydra: Evolutionary and Biological Mechanisms for Non-Senescence

resources allocated to reproduction cannot be used for maintenance, which thus leads to a reduced life span. In Hydra, both the maintenance of ramets and asexual reproduction require the same physiological process, namely, an exceptionally proliferative machinery for stem cell renewal (Schaible et al. 2014). It is remarkable that so far studies on maintenance and reproduction in Hydra have failed to detect trade-offs. As an example, Schaible et al. (2011) showed in H. magnipapillata that budding increased linearly with food intake, while starvation survival (a proxy for the investment in maintenance) remained constant. Yoshida et al. (2006) found that after a drop in temperature, H. oligactis ramets showed signs of senescence and initiated sexual reproduction, but non-senescence could be restored after the polyps were subject to an increase in temperature. This strategy resembles that of semelparous organisms (see chapter 14), where H. oligactis invests all resources into one life history trait, ‘sexual reproduction’, at the expense of somatic maintenance (no production of new stem cells) and asexual reproduction (no budding). Further experiments are needed to understand the complicated trade-off system in Hydra. It would be important to know whether H. oligactis really undergoes physiological senescence followed by a shift from budding to sexual reproduction or whether starvation and depletion of resources led to the observed mortality increase. Further, the influence of other environmental impacts such as drought, overcrowding or biotic competition on changes in mortality and reproduction patterns should be experimentally shown to understand which factors may operate in the switch between senescence and non-senescence pathways.

Conclusion Here, we have argued that Hydra’s non-senescence and low mortality patterns found in laboratory conditions can be understood as a consequence of adaptations that bear immediate benefits in the face of the ecological conditions during Hydra’s phylogenetic history. First of all, the genet-ramet dual system allows the maximisation of fitness in environments with unpredictable but huge resources that may also vanish catastrophically. A genet-ramet dual system provides an ideal life history solution in this environment. Second, the genetramet system provides high levels of maintenance capabilities, which promote longevity. These specific properties of Hydra include regeneration capabilities, a high turnover of cells and the omnipotence of stem cells. All of these properties provide immediate benefits but also promote constant and low mortality as a by-product.

References Abele, D., Brey, T. & Philipp, E. (2009). Bivalve models of aging and the determination of molluscan lifespans. Experimental Gerontology, 44, 307–15. Adami, C. (2002). What is complexity? BioEssays, 24, 1085–94. Ally, D., Ritland, K. & Otto, S. P. (2010). Aging in a long-lived clonal tree. PLoS Biology, 8(8): e1000454.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:37:38, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.012

References

249

Asada, K. (2006). Production and scavenging of reactive oxygen species in chloroplasts and their functions. Plant Physiology, 141, 391–6. Augustin, R., Fraune, S. & Bosch, T. C. G. (2010). How Hydra senses and destroys microbes. Seminars in Immunology, 22, 54–8. Bell, G. & Wolfe, L. M. (1985). Sexual and asexual reproduction in a natural population of Hydra pseudoligactis. Canadian Journal of Zoology, 63, 851–6. Blanpain, C., Mohrin, M., Sotiropoulou, P. A. & Passegue, E. (2011). DNA-damage response in tissue-specific and cancer stem cells. Cell Stem Cell, 8, 16–29. Bode, H., Berking, S., David, C. N., et al. (1973). Quantitative analysis of cell types during growth and morphogenesis in hydra. Wilhelm Roux’ Archiv fuer Entwicklungsmechanik der Organismen, 171, 269–85. Bode, H. R. (1996). The interstitial cell lineage of Hydra: a stem cell system that arose early in evolution. Journal of Cell Science, 109, 1155–64. Bosch, T. C. G. (2012). Understanding complex host-microbe interactions in Hydra. Gut Microbes, 3, 345–51. Bosch, T. C. G., Anton-Erxleben, F., Hemmrich, G. & Khalturin, K. (2010). The Hydra polyp: nothing but an active stem cell community. Development Growth and Differentiation, 52, 15–25. Bosch, T. C. G. & David, C. N. (1984). Growth-regulation in Hydra: relationship between epithelial-cell cycle length and growth-rate. Developmental Biology, 104, 161–71. Branzei, D. & Foiani, M. (2008). Regulation of DNA repair throughout the cell cycle. Nature Reviews Molecular Cell Biology, 9, 297–308. Brien, P. (1953). La Perennité Somatique. Biological Reviews of the Cambridge Philosophical Society, 28, 308–49. Brock, M. A. (1974). Growth, developmental, and longevity rhythms in Campanularia flexuosa. American Journal of Zoology, 14, 757–71. Buss, L. W. (1985). The uniqueness of the individual revisited. In Population Biology and Evolution of Clonal Organisms, ed. J. B. C. Jackson, L. W. Buss & R. E. Cook (pp. 467–506 ) (New Haven, CT: Yale University Press). Campbell, R. D. (1967). Tissue dynamics of steady state growth in Hydra littoralis: I. Patterns of cell division. Developmental Biology, 15, 487. Campbell, R. D. (1987). A new species of Hydra (Cnidaria, Hydrozoa) from North America with comments on species clusters within the genus. Zoological Journal of the Linnean Society, 91, 253–63. Caswell, H. (1985). The evolutionary demography of clonal reproduction. In Population Biology and Evolution of Clonal Organisms, ed. J. B. C. Jackson, L. W. Buss & R. E. Cook (pp. 187–224) (New Haven, CT: Yale University Press). Changizi, M., Mcdannald, M. & Widders, D. (2002). Scaling of differentiation in networks: nervous systems, organisms, ant colonies, ecosystems, businesses, universities, cities, electronic circuits, and Legos. Journal of Theoretical Biology, 218, 215–37. Chera, S., Buzgariu, W., Ghila, L. & Galliot, B. (2009). Autophagy in Hydra: a response to starvation and stress in early animal evolution. Biochimica et Biophysica Acta, 1793(9), 1432–43. Congdon, J. D., Nagle, R. D., Kinney, O. M., et al. (2003). Testing hypotheses of aging in long-lived painted turtles (Chrysemys picta). Experimental Gerontology, 38, 765–72.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:37:38, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.012

250

Hydra: Evolutionary and Biological Mechanisms for Non-Senescence

Cuker, B. E. & Mozley, S. C. (1981). Summer population fluctuations, feeding, and growth of Hydra in an arctic lake. Limnology and Oceanography, 26, 697–708. Damman, H. & Cain, M. L. (1998). Population growth and viability analyses of the clonal woodland herb, Asarum canadense. Journal of Ecology, 86, 13–26. Dańko, M. J., Kozłowski, J. & Schaible, R. (2015). Unraveling the non-senescence phenomenon in Hydra. Journal of Theoretical Biology, 382, 137–49. De Witte, L. C., Scherrer, D. & Stoecklin, J. (2011). Genet longevity and population age structure of the clonal pioneer species Geum reptans based on demographic field data and projection matrix modelling. Preslia, 83, 371–86. De Witte, L. C. & Stoecklin, J. (2010). Longevity of clonal plants: why it matters and how to measure it. Annals of Botany, 106, 859–70. Ender, A. (1997). Untersuchungen zur Evolutionsgenetik des athekaten Hydrozoons Eleutheria dichotoma (Quatrefages 1842). PhD dissertation, Frankfurt am Main, University, Germany. Fraune, S. & Bosch, T. C. G. (2007). Long-term maintenance of species-specific bacterial microbiota in the basal metazoan Hydra. Proceedings of the National Academy of Sciences of the United States of America, 104, 13146–51. Gierer, A., Bode, H., Berking, S., et al. (1972). Regeneration of hydra from reaggregated cells. Nature-New Biology, 239, 98. Grassi, M., Tardent, R. & Tardent, P. (1995). Quantitative data about gametogenesis and embryonic development in Hydra vulgaris Pall (Cnidaria, Hydrozoa). Invertebrate Reproduction and Development, 27, 219–32. Hamilton, W. D. (1966). Moulding of senescence by natural selection. Journal of Theoretical Biology, 12, 12–45. Harper, J. L. (1977). Population Biology of Plants (London: Academic Press). Hartnett, D. & Bazzaz, F. (1985). The genet and ramet population dynamics of Solidago canadensis in an abandoned field. Journal of Ecology, 407–13. Hase, A. (1909). Über die deutschen Süßwasserpolypen Hydra fusca. L., Hydra grisea L. und Hydra viridis L: Eine biologische Vorarbeit, zugleich ein Beitrag zur Vererbungslehre. Archiv für Rassen- und Gesellschafts-Biologie 6, 721–53. Hayflick, L. (1965). Limited in vitro lifetime of human diploid cell strains. Experimental Cell Research, 37, 614–36. Hobmayer, B., Jenewein, M., Eder, D., et al. (2012). Stemness in Hydra: a current perspective. International Journal of Developmental Biology, 56, 509–17. Holstein, T. & Emschermann, P. (1995). Cnidaria, Hydrozoa (G. Fischer). Hughes, R. G. (1987). The loss of hydranths of Laomedea flexuosa Alder and other hydroids, with reference to hydroid senescence. In Modern Trends in the Systematics, ecology, and Evolution of Hydroids and Hydromedusae, ed. J. F. Bouillon, F. Boero, F. Cicogna & P. F. S. Cornelius (pp. 171–84 )(Oxford: Clarendon Press). Imlay, J. A. (2013). The molecular mechanisms and physiological consequences of oxidative stress: lessons from a model bacterium. Nature Reviews Microbiology, 11, 443–54. Jones, O. R., Scheuerlein, A., Salguero-Gómez, R., et al. (2014). Diversity of ageing across the tree of life. Nature, 270, 301–4. Karlson, R. H. (1986). Disturbance, colonial fragmentation, and size-dependent life history variation in two coral reef cnidarians. Marine Ecology Progress Series, 28, 245–9.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:37:38, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.012

References

251

Kawaida, H., Shimizu, H., Fujisawa, T., et al. (2010). Molecular phylogenetic study in genus Hydra. Gene, 468, 30–40. Kirkwood, T. B. L. (1977). Evolution of aging. Nature, 270, 301–4. Kirkwood, T. B. L. (1991). Longevity, senescence, and the genome: Finch, Ce. Science, 252, 1864–5. Kirkwood, T. B. L. (2005). Understanding the odd science of aging. Cell, 120, 437–47. Kirkwood, T. B. L. & Austad, S. N. (2000). Why do we age? Nature, 408, 233–8. Krafts, K. P. (2010). Tissue repair: the hidden drama. Organogenesis, 6, 225–33. Kramer, B. H. & Schaible, R. (2013). Life span evolution in eusocial workers: a theoretical approach to understanding the effects of extrinsic mortality in a hierarchical system. Plos ONE, 8(4): e61813. Martínez, D. E. (1998). Mortality patterns suggest lack of senescence in Hydra. Experimental Gerontology, 33, 217–25. Martínez, D. E. (2002). Senescence and rejuvenation in asexual metazoans. In Reproductive Biology of Invertebrates, Vol. XI: Progress in Asexual Reproduction, ed. R. N. Hughes (pp. 115–40 ) (New York: Wiley). Martínez, D. E., Iniguez, A. R., Percell, K. M., et al. (2010). Phylogeny and biogeography of Hydra (Cnidaria: Hydridae) using mitochondrial and nuclear DNA sequences. Molecular Phylogenetics and Evolution, 57, 403–10. Martínez, D. E. & Levinton, J. S. (1992). Asexual metazoans undergo senescence. Proceedings of the National Academy of Sciences of the United States of America, 89, 9920–3. Medawar, P. B. (1952). Uniqueness of the Individual: An Unsolved Problem of Biology (pp. 44–70 ) (London: Lewis). Moore, L. B. & Campbell, R. D. (1973). Non-budding strains of Hydra: isolation from sexual crosses and developmental regulation of form. Journal of Experimental Zoology, 185, 73–81. Nilsson Skold, H. & Obst, M. (2011). Potential for clonal animals in longevity and ageing studies. Biogerontology, 12, 387–96. Orive, M. E. (1995). Senescence in organisms with clonal reproduction and complex life-histories. American Naturalist, 145, 90–108. Otto, J. J. & Campbell, R. D. (1977). Budding in Hydra attenuata: bud stages and fate map. Journal of Experimental Zoology, 200, 417–28. Quinn, B., Gagne, F. & Blaise, C. (2012). Hydra, a model system for environmental studies. International Journal of Developmental Biology, 56, 613–25. Reznick, D. N., Bryant, M. J., Roff, D., et al. (2004). Effect of extrinsic mortality on the evolution of senescence in guppies. Nature, 431, 1095–9. Ribi, G., Tardent, R., Tardent, P. & Scascighini, C. (1985). Dynamics of Hydra populations in Lake Zurich, Switzerland, and Lake Maggiore, Italy. Schweizerische Zeitschrift fur Hydrologie, 47, 45–56. Ricklefs, R. E. (1979). Adaptation, constraint, and compromise in avian postnatal-development. Biological Reviews of the Cambridge Philosophical Society, 54, 269–90. Robert, K. A. & Bronikowski, A. M. (2010). Evolution of senescence in nature: physiological evolution in populations of garter snake with divergent life histories. American Naturalist, 175, 147–59.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:37:38, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.012

252

Hydra: Evolutionary and Biological Mechanisms for Non-Senescence

Roos, W. P. & Kaina, B. (2006). DNA damage-induced cell death by apoptosis. Trends in Molecular Medicine, 12, 440–50. Salo, E. (2006). The power of regeneration and the stem-cell kingdom: freshwater planarians (platyhelminthes). Bioessays, 28, 546–59. Sancar, A., Lindsey-Boltz, L. A., Unsal-Kacmaz, K. & Linn, S. (2004). Molecular mechanisms of mammalian DNA repair and the DNA damage checkpoints. Annual Review of Biochemistry, 73, 39–85. Schaible, R., Ringelhan, F., Kramer, B. H. & Miethe, T. (2011). Environmental challenges improve resource utilization for asexual reproduction and maintenance in Hydra. Experimental Gerontology, 46, 794–802. Schaible, R., Scheuerlein, A., Dańko, M. J., et al. (2015). Constant mortality and fertility over age in Hydra. Proceedings of the National Academy of Sciences of the United States of America, 112(51), 15701–6. Schaible, R. & Sussman, M. (2013). FOXO in aging: did evolutionary diversification of FOXO function distract it from prolonging life? Bioessays, 35, 1101–10. Schaible, R., Sussman, M. & Boris, K. H. (2014). Aging and the potential of selfrenewal: Hydra living in the age of aging – a mini-review. Gerontology, 60, 548–56. Schierwater, B., Eitel, M., Jakob, W., et al. (2009). Concatenated analysis sheds light on early metazoan evolution and fuels a modern ‘urmetazoon’ hypothesis. PLoS Biology, 7(1): e1000020. Shapiro, F. (2008). Bone development and its relation to fracture repair: the role of mesenchymal osteoblasts and surface osteoblasts. European Cells and Materials, 15, 53–76. Shimizu, H., Sawada, Y. & Sugiyama, T. (1993). Minimum tissue size required for Hydra regeneration. Developmental Biology, 155, 287–96. Sparkman, A. M., Vleck, C. M. & Bronikowski, A. M. (2009). Evolutionary ecology of endocrine-mediated life-history variation in the garter snake Thamnophis elegans. Ecology, 90, 720–8. Stearns, S. C. (1992). The Evolution of Life Histories (New York, Oxford University Press). Steele, R. E. (2002). Developmental signalling in Hydra: what does it take to build a ‘simple’ animal? Developmental Biology, 248, 199–219. Stewart, E. J., Madden, R., Paul, G. & Taddei, F. (2005). Aging and death in an organism that reproduces by morphologically symmetric division. PLoS Biology, 3, 295–300. Sugiyama, T. & Fujisawa, T. (1977). Genetic analysis of developmental mechanisms in hydra: I. Sexual reproduction of Hydra magnipapillata and isolation of mutants. Development, Growth and Differentiation, 19, 187–200. Tanner, J. E. (2001). The influence of clonality on demography: patterns in expected longevity and survivorship. Ecology, 82, 1971–81. Technau, U. & Steele, R. E. (2011). Evolutionary crossroads in developmental biology: Cnidaria. Development, 138, 1447–58. Tumarkin-Deratzian, A. R., Vann, D. R. & Dodson, P. (2007). Growth and textural ageing in long bones of the American alligator Alligator mississippiensis (Crocodylia: Alligatoridae). Zoological Journal of the Linnean Society, 150, 1–39. Von Bertalanffy, L. (1948). Das organische Wachstum und seine Gesetzmäßigkeiten. Experientia, 4, 255–69.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:37:38, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.012

References

253

Wang, P., Robert, L., Pelletier, J., et al. (2010). Robust growth of Escherichia coli. Current Biology, 20, 1099–1103. Warren, A. & Robson, E.A. (1998). The identity and occurrence of Kerona pediculus (Ciliophora: Hypotrichida), a well-known epizoite of Hydra vulgaris (Cnidaria: Hydrozoa). Zoologische Verhandelingen (Leiden), 323, 235–45. Watkinson, A. (1992). Plant senescence. Trends in Ecology and Evolution, 7, 417–20. Watson, M. A. & Casper, B. B. (1984). Morphogenetic constraints on patterns of carbon distribution in plants. Annual Review of Ecology and Systematics, 15, 233–58. Welch, P. S. & Loomis, H. A. (1924). A limnological study of Hydra oligactis in Douglas Lake, Michigan. Transactions of the American Microscopical Society, 43, 203–35. Williams, P. D. & Day, T. (2003). Antagonistic pleiotropy, mortality source interactions, and the evolutionary theory of senescence. Evolution, 57, 1478–88. Williams, P. D., Day, T., Fletcher, Q. & Rowe, L. (2006). The shaping of senescence in the wild. Trends in Ecology and Evolution, 21, 458–63. Yoshida, K., Fujisawa, T., Hwang, J. S., et al. (2006). Degeneration after sexual differentiation in Hydra and its relevance to the evolution of aging. Gene, 385, 64–70.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:37:38, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.012

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:37:38, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.012

13 Physiological and Biochemical Processes Related to Ageing and Senescence in Plants Maurizio Mencuccini and Sergi Munné-Bosch

Short Summary Much progress has been made in understanding the physiology and biochemistry of organ-level senescence. By comparison, ageing and senescence at the organismal level remain poorly known, especially so for perennial plants, for which model systems have been developed only recently. It has become clear that in perennials, ageing is accompanied by meristem-extrinsic size-related effects, even though separating these effects from meristem endogenous senescence is often challenging. Several experimental approaches have been identified that allow exploring whether the physiology and biochemical processes in perennials change systematically throughout the ontogenetic cycle and whether these changes are primarily age driven or size driven. Further progress can come from a more basic understanding of the metabolic, hormonal and molecular controls of these transitions in perennial plants and by framing our physiological and biochemical understanding within a clear evolutionary framework of the evolution, or lack thereof, of senescence.

Introduction The study of the changes over time in demographic processes (recruitment, growth and mortality) is central to the analysis of the occurrence of ageing and senescence in plants and animals (Hamilton 1966; Jones et al. 2014; Medawar 1952; Williams 1957). These changes are caused by endogenous modifications in the balance of physiological and biochemical processes. Significant progress has taken place in the study of the physiology of senescence in the species Homo sapiens and other mammals, whereas our understanding of the changes occurring over time in plants is much more limited (e.g. Watson & Lu 2004). In this chapter we review the progress in understanding the physiological and biochemical processes that may form the basis of ageing and senescence in perennial plants. Because perennial plants grow modularly (Harper 1977), it is necessary to distinguish between processes driven primarily by size from those driven primarily by age. Our examples include a range of plant groups including perennial herbs, shrubs and trees. Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:39:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.013

258

Physiological and Biochemical Processes Related to Ageing and Senescence in Plants

Organ Senescence, Organismal Senescence and Size-Related Ageing ‘Perenniality’ (i.e. the capacity of an organism to survive for more than two years) is in many species inextricably linked to organ modularity. ‘Modularity’ describes the occurrence of units (metamers) of growth and/or reproduction centred on one or more meristems and a vascular transport system that replicate semi-autonomous units integrated within a single organism. Modularity necessitates a defined life span for the component metamers made up of shoot and root apices, xylem, phloem, leaves and meristems of the module. The turnover of these organs over time may be physiologically necessary to avoid the inevitable and progressive wear and tear. Enzymes and organelles have turnover times on the order of minutes to days, leaves typically have life spans on the order of months to several years, roots typically have longevities varying from a few months to several years and xylem conduit and parenchyma cells have life spans from several months to a few decades (e.g. Chabot & Hicks 1982; Peek 2007; Westoby et al. 2000). The turnover of these organs is normally a highly regulated process involving genetically controlled sequences of programmed cell death (PCD). Evolutionarily, PCD seems necessary and useful for the organism as a whole because it can lead to optimisation in the recovery of important and scarce nutrients from senescing leaves and roots, the deposition of chemicals that increase the resistance of wood to decay by insects, fungal and bacterial organisms in dead conduit cells and the withdrawal of starch and other compounds from the parenchyma ray cells of senescing secondary wood (e.g. Ameisen 2002). On the contrary, whole-organism ageing in perennial plants goes beyond the PCD of individual organs and involves a wide range of processes, some including endogenous senescence and some related to the ontogenetic development of plants into large and complex networks. Because modularity results in structures with a necessarily limited life span, there is no single universal scaling predicted to occur between the age and size of perennial plants. Accurately predicting one from the other is normally difficult as a result of the confounding and overlapping effects of several variables (e.g. Tuljapurkar 1990). Nonetheless, older plants are generally larger than younger ones in perennial organisms (but see Salguero-Gómez & Casper 2010). Age-for-size substitutions are typically employed in ecology and other applied disciplines (e.g. forestry, fishery, agriculture) to judge when certain important physiological events occur during the organismal life cycle and gauge their significance for management (e.g. Goodyear 1997; West 2009). Several tools have been developed to allow an understanding of the links between size and age in plant demography (e.g. Caswell 2001; Caswell & Salguero-Gómez 2013; Cochran & Ellner 1992; Metcalf et al. 2013; West 2009; Zuidema et al. 2010). It is now generally recognised that size can exert profound effects on organismal physiology. Some of these effects, such as the reductions in physiological performance of several key processes or the accumulations of deleterious effects on the physiology of metamers, can mimic age-related senescence patterns (e.g. Greenwood 1995, Day et al. 2001). In other cases, however, size-related processes may have substantial positive effects on organismal physiology. Size itself is difficult to define in the context of whole-

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:39:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.013

Organ Senescence, Organismal Senescence and Size-Related Ageing

259

plant physiology. A fundamental set of processes consists of those related to the carbon balance of the organism. Using these criteria, a fundamental distinction should be drawn between the amount of autotrophic and heterotrophic tissues (Chapin et al. 2012; Mooney 1972). A germinating seed is entirely made up of heterotrophic tissues, and the small developing cotyledons are primarily sinks of carbon and nutrients arriving from the maternal embryo reserves, with the checkpoint of transition to the autotrophic phase under strong genetic control (Rajjou et al. 2012). Defining ‘autotrophism’ as the balance between whole-plant photosynthesis and respiration (e.g. Lambers & RibasCarbó 2005), a seedling will rapidly reach the autotrophic stage and will normally remain autotrophic for the rest of its life cycle, at least at yearly time scales. Some perennial plants may turn heterotrophic again during short periods (e.g. during periods of seed and fruit growth) and during the seasonal pauses of growth (e.g. under climates when frost or drought occur yearly), when dormancy develops. During dormant periods, metabolic rates (dark and light respiration primarily and photosynthesis in evergreen plants) are normally down-regulated, and respiration is primarily fed from stored nonstructural carbohydrates (Dietze et al. 2014). If one expresses whole-plant photosynthesis and respiration of a plant per unit leaf area (to standardise the fluxes to a common size-independent measure of autotrophic production), then the carbon balance – that is, photosynthesis minus respiration – will go from quasi-infinite large negative number for the germinating seed stage (when leaf area is zero), to a large, positive number when the ratio of autotrophic to heterotrophic tissues reaches a maximum in young, actively growing plants, to a much lower but still positive number in older, less fast-growing plants. How much lower this number will become does indeed depend a great deal on whether large organisms suffer the negative consequences of their large size. Similarly to the carbon balance, the nutrient balance of a plant may be the result of the balance between the uptake of nitrogen (N), phosphorus (P) and potassium (K) in photosynthesising leaves versus their accumulation in other reserve tissues (roots, parenchyma), seeds and fruits or in support tissues and their losses from senescing tissues (e.g. Fageria 2009; Kozlowski & Pallardy 2010). Conceivably, nutrient stress in very large plants could result from the depletion of litter and soil nutrient stocks as large capitals are stocked and immobilised in growing biomass compartments relative to the amounts needed to regenerate the leaf canopy. Standard metabolic scaling theory (e.g. Brown et al. 2004; West et al. 1999) predicts that leaf area should scale with the square of a plant diameter, whereas plant mass should scale with a power of 8/3 of its diameter. Therefore, almost inevitably, an accumulation of heterotrophic and nutrientaccumulating tissues accompanies development. Beyond the carbon and nutrient balance, size also has implications for other components of a plant, such as water relations. In trees, the progressive height increases inevitably lead to hydrostatic gravity effects on the menisci at the upper surface of the water columns inside leaves. These gravity effects can be quantified exactly as 0.01 MPa of additional tension (i.e. negative pressure) inside the vertical water columns of the apoplast (i.e. conduits and plant cell walls) caused by each meter of height increase (Hellkvist et al. 1974; Scholander 1964). In addition to this hydrostatic effect, a hydrodynamic effect of roughly equivalent magnitude occurs as a result of the effects of the Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:39:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.013

260

Physiological and Biochemical Processes Related to Ageing and Senescence in Plants

viscosity of water and of the adhesive forces between water molecules and conduit walls. This second effect occurs as water moves towards the sites of evaporation in the leaves and is caused by the resistance to water transport inside and outside the plant xylem (Mencuccini 2002). Therefore, relative to a seedling of height of approximately 1 cm, typically a plant of 1 m height will have a leaf water potential that is lower (i.e. more negative) by about 0.02 MPa as a consequence of its height, a plant of 10 m will have a leaf water potential that is lower by about 0.2 MPa and a plant of 100 m will have a leaf water potential that is lower by about 2 MPa. Even though compensating processes occur during development (e.g. Anfodillo et al. 2006; McDowell et al. 2002), normally resistance to water transport continues to increase with plant height (Mäkelä & Valentine 2006; Mencuccini 2002), implying either more negative water potentials and/or lower cellular metabolism. These reductions in the leaf water potential affect several important processes, the most basic of which is cellular turgor, which is fundamental for cell division in plants and thus for growth. In addition to direct height and path-length effects, large plants normally have more complex network structures, in which water and nutrients are transported over longer distances and via a larger number of hydraulic resistances. Over time, these transport pathways may deteriorate as a result of the turnover of transport and support tissues. The resulting sectoriality in roots, stems and branches (e.g. Schenk et al. 2008) reduces the lateral flows in xylem and phloem tissues with strong impacts on growth, water, nutrient and sugar transport and diffusion of metabolites, hormones, pathogens and parasites across the semi-autonomous modules of the organism. One could therefore regard increased sectoriality in large and old plants as a fundamental whole-plant process that may reduce organismal integration and cross-communication between the various organs of the same individual (Mencuccini 2003; Salguero-Gómez & Casper 2011; Schenk et al. 2008). Sectoriality may also hold clear advantages. In some species, the breakup of the initial organism into smaller semi-autonomous modules reduces the amount of heterotrophic tissue that would be required to support a single large unit and allows the individual genet to grow larger while the individual ramets remain small (e.g. Larson 2001). Beyond these physiological changes, other size-related processes act to stabilise plant performance and increase organismal resilience to environmental stresses. Large plants can build up large reserves of water in their roots, stems, bark and leaves which help buffer the fluctuations in water availability that can occur during droughts (e.g. Scholz et al. 2011). Similarly, large reserves of nutrients and carbohydrates help to avoid long periods of stomatal closure and rebuild leaf area after intense defoliation events in large trees (Sala & Mencuccini 2014; Sala et al. 2011). Large trees often show phenotypes that are akin to those found in stressful environments, for example, increased water transport capacity in roots and stems relative to the leaf area in their crowns (McDowell et al. 2002), higher mass/area ratios in leaves, lower photosynthetic rates and higher recovery of nutrients from senescing leaves (Ishii 2011; King 2011; Steppe et al. 2011). All these traits are suggestive of organisms exhibiting a near-homeostatic behaviour despite the stresses associated with ageing. Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:39:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.013

Physiological Changes with Age and Size in Leaves

261

Physiological Changes with Age and Size in Leaves The most dramatic evidence of the impact of organismal ageing on leaf anatomy, morphology and chemistry is given by the phenomenon of heteroblasty, which causes foliar phenotypic modifications to occur very rapidly (i.e. as a phase change) during the first few years of a plant’s life (e.g. Zotz et al. 2011). Genetically controlled phase changes occurring during maturation inside the apical meristems are responsible for leaf primordia with widely different morphologies. The leaves of eucalypts provide a clearcut example of this phenomenon. In this group of species, the leaf area can vary by up to 100 per cent from five- to six-year-old to 250-year-old individuals (England & Attiwill 2006), with other examples coming from a wide range of species (e.g. acacias, European ivy, aroid vines and gorse) (cf. Zotz et al. 2011). Apart from the obvious phase change typical of heteroblasty, similar transitions have also been described in relation to gradual ontogenetic changes in leaf properties associated with changes in plant size (‘ontogenetic drift’) (Evans 1972), as well as with the phase change directly associated with the switch from vegetative to sexually mature organisms. During the gradual transition of the ontogenetic drift, several other properties will vary concurrently with the area of a leaf, for example, the ratio of leaf area to mass, the proportion of mesophyll tissue relative to total volume and the proportion of internal air spaces, leaf thickness, toughness and density, cuticle thickness and composition, all of which are likely to affect whole-leaf properties such as photosynthetic rates, stomatal conductance and respiration rates (e.g. Ishii 2011; Steppe et al. 2011). Similarly, concentrations of various nutrients, such as N and P, are known to change during ageing, similarly to changes in foliar lignin content. Changes have also been documented in the developmental trajectories of plant defences against herbivore damages (e.g. Boege et al. 2011). In Ryparosa kurrangii, a lowland sub-canopy rainforest tree of the Australian wet tropics, levels of cyanogenic glycosides decrease from the cotyledon to the autotrophic seedling stage and further decrease in young and then mature plants (Webber and Woodrow 2009). Ontogenetic changes have also been documented for trembling aspen, in which concentrations of condensed tannins have been shown to increase, while phenolic compounds decrease from rates at age zero to those at age twenty-five years (Donaldson et al. 2006). Consistent increases in levels of plant chemical defences with age also were found in several studies comparing tree seedlings and mature plants, especially for phenolics and terpenoids (e.g. Bryant & Julkunen-Tiitto 1995; Close et al. 2001; Fritz et al. 2001). While this is consistent with the theory that allocation to defence and reproduction increases relative to allocation to growth in mature plants, it contrasts with the idea that selective pressure for herbivory damage should be highest at the seedling stage, when mortality also tends to be highest (Karban & Myers 1989). In addition to changes caused by endogenous processes of senescence and those linked to processes of size-related ageing, plants also show changes related to the structural modifications occurring during the growth and stratification of their canopies. This highlights the need to distinguish ontogenetic and size-related effects clearly. Compared to younger and more open canopies, closed canopies will result in changes

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:39:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.013

262

Physiological and Biochemical Processes Related to Ageing and Senescence in Plants

in the proportion of sun to shade leaves, with well-known physiological consequences such as changes in leaf mass per area, per cent N and amount of photosynthetic machinery (e.g. Poorter & Rozendaal 2008). Separating these ecosystem-level changes from intrinsic individual-level properties presents specific challenges, especially in the case of deep forest canopies. Similarly, many perennial species, especially but not only trees, undergo a transition from phenotypes characterised by traits typical of shady understory conditions to phenotypes characterised by traits, both above and below ground, typical of high-irradiance conditions. This example is important because it showcases a scenario in which changes primarily driven by external factors linked to canopy height are to a large degree genetically controlled via the evolution of traits in species adapted to undergo the transition from under-storey regeneration to canopy emergence in older organisms (e.g. Poorter & Rozendaal 2008; Westoby et al. 2002). In turn, inter-specific differences in these traits and how they change with size relate to demographic properties such as growth and mortality rates (e.g. Poorter et al. 2008).

