The Nature of Nutrition: A Unifying Framework from Animal Adaptation to Human Obesity [Course Book ed.] 9781400842803

Citation preview

The Nature of Nutrition

The Nature of Nutrition A Un i f yi n g Fr am ewo rk fro m Animal A dap tat i on to H u man O besity

Stephen J. Simpson and David Raubenheimer

Princeton University Press Princeton and Oxford

Copyright © 2012 by Princeton University Press Published by Princeton University Press, 41 William Street, Princeton, New Jersey 08540 In the United Kingdom: Princeton University Press, 6 Oxford Street, Woodstock, Oxfordshire OX20 1TW press.princeton.edu Jacket Art: Henri Rousseau, The Dream, 1910, oil on canvas. Digital image © The Museum of Modern Art/Licensed by SCALA/Art Resource, NY All Rights Reserved Library of Congress Cataloging-in-Publication Data Simpson, Stephen J.   The nature of nutrition : a unifying framework from animal adaptation to human obesity / Stephen J. Simpson and David Raubenheimer.    p. cm.   Includes bibliographical references and index.   ISBN 978-0-691-14565-5 (hardback : alk. paper)  1.  Nutrition.  2.  Nutrition— Research.  3.  Animal nutrition.  4.  Adaptation (Physiology)  5.  Bioenergetics.  6.  Physiology, Experimental.  7.  Obesity.  8.  Energy metabolism.  I.  Raubenheimer, David, 1960–  II.  Title.   QP141.S534 2012   612.3—dc23    2011042321 British Library Cataloging-­in-­Publication Data is available This book has been composed in Sabon LT Std Printed on acid-­free paper. ∞ Printed in the United States of America 10  9  8  7  6  5  4  3  2  1

Contents

Acknowledgments

ix

ONE  Nutrition and Darwin’s Entangled Bank

1

1.1  Nutrition Touches and Links All Living Things

3

1.2  Nutrition Is Complex

5

1.3  Dealing with Nutritional Complexity: Enough but Not Too Much

7

1.4  Charting the Void between Nutritional Detail and Generality: The Geometric Framework

TWO The Geometry of Nutrition

10 11

2.1  The Geometric Framework: Basic Theory

11

2.2  The Geometric Framework in Practice

22

2.3  Conclusions

34

THREE Mechanisms of Nutritional Regulation

35

3.1  How to Defend an Intake Target

35

3.2  Postingestive Regulation

48

3.3  Conclusions

56

FOUR Less Food, Less Sex, Live Longer?

57

4.1  How Does Macronutrient Balance Affect Life Span?

62

4.2  Less Sex, Live Longer?

66

4.3  Conclusions

70

FIVE  Beyond Nutrients

71

5.1  The Distinction between Nutrients and Toxins

72

5.2  Self-­medication and Ecological Immunology: The Distinction between Nutrients and Medicines

79

vi  |  Contents

5.3  Toxins and Nutrients Interact

84

5.4  Conclusions

87

SIX Moving Targets

88

6.1  Moving Targets in the Short Term

88

6.2  Moving Targets in Developmental Time

91

6.3  From Parents to Offspring—­Epigenetics

95

6.4  Evolving Targets

97

6.5  Evolving Rules of Compromise: Nutrient Specialists and Generalists

99

6.6  Evolving Postingestive Responses

105

6.7  Conclusions

106

seven From Individuals to Populations and Societies

108

7.1  Cannibal Mormon Crickets

109

7.2  Locusts Are Cannibals Too

113

7.3  Communal Nutrition in Ants

114

7.4  The Blob

117

7.5  Conclusions

119

EIGHT How Does Nutrition Structure Ecosystems?

120

8.1  From Individual Fitness to Population Growth Rates

121

8.2  Interactions among Organisms and the Environment

122

8.3  Do Predators Regulate Nutrient Intake?

124

8.4  The Nutritional Geometry of Food Webs

130

8.5  The Nutritional Niche

138

8.6  Agent-­Based Modeling of Nutritional Interactions: From Individuals to Ecosystems

144

8.7  Conclusions

145

NINE Applied Nutrition

147

9.1  Domestication

147

9.2  Wildlife Conservation

157

9.3  Conclusions

165

Contents  |  vii

TEN The Geometry of Human Nutrition

167

10.1  The Modern Human Nutritional Dilemma

167

10.2  Do Humans Regulate to an Intake Target?

170

10.3  What Is the Human Rule of Compromise?

175

10.4  What Are the Implications of Protein Leverage?

182

10.5  How Do Humans Deal with Nutrient Excesses?

191

10.6  Conclusions

191

ELEVEN  Perspectives

194

11.1  Expanding GF into Further Dimensions of Nutrition

194

11.2  GF and “Omics”

195

11.3  Nutritional Epigenetics and Early-­Life Prevention   of Metabolic Disease

196

11.4  Human Obesity

196

11.5  Nutritional Immunology

197

11.6  Modeling Nutritional Interactions: From Individuals   to Ecosystems

198

11.7  Conclusions

199

References

201

Index

229

This page intentionally left blank

This page intentionally left blank

Acknowledgments

We would like to express our thanks to all of our students, postdoctoral associates, and colleagues, many of whom are named in the book at the points where we discuss their work. Pedro Telleria-­Teixeira provided invaluable support in preparation of the manuscript, and various chapters were read and commented upon by Fiona Clissold, Arthur Conigrave, Audrey Dussutour, Annika Felton, Alison Gosby, Ron Moorhouse, Jessica Rothman, Alice Tait (née Coveney), Shawn Wilder, and Ken Wilson. Thanks too to the funding agencies that have supported our research over the years, including the Australian Research Council, Australian National Health & Medical Research Council, Marsden Fund, Massey University Research Fund, National Research Centre for Growth and Development (NZ), The University of Sydney, UK Biotechnology and Biological Sciences Research Council, UK Engineering and Physical Sciences Research Council, UK Natural Environment Research Council, UK Medical Research Council, United Nations Development Programme, and the European Union COST Action Programme. We owe a considerable debt to the Wissenschaftskolleg zu Berlin (WIKO) for hosting us during a stimulating year in which this book was germinated. Finally, heartfelt thanks to our families: to Lesley, Nicholas, and Alastair Simpson, and to Jacky, Gabriel, and Julian Raubenheimer.