Physiological Changes with Age and Size in Stems and Roots Anatomical, morphological and physiological differences among plants at different stages of ontogenetic development are relatively well known, especially for trees, given the commercial importance of timber production. Several trends have been described across many model species and have been summarised before in other reviews (e.g. Gartner & Meinzer 2005). Typical patterns occur axially within a growth ring (xylem conduits taper from very wide in roots and then stem base to very narrow in branch tips and foliage), across growth rings radially (with patterns varying from species to species but normally involving maturational changes taking place from an internal juvenile wood to an exterior mature wood) and vertically along the tips of branches at different tree heights and canopy light exposures as a result of the combined changes in light regime and the gravitational hydrostatic effects on turgor that were mentioned earlier. Similar tapering patterns are also seen in phloem conduits, partly as a consequence of the fact that both xylem and phloem cells derive from the same cambial initials (e.g. Hölttä et al. 2013; Jyske & Hölttä 2014; Mencuccini et al. 2011). The most dramatic senescence process seen in the secondary wood of plants is the transition from the occurrence of living parenchyma ray cells and functionally active xylem conduits (vessels or tracheids) in the sapwood to the complete absence of live cells and the complete embolisation of xylem conduits in the central heartwood. In the living woody plant, the sapwood is responsible for water transport and nutrient and carbohydrate storage (Tyree & Zimmerman 2002), whereas the heartwood shows a much reduced functionality and is partitioned off from the functional xylem by chemical and structural changes occurring during its formation. Heartwood forms as a result of changes mediated by the ray parenchyma in the transition zone (Hillis 1987). During heartwood formation, the bordered pits of conifer conduits aspirate (Kitin et al. 2009; Petty 1972), starch is consumed (Pandalai et al. 1985), extractives are synthesised (Hillis 1968) and sequestered to the vacuole (Higuchi 1997) and the nucleus degenerates Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:39:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.013

Physiological Changes with Age and Size in Stems and Roots

263

(Nakaba et al. 2006, 2008; Yang 1993). The final stage of heartwood formation occurs when the vacuole disintegrates (Higuchi 1997) and the cell ruptures, depositing extractive chemicals in the adjacent conduits (Nobuchi & Hasegawa 1994). This also renders the aspiration of the bordered pits irreversible as they become encrusted with extractives (Kitin et al. 2009; Petty 1972). It is the deposition of extractives in the xylem that conveys the colour change and also the increase in durability associated with heartwood formation, as many of the chemicals are toxic to bacteria, fungi and insects (Hillis 1968), a post-mortem event that has clearly adaptive functions. The aspiration and encrustation of the bordered pits reduce permeability and the spread of micro-organisms. These changes protect the living tree from decay, slowing down the processes of wear and tear. Whilst the functional sapwood is afforded protection by its high moisture content, which many micro-organisms are unable to penetrate, the lower moisture content of the heartwood would leave it susceptible to attack were it not for these changes. Despite the loss of hydraulic capacity, the heartwood continues to maintain an important physiological role. The mechanical support afforded by the central heartwood is, however, substantially reduced compared to the outer rings, a result of the mechanical load being supported primarily at the outer surfaces of cylindrical structures (Niklas 1992). Although many old trees are hollow in the centre, they remain capable of self-support. Finally, the chemical changes associated with the formation of the heartwood are almost non-existent in some species (Hillis 1987), suggesting that their occurrence is probably under genetic control, reflecting an adaptation to the tree’s growing conditions. Two different sets of theories have been proposed to explain the formation of heartwood in woody plants. The first set focuses on the biophysics of water transport and stresses especially the importance of physiological plant size effects, whereas the second focuses on the importance of direct endogenous controls by age and development and therefore on metabolic, biochemical and genetic controlling processes. The most widely accepted theory of heartwood formation is the pipe model theory (Shinozaki 1964a, 1964b), whereby an area of sapwood is needed to support the transpiration of the canopy leaf area, while redundant sapwood is converted to heartwood. The central elements of the pipe model theory are retained within the more advanced metabolic scaling theory of ecology, which was alluded to earlier (e.g. Brown et al. 2004; West et al. 1999). Redundancy here needs to be interpreted in terms of the respiratory carbon costs necessary for the maintenance of expensive support and conductive structures. Since the proposition of the pipe model theory, canopy area has been shown to correlate with sapwood basal area in many species, including Scots pine (Mencuccini & Grace 1995), oak (Szymanski et al. 2008), Douglas fir (McDowell et al. 2002), ponderosa pine (Ryan et al. 2000), beech (Bartelink 1997) and balsam fir (Coyea & Margolis 1992). However, it is not clear how the area of sapwood is maintained in relation to leaf area index or how heartwood formation is regulated. The role that water transport plays in heartwood formation and how this relates to the maintenance of functional sapwood area are also not understood. Whether the cessation of water transport is the result of or cause of heartwood formation has not been ascertained. Some authors, including Spicer (2005), have argued that heartwood formation is an actively regulated process, also a form of Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:39:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.013

264

Physiological and Biochemical Processes Related to Ageing and Senescence in Plants

PCD, and not a passive process. The number of rings that are typically maintained in the sapwood by a growing woody plant – which can vary from one to several tens – is more closely related to the cambial age of the plant than to the canopy size, supporting a direct developmental regulation of heartwood formation (Gjerdrum 2003). The strong relationship between heartwood ring number and cambial age (Yang and Hazenberg 1991a; 1991b; Yang et al. 1985) also leads to the theory that heartwood formation proceeds at a fixed rate in terms of rings per year and that changes in ring width could influence the amount of heartwood (Wilkes 1991).

Experimental Approaches to Separate Age from Size Effects in Trees Over the last fifteen years, several groups have documented the physiological differences that are apparent under field conditions as trees grow in size and complexity (see Mencuccini & Grace 1996; Ryan et al. 2000, 2006; Yoder et al. 1994). An example of these striking variations is given in Figure 13.1 (Korakaki & Mencuccini, unpublished data), which reports the results of a study on the sap flux density of a size sequence of Scots pine spanning several decades. A plot of sap flux density relativised per unit of leaf area against mean daytime vapour pressure deficit (the main environmental driver of sap flux density in coniferous trees) shows a strong size- and/or age-related decline, with remarkably lower values in the older trees compared to the younger ones. The hyperbolic decline in the slope of this response function plotted as a function of tree age in the inset is typical of responses that have a strong size-related scaling component.

5

18 years old R2 = 0.81

4

4

Max Q

Sap flow Q per unit of leaf area (kg m−2 day−1)

5

3 2 1

3

0 0

20 40 60 80 100 Tree age (years)

25 years old R2 = 0.77

2

50 years old R2 = 0.88

1 80 years old R2 = 0.82

0 0.0 Figure 13.1

0.0

0.4 Daytime D, kPa

0.6

0.8

Changes in sap flow Q per unit of leaf area in the stem xylem of Scots pine trees in four stands of varying ages and sizes as a function of mean daytime air vapour pressure deficit D, an index of the evaporative power of the atmosphere. Each set of points refers to measurements conducted on the trees of one age class only. The ages of each stand and the variance explained by the regression are given in the rectangle alongside each line. The inset shows a plot of the slope of these regressions against stand age.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:39:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.013

Experimental Approaches to Separate Age from Size Effects in Trees

265

Several experimental techniques have been employed to help isolate size-related from age-related effects on growth, leaf-level gas exchange and the xylem and phloem properties of branches, stems and roots. These techniques are frequently borrowed from standard horticultural practices, such as grafting or air layering (e.g. Goldschmidt 2014) and allow establishing twigs of different origins under commongarden conditions so that their performance can be compared. As for all experimental manipulations of this kind, every technique has advantages and disadvantages, and it is only by combining several of them that a complete picture can be obtained. Grafting has been employed by several groups (Bond et al. 2007; Matsuzaki et al. 2005; Mencuccini et al. 2005) to isolate the direct effects of age from those of size and the environment. The principle behind these manipulations is that twigs (the ‘scion’) cut from donor plants of varying ages are grafted onto a common rootstock, in many cases a juvenile one. This approach aims to retain the age ‘signal’ – if any – that should be present inside the meristems of the original donor trees, whereas the environmental and the size-related signals present outside the meristems in the field should disappear when juvenile rootstock is employed. Manipulative experiments using grafting have to be carefully planned and analysed to avoid potential interpretational challenges. Two issues must be considered: (1) the potential constraints on xylem water movement created by the graft union itself and (2) the physiological and morphological effects caused by the fact that the grafted seedlings are composed of two different genotypes: the scion plus the rootstock. These aspects were examined by Vanderklein et al. (2007) and Mencuccini et al. (2007). The hydraulic resistance of the graft union was measured and compared to the resistances of intact shoots and roots of Scots pine, sycamore and ash as a function of time since grafting in plants of different meristematic ages. These three species were chosen because they represent a gradient, ranging from a gymnosperm (pine) in which several rings of functional sapwood exist and for which substitution of the damaged grafted region is very gradual to a ring-porous angiosperm (ash) at the other extreme, where only one functional water-conducting ring exists that is replaced yearly, thereby accelerating the recovery from grafting shock. In general, no significant differences in the hydraulic resistance of the grafted unions were found among different age classes for any of the species investigated, suggesting that the recovery process did not directly depend on ageing (Table 13.1). However, the data for Scots pine (Vanderklein et al. 2007) and the data for sycamore (cf. Table 13.1) showed that the resistance determined by the grafting unions was still considerable seven and two years after grafting, respectively. In the case of the ring-porous ash, only a very small proportion of resistance was found in the grafting union two years after grafting, suggesting, as anticipated, a very rapid recovery for this species (Table 13.1). Therefore, it is to be expected that even in the case of successful grafting, time is required for the shock to be fully absorbed by the plant. The second potential issue with regards to grafting, the involvement of two genotypes, also needs to be carefully considered. One can approach this problem by combining two different approaches. A first approach consists in including appropriate controls in the experimental design. Ungrafted young seedlings of the same size as the grafted material Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:39:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.013

266

Physiological and Biochemical Processes Related to Ageing and Senescence in Plants

Table 13.1 Means and Standard Errors of Hydraulic Resistance for Various Aboveground Compartments of Grafted Seedlings of Sycamore and Ash Age class

N

Rstem

Rscion

Rgraft (% of Rstem)

Sycamore 1 2 3

4 4 4

5.89 ± 1.14a 6.58 ± 0.68a 6.62 ± 1.51a

4.72 ± 0.95a 3.74 ± 0.44a 5.42 ± 1.34a

1.17 ± 0.23a (20.2a) 2.84 ± 0.45a (42.8a) 1.20 ± 0.17a (19.2a)

4

4

8.18 ± 2.37a

4.33 ± 0.45a

3.86 ± 2.55a (36.0a)

5 5 5 5

5.56 ± 0.18a 5.82 ± 0.53a 5.35 ± 0.75a 6.39 ± 1.29a

5.36 ± 0.17a 5.51 ± 0.48a 5.17 ± 0.72a 5.90 ± 1.14a

0.20 ± 0.04a (3.7a) 0.31 ± 0.12a (5.1a) 0.19 ± 0.03a (3.6a) 0.49 ± 0.20a (7.0a)

Ash 1 2 3 4

Note: Rstem = stem hydraulic resistance for the whole tree (excluding leaves and root system), Rscion = scion hydraulic resistance (excluding leaves) and Rgraft = graft section hydraulic resistance. a Letters within the same column refer to post-hoc ANOVA tests for the significance of the difference among age classes for each species. All resistances are in units of 103 MPa s kg−1 (Hamid & Mencuccini, unpublished data).

can be combined with self-grafted material, in which the grafting shock occurs in plants of a single, not of two different, genotypes. This allows isolating the impacts of the act of grafting on the selected genotype (e.g. Hamid & Mencuccini 2009). A second approach to separate shoot-related from root-related ageing processes uses a technique that avoids the potential artefacts of grafting altogether. Air layering is employed to create conditions favouring direct rooting of a branch in situ. This technique was exploited by Hamid and Mencuccini (unpublished data), who employed air-layered branches from sycamore trees of two age classes: 27 and 144 years. The results generally showed no differences in physiological performance among branches coming from the two age classes of propagated material (Table 13.2). Coupled with the observations that field plants of the same two age classes showed significant physiological differences in all observed parameters, these results suggest that (1) differences in growth and physiological characteristics between 27- and 144-year-old trees were not due to differences in age per se of donor trees, and (2) roots of sycamore do not show any evidence of age-related decline in their physiological performance. Simultaneous measurements of grafted sycamore scions obtained from the same field trees also failed to show evidence of agerelated declines in physiological properties (Hamid & Mencuccini 2009). The fact that two independent propagation techniques, both aimed at separating size-related from age-related effects but characterised by very different advantages and disadvantages, failed to show differences across different age classes reinforces the conclusion that the systematic physiological differences observed in the field were primarily size related and not age related. Interestingly, in the air-layering experiment, a strong size effect was detected in the rooted branches, as both hydraulic and gas-exchange variables were significantly and Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:39:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.013

Experimental Approaches to Separate Age from Size Effects in Trees

267

Table 13.2 Means and Standard Errors of Leaf-Level Gas Exchange, Transpiration Rate and Leaf-Specific Hydraulic Conductance on the Two Age Classes of Sycamore Air-Layered Plants Age (years) Parameter −2 −1

Anet (μmol m s ) Ci (μmol mol−1) E (mmol m−2 s−1) GS (mmol m−2 s−1) EL (mmol m−2 s−1) Ψpredawn (–MPa) Ψmidday (–MPa) KL (mmol m−2 s−1 MPa−1)

~27

~140

F-ratio

10.82 ± 0.44 174.43 ± 7.69 2.20 ± 0.12 133.70 ± 11.61 1.74 ± 0.22 0.482 ± 0.04 0.755 ± 0.06 6.91 ± 1.09

10.44 ± 0.48 164.36 ± 8.19 2.06 ± 0.13 127.78 ± 12.49 1.66 ± 0.09 0.476 ± 0.04 0.915 ± 0.09 4.70 ± 0.97

0.33ns 0.80ns 0.59ns 0.12ns 0.12ns 0.01ns 2.16ns 2.29 ns

Note: ns = not significantly different. Source: Hamid & Mencuccini, unpublished data.

negatively related to both branch diameter and height (i.e. length) (Figure 13.2). Such a strong impact of height and diameters on gas-exchange properties may seem unlikely in small plants. However, ancillary anatomical and hydraulic measurements confirmed that branch length does play a strong role in determining hydraulic conductance. We measured vessel diameters at various distances from the branch tip down to the base, and we plotted them as a function of distance from the apex. Vessel diameters scaled as power functions of the distance from the leader, with a steep curve in the first 2 to 3 m from the apex, suggesting that small increases in branch length may indeed affect plant physiology. Parallel hydraulic measurements led to the same conclusion (Petit et al. 2008). The techniques of grafting and air layering allow the creation of an ‘artificial’ organismal mosaic where its components may differ in meristematic age aboveground but, if carefully chosen, not in their size. When the physiology of these plants is compared against the original donor plants in the field, these techniques allow isolating, by comparison, the effects of age. Another equally powerful technique consists of exploiting material coming from clonal size (not chrono-) sequences (e.g. Mencuccini et al. 2005). Poplar is frequently propagated for commercial reasons, thanks to the facility by which cuttings produce adventitious roots when inserted in a wet soil medium. Using rooted cuttings from clonal material allows exploring whether there is direct genetic control on physiological properties that is reflective of size per se, not age per se. In one experiment (Korakaki & Mencuccini, unpublished material), we compared ortets (donor trees from which twigs or scions were propagated by direct rooting) of identical meristematic age and different sizes in the field against rooted cuttings coming from the same ortets but now of constant age and size (cf. Figure 13.3 for further explanation on this technique). Trees coming from seven to eight different tree ‘ages’ of original ortets (depending on clone availability) were propagated, with each size sequence belonging to only one specific clone. Measurements were made on both the propagated material and the original donor trees in Belgium to compare the behaviour of the same-sized rooted cuttings with the behaviour of the original donor trees. Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:39:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.013

268

Physiological and Biochemical Processes Related to Ageing and Senescence in Plants

KL (mmol m2 s−1 MPa−1)

EL (mmol m2 s−1)

3.0

(a)

2.5

R2 = 0.34**

2.0 1.5 1.0 0.5 14 12

(c)

(d) R2 = 0.76***

R2 = 0.59***

10 8 6 4 2 0 4

Figure 13.2

(b) R2 = 0.43**

6

8

10 12 14 16 18 20 22 20 Diameter (mm)

40

60

80 100 120 140 160 Height (cm)

Relationships between transpiration rate and total diameter (A) or total height (B) and between leaf-specific hydraulic conductance and total diameter (C) or total height (D) across two age classes. ***Correlations significant at P < 0.001; **correlations significant at P < 0.01.

This approach isolates size-related factors, as planted cuttings in the field were all of the same age, since the time of the last meiosis event, but of different sizes. By examining the behaviour of the material re-sampled from the field and grown under common-garden conditions, we obtained propagated plants that were all of a similar small size, identical meristematic age and grown under identical conditions. In the field experiments, all the physiological variables varied significantly with plant size across at least three or even four clones, with the notable exception of the ratio of leaf area to mass (Table 13.3). Comparison of net assimilation rates (NAR, the ratio of aboveground biomass growth to the plant leaf area) across these pure-size sequences showed consistent peaks at intermediate sizes classes followed by significant declines in the larger size classes in three clones and a continuous decline in the fourth clone. These trends in NAR in the field disappeared in three of the four clones when the equal-age, equal-size cuttings were compared in the common garden (Figure 13.3), suggesting that strong pure-size effects reduced the growth potential of large trees in the field. Similarly, monitoring of physiological variables showed strong and consistent trends in the field that entirely disappeared or were even reversed in the constant-age, constant-size common-garden experiments (Figure 13.4). Size changes, therefore, however defined, appear to have powerful effects on tree growth and physiology.

Connecting Hormones with Plant Size and Reduced Vigour Although the role of hormones in many aspects of plant development, such as cell division and expansion, germination, flowering and leaf senescence, has been studied in Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:39:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.013

Connecting Hormones with Plant Size and Reduced Vigour

Cross (1971) Figure 13.3

2003

2002

1998

1994

269

1975

Schematic example of the experimental design employed to isolate the effects of size per se using size sequences of poplar clonal material. Material from four poplar clones was obtained from the Belgian Poplar Research Institute. These crosses were (1) P. trichocarpa × P. deltoides, a controlled cross of Populus balsamifera L. ssp. trichocarpa (Torr. & Gray ex Hook.) Brayshaw and Populus deltoides Bartr. ex Marsh; (2) (P. trichocarpa × P. deltoides) × P. deltoides, a controlled cross of (P. trichocarpa × P. deltoides) S.910–2 (= cv Beaupré) × P. deltoides (S. 333–44, Michigan × S.336–16, Connecticut/31) S.910 = P. trichocarpa V.235 Washington (‘Fritzi Pauley’) × P. deltoides S.1–173, S.1 = V.5, Iowa × V.9, Missouri; (3) P. deltoides × (P. trichocarpa × P. deltoides), a controlled cross of P. deltoides (S.333–44, Michigan × S.336–16, Connecticut) × S.910–1 (= cv ‘Unal’) S.910 =P. trichocarpa V.235 Washington (‘Fritzi Pauley’) × P. deltoides S. 1–173 S.1 = V.5, Iowa × V.9, Missouri; and (4) P. trichocarpa × P. trichocarpa, a controlled cross of P. trichocarpa V.235 Washington (‘Fritzi Pauley’) × P. trichocarpa V.24 Oregon (‘Columbia River’). For each of these four cases, we tracked the original cross planted around 1971. We then looked for plantings made from cuttings coming from the original cross but taken at different times since the original cross (see years under each drawing). For all four size sequences, we sampled these plantings in the field and determined how their physiological characters changed as a function of size only (for all plantings, meristematic age is the age from 1971). In addition, we repropagated all the plantings by taking scions from each of the various size classes. This allowed us (1) to avoid genetic differences across compared individuals and sizes within each size sequence and (2) to compare plants of the same meristematic age but different sizes (Korakaki & Mencuccini, unpublished data).

detail, much less is known about the implication of hormones in the ageing process. Little experimental evidence has been obtained thus far investigating the levels of hormones in plants of different ages, but information available thus far in some species, particularly in conifers, suggests a role for the hormones in the regulation of the ageing process in perennial plants. In this section we will focus on summarising current evidence indicating that hormones play a role in the regulation of ageing-related

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:39:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.013

270

Physiological and Biochemical Processes Related to Ageing and Senescence in Plants

Table 13.3 Summary of the ANOVA Results on a Set of Physiological Variables Measured in the Field Across Size Classes of Poplar Donor Ortets of Four Different Crosses. PH, PA, PG and PT are the four crosses employed (See legend of Figure 13.3 for information on the individual crosses) F value/parameters

PH

PA

PG

PT

Net assimilation rate Anet Intercellular CO2 Ci Stomatal conductance gs Specific leaf area (SLA) Water potential difference ΔΨ Leaf specific hydraulic conductance KL Carbon isotope discrimination δ13C Foliar nitrogen concentration Nm Foliar phosphorous concentration Pm

11.89ns 3.9*** 28.6*** 36.9ns 285.4*** 84.9*** 7.1ns 12.4ns 3.2ns

76.16*** 33.67** 28.77** 17.85ns 90.84ns 14.11*** 47.52*** 29.43*** 8.27***

24.4*** 22.5*** 27.7*** 2.1ns 128.1ns 33.6* 28.9*** 4.9*** 19.5***

7.4*** 26.2*** 10.9* 5.8ns 6.3*** 11.6*** 31.3** 11.4ns 4.7ns

Notes: Levels of significance refer to the overall test of significance of the differences across age classes for each ANOVA. *** Significantly different at P < 0.001; ** P < 0.01; * P < 0.05; ns = not significant. 1.6

PH R2 = 0.31*

NAR (kg m2 year−1)

1.2

1.6

PA R2 = 0.51***

1.2

0.8

0.8

0.4

0.4

0.0 0.6

0.0 PT

PG 1.2

0.5

R2 = 0.27**

0.4

R2 = 0.94***

0.3 0.2

R2 = 0.79***

0.8 0.4

0.1 0.0

0.0 0

Figure 13.4

10

20

30

40 50 0 5 Tree ‘age’ (years)

10

15

20

25

Changes in net assimilation rates (ratio between aboveground biomass production and leaf area) of trees of four poplar clones of identical meristematic ages. Each panel shows the result from a different clone (indicated by the two capital letters at the top right of each panel). Black square points refer to measurements conducted on the field donor trees (in which size differs but meristematic age does not), whereas the white points refer to the measurements conducted on the propagated rooted cuttings in a common garden (in which neither age nor size differs). For PH, measurements on the cuttings were conducted over two years (white triangles and circles). A peak and a subsequent decline in NAR with tree size were evident in all four chrono-sequences, whereas in three of the four cases the trend disappeared entirely one year after rooting for the rooted cuttings experiment.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:39:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.013

Connecting Hormones with Plant Size and Reduced Vigour

Anet (μmol g−1 s−1)

0.4

PH

271

PA

0.5

0.3

0.4

0.2

0.3 R2 = 0.56***

0.1 0.0 0.5

PT

0.2

2

R = 0.74*** PG

0.1 0.4

0.4 0.3 0.3 0.2

0.2

R2 = 0.43** R2 = 0.88***

0.1 0

Figure 13.5

10

20

30

40 50 0 5 Tree ‘age’ (years)

10

15

20

25

0.1 30

Changes in net photosynthetic rates in trees of four poplar clones of identical meristematic ages. Each panel shows the result from a different clone (indicated by the two capital letters at the top right of each panel). The black square points refer to the measurements conducted on the field donor trees (in which size differs but meristematic age does not), whereas the white points refer to the measurements conducted on the propagated rooted cuttings in a common garden (in which neither age nor size differs). For PH, measurements on the cuttings were conducted over two years (white triangles and circles). A significant decline in net photosynthesis with tree size was evident in the field for three of the four chrono-sequences, whereas the trend disappeared entirely or was reversed in all four clones one year after rooting for the common-garden experiment on rooted cuttings.

processes in plants, with a particular emphasis in the modulation of growth as plants increase in size and age. As discussed in previous sections of this chapter, it is well documented that growth rates decrease at the shoot level with ageing. Plant growth regulators (PGRs) are good candidates to explain the causes of this physiological phenomenon. PGRs include several classes of compounds present at low concentrations within the plant, including auxins, cytokinins (CKs), gibberellins (GAs), abscisic acid (ABA), ethylene, brassinosteroids (BRs), strigolactones, jasmonates (JA), salicylates (SA) and polyamines. Among them, auxins, CKs and GAs have been better documented at the biochemical, cellular and molecular levels to exert a direct role on plant growth, particularly promoting cell division and expansion (Davies 2010). While CKs promote cell division and the sink capacity of organs, auxins promote cell expansion and play a synergic role with CKs in the stimulation of cell division. GAs also promote growth but exert a role on intercalary meristems, therefore leading to increased plant size (Davies 2010). It has been shown that changes in auxin and CK levels occur during ageing in perennial plants. Age-induced reductions in leaf growth have been associated with reduced indole-3-acetic acid and CK levels, therefore leading to reduced leaf

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:39:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.013

272

Physiological and Biochemical Processes Related to Ageing and Senescence in Plants

growth. The ability of newly emerged leaves to produce auxins and CKs declines with the age of pine trees (Valdés et al. 2002, 2004, 2005). Such findings support the link between reduced growth and deceased growth-stimulating hormone levels during ageing in perennials. Another feature of plant ageing is a reduction in the photosynthetic rates, and CKs, in conjunction with light, induce the production of chlorophyll, chloroplast differentiation and the maintenance of the photosynthetic apparatus in leaves (Haberer and Kieber 2002; Zaffari et al. 1998). Therefore, the reduction in photosynthetic rates associated with plant ageing observed in conifers may also be under hormonal control, particularly governed by a decline in CKs levels. Abscisic acid (ABA) may also play a role in the regulation of the ageing process in perennials. ABA has been shown to be involved in regulating the transition from the vegetative to reproductive phase in several species (Finkelstein et al. 2002), higher levels being characteristic of the mature phase compared to the juvenile one in conifers (Haffner et al. 1991; Valdés et al. 2004). The occurrence of changes in ABA in relation to differences in the rate of vegetative growth is consistent with the increase of this growth inhibitor observed in mature trees, in which the growth rate is reduced. Although leaves accumulate ABA during ageing, particularly during the transition phase (i.e. the transition to a reproductive, mature stage), relatively low levels of this hormone have been observed in terminal buds of fully mature trees (Valdés et al. 2002). These authors suggested that this might imply that this hormone participates in the first stage of the acquisition of the flowering ability in the pine trees. However, once the plant reaches a fully mature state, ABA might not have the same biological significance. Studies carried out in birch similarly reported equivalent increases with the transition phase, with results also disclosing comparable levels between juvenile and fully mature trees (Galoch 1985). It appears, therefore, that after a significant increase of endogenous ABA concentrations during the transition phase in trees, the levels of this hormone decrease to levels similar to those observed in juvenile trees. Unfortunately, no data are presently available for mature trees at advanced developmental stages and grown under similar conditions to unravel whether or not ageing increases ABA levels in mature trees. In shrubs, it has been shown that endogenous ABA concentrations were higher in seven- than in two-year-old male rosemary plants (Munné-Bosch & Lalueza 2007). In this case, two-year-old plants have already reached the mature stage, thus suggesting an increase in ABA levels associated with ageing in mature shrubs. Interestingly, two- and seven-year-old plants differed in size, thus indicating that enhanced ABA levels may result from a need to reduce transpiration at the whole-plant level. Aside from auxins, CKs and ABA, other PGRs may also be involved in the modulation of the ageing process in perennials. An example is polyamines, which have generally been associated both with leaf senescence and with young, rapidly growing tissues (Tiburcio et al. 1993). Rey et al. (1994) investigated endogenous polyamine concentrations in recently emerged leaves from juvenile and mature hazel shoots, as well as leaves from shoots obtained by propagation of adult tissues, and they found that polyamines, mostly free putrescine, accumulate in juvenile tissues and decrease significantly in Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:39:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.013

Connecting Hormones with Plant Size and Reduced Vigour

273

recently emerged leaves as plants age. This finding is consistent with the role of polyamines in the stimulation of cell division (Paschalidis & Roubelakis-Angelakis 2005). Thus, a reduction in these compounds would decrease the capacity for growth as plants age. More recently, Mencuccini et al. (2014) found an increased percentage of total methylated loci in nuclear DNA with increasing meristematic age. This study was performed with trees of ages between 129 and 534 years in one of the oldest extant populations of Scots pine. Apical branches were propagated by grafting onto homogeneous juvenile rootstock to eliminate the effects of size and environmental variability and isolate those due to age. The hormonal profile of leaves and seeds along with markers of the physiological status of leaves and their pattern of DNA cytosine methylation were measured fifteen years after grafting. Very few significant relationships were found between levels of hormones, pigments or physiological markers either in leaves or in seeds and age of the meristem. These findings support the theory that many of the age-related patterns in growth, reproduction and biochemical properties reported in the literature are primarily size related and have little ontogenetic basis. It is therefore of high interest to study age-related changes in hormones in plants with a very small size because possible size-related effects are kept to a minimum. Oñate et al. (2012) investigated the role of hormones in the regulation of ageingrelated processes in Borderea pyrenaica, a small dioecious geophyte relict of the Tertiary with one of the longest life spans ever recorded for any non-clonal herb (>300 years). Aside from the eco-physiological interest of this species (García & Antor 1995), investigating age-related changes in hormones in this species is particularly interesting because this is an herb with a extremely low plant growth rate, so plant size effect may be negligible, if at all, in this species, at least compared to shrubs and trees. Biomass allocation, together with levels of auxins, CKs and ABA, and other indicators of leaf physiology (chlorophylls, lipid peroxidation and the maximum efficiency of photosystem II photochemistry – Fv/Fm ratio) was measured in juvenile and mature plants, including both males and females of three age classes (up to 50 years, 50 to 100 years, and more than 100 years). Plants maintained their capacity for vegetative growth and reproductive potential intact. CKs levels decreased with age, but only in females. Such sex-related differences, however, were not associated with symptoms of physiological deterioration in leaves but with an increased reproductive effort in females. This study further confirms that age-related changes in hormonal levels are primarily size related so that age- and size-related reductions in photosynthesis and relative growth rates may be under hormonal control, but age-related changes in the levels of such PGRs are mostly modulated by size effects. In a study performed using cuttings of the Mediterranean shrub Cistus clusii to separate size from meristemrelated effects, it was found that meristems are not responsible from ABA reductions with plant ageing. Compelling evidence therefore suggests that plant size is the intrinsic phenomenon associated with ageing responsible for ABA-related reductions in relative growth rates (Oñate & Munné-Bosch 2008).

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:39:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.013

274

Physiological and Biochemical Processes Related to Ageing and Senescence in Plants

Free Radicals, ROS and Ageing/Senescence It is generally believed that antioxidants (AOXs) and AOX-rich fortified foods can help us combat the damaging effects of our intrinsic age-related increased rate of formation of reactive oxygen species (ROS) such as superoxide anion (O2−), hydrogen peroxide (H2O2), hydroxyl radicals (OH−) and singlet oxygen (1O2). Harman (1956) proposed the free-radical theory of ageing, which postulates that the physiological deterioration with ageing in humans is the result of the accumulation of alterations induced by free radicals within the cell, thus explaining the increased sensitivity of humans to diseases associated with oxidative stress with age, such as cardiovascular diseases, cancer, chronic inflammation and neurological disorders. In mammals, age-associated disorders in the cell are therefore believed to be connected to the time-dependent shift in the prooxidant/antioxidant balance in favour of oxidative stress, either caused by an increase in the production of free radicals or a depletion of antioxidant defences (Ashok and Ali 1999; Rustin et al. 2000). Furthermore, it seems that age-associated changes in the antioxidant/pro-oxidant balance are small early in the life of animals but rapidly increase with age because of the exponential nature of the process (Harman 1981). Therefore, consuming AOXs is considered healthy, particularly at advances ages. Obviously, the consumption of natural AOXs should be moderated, since AOXs can also play a role in cellular signalling, so a delicate balance between pro-oxidants and AOXs is what is needed, not just consuming lots of AOX-rich pills. But what happens in plants? Can this theory be applied to explain ageing/senescence in plants? And if so, is it valid for all plant species or can only be used to explain the ageing phenomenon in specific taxonomic groups of plant life forms? The role of oxidative stress in plant senescence of annuals, such as the model plant Arabidopsis thaliana, has been extensively documented and can nowadays be explained in detail at the physiological, biochemical and molecular levels (Dangl et al. 2000; Juvany et al. 2013; Quirino et al. 2000). Beyond this model plant, the role of oxidative stress in other monocarpic plants, such as rice, barley and wheat, is also well understood, and it has generally been found that oxidative stress increases as senescence progresses simply because senescence of leaves at the latest stages of development is accompanied by senescence at the whole-plant level once seed production has been accomplished. But what happens in iteroparous, polycarpic plants, in which reproduction occurs more than once during the plant’s life cycle? In plants, chloroplasts are one of the organelles most exposed to oxygen toxicity because they function at high oxygen concentrations and in the presence of light. Therefore, chloroplasts produce high concentrations of ROS, particularly singlet oxygen, which can cause rapid oxidation of polyunsaturated fatty acids (PUFAs), proteins and DNA. Furthermore, peroxisomes are one of the most important intracellular generators of ROS, particularly H2O2, together with mitochondria, which produce important amounts of superoxide anions due to cellular respiration (Halliwell & Gutteridge 1989). Therefore, it has been hypothesised that chloroplasts could also play a role in plant ageinduced oxidative stress in iteroparous perennials. In this respect, Munné-Bosch and

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:39:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.013

Free Radicals, ROS and Ageing/Senescence

275

Alegre (2002) showed that an age-dependent and transient photoinhibitory damage to the photosynthetic apparatus occurred in the Mediterranean shrub male rosemary, concomitantly with an enhanced oxidative stress in chloroplasts as plants aged, as indicated by enhanced lipid peroxidation and reduced AOX levels, such as tocopherols (vitamin E) and carotenoids (pro-vitamin A). The authors argued that aside from the ageinduced stomatal limitation of photosynthesis, age-induced oxidative stress in chloroplasts may result in a deregulation of photosystem II photochemistry that may reduce the photosynthetic activity of leaves. The same research group confirmed the results later, showing higher reductions in the efficiency of photosystem II and increases in the extent of lipid peroxidation as male rosemary plants aged, which is indicative of increased inhibition of photosystem II and oxidation of lipids caused by reduced AOX defences in the oldest plants (Munné-Bosch & Lalueza 2007). Furthermore, oxidative stress was not only associated with chloroplasts but also with mitochondria, thus suggesting that the overall oxidative metabolism is affected by ageing. Possible age × environmental interaction effects should be carefully considered to better understand physiological processes associated with ageing in perennial plants. In the aforementioned study in male rosemary plants (Munné-Bosch & Lalueza 2007), it is interesting to note that age-induced differences in these oxidative stress markers were mostly noticeable during spring and summer and not during autumn. During spring and summer, plants were exposed to high light stress coincident with periods of drought under Mediterranean climate, which might lead to excess excitation energy in chloroplasts and therefore to the formation of ROS by increasing stomatal closure and consequently decreasing the activity of the Calvin cycle as a sink for ATP and reducing equivalents produced in photosynthetic electron transport. It is therefore likely that an age-related enhanced ABA concentration in leaves favour the photo-oxidative stress observed during periods of environmental stress in the oldest plants. Furthermore, accelerated leaf senescence was observed in this study as plants aged, thus supporting further the role of ABA and oxidative stress in the regulation of ageing in perennials, since both ABA and oxidative stress are known to promote leaf senescence in plants (Munné-Bosch et al. 2001; Yang et al. 2003). It is interesting to note, however, that oxidative stress is apparent during periods of environmental stress only and that, although it may shorten the life span of leaves as plants age, the plants keep the capacity to make new, young leaves without any symptom of oxidative stress (unless when exposed to environmental stress) along the years due to the activity of meristems. Oxidative stress and accelerated leaf senescence associated with plant ageing may therefore be considered adaptive strategies of ageing plants to the environment. To disentangle size- from age-related effects, studies of age-related changes in oxidative stress markers in plants with small size can also provide interesting results. Morales et al. (2013) tested age- and sex-related changes in several photo-oxidative stress markers in Borderea pyrenaica. In three field samplings performed during 2008, 2010 and 2011 in the Central Pyrenees (north-east Spain), the effects of ageing and sex on photosynthetic pigment levels, the Fv/Fm ratio, lipid peroxidation and the extent of photo- and antioxidant protection in chloroplasts were examined. It was found that both Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:39:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.013

276

Physiological and Biochemical Processes Related to Ageing and Senescence in Plants

male and female plants maintained chlorophyll levels intact, as well as the Fv/Fm ratio and the levels of lipid peroxidation irrespective of age. This finding suggests the absence of age-associated oxidative stress at the organismal level in this long-lived small geophyte. Furthermore, photo-protection mechanisms were found to be similarly efficient in the oldest individuals as in juvenile plants in terms of xanthophyll cycle deepoxidation and accumulation of low-molecular-weight antioxidants (carotenoids and tocopherols). It is very likely that, among other factors, the small size of this plant species contributes to its longevity (Morales & Munné-Bosch 2015).