This page intentionally left blank

The Nature of Nutrition

This page intentionally left blank

one

Nutrition and Darwin’s Entangled Bank

Charles Darwin (1859)  famously ended his revolutionary book The Origin of Species with a paragraph that opened: It is interesting to contemplate an entangled bank, clothed with many plants of many kinds, with birds singing on the bushes, with various insects flitting about, and with worms crawling through the damp earth, and to reflect that these elaborately constructed forms, so different from each other, and dependent on each other in so complex a manner, have all been produced by laws acting around us. Darwin demonstrated in his book that a few biological facts—what he refers to as “laws”—combine to provide an elegantly simple natural mechanism that can explain the origin of the diverse and elaborately constructed plants and animals in his “entangled bank.” The facts are reproduction with inheritance, variability, and competition for resources; the mechanism is natural selection. The theory of natural selection provided a framework that encompassed all of biology. But Darwin was well aware that within this framework there were daunting webs of entangled complexity that remained to be unraveled. The “elaborately constructed” organisms—the meshwork of interactions between molecules, organelles, tissues, and organs that furnished Darwin with clear evidence of adaptation to the environment— remained poorly understood, as did the ecological interactions through which these organisms were “dependent on each other in so complex a manner.” Much of biology over the past 150 years has been focused on unraveling this complexity. Armed with progressively more powerful technologies, and sophisticated numerical and conceptual tools, functional biologists, ecologists, and applied biologists have worked away at the task, sometimes with incremental gains, sometimes with transformational advances. Darwin would be astounded by the progress that has been made. But an important opportunity has been neglected: the potential offered by following the connections provided by nutrition. Nutrition

2  |  Chapter One

touches, links, and shapes all aspects of the biological world. It builds the components of organisms, and fuels the dynamic interactions between these components; it determines whether or not wild animals thrive, how their populations grow, decline, and evolve, and how assemblages of interacting species (ecological communities) and ecosystems are structured. Nutrition also drives the affairs of humans, from individuals to global geopolitics. Food security and the burden of famine and disease from undernutrition have been pervasive in history, and recently overnutrition has emerged as a major cause of preventable death and disease. Climate change, population growth, urbanization, environmental degradation, and species extinctions all are in one way or another linked to the need for nutrients. In short, nutrients are the interconnecting threads in the web of life. And yet the science of nutrition remains fragmented. Because of its direct importance to human health and food-animal production, nutrition has traditionally been considered the domain of the medical and agricultural sciences. Research in these areas has produced a tremendously detailed account of the nutritional biology of a few species. By contrast, with some exceptions, nutrition in the ecological sciences has tended to adopt simpler, more general approaches that are applicable across the diversity of animals. Foraging might, for example, be considered a process of acquiring energy or minimizing time exposed to predators, rather than a complex balancing act of obtaining enough—but not too much—of the many nutrients that are needed for sustaining health and reproduction. The advantage of this simplified approach is that it has supported the development of powerful general frameworks for biological processes, unhindered by the staggering nutritional complexity that has been uncovered in the more applied nutritional sciences. We believe, however, that considerable potential for unraveling Darwin’s entangled bank lies unutilized in the void between the blinding detail of nutrition in the applied sciences and the conveniently simplistic nutritional frameworks of the ecological sciences. Our aim in this book is to present an approach that can help to realize this potential, by systematically introducing nutritional complexity into the ecological sciences, while providing a scaffold for extracting generalities from the mass of detail in applied nutrition. We hope that this approach, called the “Geometric Framework,” will help to disentangle the rich and complex network of interconnections that bind the web of life, and to elucidate how mismanagement in one area can lead to intractable tangles elsewhere. In this chapter we briefly expand on three important themes that form the backdrop to our story: nutrition touches and links all living things; nutrition is complex; and there have been benefits both from the highly specific and detailed approach of applied nutritional sciences and the

Nutrition and Darwin’s Entangled Bank  |  3

simplified, general approaches adopted in the ecological sciences. In the rest of the book we show how simple geometry can be used to explore the middle ground, by systematically introducing enough complexity to help navigate the extensive network from detailed biological mechanisms to large-scale ecosystem processes.

1.1 Nutrition Touches and Links all Living Things At the most conspicuous level, nutrition is a primary factor defining the geographic distribution and temporal pattern of activity for many animals (Raubenheimer 2010). It is true that the geographical and temporal patterns of animals are also governed by factors other than nutrition, such as the location of mates and the activities of predators. But mates, too, need to feed, and the problem with predators is that they aim to do just that. Indeed, in many animal systems the spatial and temporal patterns of reproduction are tightly geared toward resource availability. Extreme examples include the movements of caribou, wildebeest, and locusts, where hordes of animals migrate over vast distances to coordinate breeding with food availability. In other cases, such as the critically endangered New Zealand kakapo parrot, breeding occurs on average once every two to five years, when the fruits needed for rearing chicks are superabundant (Elliott et al. 2001). In yet other animals, breeding success has been linked not to food availability but specifically to the nutritional composition of prey. For example, recruitment of kittiwake gulls is higher when lipidrich fish are available compared with lipid-poor species. Experiments suggest that a lipid-poor diet does not support normal brain development, resulting in cognitively impaired chicks (Kitaysky et al. 2006). The role of nutrition in brain development has also been linked to mating success. It has been suggested, for example, that complex song learning has evolved in birds as a means of demonstrating to potential mates a high degree of cognitive competence—showing off, as it were. Nowicki and colleagues (1998) have argued that the ability to learn complex songs is dependent on good brain development, which in turn is influenced by nutrition. A complex song repertoire can thus provide an indication to females that a male has been well nourished in development, and therefore that she is mating into a family that is competent at foraging. Alternatively, a complex song repertoire could indicate that the male has genes that direct good development regardless of whether there have been nutritional perturbations during development. Either way, nutrition is central. And this is true not only for birdsong but also for other animal signals. Carotenoids, for example, are color pigments widely used in visual sig-

4  |  Chapter One

nals by birds. Since birds cannot synthesize these compounds but must obtain them from the foods they eat, a good supply of carotenoids is dependent on their foraging ability. If carotenoids are limiting in the environment, then by selecting a bright male a female can ensure that she mates with a competent forager that will be able to provide for her offspring. Furthermore, carotenoids are used not only as signals but also for a range of physiological functions including immunoregulation. Bright coloration might therefore indicate to a potential mate not only good foraging ability but also good health. This is believed to underlie the evolution of bright flanges in the mouths of nestlings of some species of birds. By preferentially feeding the chicks with brighter mouths, the parents can ensure that they direct the profits of their foraging efforts toward healthy offspring (Dugas 2010). Carotenoids are important not only in signaling and health but also in vision (Toomey and McGraw 2009). In birds, for example, carotenoids accumulate in the oil droplets of retinal cones and act as selective filters that enhance color vision. They also protect the retina, by absorbing harmful ultraviolet radiation. The UV-protective function of carotenoids can be an important determinant of the geographical distribution of animals. Sommaruga (2010), for example, has shown that the concentration of carotenoids in crustacean zooplankton species is strongly related to the extent to which their lake habitats expose them to UV radiation—plankton in clear, shallow, high-altitude lakes have higher carotenoid concentrations than those in deeper, more turbid lakes. Another class of diet-derived photoprotective compounds that has been related to the degree of UV exposure in plankton is the mycosporine-like amino acids (MAA). The balance of carotenoids to MAAs varies widely in zooplankton populations, and many interesting studies have addressed the question of what determines this balance (Hylander et al. 2009). One primary determinant is availability: MAAs occur only in some of the algal foods of zooplankton, and the availability of these varies among lakes; carotenoids, by contrast, are more widely available. A second important determinant has to do not with the trophic level below the zooplankton but that above—MAAs are colorless and therefore, unlike carotenoids, do not increase the conspicuousness of plankton to predators. Consequently, zooplankton exposed to a high risk of predation by visual predators like fish tend to adopt MAAs as the chosen sunscreen. If MAAs are not available, however, the zooplankton need to resolve the dilemma of whether to protect against UV damage but suffer increased predation, or avoid predation and suffer the ravages of sunburn. In experiments where copepods were exposed to a combination of predation, high UV, and low MAA supply, they opted for the sunburn over increased predation (Hylander et al. 2009). Other experiments have demonstrated that exposure to predation can alter the balance of macronutrients required by animals. Hawlena and

Nutrition and Darwin’s Entangled Bank  |  5

Schmitz (2010) compared the balance of protein to carbohydrate selected by grasshoppers in the presence and absence of spider predators. Their results showed that the presence of spiders caused the grasshoppers to select a diet higher in carbohydrate relative to protein, as opposed to when spiders were absent. A separate experiment suggested a reason for this: the stress caused by the presence of the predators resulted in a 32% increase in the metabolic rate of the grasshoppers, and the shift in the selected diet was evidently a compensatory response to meet these added energy costs. Other experiments have shown that locusts also compensate in this way to meet the energetic costs of flight, and rats do so to meet the costs of thermoregulation in cold environments (Raubenheimer and Simp­son 1997). Similar nutrient-specific responses have been observed in predatory beetles as they emerge from winter diapause: they initially select a diet high in fat relative to protein, and as the body fat that was depleted during the previous winter is replenished, they increase their intake of protein to meet reproductive demands (Raubenheimer et al. 2007). We have tried to illustrate in the preceding paragraphs how pervasive nutrition is: start with the habitat selection and activity patterns of animals, and you can seamlessly transition via a network of nutritional interconnections to brain development, birdsong, animal coloration, parental care, retinal function, natural sunscreens, diapause, flight, stress responses, thermoregulation, and an illustration of how nutrition can mediate complex relationships between food availability, predation risk, and threats from solar radiation. We could continue indefinitely, expanding the range of nutrients, contexts, and problems. However, the chapters that follow provide many further examples illustrating just how pervasive nutrition is, and elaborate on some of those introduced above. For now we will leave this topic and turn to the related issue of the complexity of nutrition.

1.2 Nutrition Is Complex For some animals, the challenges of feeding appear relatively straightforward. Many species of butterflies, for example, forage only for nectar, which consists largely of energy-rich sugars and water. For them, nutrition appears to be a simple process of matching carbohydrate acquisition to carbohydrate requirements. Most animals, by contrast, need to forage for more complex resources, comprising also amino acids, vitamins, minerals, and a range of other food components. But even here, foraging need not be a complex task, if the available foods contain all these nutrients in the required balance. It is widely believed that this is the case for predators. These animals are considered to feed on high-quality foods that are relatively similar to the predator and to one another in composition (i.e.,

6  |  Chapter One

the bodies of other animals), and consequently the principal challenge they face is to capture enough of these high-quality foods to satisfy their needs (Stephens and Krebs 1986). On closer inspection, however, even these apparently simple cases are deceptively complex. Butterflies, for example, are not physiologically exempt from the requirements for amino acids, vitamins, and the full range of nutrients that other animals need to survive and reproduce. Rather, in those species that as adults feed only on nectar, the task of acquiring the broader range of nutrients falls to the larval (caterpillar) stage—the adult draws on stores accumulated in its youth. The caterpillars therefore face the doubly complex foraging task of ensuring that they acquire enough of the various nutrients to satisfy their immediate needs as well as their future needs, both as adults and in the nonfeeding pupal phase, during which larval tissues are reconstructed into the adult body form. Likewise, it is also almost certainly the case that the foraging challenges of carnivores are more complex than meets the eye. First, accumulating evidence suggests that the body composition of prey animals can be highly variable (Fagan et al. 2002; Raubenheimer et al. 2007; Spitz et al. 2010; Raubenheimer 2011). Second, in common with other animals, the nutrient needs of predators are not fixed, but change—for example, as they grow, and with different levels of activity, with changes in their health status, and so forth. It is therefore unlikely that any one food will provide the right balance of nutrients throughout the life of the animal, and most predators will need to actively balance their nutrient gain by selecting foods appropriately and/or physiologically regulating the relative efficiency of nutrient retention. Third, it has been demonstrated in laboratory experiments that both vertebrate (Sánchez-Vázquez et al. 1999; Mayntz et al. 2009; Hewson-Hughes et al. 2011) and invertebrate predators (Mayntz et al. 2005) feed selectively in relation to the nutrient composition of foods, suggesting that they are adapted to dealing with variation in the match between nutrient needs and food compositions. The predatory ground beetle mentioned in the previous section illustrates all these points: its nutrient requirements change during the time it emerges from diapause and approaches reproductive maturity; its pattern of nutrient selection tracks these requirements; and its body composition (and hence its suitability as food for other predators) changes markedly during this period (Raubenheimer et al. 2007). In general, therefore, achieving nutritional homeostasis involves a complex interplay between multiple and changing nutrient needs and variable foods. Evolution has ensured that animals are equipped with mechanisms to deal with these complexities, but considerable challenges remain for nutritional biologists to understand these processes. Many aspects of the relationship between animals and their nutritional environ-

Nutrition and Darwin’s Entangled Bank  |  7

ments underscore this challenge. First, foods are complex mixtures of multiple components, each of which has its own functional implications for the animal. Some are necessary for maintaining health (e.g., essential nutrients), while others are hazardous and best avoided (e.g., the antipredator chemical defenses produced by some plants and animals [Sotka et al. 2009]). To further complicate issues, some toxins can be beneficial if ingested in low quantities, and even essential nutrients can be toxic if overingested (Raubenheimer and Simpson 2009). Second, a given food component (at a stipulated dose) can have multiple influences on an animal; we encountered an example above, where carotenoids in zooplankton influenced both resistance to the harmful effects of solar radiation and the risk of predation. Third, most aspects of animal function are influenced by many food components—predation risk, for example, is influenced both by the level of carotenoids stored in the body of a zooplankter (i.e., how conspicuous it is) and its energy stores (ability to escape if pursued). Finally, and in some respects most challenging of all, is that food components interact in intricate ways in their effects on animals. At the most fundamental level, the amount of a nutrient that can be ingested in a food depends critically on the animal’s relationship with other nutrients in the food. For example, if the food contains low levels of protein relative to carbohydrate, then the ability of the animal to satisfy its need for protein from that food depends on its capacity to overeat carbohydrate; conversely, its ability to avoid a carbohydrate overdose would depend on its capacity to endure a protein shortage. Likewise, the functional impact of toxins often depends critically on the nutritional status of the animal (Simpson and Raubenheimer 2001). Disentangling this web of interconnections can be as complex and daunting as it is important for understanding the biology of animals, and for managing the relationships between our own species and the world that we inhabit. To succeed, an approach is needed that systematically deals with each of the challenges mentioned above. Namely, it should provide a framework within which multiple food components and animal attributes can be distinguished, and the relationships among components and attributes disentangled and linked to the performance of individuals, the resulting ecological outcomes, and the evolutionary consequences.