Conclusion and Perspectives Our knowledge about the physiological and biochemical processes involved in ageing and senescence in perennials is rapidly increasing. Nonetheless, we still do not have a clear understanding of which processes and metabolic pathways are likely to be central to senescence in perennial plants (Watson & Lu 2004), a topic of clear relevance to ageing in other organisms, including vertebrates (e.g. Angelier et al. 2007; Austad 2001; Jones et al. 2008). Recent research has clearly highlighted the need to distinguish sizerelated ageing from endogenous senescence processes (Mencuccini et al. 2005). However, while this can now be achieved with model systems and experimental techniques developed for this purpose, it still remains unclear whether the age-related changes in the rates of demographic processes observed in the field – survival, growth, reproduction and recruitment – are primarily affected by size-related ageing, endogenous age-controlled senescence or a combination of the two. It is also unclear whether these effects are common to all perennial organisms, or whether differences exist across different groups of species. A closer interaction between the disciplines that are specialised in the identification and analysis of the detailed metabolic and physiological pathways of a plant and those that are focused on the analysis of demographic processes is likely to bring new insights into the comparative understanding of ageing and senescence.

Acknowledgements One anonymous referee and Roberto Salguero-Gómez reviewed this chapter thoroughly.

References Ameisen, J. C. (2002). On the origin, evolution, and nature of programmed cell death: a timeline of four billion years. Cell Death Differentiation, 9, 367–93. Anfodillo, T., Carraro, V., Carrer, M., et al. (2006) Convergent tapering of xylem conduits in different woody species. New Phytologist, 169, 279–90.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:39:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.013

References

277

Angelier, F., Weimerskirch, H., Dano, S. & Chastel, O. (2007). Age, experience and reproductive performance in a long-lived bird: a hormonal perspective. Behavioural Ecology and Sociobiology, 61, 611–21. Ashok, B. & Ali, R. (1999). The aging paradox: free radical theory of aging. Experimental Gerontology, 34, 293–303. Austad S. N. (2001). An experimental paradigm for the study of slowly aging organisms. Experimental Gerontology, 36, 599–605. Bartelink, H. H. (1997). Allometric relationships for biomass and leaf area of beech (Fagus sylvatica L). Annales des Sciences Forestieres, 54, 39–50. Boege, K., Barton, K. E. & Dirzo, R. (2011). Influence of tree ontogeny on plantherbivore interactions. In Size and Age-Related Changes in Tree Structure and Function, ed. F. C. Meinzer, B. Lachenbruch & T. E. Dawson (pp. 193–214) (Dordrecht: Springer). Bond, B. J., Czarnomski, N. M., Cooper, C., et al. (2007). Developmental decline in height growth in Douglas fir. Tree Physiology, 27, 441–53. Brown, J. H., Gillooly, J. F., Allen, A. P., et al. (2004). Toward a metabolic theory of ecology. Ecology, 85, 1771–89. Bryant, J. P. & Julkunen-Tiitto, R. (1995). Ontogenetic development of chemical defense by seedling resin birch: energy cost of defense production. Journal of Chemical Ecology, 21, 883–96. Caswell, H. (2001). Matrix Population Models: Analysis, Construction and Interpretation (2nd edn.) (Sunderland, MA: Sinauer Associates). Caswell, H. & Salguero-Gómez, R (2013). Age, stage, and senescence in plants. Journal of Ecology, 3, 585–95. Cochran, M. E. & Ellner, S. P. (1992). Simple methods for calculating age-based life history parameters for stage-structured populations. Ecological Monographs, 62, 345–64. Chabot, B. F. & Hicks, D. J. (1982). The ecology of leaf life span. Annual Review of Ecology and Systematics, 13, 229–59. Chapin III, F. S., Matson, P. A. & Vitousek, P. M. (2012). Plant carbon budgets. In Principles of Terrestrial Ecosystem Ecology (pp. 157–81) (New York: Springer). Close, D. C., Davies, N. W. & Beadle, C.L. (2001). Temporal variation of tannins (galloylglucoses), flavonols and anthocyanins in leaves of Eucalyptus nitens seedlings: implications for light attenuation and antioxidant activities. Australian Journal of Plant Physiology, 28, 269–78. Coyea, M. R. & Margolis, H. A. (1992). Factors affecting the relationship between sapwood area and leaf area in balsam fir. Canadian Journal of Forest Research, 22, 1684–93. Dangl, J. L., Dietrich, R. A. & Thomas, H. (2000). Senescence and programmed cell death. In Biochemistry and Molecular Biology of Plants, ed. B. B. Buchanan, W. Gruissem & R. L. Jones (pp. 1044–1100) (Rockville, MD: ASPP). Davies, P. J. (2010). Plant Hormones: Biosynthesis, Signal Transduction, Action! (Dordrecht: Springer). Dietze, M. C., Sala, A., Carbone, M. S., et al. (2014). Nonstructural carbon in woody plants. Annual Review of Plant Biology, 65, 667–87. Donaldson, J. R., Stevens, M. T., Barnhill, H. R. & Lindroth, R. L. (2006). Age-related shifts in leaf chemistry of clonal aspen (Populus tremuloides). Journal of Chemical Ecology, 32, 1415–29.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:39:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.013

278

Physiological and Biochemical Processes Related to Ageing and Senescence in Plants

England, J. R. & Attiwill, P. M. (2006). Changes in leaf morphology and anatomy with tree age and height in the broadleaved evergreen species, Eucalyptus regnans F. Muell. Trees, 20, 79–90. Evans, G. C. (1972). The Quantitative Analysis of Plant Growth (Berkeley: University of California Press). Fageria, N. K. (2009). The Use of Nutrients in Crops Plants (Boca Raton, FL: CRC Press, Taylor & Francis Group). Finkelstein, R., Gampala, S. & Rock, C. (2002). Abscisic acid signaling in seeds and seedlings. Plant Cell, 14, S15–45. Fritz, R. S., Hochwender, C. G., Lewkiewicz, D. A., et al. (2001). Seedling herbivory by slugs in a willow hybrid system: developmental changes in damage, chemical defense, and plant performance. Oecologia, 129, 87–97. Galoch, E. (1985). Comparison of the content of growth regulators in juvenile and adult plants of birch (Betula verrucosa Ehrh.). Acta Physiologia Plantarum, 7, 205–15. García, M. B. & Antor, R. J. (1995). Age and size structure in populations of a long-lived dioecious geophyte: Borderea pyrenaica (Dioscoreaceae). International Journal of Plant Sciences, 156, 236–43. Gartner, B. L. & Meinzer, F. C. (2005). Structure-function relationships in sapwood water transport and storage. In Vascular Transport in Plants, ed. M. Zwieniecki & N. M. Holbrook (pp. 307–31) (Oxford: Elsevier/Academic Press). Goldschmidt, E. E. (2014). Plant grafting: new mechanisms, evolutionary implications. Frontiers in Plant Science, 5, 727. Goodyear, C. P. (1997). Fish age determination from length: an evaluation of three methods using simulated data. Fisheries Bulletin, 95, 39–46. Gjerdrum, P. (2003). Heartwood in relation to age and growth rate in Pinus sylvestris L. in Scandinavia. Forestry, 76, 413–24. Haberer, G. & Kieber, J. J. (2002). New insights into a classic phytohormone. Plant Physiology, 128, 354–62. Haffner, V., Enjalric, F., Lardet, L. & Carron, M.P. (1991). Maturation of woody plants: a review of metabolic and genomic aspects. Annals of Forest Science, 48, 616–30. Halliwell, B. & Gutteridge, J. M. C. (1989). Free Radicals in Biology and Medicine. (Oxford University Press). Hamid, H. & Mencuccini, M. (2009). Age- and size-related changes in physiological characteristics and chemical composition of Acer pseudoplatanus and Fraxinus excelsior trees. Tree Physiology, 29, 27–38. Hamilton, W. D. (1966). Moulding of senescence by natural selection. Journal of Theoretical Biology, 12, 12–45. Harman, D. (1956). Aging: a theory based on free radical and radiation chemistry. Journal of Gerontology, 11, 298–300. Harman, D. (1981). The aging process. Proceedings of the National Academy of Sciences of the United States of America, 78, 7123–8. Harper, J. L. (1977). Population Biology of Plants (New York: Academic Press). Hellkvist, J., Richards, G. P. & Jarvis, P. G. (1974). Water potential and tissue water relations in Sitka spruce trees measured with the pressure chamber. Journal of Applied Ecology, 11, 637–67. Higuchi, T. (1997). Biochemistry and Molecular Biology of Wood (Berlin: Springer).

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:39:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.013

References

279

Hillis, W. (1968). Chemical aspects of heartwood formation. Wood Science and Technology, 2, 241–59. Hillis, W. (1987). Heartwood and Tree Exudates (Berlin: Springer-Verlag). Hölttä, T., Kurppa, M. & Nikinmaa, E. (2013). Scaling of xylem and phloem transport capacity and resource usage with tree size. Frontiers in Plant Science, 4(496), 1–19. Ishii, H. (2011). How do changes in leaf/shoot morphology and crown architecture affect growth and physiological function of tall trees? In Size and Age-Related Changes in Tree Structure and Function, ed. F. C. Meinzer, B. Lachenbruch & T. E. Dawson (pp. 215–32) (Dordrecht: Springer). Jones, O. R., Scheuerlein, A., Salguero-Gómez, R., et al. (2014). Diversity of ageing across the tree of life. Nature, 505, 169–73. Jones, O. R., Gaillard, J.-M., Tuljapurkar, S., et al. (2008). Senescence rates are determined by ranking on the fast–slow life-history continuum. Ecology Letters, 11, 664–73. Juvany, M., Müller, M. & Munné-Bosch, S. (2013). Photo-oxidative stress in emerging and senescing leaves: a mirror image? Journal of Experimental Botany, 64, 3087–98. Jyske, T. & Hölttä, T. (2014). Comparison of phloem and xylem hydraulic architecture in Picea abies stems. New Phytologist, 205, 102–15. Karban, R. & Myers, J. H. (1989). Induced plant responses to herbivory. Annual Review of Ecology and Systematics, 20, 331–48. King, D. A. (2011). Size-related changes in tree proportions and their potential influence on the course of height growth. In Size and Age-Related Changes in Tree Structure and Function, ed. F. C. Meinzer, B. Lachenbruch & T. E. Dawson (pp. 165–91) (Dordrecht: Springer). Kitin, P., Fujii, T., Abe, H. & Takata, K. (2009). Anatomical features that facilitate radial flow across growth rings and from xylem to cambium in Cryptomeria japonica. Annals of Botany, 103, 1145–57. Lambers, H. & Ribas-Carbo, M. (eds.) (2005). Plant respiration: from cell to ecosystem. In Advances in Photosynthesis and Respiration, Vol.18 (Dordrecht: Springer). Larson, D. W. (2001). The paradox of great longevity in a short-lived tree species. Experimental Gerontology 36, 651–73. Mäkelä, A. & Valentine. H. T. (2006). The quarter-power scaling model does not imply size-invariant hydraulic resistance in plants. Journal of Theoretical Biology, 243, 283–5. Matsuzaki, J., Norisada, M., Kodaira, J., et al. (2005). Shoots grafted into the upper crowns of tall Japanese cedar (Cryptomeria japonica D. Don) show foliar gas exchange characteristics similar to those of intact shoots. Trees, 19, 198–203. McDowell, N., Barnard, H., Bond, B. J., et al. (2002). The relationship between tree height and leaf area: sapwood area ratio. Oecologia, 132, 12–20. Medawar, P. B. (1952). An Unsolved Problem in Biology (London: Lewis). Mencuccini, M. & Grace, J. (1995). Climate influences the leaf-area sapwood area ratio in Scots pine. Tree Physiology, 15, 1–10. Mencuccini, M. & Grace, J. (1996). Developmental patterns of aboveground xylem conductance in a Scots pine (Pinus sylvestris L.) age sequence. Plant, Cell and Environment, 19, 939–48. Mencuccini, M. (2002). Hydraulic constraints in the functional scaling of trees. Tree Physiology, 22, 553–65.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:39:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.013

280

Physiological and Biochemical Processes Related to Ageing and Senescence in Plants

Mencuccini, M. (2003). The ecological significance of long-distance water transport: short-term regulation, long-term acclimation and the hydraulic costs of stature across plant life forms. Plant, Cell and Environment, 26, 163–82. Mencuccini, M., Martinez-Vilalta, J., Vanderklein, D., et al. (2005). Size-mediated ageing reduces vigour in tall trees. Ecology Letters, 8, 1183–90. Mencuccini, M., Martínez-Vilalta, J., Hamid, H. A., et al. (2007). Evidence for age- and size-mediated controls of tree growth from grafting studies. Tree Physiology, 27, 463–73. Mencuccini, M., Hölttä, T. & Martínez-Vilalta, J. (2011). Comparative criteria for models of the vascular transport systems of tall trees. In Size and Age-Related Changes in Tree Structure and Function, ed. F. C. Meinzer, B. Lachenbruch & T. E. Dawson (pp. 309–39) (Dordrecht: Springer). Mencuccini, M., Oñate, M., Rico, L., et al. (2014). No signs of meristem senescence in old Scots pine. Journal of Ecology, 102, 555–65. Metcalf, C. J. E., McMahon, S. M., Salguero-Gómez, R. & Jongejans, E. (2013). IPMpack: an R package for Integral Projection Models. Methods in Ecology and Evolution, 4, 195–200. Mooney, H. A. (1972). The carbon balance of plants. Annual Review of Ecology and Systematics, 3, 315–46. Morales, M., García, M. B., Munné-Bosch, S. & Alegre, L. (2013). Photo-oxidative stress markers reveal absence of physiological deterioration with ageing in Borderea pyrenaica, an extraordinarily long-lived herb. Journal of Ecology, 101, 555–65. Morales, M., Munné-Bosch, S. & Alegre, L. (2015). Secret of long life lies underground. New Phytologist, 205, 463–7. Munné-Bosch, S., Jubany-Marí, T. & Alegre, L. (2001). Drought-induced senescence is characterized by a loss of antioxidant defences in chloroplasts. Plant, Cell and Environment, 24, 1319–27. Munné-Bosch, S. & Alegre, L. (2002). Plant aging increases oxidative stress in chloroplasts. Planta, 214, 608–15. Munné-Bosch, S. & Lalueza, P. (2007). Age-related changes in oxidative stress markers and abscisic acid levels in a drought-tolerant shrub, Cistus clusii grown under Mediterranean field conditions. Planta, 225, 1039–49. Nakaba, S., Sano, Y., Kubo, T. & Funada, R. (2006). The positional distribution of cell death of ray parenchyma in a conifer, Abies sachalinensis. Plant Cell Reports, 25, 1143–8. Nakaba, S., Kubo, T. & Funada, R (2008). Differences in patterns of cell death between ray parenchyma cells and ray tracheids in the conifers Pinus densiflora and Pinus rigida. Trees, 22, 623–30. Niklas, K. J. (1992). Plant Biomechanics (University of Chicago Press). Nobuchi, T. & Hasegawa, J. (1994). Radial distribution of heartwood phenols and the cytological changes of ray parenchyma cells associated with heartwood formation in Japanese red pine (Pinus densiflora Sieb. et Zucc.). Bulletin of the Kyoto University Forests, 66, 132–42. Oñate, M., García, M. B. & Munné-Bosch, S. (2012). Age and sex-related changes in cytokinins, auxins and abscisic acid in a centenarian relict herbaceous perennial. Planta, 235, 349–54.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:39:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.013

References

281

Oñate, M. & Munné-Bosch, S. (2008). Meristem aging is not responsible for age-related changes in growth and abscisic acid levels in the Mediterranean shrub, Cistus clusii. Plant Biology, 10, 148–55. Kozlowski, T. T. & Pallardy, S.G. (2010). Physiology of Woody Plants (Burlington, MA: Academic Press). Pandalai, R. C., Nair, G. M. & Shah, J. J. (1985). Ultrastructure of ray parenchyma cells in the wood of Melia azedarach L. (Meliacae). Wood Science and Technology, 19, 201–9. Paschalidis, K. A. & Roubelakis-Angelakis, K. A. (2005). Spatial and temporal distribution of polyamine levels and polyamine anabolism in different organs/tissues of the tobacco plant: correlations with age, cell division/expansion, and differentiation. Plant Physiology, 138, 142–52. Peek, M. S. (2007). Explaining variation in fine root life span. Progress in Botany, 68, 382–98. Petit, G., Anfodillo, T. & Mencuccini M. (2008). Tapering of xylem conduits and hydraulic limitations in sycamore (Acer pseudoplatanus L.) trees. New Phytologist, 177, 653–64. Petty, J. A. (1972). Aspiration of bordered pits in conifer wood. Proceedings of the Royal Society of London Series B: Biological Sciences, 181, 395–406. Poorter, L. & Rozendaal, D. M. A. (2008). Leaf size and leaf display of thirty-eight tropical tree species. Oecologia, 158, 35–46. Poorter, L., Wright, S. J., Paz, H., et al. (2008). Are functional traits good predictors of demographic rates? Evidence from five neotropical forests. Ecology, 89, 1908–20. Quirino, B. F., Noh, Y., Himelblau, E. & Amasino, R.M. (2000). Molecular aspects of leaf senescence. Trends in Plant Science, 5, 278–82. Rajjou, L., Duval, M., Gallardo, K., et al. (2012). Seed germination and vigor. Annual Review in Plant Biology, 63, 507–33. Rey, M., Tiburcio, A. F., Díaz-Sala, C. & Rodríguez, R. (1994). Endogenous polyamine concentrations in juvenile, adult and in vitro reinvigorated hazel. Tree Physiology, 14, 191–200. Rustin, P., Kleist-Retzow, J., Vajo, Z., et al. (2000). For debate: defective mitochondria, free radicals, cell death, aging-reality or myth-ochondria? Mechanisms of Aging and Development, 114, 201–6. Ryan, M. G., Bond, B. J., Law, B. E., et al. (2000). Transpiration and whole-tree conductance in ponderosa pine trees of different heights. Oecologia, 124, 553–60. Ryan, M. G., Phillips, N. & Bond, B. J. (2006). The hydraulic limitation hypothesis revisited. Plant, Cell and Environment, 29, 367–81. Sala, A., Fouts, W. & Hoch, G. (2011). Carbon storage in trees: does relative carbon supply decrease with tree size. In Size and Age-Related Changes in Tree Structure and Function, ed. F. C. Meinzer, B. Lachenbruch & T. E. Dawson (pp. 287–30) (Dordrecht: Springer). Sala, A. & Mencuccini, M. (2014). Plump trees win under drought. Nature Climate Change, 4, 666–7. Salguero-Gómez, R. & Casper, B. B. (2010). Keeping plant shrinkage in the demographic loop. Journal of Ecology, 98, 312–23. Salguero-Gómez, R. & Casper, B. B. (2011). A hydraulic explanation for size-specific plant shrinkage: developmental hydraulic sectoriality. New Phytologist, 189, 229–40.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:39:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.013

282

Physiological and Biochemical Processes Related to Ageing and Senescence in Plants

Schenk, H. J., Espino, S., Goedhart, C. M., et al. (2008). Hydraulic integration and shrub growth form linked across continental aridity gradients. Proceedings of the National Academy of Sciences of the United States of America, 105, 11248–53. Shinozaki, K., Yoda, K., Hozumi, K. & Kira, T. (1964a). A quantitative analysis of plant form-the pipe model theory: I. Basic analysis. Japanese Journal of Ecology, 14, 97–105. Shinozaki, K., Yoda, K., Hozumi, K. & Kira, T. (1964b). A quantitative analysis of plant form-the pipe model theory: II. Further evidence of the theory and its application in forest ecology. Japanese Journal of Ecology, 14, 133–9. Scholander, P. F., Hammel, H. T., Hemmingsen, E. A. & Bradstreet, E. D. (1964). Pressure and osmotic potential in leaves of mangroves and some other plants. Proceedings of the National Academy of Sciences of the United States of America, 52, 119–25. Scholz, F. G., Phillips, N. G., Bucci, S. J., et al. (2011). Hydraulic capacitance: biophysics and functional significance of internal water sources in relation to tree size. In Size and Age-Related Changes in Tree Structure and Function, ed. F. C. Meinzer, B. Lachenbruch & T. E. Dawson (pp. 341–61) (Dordrecht: Springer). Spicer, R. (2005). Senescence in secondary xylem: heartwood formation as an active developmental program. In Vascular Transport in Plants, ed. N. Holbrook & M. Zwieniecki (pp. 457–75) (Oxford: Elsevier/Academic Press). Steppe, K., Niinemets, U. & Teskey, R. O. (2011). Tree size- and age-related changes in leaf physiology and their influence on carbon gain. In Size and Age-Related Changes in Tree Structure and Function, ed. F. C. Meinzer, B. Lachenbruch & T. E. Dawson (pp. 235–53) (Dordrecht: Springer). Szymanski, M., Pazdrowski, W., Kazmierczak, K., et al. (2008). Axial and radial variation in the proportions of sapwood and heartwood in stems of common oak (Quercus robur L.) depending on site type, age class and social class of tree position. Acta Scientiarum Polonorum, 7, 45–58. Tiburcio, A. F., Campos, J. L., Figueras, X. & Besford, R. T. (1993). Recent advances in the understanding of polyamine functions during plant development. Plant Growth Regulation, 12, 331–40. Tyree, M. T. & Zimmerman, M. H. (2002). Xylem Structure and the Ascent of Sap (Berlin: Springer). Tuljapurkar, S. (1990). Population Dynamics in Variable Environments (Berlin: Springer). Valdés, A. E., Centeno, M. L., Espinel, S. & Fernández, B. (2002). Could plant hormones be the basis of maturation indices in Pinus radiata? Plant Physiology and Biochemistry, 40, 211–16. Valdés, A. E., Fernández, B. & Centeno, M. L. (2004). Hormonal changes throughout maturation and ageing in Pinus pinea. Plant Physiology and Biochemistry, 42, 335–40. Valdés, A. E., Centeno, M. L. & Fernández, B. (2005). Age-related changes in the hormonal status of Pinus radiata needle fascicle meristem. Plant Science, 167, 373–8. Vanderklein, D., Martínez-Vilalta, J., Lee, S. & Mencuccini, M. (2007). Plant size, not age, regulates growth and gas exchange in grafted Scots pine trees. Tree Physiology, 27, 71–9. Watson M. A. & Lu Y. (2004). Factors regulating senescence in the annual shoots of perennial plants. In Cell Death in Plants, ed. L. D. Nooden (pp. 259–69) (New York: Academic Press).

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:39:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.013

References

283

Webber, B. L. & Woodrow, I. E. (2009). Chemical and physical plant defence across multiple ontogenetic history stages in a tropical rainforest understorey tree. Journal of Ecology, 97, 761–71. West, G. B., Brown, J. H. & Enquist, B. J. (1999). The fourth dimension of life: fractal geometry and allometric scaling of organisms. Science, 284, 1677–9. West, P. W. (2009). Tree and Forest Measurement (Berlin: Springer-Verlag). Westoby, M., Warton, D. & Reich, P. (2000). The time value of leaf area. American Naturalist, 155, 649–56. Westoby, M., Falster, D. S., Moles, A., et al. (2002). Plant ecological strategies: some leading dimensions of variation between species. Annual Review of Ecology and Systematics, 33, 125–59. Wilkes, J. (1991). Heartwood development and its relationship to growth in Pinus radiata. Wood Science and Technology, 25, 85–90. Williams, G. (1957). Pleiotropy, natural selection, and the evolution of senescence. Evolution, 11, 398–411. Yang, J. C., Zhang, J. H., Wang, Z. Q., et al. (2003). Involvement of abscisic acid and cytokinins in the senescence and remobilization of carbon reserves in wheat subjected to water stress during rain filling. Plant, Cell and Environment, 26, 1621–31. Yang, K. C., Hazenberg, G., Bradfield, G. E. & Maze, J. R. (1985). Vertical variation of sapwood thickness in Pinus banksiana lanb and Larix laricina (Du Roi) K. Koch. Canadian Journal of Forest Research, 15, 822–8. Yang, K. C. & Hazenberg, G. (1991a). Relationship between tree age and sapwood heartwood width in Populus tremuloides Michx. Wood and Fiber Science, 23, 247–52. Yang, K. C. & Hazenberg, G. (1991b). Sapwood and heartwood width in relationship to tree age in Pinus banksiana. Canadian Journal of Forest Research, 21, 521–5. Yang, K. C. (1993). Survival rate and nuclear irregularity index of sapwood ray parenchyma cells in 4 tree species. Canadian Journal of Forest Research, 23, 673–9. Yoder, B. J., Ryan, M. G., Waring, H., et al. (1994). Evidence of reduced photosynthetic rates in old trees. Forest Science, 40, 513–27. Zaffari, G. R., Peres, L. E. P. & Kerbauy, G. B. (1998). Endogenous levels of cytokinins, indolacetic acid, abscisic acid and pigments in variegated somaclones of micropopagated banana leaves. Journal of Plant Growth Regulation, 17, 59–61. Zotz, G., Wilhelm, K. & Becker, A. (2011). Heteroblasty: a review. Botanical Review, 77, 109–51. Zuidema, P. A., Jongejans, E., Chien, P. D., et al. (2010). Integral Projection Models for trees: a new parameterization method and a validation of model output. Journal of Ecology, 98, 345–55.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:39:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.013

14 The Evolution of Senescence in Annual Plants The Importance of Phenology and the Potential for Plasticity Liana T. Burghardt and C. Jessica E. Metcalf

Short Summary Senescence, the decline in survival or fecundity with age, is a predicted outcome of the declining force of selection at late ages. In this chapter we suggest that in semelparous plants, where reproduction is fatal, seasonal limits on the opportunity for reproduction may also influence the timing of senescence. Further, the timing, or phenology, of germination and flowering can influence these seasonal limits. To support this hypothesis, first we use an integrated life-cycle model of the semelparous annual Arabidopsis thaliana to show that under reasonable assumptions about seasonal drivers of mortality, the window for reproduction (and therefore senescence) depends on (1) where in Europe the plant lives and (2) genetic variation that determines life-cycle phenology. In particular, we find that genetic variation that influences the timing of an early lifestage transition (germination timing) can have ramifying effects on the time available for reproduction within each environment. Next, we review recent research on the molecular and genetic basis of semelparity and senescence. This work emphasises the environmental dependence and tight molecular regulation of many of the component processes that ultimately lead to whole-plant senescence. In combination, both the modelling and the review emphasise the possibility of phenotypic plasticity in the regulation of the component processes that underlie survival and fecundity and therefore of senescence. Future work should focus on dissecting the genetic basis of the environmental dependence of the survival/fecundity relationship and creating a better understanding of how the senescence of individual tissues relate to senescence on the wholeplant level.

Introduction Senescence, at the demographic level, is measured as a decline in survival (or fertility) with age in a given environment and is observed in many species across the Tree of Life (Jones et al. 2014; Nussey et al. 2013) although is certainly not ubiquitous, particularly Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:40:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.014

Introduction

285

in plants (Baudisch et al. 2013). This at first seems surprising from an evolutionary perspective because we might expect strong selection for organisms to live and reproduce for longer and longer time periods. But, as initially noted by Medawar (1952), the size of a birth cohort inevitably declines with age, an observation that provides the basis for classical understanding of the evolution of senescence (reviewed in Roach 2003; Flatt & Schmidt 2009). The rarity of older individuals means that, by chance, deleterious mutations that increase mortality or decrease fecundity in older individuals can go to fixation (‘mutation accumulation’) in a population. Similarly, genetic changes that favour survival or reproduction at younger ages at the expense of later survival or reproduction (‘antagonistic pleiotropy’) will be able to spread in the population because selection is not as strong in later cohorts (Williams 1957). In addition, recent models by Baudisch (2008) highlight the importance of trade-offs between survival and fecundity in determining the diversity of senescence patterns observed. In this chapter we outline how seasonal environments and these trade-offs may influence patterns of survival with age. Semelparous plants, in which reproduction is fatal, are a useful system for exploring the potential importance of trade-offs in plant senescence because, in some sense, they reflect the extreme of the trade-off continuum: very high returns of fertility are thought to result in a lack of investment in survival and therefore for extraordinarily high rates of senescence (Young 1990). However, recent research suggests that the timing of the onset and rate of senescence can vary even for semelparous plants. For instance, some species do not stop photosynthesising when they begin reproducing, as would be expected if plants allocated all resources towards reproduction upon flowering. In fact, some semelparous plants gain the majority of their carbon while they are reproducing (Aschan & Pfanz 2003; Earley et al. 2009; Gammelvind et al. 1996). Therefore, even within a semelparous life cycle, continued ability to survive (delayed onset of senescence) will ultimately create more resources for reproduction and thus greater fertility. In sum, despite the fact that reproduction is fatal in semelparous organisms, there is still the potential for selection for later onset of senescence or changes in senescence progression. Seasonal environmental variation can counter selection for longer reproduction and delayed senescence. Environmental conditions are not static in natural environments, and conditions that lead to low survival of reproducing plants (e.g. frosts in winter, drought conditions in summer) occur with high predictability in some habitats. These changes in the environment can shift mortality and fecundity rates, potentially creating enormous selection to complete reproduction within a given time horizon. Therefore, the environment can counteract selection for longer reproductive periods. The length of time a plant has to reproduce is strongly influenced by the timing of flowering. Thus, flowering time could also shape the conditions during which reproduction (and senescence) occurs (Burghardt et al. 2015). Intriguingly, genetic cross-talk has been found between genes that influence flowering time and those implicated in senescence at the whole-rosette level (Wingler 2011). Because flowering time strongly depends on germination time for many annual species, genetic or environmental variation that influences germination time could also influence senescence dynamics. To our Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:40:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.014

286

The Evolution of Senescence in Annual Plants

knowledge, no one has yet explored this possibility. Here we suggest that (1) the environmental context of reproduction and (2) the phenology of previous life stages may contribute to observed senescence patterns in semelparous annual plants. Due to their genetic and experimental tractability, plants provide an unparalleled opportunity to tease apart the complex factors that determine the relationship between environmental context and senescence. In this chapter we use an integrated life-cycle model for Arabidopsis thaliana to illustrate how seasonal and geographical variation and phenology shape the size of the window of opportunity for reproduction. These results suggest the potential for the environment to mediate mechanisms underlying senescence. Next, we report on recent research into mechanisms of semelparity and tissue-specific senescence processes, highlighting the environmental dependency of these mechanisms. We close by outlining next steps in evaluating the mechanisms by which the environment influences the evolution of senescence.

Semelparous Plants as a Model for the Environmental Dependence of Senescence Semelparity (also known as ‘monocarpy’ in plants), or a life history in which reproduction is fatal, generates a classic life-history paradox – Lamont Cole first suggested in 1954 that semelparity, rather than iteroparity (multiple reproductive bouts), should be the norm: a semelparous plant only needs to produce one more seed than an iteroparous plant to match its reproductive output (Cole 1954). Yet, both life histories can be observed in nature. Charnov and Schaeffer (1973) resolved the paradox, pointing out that survival of adults in many species is higher than survival of juveniles; consequently, returns from retaining adults are greater than producing one extra seed. This theory suggests that the environment in which a species is found could influence whether it has a semelparous or iteroparous life cycle. As predicted, semelparous organisms are often found in contexts where survival of adults is low, such as in dry lowlands (Kim & Donohue 2011) or in life cycles with risky long-distance migration, as is the case in salmonid fishes (Crespi & Teo 2002). Additionally, evidence has been found of repeated transitions to the semelparous habit in association with season-specific climate variation (Evans et al. 2005). Beyond the intriguing evolutionary fact of their existence, the simplicity of the semelparous life cycle makes them a unique resource for studies of life-history evolution (Jong et al. 2000; Klinkhamer et al. 1996). Much of this research has been on the evolution of the timing of flowering. The length of time spent in the vegetative stage before flowering determines whether a plant is a semelparous annual (1 year). Because flowering is fatal, flowering time is both under strong selection and generally the result of a relatively straightforward trade-off. The longer individuals live before flowering, the larger they grow, and the more seeds they produce due to an allometric relationship between size and seed output (Klinkhamer et al. 1992). However, if the delay before flowering is too long, there is a risk that individuals will die without ever having reproduced (Metcalf et al. 2003). Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:40:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.014

An Integrated Life-cycle Model to Explore Phenology

287

Thus, given sufficient demographic data on a particular species, the size at flowering can be predicted with considerable accuracy (Rees et al. 2006). As a semelparous plant and a uniquely genetically well-characterised species, A. thaliana provides the tantalising possibility of linking these evolutionary predictions (Rees et al. 2006) to the underlying genetic mechanisms that mediate these fundamental trade-offs (Metcalf & Mitchell-Olds 2009). However, even for a species as well understood as A. thaliana, the complexity of the map linking genotype to phenotype complicates this effort. For instance, one allele does not result in one particular flowering phenotype; the current environment, the sequence of past environments and the genetic context of the allele (i.e. epistasis) all shape the timing of flowering. Recently, a large body of experimental data, encompassing many genotypes and environments, has been combined into mathematical models that allow accurate prediction of flowering time for a particular genotype (Wilczek et al. 2009). Similar methods have been used to predict flowering in other species (Satake et al. 2013) and other phenological transitions such as germination (Alvarado & Bradford 2002) and budburst (Chuine & Beaubien 2001). Due to these advances, A. thaliana is an ideal system for examining how phenological traits function together to influence the time available for the plant to reproduce.