1.3 Dealing with Nutritional Complexity: Enough but Not Too Much Above, we drew a contrast between the depth of detail with which animal and human nutritionists, on the one hand, and ecologists, on the other,

8  |  Chapter One

view nutrition, and we highlighted the middle ground as a fruitful area for further exploration. We do not mean to imply that the detailed and general approaches should be replaced by an “intermediate complexity” approach. Rather, they should be completed by such an approach. The concept we wish to emphasize in this section is that of “appropriate complexity”: the level of complexity that best suits the research question at hand. The detailed studies of human and animal nutritionists have yielded a wealth of information for deriving dietary recommendations for human health, formulating animal feeds, designing dietary regimes for captive animals, supplementing the nutrition of free-ranging animals—and many other important practical applications (e.g., Robbins 1994; Halver and Hardy 2002; National Research Council 2003; Mann and Truswell 2007; Klasing 2008). Unfortunately, such depth of detail is available only for a few species. This is understandable, because the immense amount of work needed to measure the nutritional requirements of a species—not to mention different sexes and stages in the life cycle—means that, in general, it is only those with direct value to humans that are represented among the chosen few. Similarly detailed studies are lacking on animals for making broader ecological comparisons to address, for example, the question of how nutrition might mediate between latitude and species diversity (Clements et al. 2009; Kissling et al. 2011). For such questions, a framework is needed to identify those nutrients and combinations of nutrients that are worth measuring, considering the species involved, and thus those that can be ignored in approaching complex, potentially multispecies ecological processes. At the other end of the spectrum are the foraging models employed in the ecological sciences for generalizing across a broad range of organisms (see Raubenheimer et al. 2009 for a more detailed discussion of these). Some ecologists have assembled data to argue that a particular nutrient is generally limiting to animals in the wild—for example, White (1983) has suggested that in many species a shortage of protein limits the reproductive success of populations. If true, this would justify taking a nutritionally bare-bones approach to ecology, which would exclude the messy details generated in applied nutrition. A related approach is the prominent field of optimal foraging theory (OFT) (Stephens and Krebs 1986; Stephens et al. 2007). OFT is based on the premise that foraging has evolved by natural selection to optimize food gain while minimizing costs such as time spent foraging and exposure to predation, and therefore optimization mathematics provides a useful tool for understanding the evolution of foraging. OFT models usually assume that energy is the primary foraging target of animals (called the “foraging currency”), but oc-

Nutrition and Darwin’s Entangled Bank  |  9

casionally they have focused on protein (Berteaux et al. 1998). Other food components, such as toxins and nutrients that are not represented as the foraging currency, are considered in the models as constraints with which the animal must cope in its pursuit of the chosen nutritional currency. This is often done using a technique known as linear programming (Westoby 1974; Belovsky 1990). Such unidimensional approaches to nutrition have generated valuable insights into foraging behavior and have provided a heuristic framework for thinking about ecological processes (White 1983; Stephens and Krebs 1986; Stephens et al. 2007). They have, however, contributed little to understanding which nutrients or combinations of nutrients actually do influence foraging, and how the requirements of animals for these influence ecological processes. For this, models are needed that are nutritionally explicit, in the sense that they enable a study to address these questions directly (Raubenheimer et al. 2009). Optimal foraging theorists have appreciated this need in recent years and begun to formulate models that consider more than one currency simultaneously (Hengeveld et al. 2009; Houston et al. 2011). An influential development in ecology is the framework of Ecological Stoichiometry (ES) (Sterner and Elser 2002). ES differs from OFT in that a single model usually includes two or more food components and specifically focuses on the balance of these components. ES is, therefore, nutritionally explicit in a sense, but to buy the generality needed to encompass large-scale ecological processes, ES has made a different simplifying assumption; namely, that chemical elements (which, unlike nutrient molecules, are common to all interacting species in an ecosystem) can represent nutrients. This enables the proportional “nutrient requirements” of an animal to be estimated by measuring the elemental composition of its body, and correcting for an estimate of elements lost through excretion and respiration. The suitability of a habitat for an animal can then be judged by comparing the elemental composition of available foods with the animal’s estimated requirements. Undoubtedly, ES has served as a useful guiding framework in generating an impressive body of research. But the detailed work of, among others, applied nutritionists has told us that elements do not faithfully represent nutrients. For example, a carbon atom in a molecule of sugar, in an amino acid, and in a molecule of hydrogen cyanide will not be distinguished by an elemental analyzer used to populate an ES model with numbers, but most likely would have been distinguished by the animal before it was killed, dried, and crushed for analysis. The extent to which ES models can serve as a general guiding framework for unraveling biological interactions will depend on the extent to which these inaccuracies conceal important ecological processes. This we can know only by studying the effects of nutrients in ecology.

10  |  Chapter One

1.4 Charting the Void between Nutritional Detail and Generality: The Geometric Framework In the chapters that follow we present a graphical approach that we believe can help to introduce appropriate nutritional complexity into the broader biological sciences, and generality into the applied nutritional sciences. This “Geometric Framework,” which we introduce in the next chapter, takes account of the fact that animals need multiple nutrients in changing amounts and balance, and that nutrients come packaged in foods that are often hard to find, dangerous to subdue, and costly to process. In subsequent chapters we show how the Geometric Framework has been used to understand the links between nutrition and relevant aspects of the biology of individual animals. These aspects include the physiological mechanisms that direct the nutritional interactions of the animal with its environment, and the consequences (in terms of immune responses, health, and life span) of these interactions. Having considered the implications of diet for individuals, we show that these effects can translate into the collective behavior of groups and societies, and in turn ramify throughout food webs to influence the structure of ecosystems. We then show how our framework can be used to address problems in applied nutrition, including the challenge of optimizing diets for domestic animals and for conserving endangered species. Thereafter we turn to a specific problem in applied nutrition, showing how the epidemic of human obesity and metabolic disease is linked to changes in the nutritional balance of our diet, with a primary role for protein appetite driving excess energy intake when people adopt a modern Western diet. We close with a discussion of what we consider to be the priority issues for the way forward, if nutrition is to reach its potential contribution toward unraveling Darwin’s entangled bank.

two

The Geometry of Nutrition

We have seen in chapter 1 that animals face complex challenges in satisfying their multiple nutrient needs. These involve choosing the right foods, deciding how much of each to eat, regulating the efficiency with which nutrients are retained in the body, and regulating the amounts of retained nutrients that are allocated to different functions such as energy metabolism, energy and nutrient storage, tissue growth, secretions, and so on. How can we go about understanding this complex set of interrelations between an animal’s nutrient needs, its nutritional environment, and its responses to the nutritional environment? A common approach in nutritional studies is to assume, for simplicity, that one environmental factor, such as the energy, protein, or the toxin content of foods, and one animal attribute, for example, intake rate or food choice, dominates in the relationship between animals and their nutritional environment. In many cases, however, important questions about animal nutrition can be answered only by using an integrative approach, which takes into account several attributes of the environment and of the animal, and enables the researcher to study the interactions among these components. In this chapter we describe a geometric approach that we first introduced in 1993 for this purpose, known as the Geometric Framework for nutrition (GF) (Raubenheimer and Simpson 1993; Simpson and Raubenheimer 1993b). In subsequent chapters we show how GF has helped to answer a range of important questions in nutritional biology, from the relationships between nutrition, aging, and reproduction, to the vulnerability of humans to obesity.

2.1 The Geometric Framework: Basic Theory What are the core requirements for an integrative framework of nutrition? Since the job of such a framework is to help us to understand the ways that animals relate to their environment through nutrition, an integrative framework must be able to represent the animal, the environ-

12  |  Chapter two

ment, and the nutritional basis for the interaction between animal and environment (Raubenheimer et al. 2009). A second requirement is that an integrative framework should take account of the fact that the nutritional interactions between animals and their environment take place on a stage that is constructed of many food components. Third, if the framework is to be grounded within the powerful paradigm of evolutionary biology, it is imperative that the consequences for the animal of its behavioral and physiological responses to the nutritional environment can be represented. These consequences—which include such factors as reproductive rates, development rates, and the risk of premature death (e.g., the strength of the immune system)—are relevant not only to individual well-being but also to population sizes. A framework that takes into account consequences therefore extends its reach both to evolution and also to nutritional aspects of population ecology. A further step, to community ecology, is achievable if the framework fulfills a fourth requirement, that of being able to incorporate the nutritional basis of interactions among multiple species in food webs. GF was designed with these core requirements in mind. It satisfies the multiple-food-components requirement using a simple device known as a nutrient space. A nutrient space is a geometric space built of two or more axes, where each axis represents a food component that is suspected to play a role in influencing the animal’s responses to its environment. In most cases these food components will be nutrients but, as discussed in chapter 5, this is not invariably the case. The nutrient space provides the common context in which to describe the pertinent aspects of the animal, its environment, the interactions between animal and environment, and the consequences of these interactions. In the sections that follow we describe how this is achieved.

Nutrient Needs: The Intake Target An aspect of the animal that is fundamental to its interactions with the environment is its nutrient requirements—the amounts and balance of nutrients that it must eat to gain maximal benefit (usually evaluated in terms of evolutionary fitness). Nutrient requirements are fundamental in nutritional models because they provide a reference point for predicting how an animal should respond to its environment and for understanding the relationships between foods and the performance consequences of eating those foods. For example, if we know the nutrient needs of an animal and the nutritional composition of the foods available to it, we can make a prediction about which foods it would eat and which it should avoid, and if it did otherwise we would be justified in wondering why. Likewise, if the animal persisted in eating foods that did not satisfy its

The Geometry of Nutrition  |  13

nutrient needs, we would have good cause to wonder about the long-term consequences of this—would it become obese, have impaired reproduction, and/or become susceptible to infectious diseases? In GF models, the optimal nutrient requirement of an animal is represented as a point in a nutrient space, called an intake target. For example, figure 2.1A shows a nutrient space constructed from axes for protein (the x axis) and carbohydrate (the y axis). We could have chosen any two nutrients, or expanded the space to include more nutrient dimensions (including splitting protein into its 20 amino acids), but as we shall see in subsequent chapters, the macronutrients (protein, carbohydrate, and lipid) are a good place to start. The point labeled IT shows the optimal amount of protein (3.5 g) and carbohydrate (3.5 g) that should be eaten by the hypothetical animal over a given time period (e.g., a day, or the duration of the experiment). In some cases the intake target might more accurately be represented as a small region, but for illustrative purposes we will for now stick to the simpler case of a discrete point.

Foods Except for special cases such as salt licks, nutrients in the environment come packaged together as mixtures—foods. The diet of the animal may comprise a single food or, more usually, a combination of different amounts of several foods. Foods are modeled in a nutrient space in two ways: by the amounts of nutrients they contain and, more generally, by the balance of nutrients they contain. In figure 2.1A, for example, the point FIa represents a food item, such as a leaf, which contains 2 g of protein and 2 g of carbohydrate. Similarly, a larger leaf that contained 7 g of protein and 7 g of carbohydrate would be plotted as a point with coordinates (7:7) (FIb), while a food containing 3 g of protein and 6 g of carbohydrate would be plotted as point FIc. For many purposes, some of which will be discussed shortly, it is useful to disregard the precise amount of nutrients in a specific food item, considering rather the balance of the nutrients. This general property of a food type is pictured in a geometric model as a line that passes through the origin of the graph (the point 0:0) and through a point representing any quantity of that food. The slope of such a line indicates the balance of the nutrients in the food. The line labeled Foods a & b, for example, shows the balance of protein and carbohydrate in food item FIa, which is the same as the balance of the nutrients in food item FIb (1:1), while food item FIc has a balance of 6 parts carbohydrate to 3 parts protein and is therefore represented by a steeper line. Such lines representing the balance of nutrients in foods are called nutritional rails, for reasons that will become clear below.

14  |  Chapter two A.

B. Food c

10

8 Flb Flc

6

4 IT 2

0

Fla Ti 0

2

4 6 Protein eaten (g)

8

6 Food d

4

2

Tiii

IT

Tii 0

2

4 6 Protein eaten (g)

8

10

D.

10

Food e

8

Food c

10

6

Iiii Ii

4

Carbohydrate eaten (g)

Iii C++

Carbohydrate eaten (g)

8

0

10

C.

IT P--

2

0

Food c

Foods a & b Carbohydrate eaten (g)

Carbohydrate eaten (g)

10

8 6 4 2

0

2

4 6 Protein eaten (g)

8

10

0

0

2

4 6 8 Protein eaten (g)

Fig. 2.1. Protein-carbohydrate nutrient space for a hypothetical animal and several foods. The intake target (labeled IT) represents the amounts and balance of the two nutrients that are optimally required by the animal (3.5 g of each). The small solid circles in nutrient space (A) show the amounts of the two nutrients in different food items (FIa, FIb, and FIc), and the dashed radials (termed nutritional rails) represent the balance of the nutrients in each food item. In nutrient space (B), where there are no points representing specific food items, the nutritional rails represent the balance of the nutrients in the food type—­i.e., without specifying a quantity of the food. The solid arrows (Ti, Tii, and Tiii) in (A) and (B) show the trajectory over which the animal’s nutritional state changes as it eats, each step being parallel to the nutritional rail for the food being

10

The Geometry of Nutrition  |  15

Reaching the Intake Target If foods are the most fundamental aspect of an animal’s nutritional environment, then feeding is the primary nexus of interaction with this aspect of the environment. For the animal, the important thing about feeding is that it provides a means to change—and regulate—its nutritional state. Feeding is therefore represented in GF models by the change in the nutritional state of the animal that results from eating. This change is plotted as a trajectory through nutrient space. Such feeding trajectories have the

eaten. Food items FIa and FIb contain the same balance of the nutrients as the intake target—­i.e., these foods are nutritionally balanced with respect to protein and carbohydrate. The rail for food c, by contrast, does not pass through the intake target—­i.e., this food is nutritionally imbalanced, and on its own does not allow the animal to reach its intake target. However, since foods c and d fall on opposite sides of the intake target, the animal can reach its intake target by combining its intake from the two foods—­i.e., these foods are nutritionally complementary. The two sequences of arrows in (B), solid and dotted, show two alternative routes that the animal could navigate to its intake target. Nutrient space (C) shows the options available to the animal when confined to a single imbalanced food type (food c). If it feeds to intake point Ii, it gains the required amount of carbohydrate but suffers a shortfall of protein (P–­ –­); at point Iii it satisfies its protein needs but overingests carbohydrate (C++), and at point Iiii it suffers both a moderate shortage of protein and a moderate excess of carbohydrate. The way that the animal resolves this trade-­off between overingesting some nutrients and underingesting others when restricted to nutritionally imbalanced diets is known as a rule of compromise. To measure rules of compromise, an experiment is performed (nutrient space D) involving several groups of animals, each of which is confined to a food that has a different balance of the nutrients and is thus represented by a different nutritional rail. Such an experiment will yield an array of intake points, the shape of which reveals the rule of compromise. The vertical array of intake points indicates the strategy represented by intake point Iii in (C) (i.e., prioritize protein), and the horizontal array represents the strategy at Ii (prioritize carbohydrate). The third array (shaded circles) shows an instance where the intake array is asymmetrical—­i.e., the response is different for foods containing surplus protein (a line with a negative slope) and surplus carbohydrate (an arc). The former, known as the Equal Distance rule, corresponds with eating to the point on the respective rails where the deficit of one nutrient equals the surplus of the other. The arc, known as the Closest Distance rule, corresponds with eating to the point on the respective rails that minimizes the geometric distance to the intake target.