An Integrated Life-cycle Model to Explore Phenology To describe the link between phenology and the environmental context shaping the evolution of senescence, we developed an integrated life-cycle model that predicts phenology across generations by linking mathematical descriptions of how germination, flowering and seed dispersal progress as a function of the environment (Burghardt et al. in press). These phenology models are inspired by models originally used to aid crop production (Wang 1960). Mathematical functions are parameterised for a particular genotype and are used to calculate development based on the environment. Life-stage transitions occur when the organism accumulates enough progress to cross the developmental threshold. We link these individual sub-models, parameterised for A. thaliana, such that the timing of seed dispersal determines the environment seeds’ experience and thus germination time (Figure 14.1A, third panel from top); germination time determines the environment rosettes’ experience and therefore flowering time (Figure 14.1A, second panel from top); flowering time determines the environment in which plants reproduce, which determines dispersal timing (Figure 14.1A, top panel); and so on for many generations. These models are driven by hourly environmental inputs to capture how diurnal variation and extremes influence developmental rates (see Burghardt et al. 2015 for model details). Genetic variation can be incorporated into these models by changing the parameters governing how a life stage responds to the environment. For instance, by changing the parameter that describes seed dormancy, we can mimic known natural variation among populations. Figure 14.1A shows three different dormancy genotypes all occurring in the same environment of Norwich, England. These changes in dormancy can cause large Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:40:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.014

O

S A

N

J

D

1.0

0

0.5

J

J

Flowering Germination threshold Threshold

M

F

Vegetative

A Germination threshold Germination Threshold

Hourly environmental inputs

15 Temperature

5

Photoperiod

Moisture

−5 Summer

Fall

Winter Time of year

Spring

Summer

O

S

N

J

D

15

0

0.5

J

J

10 5

M Seed

25

M

A

1.0

20

F A

M O

S 250

A

N

200 Life stages:

150 100 50 0 Jan 1

Mar 15

Jun 1

Aug 15

Flowering time during the year

Oct 1

Dec 15

J

D

1.0

Low dormancy

Reproductive

Start of 2nd generation

Spring

Figure 14.1

(c)

Med. dormancy

High dormancy

0

0.5

J

J M

High dormancy

Med. dormancy

Seed dispersal threshold Germination Threshold

Day length

Dispersal progress

Temperature

Flowering progress Germination progress

1000 800 600 400 200 0

(b)

Low dormancy

300 200 100 0

2500 2000 1500 1000 500 0

# of individuals

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:40:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.014

(a) Genotypes:

F A

M

Summary of integrated life-cycle model structure and results for genotypes simulated in Norwich, England. (A) Lines in the each panel track progress towards germination, flowering and seed dispersal for genotypes reflecting three different levels of dormancy (low-solid; medium-dashed; high-double-dashed) that were dispersed on April 1; symbols in each panel show relevant environmental drivers for each life stage; the bottom panel shows the underlying environmental conditions that drive the developmental progress. (B) Corresponding flowering times across a cohort of 1,000 individuals averaged across the last twenty-five years of a forty-year simulation for each of the levels of dormancy shown in previous panel. (C) Life-cycle graphs showing the proportion of individuals in each of three key life stages averaged across the same twenty-five years. The distance from the centre of the circle indicates the proportion of individuals in the rosette (medium), reproductive (light) and seed (dark) stages each day of the year (January is at the three o’clock position, and the year progresses clockwise). All plots are scaled so the outermost circle reflects 100 per cent of individuals.

Defining Survival Criteria for Reproductive A. thaliana

289

changes in life cycle: the low-dormancy genotype ‘rapid cycles’ through multiple generations in a year, while a middle-dormancy genotype displays a classic winterannual life cycle by germinating in the fall and flowering in the spring. The most dormant genotype germinates and flowers in the spring as a spring annual. Using the model to simulate environmental variation and genetic variation that occur across the native range of A. thaliana, we can derive flowering phenology of different A. thaliana genotypes in different environments (Figure 14.1B). The model is capable of capturing the full complexity of seasonal life histories observed in this species (Lawrence 1976; Picó 2012; Wilczek et al. 2009). This diversity of life cycles results in the occurrence of flowering at multiple times of the year for some genotypes (Figure 14.1C) and suggests that the reproductive environment can vary between individuals of the same genotype. We must note that the model makes no assumptions about environment-dependent resource acquisition, survival or fertility rates of the organism or their relationship with senescence. Here our goal is to set some simple, constant rules that define how the survival of reproductive plants depends on the environment and determine how much variation in reproductive windows can occur based on the phenological variation expressed by this species.

Defining Survival Criteria for Reproductive A. thaliana To explore selection on senescence, we also need to include information on seasonal drivers likely to result in mortality. These can be life-stage specific, such as freezing tolerance, which decreases after the transition to reproduction in A. thaliana (Richter et al. 2013; Seo et al. 2009). Based on an array of empirical evidence and experience in field cultivation of A. thaliana, we set post-flowering mortality to occur if the minimum daily temperature fell below −5°C; if the average daily temperature fell below 0°C or exceeded 35°C; and if conditions were dry. Because drought is a chronic stressor that occurs over multiple days, progress towards death accumulated each day that moisture conditions were below a moisture threshold (−175 MPa) and reset to zero whenever moisture was above that threshold. Progress was also temperature dependent and scaled via the function 0.00555 × T°C such that higher temperatures in dry conditions lead to more progress towards death. Plants died if the cumulative sum reached 1. We cut off assessment of reproductive interval at 150 days post-flowering because five months of reproduction is unlikely in this species. In reality, of course, these tolerances may not be exact. Furthermore, they are likely to vary among populations adapted to different locations. For instance, natural populations from northern climes exhibit increased rosette cold tolerance in common-garden experiments (Ågren & Schemske 2012). Ideally, we would develop a set of mortality conditions fully parameterised from field data for each of the locations and life stages, but data sets reflecting the natural dynamics of A. thaliana remain rare (Metcalf & Mitchell-Olds 2009), and this information is not available. However, because our purpose is to qualitatively assess (rather than quantify) how extreme seasonal conditions translate Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:40:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.014

290

The Evolution of Senescence in Annual Plants

into selection on senescence in vastly different environments and for very different genotypes, these rules should provide some discriminatory power.

‘Windows’ for Reproduction across the Native Range Environmental conditions relevant to the A. thaliana life cycle differ across both spatial and temporal scales (Burghardt et al. 2015). To visualise how much time is available for reproduction without incorporating the influence of phenology across generations, we used our mortality rules (see earlier) to determine the mortality dates for plants bolting on each day of the year. We did this for forty-year environmental replicates extracted from a global climate model for each of four locations across the European range of the species. We used climate data spanning the years 2000 to 2020 from the A1B scenario in the Intergovernmental Panel on Climate Change (IPCC) fourth assessment (NOAA-GFDL 2004) (see Burghardt et al. 2015 for details on environmental series). We expressed the results in terms of both days and thermal time available for reproduction. Thermal time was calculated as a linear increase in development above the base temperature of 3°C. Thermal time may be an important measure because the rate of seed development often depends on temperature (Ainsworth & Ort 2010), but the measures yield qualitatively similar patterns, so we present results only in terms of days. The timing of drought, frost and extreme heat events that killed reproductive plants varied considerably across years (Figure 14.2, grey traces; mean across years, black line), but in all sites, a range of flowering times existed that would permit the maturation of one or more fruits (which contain approximately fifteen seeds) to mature (Figure 14.2, grey curves above the dashed horizontal line). Within this range, A. thaliana plants in Norwich, England, and Halle, Germany, were predicted to have the longest (in terms of days) opportunity for reproduction (Figure 14.2); Norwich had a broader range of dates with maximal reproduction than Halle due to milder winters and less summer drought. In Oulu, Finland, frosts restricted successful reproduction to late spring and summer. In Valencia, Spain, the longest opportunities for reproduction occurred early in the spring, as plants flowering in late spring and summer died quickly due to drought and heat. There also seems to be a window of opportunity for reproducing in the fall in Valencia; however, this requires germination during the summer or surviving the summer as a rosette, which is not observed (Picó 2012).

Life-cycle Phenology Determines What Portion of the ‘Window’ Is Used These calculated seasonal windows for reproduction alone will not determine selection on senescence dynamics for A. thaliana in a particular environment – flowering and germination phenology will also be important because they determine when during the window the plant will actually begin to reproduce. We used the integrated model outlined earlier to simulate how environmental context and genetic variation in seed dormancy combine to determine this phenology. Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:40:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.014

150

Life-cycle Phenology Determines What Portion of the ‘Window’ Is Used

291

0

50

Oulu, Finland

100

200

300 Halle, Germany

0

100

200

300 Norwich, England

0

Days available for reproduction 50 150 0 50

150

0

100

200

150

0

300

0

50

Valencia, Spain

0

Figure 14.2

100 200 Bolting time (Day of Year)

300

Environmentally determined ‘reproductive windows’, that is, days available for reproduction if an individual were to flower on each day, given the assumptions about fatal environmental conditions described in the text and using environments in each of the four locations extracted from a global climate model. Each grey line shows a different year (out of a total of forty years) and the solid line indicates the mean. The dashed horizontal line indicates the amount of time needed for at least one fruit to mature (one A. thaliana fruit contains ten to twenty seeds).

We simulated full life cycles for forty years to obtain flowering times for each individual each year in each location. We omitted the first fifteen years of simulations to allow populations to stabilise. We then projected the amount of time each individual had available for reproduction (and thus senescence) given its predicted flowering time and our mortality rules. An array of evidence indicates that dormancy is variable even for identical genotypes, and the model reflects this reality (Donohue et al. 2005), so each genotype is potentially reflected by a distribution of days of flowering, and we thus explore the distribution of times available for reproduction for each dormancy genotype.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:40:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.014

292

The Evolution of Senescence in Annual Plants

Density of individuals

(a)

0.06

Valencia, Spain Norwich, England Halle, Germany Oulu, Finland

0.04

0.02

0.00 0

Density of individuals

(b)

0.06

50 100 Time available for reproduction (days)

150

Low Dormancy Med. Dormancy High Dormancy

0.04

0.02

0.00 0 Figure 14.3

50 100 Time available for reproduction (days)

150

The influence of phenology on ‘time available for reproduction’. (A) Density plots of the estimated days remaining from when flowering occurs to a mortality event for every individual in twenty-five simulated years in each of four locations (Oulu-solid; Halle-uneven dash; Norwichlarge dash; Valencia-small dash). All graphs show a low-dormancy genotype. (B) Genetic variation that influences germination timing also influences the ‘time available for reproduction’. Results are shown for three genotypes that vary in dormancy level (high-solid; med-large dashes; low-small dashes) in a single location: Valencia, Spain.

To capture the full life cycle across generations, the model uses a temperaturedependent seed-maturation function to predict the timing of the release of the first 10 per cent of seeds (Burghardt et al. 2015). This ends the reproductive phase and initiates the next generation. Here we are interested in the amount of time available for reproduction (and thus senescence) given the timing of flowering and the environmental context and not the seed-maturation function that is currently implemented. Because the seasonal context of the life cycle canalises some of the variation in germination and flowering phenology, these small changes in seed dispersal time should not too greatly influence the full life cycle, as evidenced by matching between simulated life cycles and observed life cycles (Burghardt et al. 2015). Environmental variation between sites had a large influence on time available for reproduction and thus the optimal senescence strategy. Figure 14.3A shows the distribution of times available for reproduction for a low-dormancy genotype simulated at all four sites. For this genotype, Valencia provided the fewest days for reproduction, whereas Norwich provided the most. Both Halle and Oulu had bimodal distributions

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:40:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.014

Potential for Plasticity of Senescence

293

of time available for reproduction. This bimodal pattern can occur for two reasons: (1) some portion of a spring flowering cohort may flower too early and be killed by a lateseason frost; surviving rosettes that have not yet flowered create the second hump; or (2) plants could be flowering at different times of the year and therefore experiencing disparate seasons while reproducing (Figure 14.1 shows how multiple generations could occur in one year). Both dynamics have been observed in natural populations (Ågren et al. 2013; Picó 2012). To address the importance of genetic variation in determining the reproductive environment, we predicted the life cycles of three different genotypes that varied in dormancy level – seeds with higher dormancy take longer to germinate (the same genotypes used in Figure 14.1A). Genetic variation was captured by a parameter based on empirical data: higher values increase germination repression until after ripening, a moisture- and temperature-dependent process, releases the repression (Bradford 2002). Changing dormancy levels can also shift the amount of time available for reproduction and thus senescence (Figure 14.3B; see also Chapter 16). At the lowest dormancy level in Valencia, Spain, all individuals had a small window for reproduction, but as dormancy level increased, germination occurred later in the fall, shifting flowering to later in the spring and increasing the time available for reproduction. At a medium dormancy level, most individuals flowered at a time corresponding to a very short window for reproduction, but there are some peaks of individuals with a longer window corresponding to specific years where late frosts did not happen. These exceptions illustrate the importance of year-to-year environmental variation. As dormancy increased further, more individuals flowered at a time corresponding to a long window of reproduction.

Potential for Plasticity of Senescence In addition to its selective role in determining the time in which reproduction is possible, the environment has also been shown to influence the allocation trade-off between survival and reproduction (Earley et al. 2009; Remington et al. 2013). This is perhaps not surprising given that the age and size at flowering (Wilczek et al. 2009), photosynthetic rate and therefore resource acquisition (Gammelvind et al. 1996), seed production rate (Bannayan et al. 2003; Poggio et al. 2005) and survival (Ågren et al. 2013; Reinsdorf et al. 2013) all can depend on the environment. This research suggests that the existence of the trade-off between survival and reproduction may even be environmentally dependent within seasons such that no trade-off exists if conditions are favourable and nothing limits development, but, as conditions worsen, the trade-off emerges (Stearns 1992: 84–9). Senescence will only be favoured in contexts where a trade-off exists. Year-to-year variation in time available for reproduction (Figure 14.2, grey lines) suggests that if the environment occurring prior to or during reproduction is predictive of the length of the remaining reproductive window, we would expect the evolution of plasticity of senescence (Roff 2003). For instance, if long photoperiods are associated Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:40:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.014

294

The Evolution of Senescence in Annual Plants

with increased probability of drought, plants that accelerate investment in reproduction (and thus decrease survival and increase senescence) in long photoperiods might have higher fitness. A. thaliana can continue to photosynthesize and gather resources long after the transition to flowering (Earley et al. 2009), but this may divert resources from the reproductive process. Mechanisms that help to ‘tune’ the allocation of resources to continued growth versus reproduction based on the length of the reproductive window could considerably increase fitness. In addition to year-to-year variability, another perhaps more intriguing selection pressure for plasticity in senescence also emerges here; because a single genotype can reproduce at multiple times of the year (Figure 14.1; also reported empirically (Lawrence 1976; Wilczek et al. 2010)), individuals of a given genotype may experience very different conditions during reproduction even in the same year. For example, cohorts that flower in the late spring in Halle have more time for reproduction than those that flower in late summer (Figure 14.2B). Consequently, populations that flower at multiple times of the year may be under selection for increased environmental sensitivity of senescence. In contrast, populations that reproduce in only one season would be expected to optimise senescence dynamics to a single context. Experimental research could test whether plants from locations where reproduction occurs in multiple seasons senesce differently in response to environmental cues than those that reproduce in a single season.

Mechanisms of Semelparous Senescence in A. thaliana In the future it will be extremely interesting to extend the preceding modelling results to incorporate explicit molecular mechanisms that lead to senescence in A. thaliana, but unfortunately, the mechanisms that lead to whole-organism death are just beginning to be established in plants (Davies & Gan 2012). Here we highlight progress to date. Recently, it has been shown that semelparity in A. thaliana is attributable to the combination of three major developmental processes (Davies & Gan 2012). The first component is the death of somatic organs and tissues (e.g. leaves) and remobilisation of those resources to new organs, reproductive structures and storage tissues (Barth et al. 2006). This decline in function of individual tissues or organs rather than that of the entire organism is often referred to as ‘senescence’ in the physiological and molecular literature (Thomas 2013), an inconsistency in terminology across disciplines that can be confusing (see Chapter 1). This tissue ‘senescence’ of leaves and rosettes has been studied extensively – yielding evidence for variation among natural populations (Balazadeh et al. 2008a) and determination by numerous highly regulated genes, controlled at the chromatin and transcription as well as the post-transcriptional, translational and post-translational levels (Lim et al. 2007; Woo et al. 2013). In addition, microarray studies have identified hundreds of genes associated with tissue senescence (e.g. SAG12) (Grbic 2003; Hensel et al. 1993) and commensurate shifts in metabolism (Watanabe et al. 2013). While these processes may well contribute to, or be correlated with, the

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:40:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.014

Progress on Mechanisms in Iteroparous Species

295

decline in whole-organism survival with age (senescence in the evolutionary sense), they are clearly only part of the story. In addition, semelparity in A. thaliana requires both the suppression of axillary meristems to prevent the formation of new shoots and the arrest of reproductive meristems. As plants grow, they leave behind axillary meristems that generally remain inactive unless the primary meristem is damaged or the shoot apical meristem is too far away spatially to repress differentiation. In A. thaliana, the gene AtMYB2 has been shown to actively suppress auxiliary branches specifically late in development (Guo & Gan 2011). When this gene is mutated, plants continually generate new branches, and this extends the reproductive life span. Ultimately, these individual processes combine across multiple levels of organisation to lead to the complex trait of whole-organism senescence (Davies & Gan 2012). The precise manner in which this integration occurs is an active area of research.

Evidence for Environmental Sensitivity of Tissue Senescence A theme that emerges from the research highlighted earlier is that tissue senescence is highly dependent on the environment (reviewed in Lim et al. 2007; Quirino et al. 2000; Thomas 2013; Woo et al. 2013;). For example, 30 per cent of transcription factors associated with leaf senescence were differentially expressed in response to environmental cues (Balazadeh et al. 2008b). Figure 14.4 provides a summary of environmental factors known to be relevant for various tissues in A. thaliana. Environmental dependency of tissue senescence has also been found at the level of individual organs in numerous other plants (Gombert et al. 2006; Li et al. 2000; Meng et al. 2013).

Progress on Mechanisms in Iteroparous Species The same seasonal environmental scenario described for semelparous plants also plays out for iteroparous plants each year when they reproduce. Progress is being made on the mechanisms underlying iteroparous repeated flowering. Most species (e.g. Lenolium and Malus) retain vegetative or quiescent axillary meristems during reproduction (Foster et al. 2003; Onishi et al. 2003), while a few species such as Impatiens balsamina have reproductive meristems that can revert back to vegetative growth (Albani & Coupland 2010). Comparing iteroparous mechanisms to those known in semelparous life cycles has proven enlightening (Albani & Coupland 2010; Andres & Coupland 2012). For instance, in an iteroparous species related to A. thaliana, a portion of meristems each year does not transition to flowering, allowing growth in the subsequent year. The genes PEP1 and AhFLC, which are orthologs of genes important to the flowering transition in the annual A. thaliana, are key to this process (Aikawa et al. 2010; Wang et al. 2009). These similarities in the context of such a fundamentally different life cycle may provide leverage to separate the effects of selection for maximising reproduction Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:40:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.014

296

(a)

The Evolution of Senescence in Annual Plants

(b)

Env. factors

Biotic factors (e.g. pathogens)

Timing of reproduction

?

Leaf senescence

Examples of responsive genes

Tissue type

Citations

Photosynthetic rate (sugar levels)

FRI/PSII-3

Rosette

Wingler et al., 2010

SAG12

Leaf

Noh and Amasino, 1999

Cold

FLC/FRI

Rosette

Wingler et al., 2010

WRKY53

Leaf

Zentgraf et al., 2010

Light Rosette senescence

Whole plant senescence (death of all meristems)

Figure 14.4

External cue

Rosette

Weaver and Amasino, 2001

CRY2a/CIB1 AtXDHL

Leaf

Weaver and Amasino, 2001, Brychkova et al., 2008

Salt

ORE1 JUB1

Leaf

Balazadeh et al., 2010, Wu et al., 2012

Nitrogen

4-5 QTL Bay x Sha RILs

Leaf

Diaz et al., 2006

Drought

NTL4

Leaf

Lee et al., 2012

Pathogens

ELI3-2(SAG25) NIT2 (SAG27)

Leaf

Quirino et al., 1999, Quirino et al., 2000

Genetic evidence for the environmental dependence of the senescence process. (A) Schematic of known connections between the environment and tissue senescence. The environment can influence senescence directly or indirectly by altering other developmental processes such as flowering time that have cascading effects on rosette senescence. While the senescence of individual tissues clearly contributes to whole-plant senescence, the exact relationship between this process and the other major components of whole-plant senescence remains murky as well as how and if environmental factors influence these processes. (B) Table of environmental cues and examples of the downstream genes they influence in A. thaliana. Examples are only provided for leaf and rosette senescence, as those tissues are the most extensively studied.

during a particular time interval from the effects of mutation accumulation that reduce function as an organism ages (Brommer 2014; Pujol et al. 2014).

Conclusion We find that under reasonable assumptions about seasonal drivers of mortality, the window for reproduction (and therefore senescence) in a model semelparous species is location dependent. This finding suggests that there is the opportunity for selection to shape how fast A. thaliana diverts resources from maintenance and survival to seeds after the initiation of reproduction. Further, we find that genetic variation that influences the timing of an early life-stage transition (germination timing) can have ramifying effects on the time available for reproduction within each environment. This suggests that the optimal pattern of senescence will depend on phenology and, further, that genetic variation in early life-stage traits may influence selection on senescence or vice versa. Our model analysis therefore suggests the testable hypothesis that germination alleles may have correlated effects on senescence. Molecular evidence in A. thaliana emphasises the environmental dependence of many of the sub-processes that ultimately lead to senescence. This research also points

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:40:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.014

Future Directions

297

to the high level of regulation and control of these systems. Experiments that focus on how tissue-specific senescence contributes to whole-plant senescence, as well as the molecular basis and environment dependence of trade-offs between survival and fecundity, will be particularly enlightening. In sum, both molecular and modelling work suggest that the environmental dependency of trade-offs may play an important but under-explored role in determining observed senescence patterns in semelparous annual plants. Both the modelling and our review of molecular mechanisms suggest the potential for environmental sensitivity (plasticity) of the allocation of resources to survival versus reproduction (with its accompanying effects on senescence). In animals, plasticity of senescence is commonly reported (Libert et al. 2007; Mair et al. 2003), and elucidating the genetic basis of these processes was recently identified as a major direction for future work (Flatt & Schmidt 2009). In plants, there is also evidence for plasticity in longevity (Borges 2009), and further, there is empirical support for drought-dependent senescence in three semelparous plants. Desert plant populations accelerated senescence in response to drought, whereas Mediterranean populations did not (Aronson et al. 1992). Of course, environmental dependence of senescence may not be adaptive, as it may also reflect nonadaptive plasticity (developmental constraints). For example, in red deer, exposure to harsh early-life conditions constrained development and exacerbated the ageing process (Nussey et al. 2007).

Future Directions There are a number of challenges in exploiting the wealth of knowledge on A. thaliana to understand life-history evolution. In particular, its ecology in natural settings is relatively poorly known, there are few demographic data sets in natural habitats (but see Picó 2012) and knowledge of the behaviour of belowground seed and germination dynamics is especially rare. Because our results indicate that genetic changes that underlie life-stage transitions occurring much earlier than senescence can have a large influence on time available for reproduction, more detail on this part of the life history would be of great value. Many intriguing avenues for investigating the factors that influence the evolution of semelparity and inevitable senescence remain to be explored. As data sets become available on phenology and demography of natural populations of this species, the predictions of our model can be tested. Further, our modelling approach could be modified to explicitly reflect potential environment-specific trade-offs and commensurate changes in senescence to test adaptive hypotheses. Exploration of the genetic and developmental basis of the survival/reproduction trade-off in this species would greatly aid this endeavour. Lastly, another key direction includes selection experiments that test the response to selection for both late-life survival and reproductive rate simultaneously.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:40:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.014

298

The Evolution of Senescence in Annual Plants

Acknowledgements We thank Amity Wilczek, Johanna Schmitt, Kathleen Donohue, Stephen Welch, Kent Bradford, Susan Meyers and other members of the NESCent working group ‘Germination, Trait Coevolution, and Niche Limits in Changing Environments’, who were instrumental in the development of the integrated life cycle. We also thank K. Burghardt for comments and acknowledge support from NSF DDIG 1311406 and grant DEB-1020963 to K. Donohue that facilitated this research.

References Ågren, J., Oakley, C. G., McKay, J. K., et al. (2013). Genetic mapping of adaptation reveals fitness tradeoffs in Arabidopsis thaliana. Proceedings of the National Academy of Sciences of the United States of America, 110, 21077–82. Ågren, J. & Schemske, D. W. (2012). Reciprocal transplants demonstrate strong adaptive differentiation of the model organism Arabidopsis thaliana in its native range. New Phytologist, 194, 1112–22. Aikawa, S., Kobayashi, M. J., Satake, A., et al. (2010). Robust control of the seasonal expression of the Arabidopsis FLC gene in a fluctuating environment. Proceedings of the National Academy of Sciences of the United States of America, 107, 11632–7. Ainsworth, E. A. & Ort, D. R. (2010). How do we improve crop production in a warming world? Plant Physiology, 154, 526–30. Albani, M. C. & Coupland, G. (2010). Comparative analysis of flowering in annual and perennial plants. In Plant Development (San Diego: Elsevier Academic Press). Andres, F. & Coupland, G. (2012). The genetic basis of flowering responses to seasonal cues. Nature Reviews Genetics, 13, 627–39. Aronson, J., Kigel, J., Shmida, A. & Klein, J. (1992). Adaptive phenology of desert and Mediterranean populations of annual plants grown with and without water-stress. Oecologia, 89, 17–26. Aschan, G. & Pfanz, H. (2003). Non-foliar photosynthesis: a strategy of additional carbon acquisition. Flora, 198, 81–97. Balazadeh, S., Parlitz, S., Mueller-Roeber, B. & Meyer, R. C. (2008a). Natural developmental variations in leaf and plant senescence in Arabidopsis thaliana. Plant Biology, 10, 136–47. Balazadeh, S., Riano-Pachon, D. M. & Mueller-Roeber, B. (2008b). Transcription factors regulating leaf senescence in Arabidopsis thaliana. Plant Biology, 10, 63–75. Bannayan, M., Crout, N. M. J. & Hoogenboom, G. (2003). Application of the CERES-Wheat model for within-season prediction of winter wheat yield in the United Kingdom. Agronomy Journal, 95, 114–25. Barth, C., Tullio, M. D. & Conklin, P. L. (2006). The role of ascorbic acid in the control of flowering time and the onset of senescence. Journal of Experimental Botany, 57, 1657–65. Baudisch, A. (2008). Inevitable aging? Contributions to Evolutionary-Demographic Theory (Berlin, Springer).

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:40:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.014

References

299

Baudisch, A., Salguero-Gómez, R., Jones, O. R., et al. (2013). The pace and shape of senescence in angiosperms. Journal of Ecology, 101, 596–606. Borges, R. M. (2009). Phenotypic plasticity and longevity in plants and animals: cause and effect? Journal of Biosciences, 34, 605–11. Bradford, K. J. (2002). Applications of hydrothermal time to quantifying and modeling seed germination and dormancy. Weed Science, 50, 248–60. Brommer, J. E. (2014). Senescence: detecting an evolutionary fingerprint in plants. Current Biology, 24, R267–9. Burghardt, L. T., Metcalf, C. J. E., Wilczek, A. M., et al. (2015). Modeling the influence of genetic and environmental variation on the expression of plant life cycles across landscapes. American Naturalist, 185(2), 212–227. Charnov, E. L. & Schaffer, W. M. (1973). Life history consequences of natural selection: Cole’s result revisited. American Naturalist, 107, 791–3. Chuine, I. & Beaubien, E.G. (2001). Phenology is a major determinant of tree species range. Ecology Letters, 4, 500–10. Cole, L. C. (1954). The population consequences of life history phenomena. Quarterly Review of Biology, 29, 103–36. Crespi, B. J. & Teo, R. (2002). Comparative phylogenetic analysis of the evolution of semelparity and life history in salmonid fishes. Evolution, 56, 1008–20. Davies, P. J. & Gan, S. (2012). Towards an integrated view of monocarpic plant senescence. Russian Journal of Plant Physiology, 59, 467–78. Donohue, K., Dorn, D., Griffith, C., et al. (2005). The evolutionary ecology of seed germination of Arabidopsis thaliana: variable natural selection on germination timing. Evolution, 59, 758–70. Earley, E. J., Ingland, B., Winkler, J. & Tonsor, S. J. (2009). Inflorescences contribute more than rosettes to lifetime carbon gain in Arabidopsis thaliana (Brassicaceae). American Journal of Botany, 96, 786–92. Evans, M. E. K., Hearn, D. J., Hahn, W. J., et al. (2005). Climate and life history evolution in evening primroses (Oenothera, Onagraceae): a phylogenetic comparative analysis. Evolution, 59, 1914–27. Flatt, T. & Schmidt, P. S. (2009). Integrating evolutionary and molecular genetics of aging. Biochimica et Biophysica Acta, 1790, 951–62. Foster, T., Johnston, R. & Seleznyova, A. (2003). A morphological and quantitative characterization of early floral development in apple (Malus x domestica Borkh.). Annals of Botany, 92, 199–206. Gammelvind, L. H., Schjoerring, J. K., Mogensen, V. O., et al. (1996). Photosynthesis in leaves and siliques of winter oilseed rape (Brassica napus L). Plant and Soil, 186, 227–36. Gombert, J., Etienne, P., Ourry, A. & Le Dily, F. (2006). The expression patterns of SAG12/CAB genes reveal the spatial and temporal progression of leaf senescence in Brassica napus L. with sensitivity to the environment. Journal of Experimental Botany, 57, 1949–56. Grbic, V. (2003). SAG2 and SAG12 protein expression in senescing Arabidopsis plants. Physiologia Plantarum, 119, 263–9. Guo, Y. F. & Gan, S. S. (2011). AtMYB2 regulates whole plant senescence by inhibiting cytokinin-mediated branching at late stages of development in Arabidopsis. Plant Physiology, 156, 1612–19.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:40:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.014

300

The Evolution of Senescence in Annual Plants

Hensel, L. L., Grbic, V., Baumgarten, D. A. & Bleecker, A. B. (1993). Developmental and age-related processes that influence the longevity and senescence of photosynthetic tissues in Arabidopsis. Plant Cell, 5, 553–64. Jones, O. R., Scheuerlein, A., Salguero-Gómez, R., et al. (2014). Diversity of ageing across the tree of life. Nature, 505, 169–173. de Jong, T.J., Klinkhamer, P.G.L. & de Heiden, J.L.H. (2000). The evolution of generation time in metapopulations of monocarpic perennial plants: some theoretical considerations and the example of the rare thistle Carlina vulgaris. Evolutionary Ecology, 14, 213–231. Kim, E. & Donohue, K. (2011). Demographic, developmental and life-history variation across altitude in Erysimum capitatum. Journal of Ecology, 99, 1237–49. Klinkhamer, P. G. L., de Jong, T. J. & de Heiden, J. L. H. (1996). An eight-year study of population dynamics and life-history variation of the biennial Carlina vulgaris. Oikos, 75, 259–68. Klinkhamer, P. G. L., Meelis, E., de Jong, T. J. & Weiner, J. (1992). On the analysis of size-dependent reproductive output in plants. Functional Ecology, 6, 308–16. Lawrence, M. (1976). Variation in natural populations of Arabidopsis thaliana (L.) Heynh. In The Biology and Chemistry of the Cruciferae, ed. J. G. Vaghan, A. J. Macleod & B. M. G. Jones (London: Academic Press). Li, J. H., Dijkstra, P., Hymus, G. J., et al. (2000). Leaf senescence of Quercus myrtifolia as affected by long-term CO2 enrichment in its native environment. Global Change Biology, 6, 727–33. Libert, S., Zwiener, J., Chu, X. W., et al. (2007). Regulation of Drosophila life span by olfaction and food-derived odors. Science, 315, 1133–7. Lim, P. O., Kim, H. J. & Nam, H. G. (2007). Leaf senescence. In Annual Review of Plant Biology (Palo Alto, CA: Annual Reviews, 58, 115–136). Mair, W., Goymer, P., Pletcher, S. D. & Partridge, L. (2003). Demography of dietary restriction and death in Drosophila. Science, 301, 1731–3. Medawar, P. (1952). The Uniqueness of the Individual (London: Methuen). Meng, Y. Y., Li, H. Y., Wang, Q., et al. (2013). Blue light–dependent interaction between CRYPTOCHROME2 and CIB1 regulates transcription and leaf senescence in soybean. Plant Cell, 25, 4405–20. Metcalf, C. J. E. & Mitchell-Olds, T. (2009). Life history in a model system: opening the black box with Arabidopsis thaliana. Ecology Letters, 12, 593–600. Metcalf, J. C., Rose, K. E. & Rees, M. (2003). Evolutionary demography of monocarpic perennials. Trends in Ecology and Evolution, 18, 471–80. National Oceanic and Atmospheric Administration Geophysical Fluid Dynamics Laboratory (NOAA-GFDL) (2004). Intergovernmental Panel on Climate Change (IPCC) Fourth Assessment and US Climate Change Science Program (CCSP) Projects. Princeton, NJ. Nussey, D. H., Froy, H., Lemaître, J.-F., et al. (2013). Senescence in natural population of animals: widespread evidence and its implications for bio-gerontology. Ageing Research Reviews, 12, 214–25. Nussey, D. H., Kruuk, L. E. B., Morris, A. & Clutton-Brock, T. H. (2007). Environmental conditions in early life influence ageing rates in a wild population of red deer. Current Biology, 17, R1000–1.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:40:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.014