16  |  Chapter two

same angle as the nutritional rail for the food being eaten, because as the animal eats it gains the nutrients in the same proportion as they are present in the food being eaten. In figure 2.1A, for example, the arrow labeled Ti shows the intake trajectory for an animal eating food item FIb, which is exactly superimposed on the nutritional rail for this food. By feeding, animals are therefore channeled, like a train, along tracks in nutrient space set by the nutritional rails of the foods they select, with the distance of movement along these rails being determined by the amount of the food eaten. The challenge for animals is to select foods that direct them to their intake target, and ensure that they eat enough to arrive there. The simplest way to do this is to eat a food that leads directly to the target—that is, contains the same balance of the nutrients as is needed—and is plentiful enough to enable the animal to get there. For example, both food items FIa and FIb in figure 2.1A have rails that pass through the intake target, and would thus provide a direct route there. However, if the animal ate only the food item FIa, which contains 2 g protein and 2 g carbohydrate, it would fall short of its target of 3.5 g protein and 3.5 g carbohydrate. It could, on the other hand, reach the target by choosing food item FIb, or by supplementing its intake of FIa with FIb. Both foods are nutritionally balanced with respect to the animal’s protein and carbohydrate needs, but food item FIa is insufficient. Food item FIc, by contrast, is nutritionally imbalanced—it does not lead to the intake target but contains a greater proportion of carbohydrate to protein than is needed by the animal. The animal could nonetheless use this food to navigate indirectly to the intake target if it combined its intake from FIc with a second imbalanced food whose rail falls on the opposite side of the intake target. For example, figure 2.1B shows the intake trajectory of a hypothetical animal that feeds first on food c (over trajectory Tii), then switches to food d and so moves along trajectory Tiii, which is parallel to the nutritional rail for this food. Although this example involves a two-step route to the target, by mixing foods in this way the animal could alternate trajectories in any number of patterns and zigzag its way through nutrient space to the target: it might, for example, switch between foods more often, as shown by the dotted trajectory. Pairs of nutritionally imbalanced foods that jointly enable animals to reach their intake target in this way are known as nutritionally complementary foods. As we will see below, the fact that many animals are able to reach their intake target by mixing nutritionally complementary foods provides a useful means for researchers to estimate the position of the intake target, based on the testable prediction that animals will have evolved regulatory mechanisms that ensure ingestion of a balanced diet (e.g., Chambers et al. 1995; Raubenheimer and Jones 2006).

The Geometry of Nutrition  |  17

What to Eat When the Intake Target Can’t Be Reached: Rules of Compromise In some cases, the animal might have access neither to nutritionally balanced nor nutritionally complementary foods, but only to noncomplementary nutritionally imbalanced foods (e.g., food c and food e in fig. 2.1C). How should an animal in this predicament respond? At first sight, it seems reasonable to expect that the best option would be to make the most of a bad situation and get as close to the intake target as it can, given the available foods. This is, to some extent, true: in most cases the animal would do better by eating something rather than starving (not proceeding beyond the origin), and the best choice would be to eat food c because this nutritional rail would take it closer to the target. But how much of food c should it eat? Here things get a bit more complicated. The complication is that there are different measures of the distance from the target, and without performing the relevant experiments we have no way of knowing which measure is most important to the animal. For example, if the animal ate enough of the food to get to intake point Ii, then it would be as close to the target as possible with respect to its carbohydrate requirement—it would actually reach the target in the carbohydrate dimension—but in so doing would suffer a protein shortage of P– –. Alternatively, it could reach the target in the protein dimension by feeding to point Iii, but only by overeating carbohydrate by C++. A third option would be to feed to some point intermediate between these extremes (e.g., point Iiii), in which case the animal would reach the target in neither dimension, but the total distance from the target (protein distance + carbohydrate distance) would be smaller than in the other two strategies. The animal’s response in this circumstance is known as a rule of compromise, because it reflects the compromise selected by the animal between overeating some nutrients and undereating others. If we observe that when feeding on food c the animal adopts the strategy “feed to point Ii,” this tells us that the animal is prepared to accept a deficit of protein of P– – in order to avoid eating a surplus of carbohydrate. It does not, however, tell us how it would respond if it only had food e available, with a greater surplus of carbohydrate relative to protein than food c, or foods that were imbalanced in the opposite direction (e.g., food d from fig. 2.1B). To get a general description of the rule of compromise, we would therefore have to measure the responses of several treatment groups of the animal, each of which is confined to a food that has a different balance of the nutrients and is thus represented by a different nutritional rail (fig. 2.1D). The intake points for such an experiment would jointly form an intake array, the shape of which provides a

18  |  Chapter two

comprehensive description of the rule of compromise. Figure 2.1D shows how the intake arrays would look if strategies Ii and Iii applied symmetrically across all diets (the arrays of points connected by horizontal and vertical lines, respectively), as well as an example where different strategies are adopted for diets with surplus carbohydrate and surplus protein (the shaded circles). In the asymmetrical case, the arc-shaped array of points to the left of the intake target indicates that when feeding on foods that have a surplus of carbohydrate relative to protein, the animal eats an amount that minimizes the “as-the-crow-flies” distance between its actual intake and the intake target. This is known as the Closest Distance rule of compromise. To the right of the intake target, the points form a diagonal array, showing that when protein is surplus in the diet, the animal feeds to the point on its respective rails where the shortage of carbohydrate exactly equals the surplus of protein eaten. This is known as the Equal Distance rule. While there are, in theory, an infinite number of possible rules of compromise, we will return to the Closest Distance and Equal Distance rules below.

Processing Ingested Nutrients Why should animals differ in the rules of compromise they adopt? One way to understand this issue is to examine the capacity of animals to deal flexibly with nutrient excesses and shortages. We would expect, for example, that an animal might adopt the vertical rule of compromise in figure 2.1D if it had little flexibility in relation to its protein requirements but was particularly proficient at excreting or storing excess carbohydrate; the horizontal rule might be expected if the converse was true, and the Closest Distance or Equal Distance rules if the animal had some flexibility with respect to both nutrients. In thinking about these relationships, it is useful to look in more detail at the concept of “nutrient requirements.” Our use of the phrase “nutrient requirements” up to this point refers to the amounts and balance of nutrients that an animal needs to eat, which we have represented in geometric models as an intake target. There is, however, an even more fundamental context in which this term can be used, and that is in relation to the amounts and balance of nutrients the animal needs to make available to its tissues to satisfy its various demands for energy metabolism, tissue growth, nutrient and energy storage, and so forth. How does this requirement, which we call the nutrient target, differ from the intake target? It differs because animals are usually unable to allocate 100% of a nutrient they eat to useful purposes but instead lose some in the feces, urine, and, in some animals, through per-

The Geometry of Nutrition  |  19

meable membranes on the body surface (e.g., the gills of fish). To satisfy its tissue-level requirements (the nutrient target), an animal therefore needs to compensate by eating enough to cover both the nutrient target and the constrained losses. These two components—required amount + constrained losses—jointly comprise the intake target. How can we measure a nutrient target? It is a daunting task indeed to estimate the separate requirements of various tissues for two or more nutrients simultaneously. Fortunately, there is a convenient shortcut. We can be confident that millions of years of natural selection have equipped animals with a physiology that is well qualified to estimate their own nutrient requirements, and by performing simple experiments we can coax the homeostatic mechanisms into revealing these (Zanotto et al. 1993, 1997). Homeostatic mechanisms, by definition, respond to the relationship between demand and supply (Raubenheimer et al. 2012): nutrients in short supply relative to requirements are retained with high efficiency, whereas ingested nutrients that are surplus to requirements are excreted. Consequently, if the amount of a nutrient eaten is plotted on the x axis of a graph and the amount excreted on the y axis, there will be two phases in the relationship (Raubenheimer and Simpson 1994). At intakes of the nutrient that are lower than the required amount the relationship will have a shallow slope (much of what is ingested is retained), whereas for intakes that exceed the requirement the relationship will have a steep slope (the excretion rate increases). To estimate the nutrient target, therefore, we need only to perform an experiment in which the animal is coaxed into eating different amounts of the nutrient of interest—for example, as in figure 2.1C and D—and measure the amounts of the nutrient eaten and excreted over an appropriate time period. The nutrient target can then be identified as the amount of the nutrient retained (i.e., the difference between intake and excretion) at the point of intake where the excretion rate increases. An example using data from locusts is shown later in this chapter (fig. 2.5B). In some cases, however, the regulatory mechanisms might be tricked into retaining deleterious amounts of the nutrient, for example when operating in an inappropriate environment. We briefly return to this point later in the current chapter, and in chapter 10 explore it in some detail in the context of our own species. The reason why we might want to include both intake targets and nutrient targets in geometric models is that doing so helps us to better visualize the nutritional strategies of animals. For example, it might be that two animals have identical nutrient targets but one animal has an intake target that is higher on the protein axis than the other. This raises interesting questions about why the second animal is more efficient than

20  |  Chapter two

the first at processing ingested protein: has it evolved under circumstances of protein shortage, or has the other animal evolved to select high-protein foods and habitually excrete the excess? Just as the intake target can be partitioned into the nutrient target + constrained losses, so too can the nutrient target be partitioned into various subtargets. When an animal reaches its nutrient target, the total pool of available nutrients needs to be divided among several functions, including growth, metabolism, and reproduction (Raubenheimer and Simp­son 1992, 1994, 1995). How the animal allocates these nutrients is critical to fitness, and as a result natural selection has fashioned animal physiology to achieve a favorable strategy for investing its nutritional “income” across its various requirements. An animal with an annual life cycle might, for example, benefit from investing protein into growth at a high rate, and therefore have a growth target that is higher on the protein axis than a slower-growing animal of similar size. Conversely, the slower-growing animal might prioritize energy storage to a greater extent, in which case its growth target would be lower on the protein axis but higher on the carbohydrate axis. Experiments show that, within limits, animals are able to regulate nutrient utilization postingestively to defend a growth target, even when eating foods that prevent them from reaching the intake target—for example, by adjusting the digestion and absorption of nutrients from the gut (e.g., Clissold et al. 2010) and by voiding excess ingested nutrients postabsorptively (e.g., Zanotto et al. 1993, 1997). The above discussion concerns only optimal nutrient allocation to growth, but we could equally model metabolic targets, reproductive targets, and so forth. The overall point, however, is to demonstrate that the Geometric Framework can be used to construct multidimensional models of nutrient budgets, which include functionally optimal nutrient allocations (targets) as well as the allocations that are actually achieved (Rau­ ben­heimer and Simpson 1995). If we had additional information about the benefits of achieving these optima (or the costs of failing to do so), this would put us in a good position to understand how various nutritional strategies evolve. Modeling such costs and benefits is the subject to which we now turn.

The Consequences of Nutritional Imbalance If we assume that the nutritional responses of animals, including regulation to an intake target, rules of compromise, and postingestive regulation, have been fashioned by natural selection, it follows that an animal that achieves its intake target will enjoy maximal Darwinian fitness (or

The Geometry of Nutrition  |  21

put another way, suffer minimal fitness costs). The challenge is how to include such fitness consequences in geometric models. They differ from the components we have so far integrated into the models—foods, nutrient requirements, nutrient intake, and postingestive utilization—in that fitness components are not measured in terms of nutrients, but in terms of other units such as numbers of offspring, probability of premature death, longevity, and so forth. Fitness consequences cannot, therefore, be depicted in the nutrient space in the usual way. Instead, we have borrowed from evolutionary biologists the metaphor of the “fitness landscape,” in which the consequences of nutrient intake are plotted in a maplike surface superimposed on the nutrient space. This enables us to envisage the intake target state as the summit of a Darwinian fitness mountain mapped onto nutrient intake space. Hence, if we consider two nutrients, say protein and carbohydrate, a given intake of both nutrients is equivalent to a location on a topographic map, with the elevation of the fitness landscape at that point being read off the map as a contour line. Foods (represented as food rails) offer pathways across this landscape. In some nutritional environments, there will be foods that provide tracks leading to the summit, either directly (a perfectly balanced food) or zigzagging up the mountain (nutritionally complementary foods, as in fig. 2.1B). In other situations the summit of the fitness mountain cannot be reached, because there are no tracks leading there. These are nutritionally imbalanced environments. Under such circumstances evolutionary logic posits that the animal will ascend to the highest fitness point available to it. The shape of the fitness landscape will define where these “local optima” lie for different nutritionally imbalanced diets, and hence the nature of the rule of compromise. The slope, steepness, and curvature of the decline in elevation with distance from the intake target need not be the same for excesses and deficits of the same nutrient, or be the same for different nutrients, or be independent between nutrients. For example, it may be more costly to eat too little than too much carbohydrate, or more costly to eat too little protein than too much carbohydrate, or the cost of excess carbohydrate might increase with the extent to which protein is undereaten. The fitness mountain can therefore take on an infinity of shapes—it could be conical like a textbook volcano, have gentle slopes and steep faces, possess a single peak or a summit ridge, and so on. Fitness costs can be described to take account of these various possibilities using a mathematical function known as a Taylor series expansion (Simpson et al. 2004). Some simple examples of the resulting fitness landscapes and optimal feeding strategies (rules of compromise) derived from this mathematical function are illustrated and described in figure 2.2.

22  |  Chapter two

2.2 The Geometric Framework in Practice Armed with an overview of the basic concepts of nutritional geometry, we now proceed to the critical issue of how GF is used in practice. Our aim here is to present some examples where the concepts discussed above have been examined in real animals, and to elaborate on some practical issues involved in performing these experiments.