References

301

Onishi, K., Sano, Y. & Nakashima, H. (2003). Developmental fates of axillary for the pattern of life buds as a major determinant history in Lolium. Plant Production Science, 6, 179–84. Picó, F. X. (2012). Demographic fate of Arabidopsis thaliana cohorts of autumn- and spring-germinated plants along an altitudinal gradient. Journal of Ecology, 100, 1009–18. Poggio, S. L., Satorre, E. H., Dethiou, S. & Gonzalo, G. M. (2005). Pod and seed numbers as a function of photothermal quotient during the seed set period of field pea (Pisum sativum) crops. European Journal of Agronomy, 22, 55–69. Pujol, B., Marrot, P. & Pannell, J. R. (2014). A quantitative genetic signature of senescence in a short-lived perennial plant. Current Biology, 24, 744–7. Quirino, B. F., Noh, Y. S., Himelblau, E. & Amasino, R. M. (2000). Molecular aspects of leaf senescence. Trends in Plant Science, 5, 278–82. Rees, M., Childs, D. Z., Metcalf, J. C., et al. (2006). Seed dormancy and delayed flowering in monocarpic plants: selective interactions in a stochastic environment. American Naturalist, 168, e53. Reinsdorf, E., Koch, H. J. & Marlander, B. (2013). Phenotype related differences in frost tolerance of winter sugar beet (Beta vulgaris L. ). Field Crops Research, 151, 27–34. Remington, D. L., Leinonen, P. H., Leppala, J. & Savolainen, O. (2013). Complex genetic effects on early vegetative development shape resource allocation differences between Arabidopsis lyrata populations. Genetics, 195, 1087–1102. Richter, R., Bastakis, E. & Schwechheimer, C. (2013). Cross-repressive interactions between SOC1 and the GATAs GNC and GNL/CGA1 in the control of greening, cold tolerance, and flowering time in Arabidopsis. Plant Physiology, 162, 1992–2004. Roach, D. A. (2003). Evolutionary and demographic approaches to the study of whole plant senescence. In Plant Cell Death Processes, ed. L. D. Nooden (New York: Academic Press). Roff, D. A. (2003). Life History Evolution (Sunderland, MA: Sinauer Associates). Satake, A., Kawagoe, T., Saburi, Y., et al. (2013). Forecasting flowering phenology under climate warming by modelling the regulatory dynamics of flowering-time genes. Nature Communications, 4. Seo, E., Lee, H., Jeon, J., et al. (2009). Crosstalk between cold response and flowering in Arabidopsis is mediated through the flowering-time gene SOC1 and its upstream negative regulator FLC. Plant Cell, 21, 3185–97. Stearns, S. C. (1992). The Evolution of Life Histories (New York, Oxford University Press). Thomas, H. (2013). Senescence, ageing and death of the whole plant. New Phytologist, 197, 696–711. Wang, J. Y. (1960). A critique of the heat unit approach to plant-response studies. Ecology, 41, 785–90. Wang, R., Farrona, S., Vincent, C., et al. (2009). PEP1 regulates perennial flowering in Arabis alpina. Nature, 459, 423–7. Watanabe, M., Balazadeh, S., Tohge, T., et al. (2013). Comprehensive dissection of spatiotemporal metabolic shifts in primary, secondary, and lipid metabolism during developmental senescence in Arabidopsis. Plant Physiology, 162, 1290–1310.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:40:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.014

302

The Evolution of Senescence in Annual Plants

Wilczek, A. M., Burghardt, L. T., Cobb, A. R., et al. (2010). Genetic and physiological bases for phenological responses to current and predicted climates. Philosophical Transactions of the Royal Society Series B: Biological Sciences, 365, 3129–47. Wilczek, A. M., Roe, J. L., Knapp, M. C., et al. (2009). Effects of genetic perturbation on seasonal life-history plasticity. Science, 323, 930–4. Williams, G. C. (1957). Pleiotropy, natural-selection, and the evolution of senescence. Evolution, 11, 398–411. Wingler, A. (2011). Interactions between flowering and senescence regulation and the influence of low temperature in Arabidopsis and crop plants. Annals of Applied Biology, 159, 320–38. Woo, H. R., Kim, H. J., Nam, H. G. & Lim, P. O. (2013). Plant leaf senescence and death: regulation by multiple layers of control and implications for aging in general. Journal of Cell Science, 126, 4823–33. Young, T. P. (1990). Evolution of semelparity in Mount Kenya Lobelias. Evolutionary Ecology, 4, 157–71. Zentgraf, U., Laun, T. & Miao, Y. (2010). The complex regulation of WRKY53 during leaf senescence of Arabidopsis thaliana. European Journal of Cell Biology, 89, 133–7.

Downloaded from https:/www.cambridge.org/core. UCL, Institute of Education, on 19 Apr 2017 at 13:40:45, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.014

15 Demographic Senescence in Herbaceous Plants Johan P. Dahlgren and Deborah A. Roach

Short Summary For herbaceous plants, in contrast to many higher animals but similarly to other modular and sedentary life forms, it is not so much a question of how senescence progresses as it is a question of whether senescence occurs at all for most species. Overall, both the empirical evidence and theoretical predictions provide contrasting evidence. Monocarpic species and other semelparous life forms are an extreme example of senescence. For polycarpic, iteroparous plants, however, there is only limited evidence that senescence occurs at all. Our review of studies with herbaceous plants also shows that there are only a few studies with detailed age-based demographic data, and here too the evidence for senescence is limited. Moreover, the detrimental effects of ageing are hard to detect in the observational studies upon which most of our current knowledge is built. Our theoretical expectations suggest that it is likely that the evolutionary pressure that shapes senescence in other organisms also acts on plants, but there are also several aspects of plant biology that conflict with some of the assumptions in classical models of the evolution of senescence.

Introduction Definitions We use the term ‘senescence’ to mean detrimental changes in physiology and increased risk of death as organisms age. This has sometimes been called ‘whole-plant senescence’ to separate it from how the term ‘senescence’ is typically used in plant physiological literature, where it refers to the beneficial processes of regulated die-back of plant tissues and individual plant parts such as leaves. In particular, we focus on senescence in demographic rates, that is, increases in mortality or decreases in fecundity with age. We use the term ‘ageing’ to mean growing older and not as an analogous word to senescence. ‘Mortality’ is the probability that an individual will die at a given age, and ‘survivorship’ is the cumulative probability of surviving to a given age. Survivorship curves are plots of the number or proportion of surviving individuals over age. If the y-axis is logarithmic, then three different types of curves have traditionally been identified: type I (concave, opening downwards), type II (a straight line) and type III

Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:17:24, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.015

304

Demographic Senescence in Herbaceous Plants

(convex, opening upwards). For older ages (i.e. disregarding potential changes in mortality over the juvenile period), and assuming that individuals of the same age are identical and in the same constant environment, type I patterns would indicate senescence with an increased mortality with age, type II would indicate a constant mortality with age and type III would indicate a decreasing mortality with age. In natural populations, individual heterogeneity and environmental variation may cause deviations from these patterns, and actual mortality trajectories may not be monotonic (see later), but using this simplistic designation of three types of curves is still a useful way to classify patterns of mortality across species. ‘Fecundity’ is the reproductive output of an individual at a certain age. The demographic rates of mortality and fecundity, and in sizebased demography individual growth rate as well, are known as ‘vital rates’. The reproductive value quantifies how much an individual of a certain age is expected to contribute to future population growth and is thus a function of the vital rates, survivorship and fecundity, for future ages. A decrease in reproductive value with age would therefore mean ‘net senescence’, but here we focus on mortality and fecundity independently. Apart from these demographic terms, we also use some biological terms. Botanists have historically divided plants using several different class systems based on differences in life history. In this text, we usually make a distinction between ‘semelparous’ species (‘monocarpic’ plants, which reproduce at one point in time and then die) and ‘iteroparous’ species (‘polycarpic’ plants, which can reproduce at several points in time). Semelparous plants can be annual, biennial or perennial. Iteroparous species are, by definition, perennial. A third group are annual plants that are not semelparous but have indeterminate growth and are killed off by adverse environmental conditions within a year following germination. Some of these species can behave as iteroparous species and live longer than one year under favourable conditions.

Scope of This Chapter In previous studies reviewing demographic evidence for and against senescence in herbaceous plants, it has been concluded that there is some evidence both for and against senescence occurring in non-clonal iteroparous species but no evidence for senescence occurring in primarily clonal species (Pedersen 1999; Roach 1993; Silvertown et al. 2001; Watkinson 1992). We build on the previous reviews and expand these with what has been learned about ageing in recent demographic field studies where ages of plants are known and from theoretical studies. We then discuss these results in relation to evolutionary theories of senescence and to variation in plant traits. We do not include trees and shrubs, except for a few relevant examples. These woody plants may be expected to differ from herbs because the gradual accumulation of dead tissue in woody plants can cause deterioration of vital rates with size for very large, and typically old, individuals (Harper & White 1974). We also do not include studies that use stageclassified matrix projection models to calculate mortality and fecundity trajectories in this review. Such analyses have identified senescence in trees (Baudisch et al. 2013), which may be consistent with the hypothesis that large size can be detrimental for vital Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:17:24, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.015

Age Effects on Mortality

305

rates of woody plants (Mencuccini et al. 2005). For herbs, mortality typically decreases with size while fecundity typically increases, and we suspect that potential detrimental age effects may not be observed if age-from-stage methods are used. Moreover, without specific age data, the age-versus-size effects cannot be discerned. One of the advantages of using non-clonal herbs to address patterns of senescence is that individual plant age and thus age-dependent patterns are easier to determine and study demographically.

Age Effects on Mortality Semelparous Species Semelparity can be viewed as an extreme form of senescence. In semelparous species, mortality typically decreases with size, and thus with age, until after reproduction, when death of individuals in a cohort occurs in a very short time period (reviewed in Metcalf et al. 2003). Explanations for the evolution of semelparity in plants focus on trade-offs between life history traits, particularly early and late reproduction (see the discussion of trade-offs later in this chapter). It has been shown that the age of flowering in monocarpic species corresponds to the age, or size, that optimises fecundity given a particular risk of mortality (Young & Augspurger 1991).

Non-Semelparous Annuals Several annual plants keep growing and reproducing until they are killed off by adverse environmental conditions, typically after one season. We do not expect the mortality patterns of these species to be interpretable in terms of senescence because it would be hard to separate effects of age from effects of the environment. However, these species are interesting systems for studying senescence in experimental populations with no environmental deterioration. In idealised conditions, when plants can be maintained to ages beyond what which they typically experience in nature, which would be greater than one growing season for annual plants, we would expect to observe traits at late ages that have not been under any selection pressure. However, we are not aware of many such studies (but see one example suggesting that senescence may be present in an annual plant in Roach (1993)). Finally, it should be recognised that species that are annual in some habitats (e.g. with severe winters) may be perennial in others. One example of this is Poa annua. Law et al. (1977) studied perennial populations in England and Wales and reported survivorship curves that may indicate senescence in a few populations (Watkinson 1992) as being partly close to linear. However, some of the curves also indicate constant mortality (linear log-survivorship), and the time period was (naturally) short.

Iteroparous Species Showing No Mortality Senescence Several decades have passed since plant demographic studies gained popularity among ecologists, and this is reflected in the growing number of published studies from long-

Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:17:24, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.015

306

Demographic Senescence in Herbaceous Plants

term demographic monitoring projects of perennial plants. Most of these long-term field studies employed size- or stage-based demography to account for population structure, because size is typically a better predictor of mortality and fecundity than age (Caswell 2001). However, age since germination is known for many individuals in these studies, and this information is starting to be used to look at age effects. Many of the studies that have done so have shown no age pattern to mortality. One of the classic studies of several Swedish meadow species (Tamm 1956, 1972a, 1972b), which used marked individuals but of unknown age, suggested constant mortality for nearly all species. Similar patterns were found in other studies using ‘decay curves’, in other words, the loss of marked individuals of unknown age from a population (Antonovics 1972; Sarukhan & Harper 1973). More detailed long-term studies, in which individuals from different yearly cohorts have been followed through time since establishment and in which ages are thus known for individuals, have shown several different patterns for survival. One early study, by Canfield (1957), with individuals of known ages, reported survivorship data for eleven grass species, where mortality was fairly constant for older individuals, although inspection of the data indicates a weak trend of increased mortality with age in grazed plots and the opposite trend in un-grazed plots. In addition, type II survivorship curves, signifying constant mortality, have been reported for Poa annua (Law et al. 1977), and a type III pattern with decreasing mortality with age has been reported for Viola sororia (Solbrig et al. 1980). Whereas P. annua is very short-lived (often annual), and the observation period covered the entire life span in Law et al. (1977), in the study with V. sororia there were no data for individuals near the maximal recorded life span (Solbrig et al. 1980). Several detailed studies reporting age trajectories of mortality or survivorship have also been published in relatively recent years and are summarised in Table 15.1. Lauenroth and Adler (2008) report constant or decreasing mortality with age in a longterm study of twenty-nine forb and eleven tussock-forming grass species in permanent plots, where survivorship curves were of type II (forbs) and type III (grasses). Judging

Table 15.1 Studies Published between 1993 and 2014 Describing Genet Mortality Patterns in Herbaceous Perennials over Life Spans Species

Increase in mortality

Study

19 grassesa 29 forbs and 11 grassesa Borderea pyrenaica Bouteloua gracilis Gentiana pneumonanthe Lobularia maritime Ophrys sphegodes Plantago lanceolata

No No No No No Yes No Yes

Chu & Adler 2014 Lauenroth & Adler 2008 Garcia et al. 2011 Fair et al. 1999 Rose et al. 1998 Pico & Retana 2008 Hutchings 2010 Roach et al. 2009

a

Some populations were the same in these two studies. Whether or not an increase in mortality with age was reported for adult individuals is indicated.

Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:17:24, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.015

Age Effects on Mortality

307

from the data provided in the appendices of these papers, the type III patterns seem to have been driven by decreases in mortality early in life, and there seem to be no trends over age for older ages in either group. However, data for the oldest individuals were scant for most species. From a similar study based in part on some of the same long-term data for nineteen populations of perennial grasses, Chu and Adler (2014) report that mortality decreased with age in all but one population. A lack of evidence of senescence was also reported from a fourteen-year study on Gentiana pneumonanthe, in which mortality did not differ between age groups of mature plants (Rose et al. 1998). In addition, vital rates of already established plants did not differ from those of plants established during the study. Also, in a thirty-eight-year mortality data set for genets of the grass Bouteloua gracilis there was no effect of age on mortality (Fair et al. 1999). In another very long-term study (thirty-two years) of a large number of individuals of the short-lived orchid Ophrys sphegodes, the log-survivorship curve was linear over the plant’s life span (Hutchings 2010). In some species, the anatomy of belowground parts can be used to determine age. Such methods are particularly useful for species with extreme life spans. This method has been used to study senescence of individuals in the very long-lived species B. pyrenaica, with life spans of up to 300 years. Senescence was studied in this species using shoot scars on the rhizome to determine plant ages, and five years of demographic data were used to calculate vital rates (Garcia et al. 2011). The results of this study suggested that mortality stayed constant or possibly even decreased with age for old plants. These demographic results are consistent with physiological studies on this species that did not identify any deterioration of plant function with age (Morales et al. 2013; Onate et al. 2012). As in all studies that are primarily cross sectional, the patterns of mortality across different ages must be calculated using the fraction of individuals who survived to those ages (Hadfield 2007), but with these extremely longlived plant species, longitudinal studies from germination to death are clearly not possible.

Iteroparous Species Showing Mortality Senescence Studies with two herbaceous plant species directly support the hypothesis that individuals senesce, and there are a few studies in which indirect results may support this hypothesis (e.g. Tuomi et al. 2013; see also Chapter 16). With the short-lived species L. maritima, Pico and Retana (2008) followed 1,367 individuals from three-yearly cohorts and showed a gradually increasing mortality rate with age. A multiple-cohort long-term study was also done with the herbaceous perennial P. lanceolata. In this study, a large number of individuals, nearly 30 000, were planted and followed from establishment to death. These large numbers made it possible to estimate mortality precisely through to the latest ages. Four different cohorts were established over three years; thus, cohort effects could be separated from environmental effects on mortality by having different-aged individuals experiencing the same environment. The results showed that there is a large environmental influence on mortality (Roach 2003), particularly in response to changes in precipitation and temperature. Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:17:24, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.015

308

Demographic Senescence in Herbaceous Plants

Moreover, when the environment was favourable, all individuals of different ages showed the same pattern (Roach et al. 2009). However, individuals of older cohorts had higher mortality than younger individuals when mortality was high overall (Roach et al. 2009). Thus, there was an indication of senescence only when conditions were stressful. Condition-dependent ageing in P. lanceolata was also found with respect to growth rate such that older individuals showed negative growth; in other words, they were shrinking, whereas younger individuals had positive growth even during the observed periods of environmental stress (Roach 2012). The advantage of having different-aged individuals, in other words, multiple cohorts, experiencing the same environment in this study was demonstrated when these results were compared to a previous experiment with this same species. A covariate modelling analysis that included both intrinsic variables such as size and reproduction and extrinsic environmental variables such as precipitation and temperature but that only had data for one cohort could not identify any evidence for senescence (Roach & Gampe 2004). These results with P. lanceolata are of particular interest because of the detailed data and analyses. Observations of variation across cohorts in demographic patterns also highlight the problems associated with uncovering age-dependent patterns in populations with large environmental influences.

Conclusions Regarding Empirical Knowledge of Mortality Trajectories Only a few of the reviewed empirical studies report increases in mortality with age, and it is not clear whether these results reflect the true variability of senescence patterns across plant species. It is interesting to consider both the experimental protocol employed and the variation in the biology of the species studied to understand the variation in these patterns. Firstly, discerning senescence patterns from survivorship curves can be difficult because the functional form of the relationship between age and mortality can change with age (e.g. Dahlgren et al. 2011; Vaupel et al. 2004), and smaller increases in mortality with age for very old individuals can be hard to detect when looking at survivorship curves over all ages. This is particularly true because data for the oldest individuals are scarce in most studies both because few individuals in cohorts reach these ages and because the oldest individuals in most cases have life spans that are longer than the study period (but see Fair et al. 1999; Hutchings 2010). Moreover, the effects of heterogeneity on individual quality may obscure a deterioration with age if the oldest surviving individuals are of a higher quality overall and have a lower mortality (Vaupel & Yashin 1985). In indeterminately growing species, the effects of size may counteract ageing effects per se as long as individuals are still growing. Size effects therefore can obscure potential negative effects of age (Vaupel et al. 2004) unless they are corrected for. It may also be so that increases in size with age result in an escape from senescence for some period of the life cycle (this is discussed further later). Unfortunately, investigating age and size effects simultaneously has only rarely been done. Moreover, there are reasons to believe that senescence can be difficult to identify in natural populations because environmental effects, well known to affect vital rates, have to be accounted for. Environmental variation Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:17:24, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.015

Age Effects on Fecundity

309

can cause patterns that could mistakenly be assumed to be age effects, for example, if environments of individuals deteriorate over time. In studies not examining multiple cohorts, age and environment cannot be separated in this scenario. Spatial environmental heterogeneity may also make age effects hard to detect. Finally, the studies in which evidence for mortality senescence was found were among the ones where data were available for the entire life spans of the species. This may suggest that a lack of evidence in longer-lived species is due to data limitations and that more data are clearly needed.

Age Effects on Fecundity Senescence may, in addition to increased mortality, be manifest demographically as a decrease in fecundity with age. Furthermore, fecundity may be a good biomarker for general physiological decline associated with senescence (Monaghan et al. 2008), and it can be followed within individuals, unlike mortality, to avoid effects of amongindividual heterogeneity (Nussey et al. 2008). Fecundity trajectories that have been investigated in long-term monitoring studies of iteroparous herbaceous species show variable results (Table 15.2). In a fourteen-year study on G. pneumonanthe, flowering probability did not differ between age groups of mature plants, nor did it differ between plants already established at the start of the study and plants established during the study (Rose et al. 1998). Willems and Dorland (2000) report no relationship between age and flowering probability, flowering number or flowering stem height over a maximum of nine years from the start of flowering for the orchid Spiranthes spiralis (average age of flowering was 3.5 years). Using tuber anatomy to determine ages of individuals, Ehlers and Olesen (2004) found that flower number stayed constant with age in old individuals of Corydalis intermedia. Perkins et al. (2006) used root anatomy to determine plant ages, for Potentilla recta and reported no fecundity senescence. Using similar techniques, Garcia et al. (2011) reported that net reproductive output of both male and female plants increased with age in young B. pyrenaica but stayed roughly constant with age for old individuals.

Table 15.2 Studies Published between 1993 and 2014 Describing Fecundity Patterns in Herbaceous Perennials over Life Spans Species

Decrease in fecundity

Study

Borderea pyrenaica Corydalis intermedia Gentiana pneumonanthe Kosteletzkya pentacarpos Plantago lanceolata Potentilla recta Spiranthes spiralis Trillium grandiflorum

No No No Yes Yes No No No

Garcia et al. 2011 Ehlers & Olesen 2004 Rose et al. 1998 Pino & Roa 2007 Shefferson & Roach 2013 Perkins et al. 2006 Willems & Dorland 2000 Hanzawa & Kalisz 1993

Note: Whether or not a decrease in fecundity with age was reported is indicated.

Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:17:24, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.015

310

Demographic Senescence in Herbaceous Plants

There are also some studies reporting patterns suggesting fecundity senescence. A few early agronomical studies of grasses report a decrease in fecundity over age (reviewed in Roach 1993). T. grandiflorum was found to produce fewer reproductive shoots as individuals aged (Hanzawa & Kalisz 1993). In addition, van Dijk (2009) noted that some traits related to reproduction declined in Beta vulgaris as plants came close to the end of their life span. In such cases it may be interesting to determine whether patterns are due to age-dependent senescence or to terminal declines related to time to death. A recent analysis of the multiple-cohort study with P. lanceolata evaluated both age and time-to-death influences on fecundity (Shefferson & Roach 2013). The results showed that fecundity increased, for all cohorts, to a peak at age two and then declined with age. Additionally, the authors showed that there was both an age-dependent and an age-independent decline in fecundity as individuals approached death. Our understanding of reproductive declines with age and terminal reproductive declines as an individual approaches death is still very limited because most studies have not followed individuals for their entire life span. Again, we clearly need more data to understand the breadth of variation in fecundity senescence in herbaceous plants.

Survival–Fecundity Trade-Offs The existence of trade-offs between fecundity and plant vitality is a fundamental ecological assumption for which there is much evidence, although within-species tradeoffs can often be difficult to detect (e.g. Obeso 2002). For example, observations that non-semelparous annuals produce more seeds per biomass and year than perennials have been attributed to trade-offs between fecundity and life span (Silvertown & Charlesworth 2001). This notion of life history trade-offs is important in the context of senescence if reproduction has a cost in terms of decreased future reproduction and/or increased mortality. Evidence for the effects of trade-offs on senescence patterns has been found in animals (e.g. Good & Tatar 2001), and optimal resource allocation between reproduction and vitality is a major component of theories explaining how senescence evolved (see later). That reproduction can come at a cost of increased mortality for plants is clearly illustrated in semelparous species. A trade-off between survival and fecundity in these species, related to resource availability, is evident when experimental depression of seed production increases life span (Hautekeete et al. 2002; Noodén 1988). In addition, increased resource availability through fertilisation is reported to have caused iteroparous reproduction in some normally semelparous species (Young & Augspurger 1991). Fecundity–vitality trade-offs also exist in iteroparous species (Obeso 2002). For example, in P. annua, increased reproduction showed a cost in terms of decreased growth and thus decreased future reproduction (Law 1977), and in Podophyllum peltatum, reproduction has consequences for the growth pattern of future shoots (Geber et al. 1997). Life history trade-offs can also explain why iteroparous plants commonly show delayed reproduction (Silvertown & Charlesworth 2001). Miller et al. (2012) modelled Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:17:24, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.015

Connections with Theory

311

expected first flowering time using integral projection models (see also Childs et al. (2003) for monocarps) to show that the average time of first reproduction for a perennial orchid fit these patterns. These survival–fecundity trade-offs shape plant life histories and are likely to affect age trajectories of vital rates. One of the currently unanswered questions is whether survival–fecundity trade-offs change with age in herbaceous species.

Connections with Theory Theories on the Evolution of Senescence and Plants To understand whether senescence can occur in plants, it is important to compare empirical knowledge to predictions of theoretical evolutionary models. Classical evolutionary theories of senescence are based on the premise that selection will weaken with age (Medawar 1952). The antagonistic pleiotropy theory suggests that mutations giving rise to pleiotropic genes with effects that are beneficial early in life but detrimental late in life can be selected for (Williams 1957). The mutation accumulation theory predicts that a weakening selection pressure over age could also give rise to an accumulation of deleterious late-acting mutations over life (Hamilton 1966). The prediction from these classical theories is that additive genetic variance varies across ages, resulting in a genotype–age interaction. In accordance, a recent study with Silene latifolia has demonstrated an increase in the additive genetic variance of traits closely related to fitness (Pujol et al. 2014). The results of this study could not distinguish between mutation accumulation and antagonistic pleiotropy mechanisms, but this quantitative genetic experiment is a first step towards using plants to test these theories. Whereas agespecific demographic patterns can be used to indicate the presence or absence of senescence, these classic theories cannot be directly tested without genetics. Results from more realistic demographic models challenge the predictions of these classical senescence theories. Observations of indeterminate growth in herbaceous species such as P. lanceolata led to new models that have demonstrated that negligible, or even negative, senescence is possible for species that attain a size at reproductive maturity that is less than their maximum size (Vaupel et al. 2004). In addition, a declining selection pressure with age alone may not be sufficient for senescence to evolve in models that take trade-offs and interdependence among vital rates at different ages into account (Wachter et al. 2013; Wensink et al. 2014). Nevertheless, for species with germ lines that are separated from the somatic cell linages early in life, the ‘disposable-soma theory’ provides an explanation for how senescence will eventually occur as in this case there exists an age after which an investment of resources in offspring will increase fitness more than an investment in repairing somatic tissue (Kirkwood & Holliday 1979; see also Chapter 2). Models based on the disposable-soma theory have reliably predicted senescence patterns for humans and other mammals, and this theory is how senescence is currently understood to have been favoured by evolution in these species (Kaplan & Robson 2009). However,

Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:17:24, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.015

312

Demographic Senescence in Herbaceous Plants

a separation of germ cells early in life does not necessarily occur in plants. More general resource-allocation theories still apply, but a larger variation in age patterns is possible (Baudisch & Vaupel 2012). Next, we discuss several features of the biology of plants that may act against the evolution of senescence.

Unique Biology of Plants that May Cause Deviations from Theory Multiple Meristems Plants grow from multiple meristems, which are tissues consisting of undifferentiated ‘pluripotent’ stem cells. Even though the main meristems are formed during embryogenesis, new meristems typically can be formed throughout life (e.g. Laskowski et al. 1995; Long & Barton 2000). This implies that a germ cell line is not separated early in life, even in species that are not clonal, and this may be an important difference between plants and higher animals regarding the evolution of senescence (see Chapter 11 for a discussion of modular organisms). Growth from multiple meristems allows plants to be comprised of different modules that can be more or less independent – and even separate in some cases, which would result in clones (see below). Exactly what the proximate physiological causes of senescence would be is beyond the scope of this chapter (but see Chapter 13). Still, the modularity of plants may allow them to possibly cope with ‘stressed’ or damaged modules by ceasing to devote resources to damaged modules (cf. Thomas 2013). In any case, for senescence to occur, all meristems in an individual must deteriorate (e.g. Watson & Riha 2010). Given the diversity of plant growth forms and of plant life spans, the expectation that herbaceous plants will all show the same patterns of senescence does not appear to be realistic.

Clonality Many plants are to some extent clonal; in other words, they can reproduce vegetatively as well as sexually through the production of seeds. Some plants reproduce exclusively vegetatively over at least some parts of their range (one example of this is the primarily Southern European and Western Asian species Petasites hybridus, for which female plants are very rare in Northern Europe) (Mossberg & Stenberg 2003). Vegetative propagation can be achieved by the formation of specialised propagules or by fragmentation. Genetic individuals (‘genets’) of clonal species can live for a very long time, sometimes many thousands of years (e.g. Cook 1983), even though there is a turnover of physiological individuals (‘ramets’). Because of these long life spans, little is known about whether genets of clonal individuals deteriorate with age (e.g. de Witte & Stöcklin 2010), and in fact, the same is true for ramets. A heavy reliance on vegetative propagation may be a life history strategy leading to selection against senescence for the genet because even though clonality in itself may not give species the possibility to avoid senescence, clonal reproduction by separation of modules, leading to the formation of new main meristems, is essentially continuous somatic growth, which has been suggested as a mechanism of how negative senescence may evolve (Vaupel et al. 2004). Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:17:24, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.015

Connections with Theory

313

Theoretical models have shown that both genets (Gardener & Mangel 1997) and ramets (Orive 1995) may escape senescence in some but not all circumstances depending on their life histories. In these models, clonal reproduction was found to inhibit senescence but not necessarily preclude it if there was also sexual reproduction. Little is known about what actually occurs in nature. However, in one study on a clonal tree, quaking aspen (Populus tremuloides), genets were aged based on their genetic diversity and the use of a molecular clock, and a decline in pollen production was found with genet age but not with ramet age (Ally et al. 2010). Ageing effects in ramets can also be studied in tussock-forming species, even though in this case ramets are typically connected via rhizomes and do not form distinct physiological individuals. In a study of two tussock-forming herbaceous species, in which tussocks were aged based on annual rings in the rhizome, tussock age was positively related to ramet mortality and also to fecundity in one species (Münzbergova et al. 2005). In contrast, studies on grasses (e.g. Lauenroth & Adler 2008) that are also tussock forming suggest decreasing mortality rates with age for genets. In conclusion, there is theoretical and scant empirical evidence both for and against senescence of ramets and genets of clonal plants, but much is still unknown about the effect of clonality on the evolution of senescence.

Size Size has traditionally been regarded as better predictor of plant vital rates than age, and plant demographers typically describe populations in terms of size or stage structure rather than age structure (Caswell 2001). However, the relationship between size and demography is not always straightforward and may be confounded by short- or longterm growth patterns of individuals (Shefferson et al. 2014). Moreover, the fact that size, or some correlated measure of vitality, may be more important than age does not mean that there is not also an effect of age (e.g. Lauenroth & Adler 2008). However, if vital rates are mainly size dependent, then senescence may not evolve if size increases with age (Caswell & Salguero-Gómez 2013). This may be particularly true for species that grow continuously with age and thus have higher reproductive output with age. In other words, if an herbaceous species has indeterminate growth for some period of its adult life, then it may escape senescence for that period (Vaupel et al. 2004).

Sedentariness and Density Dependence Another feature of plants is that they are sedentary in the sense that established plants are typically permanently attached. This, in combination with density-dependent intraspecific competition, has been suggested to possibly counteract the evolution of senescence (Borges 2009; Seymour & Doncaster 2007). In particular, as density increases, reproductive values of older, larger individuals increase, as does their relative fitness, because younger individuals are out-competed. Selection would then favour high latelife fitness, which could act against the evolution of senescence in populations that typically occur at high densities. However, models showing this pattern assume a stable environment (Seymour & Doncaster 2007), and much is still unknown of how density dependence at different life stages affects plant populations. Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:17:24, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.015

314

Demographic Senescence in Herbaceous Plants

Future Directions The range of plant life history strategies and growth forms presents a diverse set of models for future studies to understand how trade-offs shape age trajectories of mortality and fecundity. If species are found that do not senesce, this lack of senescence may be due to the existence of multiple meristems, size effects or density dependence. To evaluate these hypotheses, rigorous demographic studies identifying age patterns of demography and physiology will be needed.

Age Determination In the majority of the reviewed studies, ages of plants have been known because individuals have been followed from germination. This method of determining ages has a clear advantage in that it is non-destructive. However, it requires long monitoring periods. A few demographic studies have used age rings in roots (e.g. Perkins et al. 2006) and other aspects of plant belowground morphology (e.g. Garcia et al. 2011) to determine ages of herbs. It has recently been shown that for a large number of species it is possible to determine ages of individuals based on root anatomy (Schweingruber & Poschlod 2005). Such methods are likely to become valuable in order to examine patterns of more species, particularly those with long life spans. Determining plant ages in this way will make it possible to take large samples from populations with stable age distributions and use these data to determine mortality trajectories (Dahlgren, Hellmann, Buentgen & Schweingruber, 2016). For extremely long-lived species, molecular techniques such as quantifying genetic diversity in neutral alleles (as used by Ally et al. 2010) may also be possible. Even though simply determining age trajectories for mortality and fecundity will not be sufficient to tell the full story of senescence, as we have shown here, more such studies are needed to determine general patterns.

Within-Species Studies There are additional approaches that we believe could prove particularly valuable for learning about senescence in plants. For example, we are aware of no studies that investigate plant senescence in multiple populations of the same species or over environmental gradients within populations. Such studies are needed to determine how age patterns of mortality and fecundity vary within species and to investigate, for example, effects of the environment and within-species genetic variability on these patterns. There are also other questions about environmental gradients; for example, herbs that occupy less nutrient-rich sites are usually more long lived (Nobis & Schweingruber 2013), but whether this is true also within species, and whether it is an effect of less pronounced senescence or of a lower mortality in poorer soils for individuals of all ages, remains to be discovered. Finally, the interactions of environment-by-age and environment-bygenetics-by-age have only minimally been explored (cf. Roach et al. 2009). To address these topics, experimental manipulations in the field will be required.

Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:17:24, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.015

Future Directions

315

In the design of future experiments, it is important to recognise that there may be longterm changes in the environment that can influence results. Even in the observational studies of natural populations over time, the length of the study period would have to be several times longer than the life span of the focal individuals if it is to be demonstrated that age patterns are in fact due to age-dependent effects or if the environment (possibly due to intra-specific effects) of different age groups differed. Temporal trends in vital rates of plants in cyclic environments are illustrative of how age effects could be confounded with environmental change. For example, in a study of Eryngium cuneifolium, germination occurred after fire, and with time, the environment deteriorated in the study population (Menges & Quintana-Ascencio 2004). Age effects are therefore not distinguishable from environmental effects in these data. Similar processes, possibly less dramatic, could lead to younger cohorts doing less well than older cohorts because of environmental change. For these reasons, cohort and age effects need to be separated (cf. Roach 2009).

Among-Species Studies Comparative demographic studies have already revealed much about the evolution of senescence across the Tree of Life (e.g. Jones et al. 2014) and have been used to study patterns within plants based on calculating age-specific parameters from size data (Baudisch et al. 2013; Horvitz & Tuljapurkar 2008; Silvertown et al. 2001). As more information on how plant vital rates depend on age becomes available, these types of studies will likely be extremely useful for determining general patterns and connections between age effects and traits relating to modularity, traits correlated with life history trade-offs and to the extent growth is determinate (versus indeterminate). As mentioned earlier, plant species are extremely variable in these regards, and among-species studies will likely be valuable to determine the effects. Finally, in the future, when information is obtained from more plant species, it will be particularly informative to do rigorous phylogenetic analyses of senescence patterns.

Linking Demography and Physiology In addition to these topics, it is time to go beyond mortality and fecundity and focus on age-specific physiological changes as an additional metric of ageing. Possible physiological metrics of ageing in plants may include age-specific analysis of functional traits such as growth rate and photosynthesis, changes in competitive ability, or changes in oxidative damage and hormone levels with age (see also Chapter 13). This may open up new ways of measuring senescence and provide insights into the mechanisms involved. Combining demographic and physiological approaches can provide stronger evidence for or against senescence, for example, as was done in the cross-sectional study with B. pyrenaica, where both types of analyses suggest no deterioration with age (Garcia et al. 2011; Morales et al. 2013; Onate et al. 2012). Physiological measurements in longitudinal studies would be particularly useful to connect physiological and demographic patterns over time and age. Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:17:24, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.015

316

Demographic Senescence in Herbaceous Plants

References Ally, D., Ritland, K. & Otto, S. P. (2010). Ageing in a long-lived clonal tree. PLoS Biology, 8, 1–8. Antonovics, J. (1972). Population dynamics of the grass Anthoxanthum odoratum on a zinc mine. Journal of Ecology, 60, 351–66. Baudisch, A. & Vaupel, J. W. (2012). Getting to the root of ageing. Science, 338, 618–19. Baudisch, A., Salguero-Gómez, R., Jones, O. R., et al. (2013). The pace and shape of senescence in angiosperms. Journal of Ecology, 101, 596–606. Borges, R. M. (2009). Phenotypic plasticity and longevity in plants and animals: cause and effect? Journal of Biosciences, 34, 605–11. Canfield, R. H. (1957). Reproduction and life span of some perennial grasses of southern Arizona. Journal of Range Management, 10, 199–203. Caswell, H. (2001). Matrix Population Models: Construction, Analysis and Interpretation (2nd edn.) (Sunderland, MA: Sinauer Associates). Caswell, H. & Salguero-Gómez, R. (2013). Age, stage and senescence in plants. Journal of Ecology, 101, 585–95. Childs, D. Z., Rees, M., Rose, K. E., et al. (2003). Evolution of complex flowering strategies: an age- and size-structured integral projection model. Proceedings of the Royal Society of London Series B: Biological Sciences, 270, 1829–38. Chu, C. & Adler, P. B. (2014). When should plant population models include age structure? Journal of Ecology, 102, 531–43. Cook, R. E. (1983). Clonal plant populations. American Scientist, 71, 244–53. Dahlgren, J. P., Garcia, M. B. & Ehrlén, J. (2011). Nonlinear relationships between vital rates and state variables in demographic models. Ecology 92, 1181–7. de Witte, L. C. & Stöcklin, J. (2010). Longevity of clonal plants: why it matters and how to measure it. Annals of Botany, 106, 859–70. Dahlgren, J. P., Rizzi, S., Schweingruber, F. H., Hellmann, L. & Büntgen, U. (2016). Age distributions of Greenlandic dwarf shrubs support concept of negligible actuarial senescence. Ecosphere, 7, e01521. Ehlers, B. K. & Olesen, J. M. (2004). Flower production in relation to individual plant age and leaf production among different patches of Corydalis intermedia. Plant Ecology, 174, 71–8. Fair, J., Lauenroth, W. K. & Coffin, D. P. (1999). Demography of Bouteloua gracilis in a mixed prairie: analysis of genets and individuals. Journal of Ecology, 87, 233–43. Garcia, M. B., Dahlgren, J. P. & Ehrlén, J. (2011). No evidence of senescence in a 300-year-old mountain herb. Journal of Ecology, 99, 1424–30. Gardener, S. H. & Mangel, M. (1997). When can a clonal organism escape senescence? American Naturalist, 150, 462–90. Geber, M. A., de Kroon, H. & Watson, M. A. (1997). Organ preformation in mayapple as a mechanism for historical effects on demography. Journal of Ecology, 85, 211–23. Good, T. P. & Tatar, M. (2001). Age-specific mortality and reproduction respond to adult dietary restriction in Drosophila melanogaster. Journal of Insect Physiology, 47, 1467–73. Hadfield, J. D. (2007). Estimating evolutionary parameters when viability selection is operating. Proceedings of the Royal Society of London Series B: Biological Sciences, 275, 723–34.

Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:17:24, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.015

References

317

Hanzawa, F. M. & Kalisz, S. (1993). The relationship between age, size, and reproduction in Trilliu grandiflorum (Liliaceae). American Journal of Botany, 80, 405–10. Hamilton, W. D. (1966). The moulding of senescence by natural selection. Journal of Theoretical Biology, 12, 12–45. Harper, J. L. & White, J. (1974). The demography of plants. Annual Review of Ecology and Systematics, 5, 419–63. Hautekèete, N.-C., Piquot, Y. & van Dijk, H. (2002). Life span in Beta vulgaris ssp. maritima: the effects of age at first reproduction and disturbance. Journal of Ecology, 90, 508–16. Horvitz, C. C. & Tuljapurkar, S. (2008). Stage dynamics, period survival, and mortality plateaus. American Naturalist, 172, 203–15. Hutchings, M. J. (2010). The population biology of the early spider orchid Ophrys sphegodes Mill: III. Demography over three decades. Journal of Ecology, 98, 867–78. Jones, O. R., Scheuerlein, A., Salguero-Gómez, R., et al. (2014). Diversity of ageing across the tree of life. Nature, 505, 169–73. Kaplan, H. S. & Robson, A. J. (2009). We age because we grow. Proceedings of the Royal Society of London Series B: Biological Sciences, 276, 1837–44. Kirkwood, T. B. & Holliday, R. (1979). The evolution of ageing and longevity. Proceedings of the Royal Society of London Series B: Biological Sciences, 205, 531–46. Lauenroth W. K. & Adler P. B. (2008). Demography of perennial grassland plants: survival, life expectancy and life span. Ecology, 96,1023–32. Laskowski, M. J., Williams, M. E., Nusbaum, H. C. & Sussex, I. M. (1995). Formation of lateral root meristems is a two-stage process. Development, 121, 3303–10. Law, R., Bradshaw, A. D. & Putwain, P. D. (1977). Life history variation in Poa annua. Evolution, 31, 233–46. Long, J. & Barton, M. K. (2000). Initiation of axillary and floral meristems in Arabidopsis. Developmental Biology, 218, 341–53. Mencuccini M., Martinez-Vilalta J., Hamid H. A., et al. (2005). Size-mediated ageing reduces vigour in trees. Ecology Letters, 8, 1183–90. Menges, E. S. & Quintana-Ascencio, F. (2004). Population viability with fire in Eryngium cuneifolium: deciphering a decade of demographic data. Ecological Monographs, 74, 79–99. Metcalf, J. C., Rose, K. E. & Rees, M. (2003). Evolutionary demography of monocarpic perennials. Trends in Ecology and Evolution, 18, 471–80. Miller, T. E. X., Williams, J. L., Jongejans, E., et al. (2012). Evolutionary demography of iteroparous plants: incorporating non-lethal costs of reproduction into integral projection models. Proceedings of the Royal Society of London Series B: Biological Sciences, 279, 2831–40. Medawar, P. B. (1952). An Unsolved Problem of Biology (London: Lewis). Monaghan, P., Charmantier, A., Nussey, D. H. & Ricklefs, R. E. (2008). The evolutionary ecology of senescence. Functional Ecology, 22, 371–8. Morales, M., Oñate, M. Garcia, M. B. and Munné-Bosch, S. (2013). Photo-oxidative stress markers reveal absence of physiological deterioration with ageing in Borderea pyrenaica, an extraordinarily long-lived herb. Journal of Ecology, 101, 555–65. Mossberg, B. & Stenberg, L. (2003). Den nya nordiska floran (Stockholm: Wahlström & Widstrand).

Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:17:24, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.015

318

Demographic Senescence in Herbaceous Plants

Münzbergova, Z., Krivanek, M., Bucharova, A., et al. (2005). Ramet performance in two tussock plants: do the tussock-level parameters matter? Flora, 200, 275–84. Nobis, M. P. & Schweingruber, F. H. (2013). Adult age of vascular plant species along an elevational land-use and climate gradient. Ecography, 36, 1076–85. Noodén L. D. (1988). Correlative controls of senescence and plant death in Arabidopsis thaliana (Brassicaceae). Journal of Experimental Botany, 52, 2151–9. Nussey D. H., Coulson T., Festa-Bianchet M. & Gaillard J.-M. (2008). Measuring senescence in wild animal populations: towards a longitudinal approach. Functional Ecology, 22, 393–406. Obeso, J. R. (2002). The costs of reproduction in plants. New Phytologist, 155, 321–48. Oñate, M., García, M. B. & Munné-Bosch, S. (2012). Age and sex-related changes in cytokinins, auxins and abscisic acid in a centenarian relict herbaceous perennial. Planta, 235, 349–58. Orive, M. E. (1995). Senescence in organisms with clonal reproduction and complex life histories. American Naturalist, 145, 90–108. Pedersen, B. (1999). Senescence in plants. In Life History Evolution in Plants (pp. 239–74) (Dordrecht: Kluwer). Perkins, D. L., Parks, C. G., Dwire, K. A., et al. (2006). Age structure and age-related performance of sulfur cinquefoil (Potentilla recta). Weed Science, 54, 87–93. Pico, F. X. & Retana, J. (2008). Age-specific, density-dependent and environment-based mortality of a short-lived perennial herb. Plant Biology, 10, 374–81. Pino, J. & Roa, E. (2007). Population biology of Kosteletzkya pentacarpos (Malvaceae) in the Llobregat delta (Catalonia, NE of Spain). Plant Ecology, 188, 1–16. Pujol, B., Marrot, P. & Pannell, J. R. (2014). A quantitative genetic signature of senescence in a short-lived perennial plant. Current Biology, 24, 744–7. Roach, D. A. (1993). Evolutionary senescence in plants. Genetica, 91, 53–64. Roach, D. A. (2003). Age-specific demography in Plantago: variation among cohorts in a natural plant population. Ecology, 84, 749–56. Roach, D. A. (2012). Age, growth and size interact with stress to determine life span and mortality. Experimental Gerontology, 47, 782–6. Roach, D. A. & Gampe J. (2004). Age-specific demography in Plantago: uncovering age dependent mortality in a natural population. American Naturalist, 164, 60–9. Roach, D. A, Ridley, C. E. & Dudycha, J. L. (2009). Longitudinal analysis of Plantago: age by environment interactions reveal aging. Ecology, 90, 1427–33. Rose, R. J., Clarke, R. T. & Chapman, S. B. (1998). Individual variation and the effects of weather, age and flowering history on survival and flowering of the long-lived perennial Gentiana pneumonanthe. Ecography, 21, 317–26. Sarukhán, J. & Harper, J. L. (1973). Studies on plant demography: Ranunculus repens L., R. bulbosus L. and R. acris L.: I. Population flux and survivorship. Journal of Ecology, 61, 675–716. Schweingruber, F. H. & Poschlod, P. (2005). Growth rings in herbs and shrubs: life span, age determination and stem anatomy. Forest Snow and Landscape Research, 79(3), 195–415. Seymour, R. M. & Doncaster, C. P. (2007). Density dependence triggers runaway selection of reduced senescence. PLoS Computational Biology, 3, e256.

Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:17:24, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.015

References

319

Shefferson, R. P. & Roach, D. A. (2013). Longitudinal analysis in Plantago: strength of selection and reverse age analysis reveal age-indeterminate senescence. Journal of Ecology, 101, 577–84. Shefferson, R. P., Warren, R. J., II & Pulliam H. R. (2014). Life history costs make perfect sprouting maladaptive in two herbaceous perennials. Journal of Ecology 102: 1318–28. Silvertown, J. & Charlesworth, D. (2001). Introduction to Plant Population Biology (Oxford: Blackwell Science). Silvertown J, Franco, M. & Perez-Ishiwara, R. (2001). Evolution of senescence in iteroparous perennial plants. Evolutionary Ecology Research, 3, 393–412. Solbrig, O. T., Newell, S. J. & Kincaid, D. T. (1980). The population biology of the genus Viola: I. The demography of Viola sororia. Journal of Ecology, 68, 521–46. Tamm, C. O. (1956). Further observations on the survival and flowering of some perennial herbs, part I. Oikos, 7, 273–92. Tamm, C. O. (1972a). Survival and flowering of perennial herbs: II. The behaviour of some orchids on permanent plots. Oikos, 23, 23–8. Tamm, C. O. (1972b). Survival and flowering of perennial herbs: III. The behavior of Primula veris on permanent plots. Oikos, 23, 23–8. Thomas, H. (2013). Senescence, ageing and death of the whole plant. New Phytologist, 197, 696–711. Tuomi, J., Crone, E. E., Gremer, J. R., et al. (2013). Prolonged dormancy interacts with senescence for two perennial herbs. Journal of Ecology, 101, 566–76. van Dijk, H. (2009). Ageing effects in an iteroparous plant species with a variable life span. Annals of Botany, 104, 115–24. Vaupel, J. W., Baudisch, A., Dolling, M., et al. (2004). The case for negative senescence. Theoretical Population Biology, 65, 339–51. Vaupel, J. W. & Yashin, A. I. (1985). Heterogeneity’s ruses: some surprising effects of selection on population dynamics. American Statistician, 39, 176–85. Wachter, K. W., Evans, S. N. & Steinsaltz, D. (2013). The age-specific force of natural selection and biodemographic walls of death. Proceedings of the National Academy of Sciences of the United States of America, 110, 10141–6. Watkinson, A. (1992). Plant senescence. Trends in Ecology and Evolution, 7, 417–20. Watson, J. & Riha, K. (2010). Telomeres, ageing, and plants: from weeds to Methuselah – a mini-review. Gerontology, 57, 129–36. Wensink, M. J., Wrycza, T. F. & Baudisch, A. (2014). No senescence despite declining selection pressure: Hamilton’s result in broader perspective. Journal of Theoretical Biology, 347, 176–81. Willems, J. H. & Dorland, E. (2000). Flowering frequency and plant performance and their relation to age in the perennial orchid Spiranthes spiralis (L.) Chevall. Plant Biology, 2, 344–9. Williams, G. C. (1957). Pleiotropy, natural selection, and the evolution of senescence. Evolution, 11, 398–411. Young, T. P. & Augspurger, C. K. (1991). Ecology and evolution of long-lived semelparous plants. Trends in Ecology and Evolution, 6, 285–9.

Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:17:24, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.015

16 Complex Life Histories and Senescence in Plants Avenues to Escape Age-Related Decline? Jennifer Gremer, Satu Ramula, Bård Pedersen, Elizabeth Crone, Peter Lesica, Anne Jäkäläniemi and Juha Tuomi

Short Summary ‘Demographic senescence’ is defined as an age-related decline in vital rates, such as survival and fecundity. For many organisms, such declines can be observed as individuals move through the life cycle, progressing forward through developmental stages with time. However, herbaceous perennial plants may grow and shrink repeatedly or even remain in the same size class for several years, resulting in complex life cycles that do not always progress linearly. In addition, the life cycle of many herbaceous perennials includes prolonged dormancy, in which plants remain alive below ground for one or more years without producing any aboveground biomass. Prolonged dormancy has the potential to interfere with senescence because dormant plants may have decreased metabolic activity, reduced exposure to elements and possible access to limiting resources while below ground. Here we review evidence for senescence in relation to prolonged dormancy in perennial herbs. We compare prolonged dormancy to other processes of developmental renewal or stasis in plants, such as seed dormancy, meristem dormancy and retrogression. We discuss the implications for understanding senescence, or the lack thereof, in herbaceous perennial plants.

Introduction Many organisms have simple life cycles, which progress forward through successive developmental stages through time. For these organisms, such as annual plants, senescence can be observed as the decline in vital rates (e.g. survival, fecundity) associated with later (i.e. older) stages. However, the life cycles of most plants are often more complex. For example, many herbaceous perennial plants die back to perennating belowground tissues every year and also have the ability to remain dormant, either as seeds or as germinated individuals. These plants not only may progress forward through life stages but also may remain in a given stage, a demographic process known as ‘stasis’, or return to previous stages of their life cycle (‘retrogression’). In other words, these plants may grow, Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:24:57, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.016

Introduction

321

shrink or remain unchanged for many years. While chronological age still increases through time for these plants, such patterns have the potential to interrupt or affect the process of senescence (Salguero-Gómez et al. 2013; Tuomi et al. 2013). Support for senescence in plants is mixed (Baudisch et al. 2013; Caswell & SalgueroGómez 2013; Garcia et al. 2011; Roach et al. 2009; Shefferson & Roach 2013; Silvertown et al. 2001; see also Chapters 15 & 20). Because germ and somatic cells are not differentiated early in the life of plants as they are in non-clonal animals, selection is expected to maintain the integrity of somatic tissues later in life for plants (Kirkwood 1977; Pedersen 1999; Roach 1993; Williams 1957). In addition, plants have modular growth, which can provide an escape from the senescence because new tissues can replace old tissues (Peñuelas & Munné-Bosch 2010; Roach 2004; Thomas 2002; see also Chapter 13). Moreover, as plants grow older, they often grow larger (Stephenson et al. 2014; see also Chapter 13). Since performance in plants is usually considered to be more strongly related to size than age (Bierzychudek 1982; Caswell 2001), this increase in size can lead to an increase in vital rates with age. Although senescence has been observed in several plant species, there is growing evidence that many plants, particularly herbaceous perennials, either do not senesce or exhibit negative senescence, which is when vital rates are observed to increase with age (Baudisch et al. 2013; Ehlers & Olesen 2004; Garcia et al. 2011; Silvertown et al. 2001; see also Chapter 20). In a recent analysis of 290 angiosperm species, Baudisch et al. (2013) demonstrated that 93 percent of species showed no evidence of senescence. The species that did exhibit senescence (Figure 16.1) were mainly trees and shrubs (i.e. phanerophytes), but herbaceous perennial plants generally did not (i.e. cryptophytes) (Baudisch et al. 2013). If herbaceous perennial plants escape senescence, how do they manage to do it? Herbaceous perennials often have complex life cycles that include stasis and retrogression, as well as dormant stages that may allow individuals to retard the process of senescence. In stasis, plants seem to remain unchanged in the same stage for one or more years. If associated vital rates, such as survival and fecundity, also remain constant during stasis, then senescence may be delayed. Retrogression represents transitions to smaller or less developed stages (e.g. non-reproductive classes) that are typically experienced earlier in the life cycle. If this return to earlier stages resets vital rates so that the plant’s expected future performance is similar to that expressed earlier in the typical life cycle, then retrogression can be an avenue for ‘resetting’ the senescence clock, or rejuvenation. Indeed, rejuvenation in plants has been reported in the past (Chen et al. 2013; Mencuccini et al. 2014). Therefore, stasis and retrogression may have important implications for the relationship between vital rates and age. In addition, many herbaceous perennial plants may enter a dormant stage, in which adult plants may remain below ground for one or more years (Lesica & Steele 1994; Shefferson 2009). This stage, known as ‘prolonged dormancy’ or ‘vegetative dormancy’, remains poorly understood (Reintal et al. 2010; Shefferson 2009; Shefferson et al. 2014) but has the potential to retard the process of senescence. In this chapter we present a demographic approach to studying senescence in plants with complex life histories and investigate possible avenues for escaping age-related decline in vital rates. We focus on prolonged dormancy here because it is relatively Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:24:57, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.016

322

Complex Life Histories and Senescence in Plants

70 Cryptophytes Hemicryptophytes Chamaephyte Epiphyte Phanerophyte

60

Number of species

50 40 30

20 10 0

Figure 16.1

0.135

0.173

0.211

0.249

0.287 0.325 Shape

0.362

0.400

0.438

Evidence for senescence in 290 species of plants (redrawn from Baudisch et al. 2013). Bars represent the frequency of shape values, which measure whether mortality increases with age (i.e. senescence) or not (i.e. negligible or negative senescence). Fills represent the distribution of shape values for different life forms: cryptophytes = shoot meristems below ground (rhizome, bulb and corm-producing plants); hemicryptophytes = shoot meristems at or near the surface (many rosetteforming plants); chamaephytes = shoot meristems < 25 cm above ground (succulents and shrubs); phanerophytes = shoot meristems > 25 cm above ground (shrubs and trees); and epiphytes = shoot meristems measured with respect to their position on the host plant. The dashed line represents the boundary for senescence to occur: senescence occurs for shape values to the right of this line, while negative or negligible senescence is observed to the left of this line (see Baudisch et al. (2013) for details). Perennial plants, which are usually cryptophytes and sometimes hemicryptophytes, did not show senescence in this analysis. Please note that for this study, vital rates were calculated from stage-based models without explicitly observing the age of individuals.

common in the life histories of herbaceous perennial plants (Reintal et al. 2010; Shefferson 2009), yet it has received little treatment in studies on senescence (but see Tuomi et al. 2013). Further, our study of prolonged dormancy illustrates a few potential mechanisms influencing senescence in plants whose life histories include stasis and retrogression. Thus, our approach has implications for understanding senescence in plants with other life histories, and we discuss examples of this later in this chapter. Finally, we conclude by discussing challenges in studying senescence in plants with complex life histories that include prolonged dormancy and suggest some future directions.

Prolonged Dormancy in Herbaceous Perennial Plants ‘Prolonged dormancy’ refers to dormant, non-seed stages that perennial plants spend below ground without re-sprouting during the subsequent growing season (Lesica &

Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:24:57, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.016

Prolonged Dormancy in Herbaceous Perennial Plants

323

Steele 1994; Reintal et al. 2010; Shefferson 2009). Such belowground periods may occur multiple times during an individual’s life, lasting from one to several years (Lesica & Steele 1994; Shefferson 2009). For example, the individuals of the perennial orchid Epipactis helleborine may stay dormant up to a remarkable period of eighteen years (reviewed in Reintal et al. 2010). In some species, prolonged dormancy seems to be a consequence of growth costs to future survival and therefore adaptive (Shefferson et al. 2014). Moreover, prolonged dormancy may increase the lifetime fitness of individuals in stochastic environments by enabling individuals to escape unfavourable conditions encountered above ground (Gremer & Sala 2013; Gremer et al. 2012; Shefferson 2009). Consequently, dormancy may be more common in long-lived perennials of temporally varying harsh environments. Prolonged dormancy has been reported in more than sixty perennial plant species and is only observed in geophytes, which are plants that die back each year to belowground organs (Lesica & Steele 1994). It has been most frequently reported among orchids (Reintal et al. 2010; Shefferson 2009), with the earliest observations dating back many decades (e.g. Rabotnov 1969; Tamm 1972). Still, the level of activity while plants are below ground is not well known (Gremer et al. 2010; Lesica & Steele 1994). Dormant plants respire (Wyka 1999) but may not be strongly metabolically or developmentally active during this phase. Therefore, prolonged dormancy has the potential to interfere with the process of senescence, which may have consequences for designing, carrying out and interpreting demographic studies. In life cycle graphs and demographic models, prolonged dormancy is often described as a single stage (Figure 16.2), making implicit assumptions about the effect of dormancy on individuals’ developmental and physiological processes that may lead to senescence (Tuomi et al. 2013). If the variability of vital rates is not modelled explicitly, a single dormant stage assumes, for instance, that all dormant individuals have identical vital rates regardless of their past history (i.e. the frequency of emergent and dormant

E1

E2

E3

E4

E5

D

Figure 16.2

Life cycle diagram traditionally used in demographic studies of plants with prolonged dormancy (redrawn from Tuomi et al. 2013). Nodes E1–E5 are emergent stages, and D is a single dormant stage. Transitions to previous stages represent retrogression. Circular arrows represent stasis, in which plants remain in the same stage for one or more years. Transitions representing reproduction while plants are emergent are not included in this diagram.

Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:24:57, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.016

324

Complex Life Histories and Senescence in Plants

periods) and that individuals moving from a dormant stage to aboveground stage(s) have identical vital rates to individuals that have never been dormant (Figure 16.2). Prolonged dormancy, when modelled as a single stage thus may enable individuals to stay dormant for any finite length of time or may enable them to rejuvenate and return to previous stages and vital rates experienced earlier in life. How does prolonged dormancy affect the vital rates and life span of individual plants? Survival and fecundity are particularly interesting because they are both expected to decline as a consequence of the physiological deterioration as individuals age. Only a few studies have explicitly investigated the relationship between prolonged dormancy and vital rates in perennial plants, mostly orchids (Table 16.1). All of these studies have observed vital rates under natural environmental conditions. While some studies have reported lower survival for individuals with dormant periods than for emergent individuals, a greater number have observed no relationship between plant survival and prolonged dormancy (Table 16.1). Relationships between fecundity (including flowering probability and reproductive output) and prolonged dormancy are also mixed, varying from positive to negative to no relationship (Table 16.1). Even fewer studies have examined the effect of prolonged dormancy on observed life span. In two orchid species, Ophrys sphegodes and Cypripedium calceolus, longer life spans were observed in individuals that spent periods in dormancy than in individuals that were emergent (Hutchings 1987; Shefferson et al. 2012) (Table 16.1). For C. calceolus, prolonged dormancy increased observed life spans regardless of experimental defoliation and shading of emergent individuals (Shefferson et al. 2012). These findings suggest that prolonged dormancy might retard developmental and physiological processes that lead to senescence and that plants could possibly slow senescence by becoming dormant. However, no relationship was found between prolonged dormancy and observed life span for the orchid Cephalanthera longifolia (Shefferson et al. 2012) (Table 16.1). In a comparative analysis of perennial plants, Salguero-Gómez and Casper (2010) found that short-lived perennials benefitted from prolonged dormancy, while prolonged dormancy was found to reduce fitness for long-lived perennials (as measured by the overall population growth rate), indicating that fitness consequences may depend on life span. Overall, these contrasting effects of prolonged dormancy on the survival, fecundity and life span of plants might be due in part to the short observation periods used (mean = 11.3 years; SD = 5.0) (Table 16.1) that rarely span the whole life of long-lived individuals but also may indicate that effects depend on environmental conditions and the resource status of individuals considered.

Prolonged Dormancy and Senescence To explore how prolonged dormancy may affect vital rates and plant senescence, we developed four demographic models. In the ‘true age model’, prolonged dormancy has no effect on plant performance, which simply depends on time (years) since germination regardless of the sequence of emergent and dormant periods experienced by the individuals. The performance of individual plants is thus described in the same way as in an age-classified life cycle graph, in which surviving individuals move to the next age class. Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:24:57, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.016

Prolonged Dormancy in Herbaceous Perennial Plants

325

Table 16.1 Effect of Prolonged Dormancy on Survival, Flowering Probability, Seed Production and Observed Life Span for Perennial Herbs

Species

Survival

Arnica angustifolia Astragalus scaphoides Cephalanthera longifolia Cleistes bifaria Corallorhiza odontorhiza Cypripedium calceolus Cypripedium calceolus ssp. parviflorum Cypripedium candidum Cypripedium parviflorum Cypripedium reginae Cypripedium × andrewsii Epipactis atrorubens Neotinea ustulata Ophrys sphegodes Silene spaldingii

0 0 – –

Flowering probability

No. of seeds

Life span

–a 0 – +

–

–

0 0 –a 0 0 – – 0

– /+ +

+

Study length (years)

Reference

7 25 7 13 11 7 7

Jäkäläniemi 2011 Gremer et al. 2012 Shefferson et al. 2012 Gregg & Kéry 2006 Shefferson et al. 2011 Shefferson et al. 2012 Shefferson 2003

11 11 11 11 8 13 9 19

Shefferson et al. 2006 Shefferson et al. 2006 Kéry & Gregg 2004 Shefferson et al. 2006 Jäkäläniemi et al. 2011 Shefferson & Tali 2007 Hutchings 1987 Lesica & Crone 2007

Indicated as a positive relationship (+), a negative relationship (–), no statistically significant (P > 0.05) relationship (0). a A tendency or marginally significant trend.

In the ‘sleeping beauty model’, prolonged dormancy retards developmental and physiological processes that may lead to senescence, and the state of individuals remains unchanged when dormant. Therefore, only the number of emergent years since germination is expected to affect plant performance and senescence. While processes that may lead to senescence are retarded in the sleeping beauty model, they are completely reset in the ‘reboot model’. This third model allows rejuvenation multiple times during an individual’s life span because after each dormant period it resets individuals to a postdormant state that does not depend on their previous history. In other words, all plants emerge from each bout of dormancy at the same demographic state regardless of how many years they had actually lived before a bout of dormancy. For instance, increased flowering probability after prolonged dormancy in the orchid Ophrys sphegodes might reflect rebooting (Hutchings 1987). Our last model is the ‘active dormancy model’, in which plant performance depends on both the number of dormant and emergent years, but the rate and pattern of ageing differ between dormant and emergent years. In this model, prolonged dormancy interacts with developmental and physiological processes that may either accelerate or retard senescence. We then fit each of these four models to long-term demographic data of two perennial herbs, A. scaphoides and S. spaldingii. The data consisted of more than twenty years, which roughly covers the mean expected life span of both species (twenty-one years for Astragalus and twelve years for Silene). The two species showed contrasting patterns of

Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:24:57, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.016

326

Complex Life Histories and Senescence in Plants

senescence in terms of individual vital rates and reproductive values, with the latter being a function of survival and fecundity. Astragalus individuals seemed to senesce as their annual survival and reproductive value declined with age (Figure 16.3). Silene showed negative senescence, with reproductive value increasing with age (Figure 16.3). The reason that Silene showed increasing reproductive value with age is primarily that older plants were more likely to flower and produced more seed pods per flowering plant. For both species, the reboot model provided the best fit for three of four vital rates examined. A good fit of the reboot model indicates that prolonged dormancy interferes with senescence and resets developmental and physiological processes, slowing the senescence of individual plants and ‘rejuvenating’ them.

Possible Mechanisms for the Delay of Senescence via Prolonged Dormancy The precise mechanisms by which prolonged dormancy interferes with senescence are not known, but there are several possibilities. It is possible that the rejuvenating effects of prolonged dormancy that were observed for A. scaphoides and S. spaldingii are the result of slowed physiological processes during prolonged dormancy. In animals, it has been shown that reduced metabolism and delayed growth can slow senescence (Masoro 2005). For instance, reproductive dormancy has been observed in wild-type Drosophila melanogaster individuals (Tatar et al. 2001). Upon leaving this quiescent stage, survival of these older flies was similar to that of newly emerged young flies, suggesting that reproductive dormancy resulted in rejuvenation (Tatar et al. 2001). However, post-dormancy reproduction was lower for the flies that entered reproductive dormancy, which indicates that there was a cost to reproduction associated with that increase in survival. Similar costs have been observed for A. scaphoides, since individuals that spent more of their lifetime in prolonged dormancy had higher survival, particularly in smaller stages, but also lower annual fruit set (Gremer et al. 2012). In contrast, S. spaldingii showed no such costs. Dormant S. spaldingii individuals had similar survival probabilities to plants that had emerged above ground but were more likely to flower than emergent plants (Lesica & Crone 2007). These patterns are consistent with lower senescence rates in A. scaphoides, but they are somewhat more puzzling in S. spaldingii because reproduction generally increases with age. A better understanding of physiological processes occurring during prolonged dormancy may explain the different impacts of prolonged dormancy on senescence in herbaceous perennial plants. Experiments have shown that caloric restriction in animals can slow physiological deterioration that leads to senescence (Colman et al. 2014; Masoro 2005; Sohal & Weindruch 1996; but see Mattison et al. 2012). These patterns are fairly consistent across a number of taxa, and research suggests that reduced caloric intake results in longer life spans (reviewed in Masoro 2005). The precise underlying mechanisms in animals remain unclear, but it may be that the reduction in metabolism associated with a restricted diet decreases damage from free radicals (Finkel & Holbrook 2000; Masoro 2005). Since plants in prolonged dormancy are not active above ground, their resource acquisition is likely to be restricted. If so, then prolonged dormancy Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:24:57, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.016

Prolonged Dormancy in Herbaceous Perennial Plants

327

Astragalus scaphoides (a) True age

(c) Reboot

(d) Active dormancy

1 0

Reproductive value

2

3

4

5

(b) Sleeping beauty

5

10

15

20

5

10

15

20

5

10

15

20

5

10

15

20

Silene spaldingii (f) Sleeping beauty

(g) Reboot

(h) Active dormancy

0 10 20 30 40 50

(e) True age

5

10

15

20

5

10

15

20

5

10

15

20

5

10

15

20

Age (years)

Figure 16.3

Age-specific reproductive value of A. scaphoides and S. spaldingii calculated for four different demographic models (see Tuomi et al. (2013) for details). Also, 95 per cent confidence intervals were calculated using parametric bootstrapping with 1,000 iterations. For the reboot model of Silene, confidence intervals increase exponentially, and therefore, only the first fifteen years are presented. For both species and most vital rates, the reboot model (panels C and G in top and bottom rows, respectively) provided the best fit to the data.

could act to restrict resource intake and slow the process of senescence in similar ways as in animals. It is not known whether dormant plants acquire resources while below ground, either through direct absorption or with the aid of microbial symbionts. Gremer et al. (2010) suggested that plants may remobilise structural carbon into available forms during prolonged dormancy, indicating that metabolism proceeds during dormancy, though it may still be at lower levels than for emergent plants. However, the reduction in photosynthesis associated with prolonged dormancy may provide benefits because chloroplasts are the major source of reactive oxygen forms, or free radicals, in plants (Taiz & Zeiger 2006; see also Chapter 13). In an experimental study, Shefferson et al. (2005) found that shading increased survival relative to control plants in one orchid species, C. longifolia, but for another species, C. calceolus, shading had no effect. It is unclear whether these patterns were a result of limited photosynthesis, reduced resource acquisition or some other mechanism. Future studies that reduce available nutrients as well as photosynthetic carbon gain may lend insight into processes related to metabolism and oxidative damage.

Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:24:57, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.016

328

Complex Life Histories and Senescence in Plants

Similarly, the benefits of prolonged dormancy for slowing senescence may result from the safety of being below ground. The ‘damage accumulation hypothesis’ for senescence suggests that environmental quality can affect the timing and speed of senescence (Dahlgren & Roach 2016; Martin & Festa-Bianchet 2011; McNamara et al. 2009; Ricklefs 2000; Shefferson & Roach 2013; see also Chapters 14 and 15). According to this hypothesis, an individual’s history, including past reproductive events and exposure to environmental hazards, affects its ability to respond to current environmental conditions. Thus, a decline in physiological condition from past events results in reduced survival and fecundity through time. A longitudinal study on Plantago lanceolata has provided support for this hypothesis in plants. In this study, Roach et al. (2009) showed an age-by-environment interaction in which older individuals were more strongly affected by environmental stress than younger ones. Prolonged dormancy has been demonstrated to buffer plants from the risks of variable environments and unfavourable conditions (Gremer & Sala 2013; Gremer et al. 2012; Reintal et al. 2010; Shefferson 2009; Shefferson et al. 2012). If so, prolonged dormancy may slow the process of senescence by decreasing exposure to damage that may accelerate the decline in vital rates through time. While our observational study of prolonged dormancy in two long-lived perennial herbs (Tuomi et al. 2013) has revealed patterns that are consistent with these possible mechanisms of retarding senescence, we cannot determine causation. Indeed, it has also been suggested that prolonged dormancy may be a symptom of senescence (Shefferson 2009). Consistent with this idea, rates of prolonged dormancy increase with age for both A. scaphoides and S. spaldingii (Tuomi et al. 2013). However, prolonged dormancy is also common among young plants of the same species, suggesting that it is not simply a sign of senescence in older plants (Tuomi et al. 2013). Further study and experimentation that supplement long-term studies are needed to identify the relationship between senescence and prolonged dormancy in these and other species.

Senescence in Relation to Other Dormant Stages Other dormant stages in plant life cycles have the potential to interfere with senescence. For many species, dormant seeds in persistent seed banks can remain viable for years, decades and possibly even centuries (Darlington & Steinbauer 1961; Moriuchi et al. 2000). Such patterns suggest that some dormant seeds do not senesce, or if they do, they do so rather slowly. However, senescence processes associated with seed dormancy remain poorly understood as few studies have examined how time in the seed bank affects vital rates later in life. Indeed, in the preceding examples, we defined ages of Astragalus and Silene individuals based on the number of years since first emergence above ground (Tuomi et al. 2013). Rice and Dyer (2001) examined the performance of Bromus tectorum seedlings that emerged from seeds of different ages (one to four years old). In one of their populations, older seeds germinated later and had lower biomass and reduced competitive ability. If these patterns scale to affect survival and reproduction, it

Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:24:57, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.016

Prolonged Dormancy in Herbaceous Perennial Plants

329

may be that the ageing clock already begins to tick during the seed stage and continues through the rest of the life cycle. Age structure of seed banks past the first year of dormancy is rarely included in demographic models (Saatkamp et al. 2009; but see Kalisz & McPeek 1993). Future studies that examine age structure in seed banks and include the age of seeds in estimates of life span and vital rates will inform demographic models of senescence. Dormant meristems may offer another possible avenue for rejuvenation. This can occur because meristems can remain indeterminate and maintain the ability for cellular division and differentiation through time (Munné-Bosch 2008; Peñuelas & MunnéBosch 2010; see also Chapter 13). In other words, meristems do not seem to age. Rejuvenation through dormant meristems may occur in some polycarpic perennial herbs, in which meristems are produced at different times and die asynchronously without gradual deterioration of the genet’s performance with age. These species thus may be able to avoid the negative consequences of the accumulation of deleterious mutations and/or structural and physiological constraints of growth leading to senescence. For example, grass species may maintain meristems of different ages, thus creating ‘bud banks’ that may buffer individuals from environmental hazards (Nilsson et al. 1996; Ott & Hartnett 2012). These types of bud banks may also provide rejuvenation through the production of new tissues, particularly if bud age does not affect later performance. There is little research that has directly examined meristem age with performance in herbaceous perennials, though studies on woody species challenge the idea that meristems age (Lanner & Connor 2001; Mencuccini et al. 2014; Munné-Bosch 2008; Oñate & Munné-Bosch 2008; but see Ally et al. 2010). For instance, in a study of a dune shrub Cistus clussi, Oñate and Munné-Bosch (2008) showed that cuttings from older plants demonstrated physiological rejuvenation despite having older meristems. These patterns suggest that meristem age did not affect performance and that meristems do not accumulate damage through time, but more studies are needed, particularly in herbaceous species. Likewise, plants that emerge from prolonged dormancy may experience rejuvenation by re-sprouting from dormant meristems. However, it is unclear why such processes would differ from plants emerging from typical winter dormancy. Other processes occurring during prolonged dormancy, such as the acquisition of limiting resources, may explain the difference (Gremer et al. 2010; Lesica & Crone 2007). For instance, Lesica and Crone (2007) suggested that S. spaldingii individuals that enter prolonged dormancy may be limited by nutrients, such as nitrogen or phosphorus, and may be able to acquire such nutrients while dormant. This may be particularly true for species that have mutualisms that can provide resources during dormancy, that is, species that rely on mycorrhizal, rhizobial or other microbial associations for nutrition (Shefferson 2009). Together these patterns indicate that understanding how and why rejuvenation differs between prolonged dormancy and meristem activation for both clonal and non-clonal plants would provide necessary insight into the process of senescence, or the slowing of it, during different types of dormancy.

Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:24:57, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.016

330

Complex Life Histories and Senescence in Plants

Senescence in Relation to Retrogression in Perennial Herbs ‘Retrogression’ is the return to earlier developmental stages or smaller size classes of the life cycle; in plants, this usually implies returning to smaller or non-reproductive stages (Silvertown et al. 1993). Since size is often considered a good measure of performance in plants (Caswell 2001), retrogression can represent a decline in individual function and thus may be evidence of senescence (Caswell 2001; Roach 2004). This seems to be true for P. lanceolata, a short-lived perennial plant, in which there was a decline in size over a period of three years before death, as well as a decline in reproduction (Shefferson & Roach 2013). Similarly, in a study on two perennial bunchgrasses, Dalgleish et al. (2011) showed that retrogression was more likely once plants reached a size threshold later in life and that retrogression often occurred before mortality. These patterns suggest that retrogression, or shrinkage, may be a sign of senescence and does not lead to rejuvenation. However, it has been shown that plants can rejuvenate when their size is artificially reduced through trimming (Caswell & Salguero-Gómez 2013; Hackett 1985; Wendling et al. 2014). This may be particularly true for herbaceous perennial plants, and shrinkage may slow the process of senescence. For Hiliaria mutica, a bunchgrass of the Chihuahuan Desert, retrogression was more important during dry years (Vega & Montana 2004), and defoliation led to increased growth in two grass species in Argentina (Becker et al. 1997). This suggests that retrogression may aid in stress response and increase longevity, though it could also be a consequence of environmental conditions without adaptive value. In a study that included eighty species of herbaceous perennial plants, retrogression to smallerstage classes was shown to result in higher survival and greater demographic buffering potential (Salguero-Gómez & Casper 2010). In the same study, the authors found that the effects of retrogression (shrinkage) on fitness were weaker for plants whose life cycles included prolonged dormancy and stronger for those without prolonged dormancy (Salguero-Gómez & Casper 2010). It may be that the benefits of retrogression and prolonged dormancy are similar, though the associated mechanisms remain unclear.

Challenges Defining senescence is the first challenge for plants with complex life cycles because it can refer to either lifetime senescence (i.e. decline in vital rates at older ages) or a decline in vital rates in relation to a previous rejuvenating event. In the latter case, senescence may not be a continuous decline of vital rates at older ages but rather a fluctuating process in which the declines and increases of vital rates may alternate within the same individual depending on their demographic state. Using demographic models constructed with different underlying assumptions of prolonged dormancy and senescence, it is possible to distinguish these two alternatives and examine both simultaneously (Tuomi et al. 2013). However, our modelling approach is not completely without

Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:24:57, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.016

Challenges

331

problems as it is based on assumptions that may be difficult to validate in practice. For instance, the reboot model assumes that rejuvenation associated with prolonged dormancy is independent of plant age and previous history (e.g. resource status) (Tuomi et al. 2013) and that the rejuvenating effect itself remains constant over the whole life of individuals. In other words, the reboot model assumes that rejuvenation works similarly in younger and in older individuals without diminishing as individuals age. This assumption ignores the fact that prolonged dormancy early in life might be adaptive, while it might be a sign of senescence later in life. Examining how prolonged dormancy affects senescence of perennial herbs requires data on the vital rates of individual plants in relation to both their age and demographic fate (e.g. Tuomi et al. 2013). This may seem a straightforward task, but there are often practical constraints related to data. Ideally, individuals should be followed annually during their whole life cycle from birth to death and preferably in multiple populations. Since the predicted average life span of many perennial herbs can be up to several decades (Ehrlén & Lehtilä 2002), such long-term data are labour-intensive to collect and therefore rare. For some plant species, it might be possible to define the age of individuals retrospectively from roots or tubers without observing the birth (Garcia et al. 2011; Jäkäläniemi et al. 2004; Schaal & Levin 1976). In addition to the lack of long-term demographic data, information on local environmental conditions is also scarce. Recording biotic and abiotic conditions together with the age and demographic fate of individuals would be necessary for a deeper understanding of environmental context in the process of plant senescence (see Chapter 3). When long-term demographic data are not available, it might be possible to use shorter-term data for studies on the role of prolonged dormancy in the process of senescence. Recently compiled demographic databases provide a valuable data source for comparative studies. For instance, the data set published by Ellis et al. (2012) contains long-term demographic data of twenty perennial herbs, and the COMPADRE Plant Matrix Database (www.compadre-db.org) provides demographic models for over 600 plant species, including species with prolonged dormancy. Although these databases contain mainly size- or stage-based matrices without explicit records of plant age, researchers have used such data to infer age from stage transitions (Baudisch et al. 2013; Caswell & Salguero-Gómez 2013; see also Cochran and Ellner 1992). This approach is a powerful way to make inferences about long-term ageing from shortterm data. However, it may be biased against detecting senescence because vital rates are assumed to be similar for all members of the same stage or size class. In other words, it is possible for this approach to detect patterns where plants become smaller with age, but it is not possible to test for differences between young small plants and old small plants. Therefore, analyses based on stage-based models may not reveal senescence unless the age of individuals is explicitly followed together with their demographic fate. Age is rarely recorded for perennial herbs, in which individuals are characterised based on their size or life stage (but see e.g. Caswell & Salguero-Gómez 2013; Chu & Adler 2014).

Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:24:57, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.016

332

Complex Life Histories and Senescence in Plants

Future Directions Empirical evidence on the effect of prolonged dormancy on plant senescence is mixed (Table 16.1) and mostly based on correlative studies on vital rates and observed life spans during a limited period of individuals’ life. So far we are lacking experimental studies that are necessary to determine the consequences of prolonged dormancy for the process of senescence (if any) and to distinguish among size-, age- and environmentdependent responses of individuals. For instance, it is unclear whether the effects of dormancy change throughout the life cycle, which would require comparing fitness among plants that experienced more bouts of dormancy at different stages (e.g. early in life versus later). Techniques to experimentally manipulate the occurrence of prolonged dormancy, without altering the resource status of individuals or environmental conditions, are needed. Finding methods to do so, such as the use of hormones, as done in seed dormancy research with abscisic acid and giberrellins (Finch-Savage & LeubnerMetzger 2006), would greatly improve our understanding of how age, size and environmental conditions mediate the effects of prolonged dormancy on fitness and senescence. If these methods can be developed, then vital rates and life spans of treated individuals could be compared to those of controls (preferably using siblings under controlled environments) in order to determine the fitness consequences of prolonged dormancy. A similar experimental approach could be applied to investigate the effect of retrogression on plant senescence, as clipping returns individuals to an earlier (smaller) stage of the life cycle. However, clipping also alters the resource status of individuals through the loss of biomass, so methods to mitigate altered resource states, such as manipulating nutrient or water availability simultaneously with clipping, would be necessary. Such long-term manipulative studies on the fitness consequences of prolonged dormancy and retrogression may be challenging but would provide necessary insight into the mechanisms associated with patterns of senescence, or lack thereof, in herbaceous perennials with complex life histories. Classical evolutionary theories of senescence are based on the age-structured life cycle, where performance is seen as a function of age. In these models, age represents the organism’s physiological and developmental progress towards maturity and possible senescence later in the life together with the associated changes in survival and reproduction. As the age of an individual necessarily increases with time, there is no opportunity in the age-structured model for rejuvenation in the sense that individuals return to a state experienced earlier in their lives and thereby attain the same expected future performance as last time entering that state. The only mechanism that resets the age clock in this way is reproduction. Complex life cycles of perennial plants challenge this framework by suggesting that individuals may be able to rejuvenate and transfer to earlier stages in the life cycle, as discussed earlier. However, we are not aware of a theoretical framework to investigate the evolution of senescence in plants with complex life cycles. As pointed out by Baudisch et al. (2013), the classical agestructured approach may not be well suited for developing evolutionary theories of senescence in plants having modular, plastic architecture. Instead, demographic models

Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:24:57, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.016

Future Directions

333

using both the size and age of individual plants as predictors might provide a basis for developing evolutionary theories of senescence. Caswell and Salguero-Gómez (2013) developed an extension of a traditional matrix model that in addition to a stage structure enables the inclusion of an age structure, as well as the interaction between stage and age. In addition, an individual’s history is likely to affect vital rates (see Chapter 4) and methods to allow current performance to depend on previous states in matrix models are relatively straightforward (Ehrlén 2000; Goodman 1969; Shefferson et al. 2014). Alternatively, integral projection models can be used for linking vital rates to multiple continuous individual-based variables (e.g. size and age) (Ellner & Rees 2006; Easterling et al. 2000). In these models, links between the characteristics and demographic fates of individuals are established through statistical models (Easterling et al. 2000). For instance, in their integral projection models, García et al. (2011) modelled vital rates as non-linear functions of both the size and age of individuals when examining the senescence of the long-lived dioecious herb Borderea pyrenaica. A clear advantage of integral projection models is the possibility to include environmental covariates (e.g. rainfall, temperature and herbivores) and investigate their effect on age-related declines of vital rates simultaneously with the size, age and history of individuals (Kuss et al. 2008; Merow et al. 2014). However, a traditional measure of size adopted in demographic studies (e.g. height, rosette diameter and number of leaves) might be a misleading indicator of the resource status of individual plants. As an example, there could be ten emergent shoots per plant before dormancy, but only two of them may become active after dormancy, gaining a greater share of common resources than the emergent shoots before dormancy. In addition, Shefferson (2009) suggested that prolonged dormancy, particularly in orchids, is an evolutionary trajectory towards increasing dependence on belowground symbionts, such as mycorrhizae, for resources. If so, then vital rates of plants may be more dependent on belowground condition and symbiotic interactions than aboveground state. Therefore, we might need to consider alternative ways to classify individuals (e.g. based on the number of dormant and active meristems or belowground size) in order to describe and assess their resource status. If such information is critical for determining vital rates of individuals, current demographic approaches may not be detailed enough to investigate plant senescence. In the worst case, they might even result in erroneous conclusions of the mechanisms of plant senescence by partly masking the role of resource status. Exploring resource sharing among meristems and shoots within individual plants, as well as symbiotic interactions, and their connections to demography is highly relevant for explaining diverse patterns observed in the senescence of perennial plants (Salguero-Gómez et al. 2013). Our exploration of complex life histories and senescence also highlights the need for a better understanding of the physiology during different stages. Future research should focus on revealing potential mechanisms of rejuvenation behind different types of dormancy and retrogression for perennial plants. Regarding prolonged dormancy, better understanding of metabolic processes that may take place during dormancy is essential. Further, investigating whether resource acquisition occurs during dormancy and comparing these patterns among species will provide insight into the causes of variation in Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:24:57, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.016

334

Complex Life Histories and Senescence in Plants

patterns of senescence and rejuvenation. We encourage similar directions for understanding how meristem dormancy and retrogression relate to rejuvenation. Modern techniques, including stable isotopes, are promising tools for understanding metabolism, resource acquisition and allocation in relation to these different life histories because researchers can follow the movement of nutrients and substrates in plants (Dawson et al. 2002). Such studies would provide a mechanistic understanding of senescence in plants and the possible avenues to escape age-related decline in vital rates.

References Ally, D., Ritland, K. & Otto, S. P. (2010). Aging in a long-lived clonal tree. PLoS Biology, 8(8). Baudisch, A., Salguero-Gómez, R., Jones, O. R., et al. (2013). The pace and shape of senescence in angiosperms. Journal of Ecology, 101(3), 596–606. Becker, G. F., Busso, C. A., Montani, T., et al. (1997). Effects of defoliating Stipa tenuis and Piptochaetium napostaense at different phenological stages: tiller demography and growth. Journal of Arid Environments, 35(2), 251–68. Bierzychudek, P. (1982). Life histories and demography of shade-tolerant temperate forest herbs: a review. New Phytologist, 90(4), 757–76. Caswell, H. (2001). Matrix Population Models: Construction, Analysis, and Interpretation (Sunderland, MA, Sinauer Associates). Caswell, H. & Salguero-Gómez, R. (2013). Age, stage and senescence in plants. Journal of Ecology, 101(3), 585–95. Chen, Y. T., Shen, C. H., Lin, W. D., et al. (2013). Small RNAs of Sequoia sempervirens during rejuvenation and phase change. Plant Biology, 15(1), 27–36. Chu, C. & Adler, P. B. (2014). When should plant population models include age structure? Journal of Ecology, 102(2), 531–43. Cochran, M. E. & Ellner, S. (1992). Simple methods for calculating age-based life-history parameters for stage-structured populations. Ecological Monographs, 62(3), 345–64. Colman, R. J., Beasley, T. M., Kemnitz, J. W., et al. (2014). Caloric restriction reduces age-related and all-cause mortality in rhesus monkeys. Nature Communications, 5. Dalgleish, H. J., Koons, D. N., Hooten, M. B., et al. (2011). Climate influences the demography of three dominant sagebrush steppe plants. Ecology, 92(1), 75–85. Darlington, H. & Steinbauer, G. P. (1961). Eighty-year period for Dr. Beal’s seed viability experiment. American Journal of Botany, 48(4), 321. Dawson, T. E., Mambelli, S., Plamboeck, A. H., et al. (2002). Stable isotopes in plant ecology. Annual Review of Ecology and Systematics, 33, 507–59. Easterling, M. R., Ellner, S. P. & Dixon, P. M. (2000). Size-specific sensitivity: applying a new structured population model. Ecology, 81(3), 694–708. Ehlers, B. K. & Olesen, J. M. (2004). Flower production in relation to individual plant age and leaf production among different patches of Corydalis intermedia. Plant Ecology, 174(1). Ehrlén, J. (2000). The dynamics of plant populations: does the history of individuals matter? Ecology, 81(6), 1675–84.

Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:24:57, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.016

References

335

Ehrlén, J. & Lehtilä, K. (2002). How perennial are perennial plants? Oikos, 98(2), 308–22. Ellis, M. M., Williams, J. L., Lesica, P., et al. (2012). Matrix population models from 20 studies of perennial plant populations. Ecology, 93(4), 951. Ellner, S. P. & Rees, M. (2006). Integral projection models for species with complex demography. American Naturalist, 167(3). Finch-Savage, W. E. & Leubner-Metzger, G. (2006). Seed dormancy and the control of germination. New Phytologist, 171(3), 501–23. Finkel, T. & Holbrook, N. J. (2000). Oxidants, oxidative stress and the biology of ageing. Nature, 408(6809), 239–47. Garcia, M. B., Dahlgren, J. P. & Ehrlén, J. (2011). No evidence of senescence in a 300-year-old mountain herb. Journal of Ecology, 99(6), 1424–30. Goodman, L. A. (1969). Analysis of population growth when birth and death rates depend upon several factors. Biometrics, 25(4), 659. Gremer, J. R., Crone, E. E. & Lesica, P. (2012). Are dormant plants hedging their bets? Demographic consequences of prolonged dormancy in variable environments. American Naturalist, 179(3), 315–27. Gremer, J. R. & Sala, A. (2013). It is risky out there: the costs of emergence and the benefits of prolonged dormancy. Oecologia, 172(4), 937–47. Gremer, J. R., Sala, A. & Crone, E. E. (2010). Disappearing plants: why they hide and how they return. Ecology, 91(11), 3407–13. Hackett, W. P. (1985). Juvenility, maturation, and rejuvenation in woody plants. Horticultural Reviews, 7, 109–55. Hutchings, M. J. (1987). The population biology of the early spider orchid, Ophrys sphegodes Mill: II. Temporal patterns in behaviour. Journal of Ecology, 75, 729–42. Jäkäläniemi, A. (2011). Narrow climate and habitat envelope affect the survival of relict populations of a northern Arnica Angustifolia. Environmental and Experimental Botany, 72(3), 415–21. Jäkäläniemi, A., Crone, E. E., Närhi, P. & Tuomi, J. (2011). Orchids do not pay costs at emergence for prolonged dormancy. Ecology, 92(7), 1538–43. Jäkäläniemi, A., Kauppi, A., Pramila, A. & Vähätaini, K. (2004). Survival strategies of Silene tatarica (Caryophyllaceae) in riparian and ruderal habitats. Canadian Journal of Botany, 82(4), 491–502. Kalisz, S. & McPeek, M. A. (1993). Extinction dynamics, population-growth and seed banks: an example using an age-structured annual. Oecologia, 95(3), 314–20. Kirkwood, T. B. L. (1977). Evolution of aging. Nature, 270(5635), 301–4. Kuss, P., Rees, M., Aegisdottir, H. H., et al. (2008). Evolutionary demography of long-lived monocarpic perennials: a time-lagged integral projection model. Journal of Ecology, 96(4), 821–32. Lanner, R. M. & Connor, K. F. (2001). Does bristlecone pine senesce? Experimental Gerontology, 36(4–6). Lesica, P. & Crone, E. E. (2007). Causes and consequences of prolonged dormancy for an iteroparous geophyte, Silene spaldingii. Journal of Ecology, 95(6), 1360–9. Lesica, P. & Steele, B. M. (1994). Prolonged dormancy in vascular plants and implications for monitoring studies. Natural Areas Journal, 14, 209–12. Martin, J. G. A. & Festa-Bianchet, M. (2011). Age-independent and age-dependent decreases in reproduction of females. Ecology Letters, 14(6), 576–81.

Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:24:57, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.016

336

Complex Life Histories and Senescence in Plants

Masoro, E. J. (2005). Overview of caloric restriction and ageing. Mechanisms of Ageing and Development, 126(9), 913–22. Mattison, J. A., Roth, G. S., Beasley, T. M., et al. (2012). Impact of caloric restriction on health and survival in rhesus monkeys from the NIA study. Nature, 489(7415), 318–21. McNamara, J. M., Houston, A. I., Barta, Z., et al. (2009). Deterioration, death and the evolution of reproductive restraint in late life. Proceedings of the Royal Society if London Series B: Biological Sciences, 276(1675), 4061–6. Mencuccini, M., Oñate, M., Peñuelas, J., et al (2014). No signs of meristem senescence in old Scots pine. Journal of Ecology, 102(3), 555–65. Merow, C., Dahlgren, J. P., Metcalf, C. J. E., et al. (2014). Advancing population ecology with integral projection models: a practical guide. Methods in Ecology and Evolution, 5(2), 99–110. Moriuchi, K. S., Venable, D. L., Pake, C. E. & Lange, T. (2000). Direct measurement of the seed bank age structure of a Sonoran Desert annual plant. Ecology, 81(4), 1133–8. Munné-Bosch, S. (2008). Do perennials really senesce? Trends in Plant Science, 13(5), 216–20. Nilsson, P., Tuomi, J. & Åström, M. (1996). Bud dormancy as a bet-hedging strategy. American Naturalist, 147(2), 269–81. Oñate, M. & Munné-Bosch, S. (2008). Meristem aging is not responsible for age-related changes in growth and abscisic acid levels in the mediterranean shrub, Cistus clusii. Plant Biology, 10, 148–55. Ott, J. P. & Hartnett, D. C. (2012). Contrasting bud bank dynamics of two co-occurring grasses in tallgrass prairie: implications for grassland dynamics. Plant Ecology, 213(9), 1437–48. Pedersen, B. (1999). Senescence in Plants: Life History Evolution in Plants, ed. T. O. Vuorisalo & P. K. Mutikainen (pp. 239–74) (Dordrecht: Kluwer). Peñuelas, J. & Munné-Bosch, S. (2010). Potentially immortal? New Phytologist, 187(3), 564–7. Rabotnov, T. A. (1969). On coenopopulations of perennial herbaceous plants in natural coenoses. Vegetatio, 19(1–6), 87–95. Reintal, M., Tali, K., Haldna, M. & Kull, T. (2010). Habitat preferences as related to the prolonged dormancy of perennial herbs and ferns. Plant Ecology, 210(1), 111–23. Rice, K.J. & Dyer, A. R. (2001). Seed aging, delayed germination and reduced competitive ability in Bromus tectorum. Plant Ecology, 155(2), 237–43. Ricklefs, R. E. (2000). Intrinsic aging-related mortality in birds. Journal of Avian Biology, 31(2), 103–11. Roach, D. A. (1993). Evolutionary senescence in plants. Genetica, 91(1–3). Roach, D. A. (2004). Evolutionary and demographic approaches to the study of whole plant senescence. In Plant Cell Death Processes, ed. L. D. Nooden (pp. 341–8) (New York: Academic Press). Roach, D. A., Ridley, C. E. & Dudycha, J. L. (2009). Longitudinal analysis of Plantago: age-by-environment interactions reveal aging. Ecology, 90(6). Saatkamp, A., Affre, L., Dutoit, T. & Poschlod, P. (2009). The seed bank longevity index revisited: limited reliability evident from a burial experiment and database analyses. Annals of Botany, 104(4), 715–24.

Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:24:57, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.016

References

337

Salguero-Gómez, R. & Casper, B. B. (2010). Keeping plant shrinkage in the demographic loop. Journal of Ecology, 98(2), 312–23. Salguero-Gómez, R., Shefferson, R. P. & Hutchings, M. J. (2013). Plants do not count . . . or do they? New perspectives on the universality of senescence. Journal of Ecology, 101(3), 545–54. Schaal, B. A. & Levin, D. A. (1976). The demographic genetics of Liatris cylindracea Michx. (Compositae). American Naturalist, 110, 191–206. Shefferson, R. P. (2009). The evolutionary ecology of vegetative dormancy in mature herbaceous perennial plants. Journal of Ecology, 97(5), 1000–9. Shefferson, R. P., Proper, J., Beissinger, S. R. & Simms, E. L. (2003). Life history trade-offs in a rare orchid: the costs of flowering, dormancy, and sprouting. Ecology, 84(5), 1199–1206. Shefferson, R. P., Kull, T. & Tali, K. (2006). Demographic response to shading and defoliation in two woodland orchids. Folia Geobotanica, 41(1), 95–106. Shefferson, R. P., Kull, Y., Tali, K. & Kellett, K. M. (2012). Linking vegetative dormancy to fitness in two long-lived herbaceous perennials. Ecosphere, 3(2). Shefferson, R. P. & Roach, D. A. (2013). Longitudinal analysis in Plantago: strength of selection and reverse age analysis reveal age-indeterminate senescence. Journal of Ecology, 101(3), 577–84. Shefferson, R. P. & Tali, K. (2007). Dormancy is associated with decreased adult survival in the burnt orchid, Neotinea ustulata. Journal of Ecology, 95(1), 217–25. Shefferson, R. P., Kull, T. & Tali, K. (2005). Adult whole-plant dormancy induced by stress in long-lived orchids. Ecology, 86(11), 3099–3104. Shefferson, R. P., Warren, R. J. & Pulliam, H. R. (2014). Life-history costs make perfect sprouting maladaptive in two herbaceous perennials. Journal of Ecology, 102(5), 1318–28. Silvertown, J., Franco, M. & Perez-Ishiwara, R. (2001). Evolution of senescence in iteroparous perennial plants. Evolutionary Ecology Research, 3(4), 393–412. Silvertown, J., Franco, M., Pisanty, I. & Mendoza, A. (1993). Comparative plant demography: relative importance of life-cycle components to the finite rate of increase in woody and herbaceous perennials. Journal of Ecology, 81(3), 465–76. Sohal, R. S. & Weindruch, R. (1996). Oxidative stress, caloric restriction, and aging. Science, 273(5271), 59–63. Stephenson, N. L., Das, A. J., Condit, R., et al. (2014). Rate of tree carbon accumulation increases continuously with tree size. Nature, 507(7490), 90–3. Taiz, L. & Zeiger, E. (eds.) (2006). Plant Physiology (Sunderland, MA, Sinauer Associates). Tamm, C. O. (1972). Survival and flowering of some perennial herbs: II. The behaviour of some orchids on permanent plots. Oikos, 23(1), 23–8. Tatar, M., Chien, S. A. & Priest, N. K. (2001). Negligible senescence during reproductive dormancy in Drosophila melanogaster. American Naturalist, 158(3), 248–58. Thomas, H. (2002). Ageing in plants. Mechanisms of Ageing and Development, 123(7), 747–53. Tuomi, J., Crone, E. E., Gremer, J. R., et al. (2013). Prolonged dormancy interacts with senescence for two perennial herbs. Journal of Ecology, 101(3), 566–76. Vega, E. & Montana, C. (2004). Spatio-temporal variation in the demography of a bunch grass in a patchy semiarid environment. Plant Ecology, 175(1), 107–20.

Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:24:57, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.016

338

Complex Life Histories and Senescence in Plants

Wendling, I., Trueman, S. J. & Xavier, A. (2014). Maturation and related aspects in clonal forestry: II. Reinvigoration, rejuvenation and juvenility maintenance. New Forests, 45(4), 473–86. Williams, G. C. (1957). Pleiotropy, natural-selection, and the evolution of senescence. Evolution, 11(4). Wyka, T. P. (1999). Storage, growth and reproduction in an alpine herbaceous plant, Oxytropis sericea. PhD dissertation, University of Missouri.

Downloaded from https:/www.cambridge.org/core. University of Warwick, on 21 Apr 2017 at 10:24:57, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.016

17 Why Some Fungi Senesce and Others Do Not An Evolutionary Perspective on Fungal Senescence Marc F. P. M. Maas, Alfons J. M. Debets, Bas J. Zwaan and Anne D. van Diepeningen

Short Summary Fungi are generally considered to be modular organisms with no clear distinction of a germ line: with the expansion of the mycelium, chances for reproduction are expected to increase, and each unit under favourable circumstances may produce offspring. Fungi with such modular body plans are expected to be long-lived, as most fungi indeed seem to be. However, fungi exist that do senesce, and their growth often seems to be limited by space or time. For these fungi, we can consider the term ‘pseudo-unitary’, as life history details and ecological conditions constrain the size of the soma and the opportunities for reproduction. We may predict the life history traits and ecological conditions that favour such evolution of fungal senescence. Known proximate mechanisms of fungal senescence can be viewed in the light of this evolutionary context.

Introduction Organisms with a ‘unitary body plan’ – such as many mobile animals – have a determinate structure composed of strict numbers of body parts, specialised organs and a separate germ line. Unitary organisms are arguably all subject to the process of senescence, resulting in death even under protected idealised conditions. Organisms with ‘modular body plans’, however – including plants, fungi and (sessile) animals such as hydroids and bryozoans – seem to escape this process and are potentially immortal. Modular organisms are composed of multiple genetically identical vegetative modules that may remain attached or become separated to form physiologically independent clones. Here all cells are in principle totipotent and capable of expansion and reproduction, so death of parts of a modular organism does not necessarily cause the death of the whole organism (Figure 17.1). In plant biology these modules are generally referred to as ‘ramets’. The ensemble of ramets that makes up a single genetic entity is generally referred to as a ‘genet’. The body plan of an organism has major implications for the way Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 14 Jun 2017 at 11:11:35, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.017

342

Why Some Fungi Senesce and Others Do Not

germ line

(a)

immortality Figure 17.1

a

som

senescence

immortality

Modular a

som

senescence

germ line

Pseudo-unitary germ line

Unitary germ line

(b)

a

som

immortality

Germ-line and soma overlap to various extents in different organisms. (A) Mushroom-forming fungi are typical modular organisms with indeterminate growth. Any somatic cell (potentially with somatic mutations) can become part of the germ line (white arrows). (B) In unitary systems such as most animals (left), a small and distinct subset of cells is reserved that contributes its hereditary material to the next generation (indicated in grey; this subset of cells must be immortal). In modular systems such as plants and fungi (right) all cells can do this. We propose that an intermediary situation of these two extremes exists as well, in which germ line and soma partially overlap, for which we propose the term (pseudo-unitary) (middle). In pseudo-unitary organisms the indeterminate modular growth of the colony is restricted e.g. by the limited availability of substrate.

selection can act. In modular organisms, selection rarely acts directly on the genet: the unit of selection is usually the ramet. Fungi are considered to be modular organisms with no clear distinction of a germ line. With the expansion of the mycelium, chances for reproduction are expected to increase, Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 14 Jun 2017 at 11:11:35, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.017

Evolutionary Theory of Senescence

343

and each unit under favourable circumstances may produce offspring (Figure 17.1B, right). Fungi with such modular body plans are expected to be long-lived, as most fungi indeed seem to be. However, some fungi exist that do senesce, and their growth often seems to be limited by space or time. For these fungi, we consider the term ‘pseudounitary’ (Figure 17.1B, centre), as life history details and ecological conditions constrain the size of the soma and the opportunities for reproduction. In this chapter we discuss why fungi are usually long-lived and why there are exceptions to this. We predict life history traits and ecological conditions that favour the evolution of fungal senescence. Finally, we discuss the proximate mechanisms of fungal senescence in the light of this evolutionary context.

Evolutionary Theory of Senescence Evolutionary adaptations occur via natural selection of heritable traits that increase fitness. Senescence, however, seems to be a highly maladaptive trait: it is thus not immediately clear why it would have evolved in the first place. Charles Darwin surprisingly never addressed this paradox. One of the earliest evolutionary explanations of senescence was that made by Alfred Russel Wallace, the co-discoverer of evolution by natural selection. In a footnote to the English translation of the 1881 essays of the German theoretician August Weismann, he essentially suggested that immortality would be sacrificed for the sake of reproduction, an idea that was embraced by Weismann and would be echoed many years later in the works of Medawar (1952), Williams (1957) and Kirkwood (1977). Paramount to the latter idea is the distinction between germ line and soma (see Figure 17.1). Weismann stated that whereas germ-line cells are able to transmit hereditary information to the next generation, somatic cells should not. In other words, there should be a unidirectional flow of hereditary information from germ line to soma, but not the other way round. This concept is commonly known as the ‘Weismann barrier’, and it implies that senescence is a property of the soma and cannot be a property of the germ line: The germ line must be evolutionarily immortal. In line with this, modular organisms do not have a clear distinction between germ line and soma, and indeed many appear not to senesce. Whereas individual modules such as fruiting bodies may outlive their usefulness and die, the organism as a whole is generally expected to be long-lived. In line with this, fungal fruiting bodies usually last at most a single season (with the exception of, for example, polypores or bracket fungi, which form woody perennial fruiting bodies), but mycelia can last for decades, producing new fruiting bodies every year or whenever local conditions are permissive. A further specification can be made for the conditions that allow for the evolution of senescence. Theoretical work has shown that an important prerequisite for senescence to evolve is that parents and offspring can be individually identified (Partridge & Barton 1993). This, for instance, can explain senescence in bacteria with asymmetrical division (Ackermann et al. 2003) and even in bacteria with binary fission (Stewart et al. 2005). In the former, parents and offspring differ in size and morphology, and in the latter, parents and offspring can be identified by the age of the pole region of the cells. Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 14 Jun 2017 at 11:11:35, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.017

344

Why Some Fungi Senesce and Others Do Not

Senescence evolves in the shadow of natural selection (Medawar 1952). This is the central tenet of the evolutionary biology of senescence. Organisms experience a certain (statistical) risk of death by predation, disease or other accidental causes of mortality. Any trait that would manifest only at an age at which the organism is long expected to have died from extrinsic causes would not be under selection. This means that regardless of the nature of this trait, which may be beneficial but also deleterious or even lethal, it would be able to drift through the population. This idea is generally known as the ‘mutation accumulation theory of ageing’ (Medawar 1952). Late-acting deleterious traits with pleiotropic, beneficial effects early in life may even be favoured by natural selection. The latter is commonly known as the ‘antagonistic pleiotropy theory of ageing’ (Williams 1957). These two ideas are not mutually exclusive. A more refined version of the antagonistic pleiotropy theory of ageing is the ‘disposable soma theory’ (Kirkwood 1977): this is based on the premise that somatic maintenance and repair are costly and states that resource allocation between maintenance and repair will be optimised so that the total reproductive output of an organism is maximised. This implies that given a certain regimen of extrinsic mortality, just enough metabolic resources will be invested to maintain an organism in a proper state only for the duration of its expected lifetime. The latter two theories use optimisation arguments balancing life span with reproduction (Partridge & Barton 1993), while the disposable soma theory translates the genetic argument of the antagonistic pleiotropy theory into candidate physiological mechanisms (Zwaan 1999). Also, for the theory on ‘negative senescence’ (Vaupel et al. 2004), reproduction is of importance: ‘negative senescence’ is defined as a decline in mortality and is generally accompanied by an increase in fecundity. Especially, indeterminate-growth species for which size and fertility increase with age are expected to experience negative senescence. Are the preceding ideas, which were developed largely with unitary organisms in mind, also applicable to fungi? The main issue may be the unit of selection: in unitary organisms, the unit of selection, by definition, is the genet. In a fungus, however, this can be both the genet and the ramet depending on its particular life history and ecological setting. The modular nature of fungi normally protects them from suffering systemic extrinsic mortality, thus preventing selection at the level of the genet. Some life histories and ecological settings may make fungi much more prone to systemic extrinsic mortality. These are the lack of a vegetative dispersal phase and the occupation of a spatiotemporally restricted (e.g. ephemeral) ecological niche. These conditions favour selection at the level of the genet and may result in a ‘pseudo-unitary’ growth pattern (Figure 17.1B).

Fungi Can Be Extremely Long-Lived Intrinsic to their way of growing, fungi experience a strong reproductive gain with age. Fungal colonies typically expand radially, and their surface is directly proportional to the amount of fruiting bodies they can form. Therefore, the reproductive value of a colony may strongly increase with age, leading to negative senescence (Vaupel et al. 2004) when mortality rates decrease. One can think, for instance, of the increasing Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 14 Jun 2017 at 11:11:35, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.017

But Some Grow Old and Die within a Matter of Weeks

345

circumference of the colony in fairy ring–forming basidiomycetes or as a function of the surface of the colony. In other words, fungi have much to gain by growing old and are unlikely to evolve senescence. This type of argument also applies, for example, to organisms such as size-indeterminate fish (Reznick et al. 2002) and mole-rats (Buffenstein 2005). These animals do not stop growing after reaching the age of maturity and strongly benefit from an ever-increasing reproductive output. One of the most famous and most-quoted examples to illustrate fungal longevity is the Honey mushroom Armillaria bulbosa, nowadays known as Armillaria gallica. Smith et al. (1992) reported a single mycelial clone of this fungus that was approximately fifteen hectares in size. These authors (very conservatively) estimated it to be 1,500 years old and about ten tons heavy, which is about the mass of an adult blue whale. Since then, many more cases of ‘humongous fungus’ have been found. Similarly striking examples of longevity are also commonly found in clonal plants. Quaking aspen (Populus tremuloides), for example, can spread clonally via root suckers, and a single (male) stand of Aspen in Utah (USA) was estimated to be 80,000 years old and 6,000 tons (Barnes 1975; Grant et al. 1992).

But Some Grow Old and Die within a Matter of Weeks Despite their reputation as long-lived organisms, some fungi do grow old and die within a matter of weeks (for an overview of fungi in which senescence/degeneration has been observed, see Table 17.1). One of the most intensively studied examples of senescing fungi is the ascomycete Podospora anserina. All natural isolates of this species that have been collected show mitotic (a.k.a. replicative or proliferative) senescence; their growth reduces over time and eventually stops (Rizet 1953; van der Gaag et al. 1998; van Diepeningen et al. 2008a). Post-mitotic senescence – loss of viability of formed mycelium – also occurs but has not been studied in great detail. The life span of a P. anserina strain is usually tested in long glass tubes referred to as ‘race tubes’ and expressed as distance (e.g. centimetres) or time (e.g. days) of growth. As cultures ‘go down the tube’, they progressively decline in vigour, usually commencing with a decline in female fertility, followed by a decline in mycelial growth rate and ending in arrest and death of the mycelial growth front. At the dying mycelial front, hyphae show various types of morphological aberrations and accumulate large amounts of lipofuscin, a yellow-brown pigment that is a rest product of fatty acid oxidation, also seen as the ‘ageing pigment’ in, for example, human tissue (Munkres & Rana 1978b). The formation of lipofuscin may be symptomatic of mitochondrial degeneration (see later). The transition of the culture to a senescent state (i.e. the onset of senescence) correlates with the appearance of an infectious element originally referred to as the ‘determinant’: when senescent and non-senescent (young) sub-cultures of the same isolate are inoculated together, the mixture typically adopts the shorter life span of the senescent culture. To date, there is no consensus on the exact nature of this infectious determinant element, but it is possible (if not likely) that multiple elements are involved, Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 14 Jun 2017 at 11:11:35, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.017

346

Why Some Fungi Senesce and Others Do Not

Table 17.1 Fungi with Observed Senescence or ‘Degeneration’

Genus Ascobolus Aspergillus

Cercophora

Senescing/ degenerating species Life style A. immersus A. amstelodami A. niger

C. gradiuscela, C. samala Chaetomium C. nigricolor C. pachypodioides Cordyceps C. militaris Cryphonectria C. parasitica Fusarium F. tricinctum, F. fujikuroi Heterobasidion H. parviporum Humicola H. variabilis Metarhizium M. anisopliae Neurospora N. crassa, N. intermedia N. tetrasperma Phomopsis P. subordinaria Podospora P. anserina P. curvicolla P. flatula, P. setosa P. tetraspora Sordaria S. macrospora Zopfiella Z. longicaudata Unknown Pezizales sp.

Coprophilic Stored products saprobic Coprophilic

Reference Francou 1981; Marcou 1960 Caten 1972; Lazarus et al. 1980; Lazarus & Kuntzel 1981; van Diepeningen et al. 2006, 2008b Geydan et al. 2012

Coprophilic, saprobic Entomopathogenic Plant pathogenic Plant pathogenic

Geydan et al. 2012

Forest pathogen Coprophilic Entomopathogenic Burned substrate

Vainio et al. 2014 Geydan et al. 2012 Wang et al. 2005 Court et al. 1991; Griffiths & Bertrand 1984, Marcinko-Kuehn et al. 1994; Rieck et al. 1982

Plant pathogenic Coprophilic

Geydan et al. 2012 Rizet 1953; Gagny et al. 1997; Geydan et al. 2012

Coprophilic Coprophilic Coprophilic

Gagny et al. 1997; Marcou 1961 Geydan et al. 2012 Geydan et al. 2012

Xiong et al. 2013 Baidyaroy et al. 2011 Leslie & Summerell 2006; Geydan et al. 2012

including both suppressive mitochondrial DNA (mtDNA) lesions and protein-based epigenetic factors. Examples of the multiple candidates for determinant elements in fungi are, for instance, mitochondrial plasmids, mobile mitochondrial introns and retrotransposon-like elements (e.g. Jamet-Vierny et al. 1980; Kudryavtseva et al. 2012; Osiewacz & Esser 1980; Tudzysnki & Esser 1979; Tudzynski et al. 1980; Wright et al. 1982). Similar phenomena have been described in other fungal genera, including Neurospora. However, fungal populations are typically polymorphic for the senescence trait (i.e. it is found in only a fraction of the strains). Senescence has, for example, been found in about a third of all the Hawaiian N. intermedia wild types and in about a fifth of N. tetrasperma isolates (Debets et al. 1995; Griffiths & Bertrand 1984; Maas et al. 2005; Rieck et al. 1982; see also Maheshwari & Navaraj 2008). In Neurospora, senescence can also be demonstrated in race tubes, as is routinely done in Podospora, but it is more commonly demonstrated by serial sub-culturing the isolates

Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 14 Jun 2017 at 11:11:35, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.017

Ecological Conditions that Favour the Evolution of Senescence

347

using vegetative spores (macroconidia). After a strain-specific number of transfers, the fertility of the cultures, their macroconidial production and growth rates decline. The last cultures typically still show conidiation, though none of the spores produced are able to germinate. Similar to Podospora, death often (but not always) coincides with morphologically aberrant hyphae and lipofuscin accumulation (Munkres & Rana 1978a). Reminiscent of the ‘senescence determinant’ from Podospora, the onset of senescence correlates with the appearance of an infectious element. Similar to Podospora, there is no consensus on the nature of this element: although the senescence process in Neurospora involves horizontally transmitted mitochondrial plasmids (see later), plasmids can be transmitted independently from senescence (Debets et al. 1994). Both in Podospora and in Neurospora, the onset of senescence is a strain-specific trait, and life span is inherited largely (but not strictly) maternally (Marcou 1961; Rizet 1953, 1957). This is because mitochondria play a major role in the senescence phenomenon. Senescence is always accompanied by a ‘mutational meltdown’ of the mitochondrial genome, whereby certain regions are amplified and/or mtDNA molecules with large rearrangements (e.g. insertions and deletions) accumulate. The energetic decline that results from this mitochondrial meltdown is likely the major proximate agent of senescence. Also within the genus Aspergillus, senescence or senescence-like phenomena have been reported. In ‘ragged’ mutants of A. amstellodami, for example, a retro-transposonlike element inside the mitochondrial genome seems to be the cause of senescence (Lazarus et al. 1980). In A. niger colonies that are heavily infected with double-stranded RNA (dsRNA) mycoviruses, a senescence-like phenotype can sometimes be observed in sectors of the mycelium. The majority of mycovirus infections (occurring in approximately 10 per cent of A. niger wild-type isolates), however, do not seem to cause such effects (van Diepeningen et al. 2006, 2008b). In a virus-free strain of Cryphonectria parasitica, an altered form of mtDNA has been associated with hypo-virulence and senescence (Baidyaroy et al. 2011). Recently, Vainio et al. (2014) described how the forest pathogen Heterobasidion parviporum – a typical candidate for immortality/postponed or negative senescence – senesces and dies through its dsRNA mycovirus infection. Also, the colony sectorisation of Metarhizium anisopliae, accompanied by oxidative stress and loss of fertility, can be seen as a sign of senescence, though there the cause of the phenomenon has not been linked to any (extra) genomic element (Wang et al. 2005).

Ecological Conditions that Favour the Evolution of Senescence Many fungi occupy niches in which they are limited neither spatially nor temporally. The earlier-mentioned species of Armillaria, for example, is usually found as an innocuous saprophyte living on organic matter in the soil. It can spread over large distances via specialised root-like bundles of hyphae called ‘rhizomorphs’. Spatially as well as temporally it is therefore potentially unlimited in its growth and expected to be extremely long-lived. Fungi that grow on more ephemeral and/or spatially restricted Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 14 Jun 2017 at 11:11:35, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.017

348

Why Some Fungi Senesce and Others Do Not

Figure 17.2

Examples of spatiotemporally restricted ecological niches in which fungal senescence is expected to evolve. (A) Mosquito (Anopheles sp.) that was killed by the fungal parasite Beauveria bassiana. (B) Spruce cone with fruiting bodies of the Sprucelon cap (Strobilurus esculentus). (C) Birch bolete (Leccinum scabrum) with a parasitic fungus. (D and E) Cow bones with an unidentified basidiomycete.

substrates, however, may be expected to be short-lived: after depletion of the short-lived substrate and spore production, the remaining mycelium has lost its usefulness and has become disposable. These ‘pseudo-unitary’ fungi include parasites of short-lived hosts, endophytes, saprophytic species specialising on dung, bones, feathers and similar substrates, etc. Several examples are illustrated in Figure 17.2. Most of the fungi that have been observed to date to senesce prove to have a coprophillic life style and belong to ascomycetous genera, including Podospora, Ascobolus, Sordaria, Chaetomium and Cercophora spp. (Böckelmann & Esser 1986; Gagny et al. 1997; Geydan et al. 2012; Marcou 1961). A broad phylogenetic approach that included also non-coprophillic members of these genera showed that fungal senescence evolved independently in several clades within the sordariomycetes as a function of the ephemerality of the substrate (Geydan et al. 2012). In the study by Geydan et al. (2012), approximately 70 per cent of the coprophilic species proved to senesce (seven of ten of the strains isolated from dung in that study and nine of thirteen of all the coprophilic fungi tested). In line with the expectations of a link between ephemeral substrate and senescence, cases of senescence or senescence-like phenomena have also been observed in both Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 14 Jun 2017 at 11:11:35, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.017

Ecological Conditions that Favour the Evolution of Senescence

349

insect and plant pathogens: for instance, in Metarhizum anisoplae (Wang et al. 2005), Phomopsis spp. (Geydan et al. 2012; A. van Harn, unpublished observation) and Botrytis spp. (F. Seegers, unpublished observation). All wild-type isolates of P. anserina senesce, but they do so in a strain-dependent manner: some strains senesce faster than others, and life spans on nutrient-rich media vary from less than one week to over three weeks (van der Gaag et al. 1998; van Diepeningen et al. 2008a). In the case of the different senescent Neurospora species, growing on burned vegetation, the fraction of senescent strains varies between species and even populations, and 70 per cent or more of a population may have a non-senescent phenotype (Debets et al. 1995; Griffiths & Bertrand 1984; Maas et al. 2005; Rieck et al. 1982; see also Maheshwari and Navaraj 2008). Thus, fungal strains can show (genetic) variation in whether or not they senesce and in their rate of senescence. Natural variation in susceptibility to senescence plasmids may exist, and restrictions in somatic fusion between mycelia can limit the spread of fungal senescence diseases (Bastiaans et al. 2014). Fungi show an extreme diversity in ecological strategies, and these are notoriously difficult to classify in a single scheme. In r/K selection theory, for example (after the growth rate parameter r and carrying-capacity parameter K from the Verhulst equation of population dynamics), a distinction is made between growth-rate-selected species (r-strategists) that exploit less crowded niches (e.g. primary colonisers) and species that exist at densities close to the carrying capacity of their ecological setting (K-strategists). According to r/K selection theory, the former are species with a high reproductive output and low survival, and vice versa. There are relatively few species that can be classified as pure r- or K-strategists, however: most fungi simultaneously exhibit features of both. It is certainly not true that fungal species with r-selected traits are always short-lived and those with K-selected traits are always long-lived. For example, in the ecological succession on dung, primary colonisers (including mainly zygomycetes) specialise on easily degradable carbon sources. They are all fast-growing species and poor competitors that invest mainly in short-lived vegetative spores. The mycelia of these primary colonisers, however, tend to be long-lived. Late species (including mainly ascomycetes and basidiomycetes) specialise on complex carbon sources such as cellulose. They are all relatively slow-growing species and good competitors that invest in robust sexual spores. These late species, however, tend to be short-lived. As discussed later, there are important differences in the life histories of these early and late species that may explain why. Also, when we consider an alternative scheme such as Grime’s triangle theory (Grime & Pierce 2012), which was developed for plant ecology, it is not easy to classify typical senescent or long-lived fungal species. Grime’s three extreme ecological strategies are competitor (C), stress tolerator (S) and ruderal (R). Looking again at the ecological succession on dung, the primary colonisers could be considered typical ruderals that are fast-growing and rapidly producing large amounts of spores. Tested in the laboratory, these species will keep growing as long as there is available substrate, but in nature they are rapidly outcompeted. The late species on dung with their specialised carbon source utilisation are these better competitors, but also they are limited in their life span by the ephemeral dung and Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 14 Jun 2017 at 11:11:35, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.017

350

Why Some Fungi Senesce and Others Do Not

need to reproduce fast. To do so, the obligate sexual species P. anserina has adapted its reproductive mode: instead of being heterothallic and needing a chance partner of opposite mating type, it has become pseudo-homothallic and solved the mate problem by containing two different nuclei of opposite mating type.

Reproductive Strategies that Favour the Evolution of Senescence Life history traits like reproductive mode or place in the succession of a new environment can also influence the evolution of senescence: P. anserina is an obligate sexually reproducing fungus and is an example of a relatively late species in the ecological succession on dung. The most relevant difference with early species (primary colonisers) is that late species such as Podospora usually do not spread via vegetative spores (Figure 17.3): all vegetative mycelium modules of an individual colony consequently share the same fate. This implies that the unit of selection will always coincide with the genetic individual (genet) as a whole. In this sense, it is similar to unitary organisms. Many coprophilic ascomycetes and basidiomycetes share these properties and are similarly expected to be short-lived. In contrast, in fungi that have an important vegetative dispersal stage, the unit of selection will usually coincide with the ramet. Species of Neurospora, for example, are primary colonisers of burnt vegetation (Figure 17.4). Although the true spatiotemporal restrictions of this niche are unknown (a burning site may remain open for many years, but it is unknown to what degree it will be suitable for Neurospora), Neurospora species can spread very efficiently across post-fire sites via wind and/or insect dispersal of vegetative spores (macroconidia). In Neurospora, the unit of selection will therefore usually coincide with the individual clone (ramet) rather than directly with the genetic individual (genet) as a whole. In line with this, Neurospora cultures are much longer-lived than those of Podospora. Contrary to expectation, however, as also discussed earlier, senescence can be found in some Neurospora strains and, depending on the geographical location, even at a relatively high frequency (e.g. in about a third of the N. intermedia isolates from Hawaii; see earlier). One possible explanation for this is that the agricultural practice of regular sugar cane burning in these locations has created conditions that would limit the fungus to a ‘pseudo-unitary’ system: the regular pre-harvest fires kill the somatic tissue (mycelium and spores) but not the sexual spores (ascospores). Instead, these fires induce germination of sexual spores, thus introducing discrete generations (separating parent from offspring) and effectively increasing the shadow of selection.

Proximate Mechanisms of Fungal Senescence: Evidence for Trade-Offs between Life Span and Reproduction The core premise of the antagonistic pleiotropy theory is that genetic trade-offs shape the evolution of senescence. In fungi, there is substantial evidence for such Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 14 Jun 2017 at 11:11:35, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.017

Proximate Mechanisms of Fungal Senescence

351

spores shot onto surrounding vegetation germination

ingestion by herbivore

perithecium dikaryotic mycelium

trichogyne ascogonial cell fertilisation ascus with four binucleate spores protoperithecium

ascus development Figure 17.3

Life cycle of Podospora anserina. P. anserina is a secondary homothallic ascomycete that grows on dung. It produces sexual spores that are shot onto the surrounding vegetation and eaten by herbivores. Passage through the gut subsequently induces germination and the cycle starts over. Because it does not spread via vegetative spores (unlike Neurospora, see Figure 17.4), there is a discrete vegetative phase.

trade-offs. First of all, fungi respond to calorie restriction by postponing reproduction for the sake of longevity: ‘calorie restriction’ refers to a dietary regimen that is low in calories but without malnutrition. Its life-span-extending effect was first noted in rodents (McCay et al. 1935), and since its initial discovery, it has been documented in many different organisms ranging from yeast (Lin et al. 2002) to primates (Ingram et al. 2004; Lane et al. 2001). Calorie restriction is associated with increased resistance to various kinds of stress, including, for example, heat and oxidative stress, and appears to forestall many late-onset diseases, including cancer (Berrigan et al. 2002; Hursting et al. 2003; Sohal & Weindruch 1996; Weindruch & Walford 1988). This kind of response allows organisms to postpone reproduction and survive unfavourable conditions. When food intake is restored, calorie-restricted individuals are typically still able to reproduce, whereas controls that are fed ad

Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 14 Jun 2017 at 11:11:35, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.017

352

Why Some Fungi Senesce and Others Do Not

germination macroconidia

monokaryotic mycelium fire induced germination

monokaryotic mycelium

macroconidia

macroconidiophore

fertilisation trichogyne perithecium

ascogonial cell

ascus with eight mononucleate spores protoperithecium ascus development Figure 17.4

Life cycle of Neurospora crassa. N. crassa is a heterothallic ascomycete that grows on burnt vegetation. It produces both sexual spores (ascospores) and vegetative spores (macroconidia). The sexual spores remain in the soil until their germination is triggered by fire. The vegetative spores are spread by wind and/or insects. Due to the vegetative dispersal phase, a single genet can spread across burning sites and potentially survive indefinitely.

libitum are post-reproductive or no longer alive. The plasticity of these (life history) traits induced by calorie restriction is therefore of clear selective value (Holliday 1989). This also applies to fungi: in P. anserina, calorie restriction increases life span by forestalling both the onset and the progression of mtDNA instability (van Diepeningen et al. 2009). Since mtDNA integrity is required for producing quality offspring, this response allows a postponement of reproduction. Podospora cultures that are grown on highly reduced glucose levels (e.g. |±0.3| indicate a relatively high contribution of the life history trait to the PCA axis.

difficulties of converting their population dynamics to an annual basis to compare with all other species’ models. 6. We only used MPMs that described the average population dynamics of the study. MPMs from contiguous temporal transitions, from multiple populations in a given study and under un-manipulated conditions were averaged element by element. 7. We only included MPMs of dimension >3 (i.e. with four or more stages in the life cycle) to avoid issues with quick convergence to stationary equilibrium, at which point the estimates of life history trait values and rates of senescence may be unreliable (Jones et al. 2014). 8. For sexually reproducing species, we only included MPMs in which sexual reproduction was explicitly modelled in order to have information about both major fitness components.

Downloaded from https:/www.cambridge.org/core. The University of British Columbia Library, on 08 Jun 2017 at 23:08:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.020

408

Life History Trade-Offs Modulate the Speed of Senescence

9. We only included MPMs where stage-specific survival was 1 (Legendre 2012). We then inspected the variation explained by the retained axes by obtaining the scree ranks of the PCA. Finally, we used the scores along each retained axis to quantify the life history strategies of each MPM/species.

Phylogeny Our phylogenetically informed analysis was founded on a phylogeny created by combining the plant tree provided in Salguero-Gómez et al. (2015) and an animal tree based on the trees provided as part of the supplementary material for Hedges et al. (2015) – the ‘Time Tree of Life’ (TTOL). For the animal tree, where the exact species for which we had an MPM was not included in the TTOL, a close relative was selected (e.g. from the same genus). Where we had more than one MPM for a given species, we added arbitrarily very short branch(es) to represent these populations. Then, because the trees have been constructed using different methods, rather than using the node heights provided in the dated phylogenies, we elected to use Grafen’s (1989) branch length transformation which we implemented using the function compute.brlen in package ape (Paradis et al. 2004), whereby each node is given a ‘height’, matching the number of leaves in the subtree minus 1, and these heights are then scaled so that root height is 1.

Exceptional Mature Longevity We calculated a metric of exceptional mature longevity Lα–ω to examine variability in adult life spans among the studied animal and plant species. To do so, we quantified the window of time in the life events of the average individual of the studied population constrained by a start and an end point. The start point was the mean age at maturity Lα (Figure 20.1), as defined earlier using the methods in Caswell (2001). The end point was set as the age at which a cohort in the population would undergo 99 per cent mortality Lω (Figures 20.1 and 20.2). To calculate this point, we iterated a population vector n with 100 individuals in the first non-seed-bank stage (see below) through the population Downloaded from https:/www.cambridge.org/core. The University of British Columbia Library, on 08 Jun 2017 at 23:08:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.020

410

Life History Trade-Offs Modulate the Speed of Senescence

(b) I

Survivorship (lx)

Age-specific pattern

(a)

Seed bank

II III

Stationary dynamics

Lα

QSD90 Age (x)

Age (x)

Survivorship lx Seed Lα bank

QSD90 Stationary dynamics Age (x)

Figure 20.1

0.8 0.6 Seed bank 0.4

Stationary dynamics

0.2

Fecundity mx

Age-specific pattern

1.0

0.0 Lα

Lω QSD90 Lα−ω Age (x)

An exemplified age-specific pattern for survivorship lx (blue solid line, left axis) and fecundity mx (red dashed line, right axis) obtained from a matrix population model, where we left-truncate values when seed banks are included in the model (only for plants) and right-truncate values in order to exclude potential artefacts of stationary dynamics. The exceptional mature longevity values Lα–ω are calculated as the time elapsed from the mean age at maturity Lα until the age at mortality of 99 per cent of the individuals of a cohort in the examined population Lω.

matrix A and tracked the age at which only 1 individual (or a fraction of it) remained alive, as per Doak and Morris (2002). There are two caveats to this seemingly straightforward approach. The first concerns the seed bank possessed by some plant populations, and the second concerns a mathematical assumption of MPMs. Firstly, seed-bank stages are common in plant MPMs but pose a problem because (1) good seed-bank data are scarce and (2) seed-bank dynamics are likely to be only known with low certainty (Baskin & Baskin 2014). Therefore, if the MPM included seed-bank stages, we ignored the dynamics resulting from the residence time in that stage (e.g. ηe was calculated as the conditionality on entering the life cycle in the first non-seed-bank stage; the population vector n described earlier started with 100 individuals in the first non-seed-bank stage). This is a common approach in comparative analysis using MPMs (Burns et al. 2010; Salguero-Gómez et al. 2016a). The second potential issue arises because MPMs are typically parameterised with a stasis loop in the oldest/largest/most-developed stage (e.g. ‘adult survival’), which means that mortality and fertility plateaus may emerge as a mathematical artefact when examining age-specific patterns (Caswell 2001). To avoid this in our

Downloaded from https:/www.cambridge.org/core. The University of British Columbia Library, on 08 Jun 2017 at 23:08:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.020

Methods

411

Animalia Plantae

Figure 20.2

Geographical distribution of the studies containing population matrix models used here to examine which and how life history traits predict rates of senescence among 571 species, including animals and plants. Only studies with available GPS coordinates from the COMADRE Animal Matrix Database and the COMPADRE Plant Matrix Database are shown (403 of the 622 studies).

calculations, we only considered species for which Lω is less than the age at which the quasi-stationary stage distribution is reached. We define this as the age at which the cohort approximates 90 per cent of its stable stage distribution (the normalised right eigenvector w of the MPM), as depicted in Figure 20.1 (see Jones et al. (2014) for further explanation).

Statistical Analyses Our statistical analyses involved three approaches. Firstly, we related life history trait values to the associated patterns of demographic senescence, and we examined the correlation between each life history trait (Table 20.1) and exceptional mature longevity Lα–ω. To account for the fact that this involves multiple testing, which can inflate the likelihood of type II errors, we adjusted the obtained p-values using the function p.adjust of the stats library in R using methods described by Benjamini and Hochberg (1995). Secondly, we conducted a post-hoc analysis of the results of our PCA (described earlier). The Kaiser criterion determined that only the first two PCA axes should be retained and that higher-order axes could safely be ignored. We therefore fitted a twoway ANOVA model to predict exceptional mature longevity Lα–ω from the scores of these two important PCA axes. Thirdly, we ran a series of pairwise t-tests on the values of exceptional mature longevity Lα–ω at the taxonomic kingdom and class levels to examine differences in the speed of senescence among those groups. For the latter, because some classes contained small sample sizes, we restricted this analysis to groups with sample size of

Downloaded from https:/www.cambridge.org/core. The University of British Columbia Library, on 08 Jun 2017 at 23:08:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.020

412

Life History Trade-Offs Modulate the Speed of Senescence

more than ten MPMs per group. This resulted in the inclusion of seven major groups: bony fish (Actinopterygii, n = 29), reptiles (Reptilia, n = 15), birds (Aves, n = 23) and mammals (Mammalia, n = 66), and, in the plant kingdom, certain conifers (Pinopsida, n = 24), monocots (Liliopsida, n = 100) and dicots (Magnoliopsida, n = 317). Again, we accounted for the multiple comparisons using the p.adjust approach described earlier.

Results Our results provide the first comparative demographic study using data spanning all continents (Figure 20.3) and an unprecedented number of animal and plant taxonomic classes. These include 162 animal species (29 Actinopterygii, 1 Adenophorea, 1 Amphibia, 2 Anthozoa, 1 Arachnida, 23 Aves, 5 Bivalvia, 3 Branchiopoda, 1 Cephalaspidomorphi, 2 Demospongia, 1 Elasmobranchii, 5 Gastropoda, 4 Insecta, 2 Malacostraca, 65 Mammalia, 1 Maxillopoda, 15 Reptilia and 1 Secernentea) and 409 plant species (8 Cycadophyta, 86 Liliopsida, 292 Magnoliopsida, 22 Pinopsida and 1 Pteridopsida). The life history traits of the 571 included species varied enormously. For instance, generation time T ranged from 1.05 years for the legume Aeschynomene virginica to 79.25 years for the cactus Mammillaria solisioides; mean life expectancy at ‘birth’ ηe ranged from 1 year for the alpine bluegrass (Poa alpina) and devil’s-bit scabious (b)

2

ηe Sσ

1

φ

0

γ LT α

−2

−2

−1

PCA 2 (21.32%)

PCA 2 − Reproductive strategy −1 0 1 2 3

3

Animalia Plantae

−3

Figure 20.3

−2

−1

0 1 2 PCA 1 (58.37%)

3

4

Animalia Plantae

Lα−ω (years)

(a)

−3

150 100 50 0

−2

−1 0 1 2 3 PCA 1 − Fast−slow continuum

4

(A) Key life history traits (Table 20.1) explain approximately 80 per cent of the variation in life history strategies in the examined species. The first axis is predominantly driven by the variation in generation time T, mean age at maturity Lα and progressive growth γ (Table 20.1) of each organism, while the second axis is regulated primarily by the degree of iteroparity S. The life history traits are colour-coded to indicate their association with population turnover (black: generation time T), longevity (grey: mean life expectancy ηe and survival probability σ, mean age at maturity Lα), reproduction (mean reproduction Φ), degree of iteroparity S, and changes in age/size (blue: progression γ, and retrogression ρ). (B) The rich range of values for the exceptional mature longevity values Lα–ω (in years) achieved by animals and plant species is predicted by their relative scores along the fast–slow continuum and reproductive strategy axis.

Downloaded from https:/www.cambridge.org/core. The University of British Columbia Library, on 08 Jun 2017 at 23:08:46, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781139939867.020

Methods

413

(Succisa pratensis) to 34.46 years for the Chinese tallow tree (Sapium sebiferum). The probability of survival σ ranged between