Position of the Intake Target Our first example concerns the important issue of measuring the position of the intake target. As mentioned previously, a convenient way to do this is to present the animal with two or more nutritionally complementary foods and let it navigate to its preferred position in the nutrient space. This approach is based on the assumption that natural selection has equipped the animal to regulate its intake to the optimal point when given the opportunity, and we can therefore use the chosen point as an indication of the position of the intake target (Simpson and Raubenheimer 1995). There are two important caveats to this approach. First, the intake target is defined as the point of nutrient intake that is most beneficial for the animal to achieve. Therefore, to be confident that the position in the nutrient space that is selected by the animal is an intake target, we should ensure that reaching this point does, in fact, provide an overall fitness benefit compared with other points. We return to this issue later in this chapter. A second caveat is that we need to ensure that the chosen point in nutrient space is indeed the outcome of nutritional regulation. It might be, for example, that the animal mixes its intake from the foods in a way that has nothing to do with their nutrient content—perhaps it randomly takes similar amounts of its total intake from each of the foods, or it simply eats more of one food because of its physical consistency or taste. In this case the animal will end up somewhere in the nutrient space, but we would be remiss to conclude that this is an intake target. To be satisfied that the chosen point is an intake target, we would need to show that it reflects the operation of mechanisms that actively regulate the nutritional state of the animal. Fortunately, demonstrating active regulation need not require a detailed understanding of the physiological mechanisms regulating feeding. Instead, we can design an experiment in which the animal is challenged not only to select but also to defend a target position in the face of variation in its circumstances. One way to do this is to test whether the animal selects the same position in nutrient space when provided with different

The Geometry of Nutrition  |  23 A. 100

80

80

60

60

40

40

20

20

Fitness cost

a. 100

0

0

50

100

150

200

0 100

80

80

60

60

40

40

20

20

Fitness cost

b. 100

0

0

50

100

150

200

0 100

80

80

60

60

40

40

20

20

Fitness cost

c. 100

0

0

50

100 150 Protein eaten

200

0

0

50

100

150

200

0

50

100

150

200

0

50 100 150 Carbohydrate eaten

200

Fig. 2.2. (A) Three different ways in which fitness costs might depend on ingested excesses and deficits of protein (left column) and carbohydrate (right column). For illustration, the target intake of each nutrient is considered to be 100 units, at which intake the fitness costs are zero. Example (a) shows linear costs (i.e., costs increase linearly with deviation above or below the target intake); example (b) shows symmetrical quadratic costs (i.e., the costs of surpluses are the same as those of deficits, and the cost functions are curved, not linear); example (c) shows asymmetrical quadratic costs (i.e., costs of equivalent-size excesses and deficits are not the same). (Fig. 2.2. continues on next page)

24  |  Chapter two B.

Carbohydrate eaten

a.

200 150 100

Carbohydrate eaten

100 50

0

50

100

150

200

150 100

T

20 10 5 2

Carbohydrate eaten Carbohydrate eaten

100

150

200

0

50

100

150

200

0

50

100

150

200

0

50 100 150 Protein eaten

200

50

0

50

100

150

200

0 200

150 T

100

10 2 5

20

150 100

50

50

0

200

50

100

150

200

100

0 200

20 10 5 2 T

150

150 100

50 0

50

100

200

d.

0

150

50

0

0 200

200

0 c.

150

50 0

b.

200 80 60 100 40 20 T

50

0

50 100 150 Protein eaten

200

0

The Geometry of Nutrition  |  25

combinations of nutritionally complementary foods. If the position of the selected point differed according to the food combination, then we would need to think again about how to identify the intake target. On the other hand, if the animal selected the same point in nutrient space regardless of which food combination it had, then this would demonstrate that the point reflects active regulation. In effect, we would have shown that the animal is prepared to change its behavior (the relative proportions of the different foods eaten) to defend an outcome (the amounts and balance of nutrients eaten), indicating that it has evolved mechanisms to achieve that outcome, as would be expected for an intake target. An example of such an experiment is presented in figure 2.3A, involving data collected by Chambers and coworkers (1995) on nymphs of the African migratory locust (Locusta migratoria). The open squares show the intakes that would be expected if the four treatment groups had simply eaten the same amount of the two foods available regardless of their nutrient content, and the filled circles show the actual intakes. The results

Fig. 2.2. (continued) (B) Fitness landscapes and associated optimal intake arrays, where the costs of exceeding or underachieving the intake target (T) for two nutrients (protein and carbohydrate) are modeled using a Taylor series expansion. Dashed lines in the left panels are fitness contours, and numbers on the contours show the associated fitness costs. Optimal fitness (zero cost) is at the intake target. In the right panels, thin lines represent selected feeding rails (food compositions), and thick lines are the optimal intake strategies, where maximum fitness is obtained along each food rail. Example (a) shows linear costs of excesses and deficits (see fig. 2.2A[a]). The fitness contours are straight parallel lines in each of the quadrants around the target intake, T. Example (b) shows symmetrical quadratic costs (see fig. 2.2A[b]). The fitness contours are ellipses, and the optimal strategies (thick lines) are also ellipses. Example (c) shows asymmetrical quadratic costs (see fig. 2.2A[c]). The fitness contours are ellipses in each of the quadrants around T, and the optimal strategies are also ellipses in each quadrant. Example (d) shows symmetrical quadratic costs with interaction costs (i.e., costs of excesses or deficits of one nutrient are dependent on the other nutrient). (The latter case is not illustrated in fig. 2.2A, where in each of the three examples the costs of surpluses and deficits for one nutrient are independent of the other nutrient). The fitness contours are now tilted ellipses, and the optimal strategies are more linear than in (b). The quadratic component of the fitness cost is the same as in (b). (From Simpson et al. 2004.)

26  |  Chapter two B.

400

14:28 or 7:14

Carbohydrate eaten (mg)

350 300 250

28:14 or 14:7

200 150 100

Carbohydrate eaten (mg)

A.

25

20

15

10

5

50 0

0

50

100 150 200 250 300 350 400 Protein eaten (mg)

0

0

2

4 6 8 Protein eaten (mg)

Fig. 2.3. Two experiments demonstrating defense of a protein to carbohydrate intake target. (A) Locusts (Locusta migratoria) given one of four food combinations (% protein to % carbohydrate = 14:28 or 7:14 and 28:14 or 14:7) reached the same point in nutrient space by altering the relative amounts of the two foods eaten (data from Chambers et al. 1995). The closed circles represent the bivariate mean (+ standard error) protein and carbohydrate intakes of the four experimental groups. The open squares indicate the expected outcome if there were no homeostatic regulation. (B) Female German cockroaches (Blattella germanica) were subjected to a 48-hour pretreatment in which they were confined to one of three nutritionally imbalanced foods. The dashed lines represent nutritional rails of the three pretreatment foods, the steepest being 4:56 (excess carbohydrate), followed by 15:45 (intermediate) and the shallowest 47:13 (excess protein). The first point in each series (i.e., that closest to the origin) shows the bivariate mean (+ standard error) for the amounts of protein and carbohydrate ingested by the end of the pretreatment period (solid triangles = low protein, hollow squares intermediate protein, and solid circles high protein). Note that these points fall on the respective nutritional rails, because during the pretreatment the insects had access only to a single food and so were constrained to ingest protein and carbohydrate in the same proportion that they were present in the food. Subsequent points in each series represent cumulative intake after a further 4, 10, 24, 48, and 120 hours in which the insects had access to all three foods, and so were no longer confined to the pretreatment rail but could move freely within the nutrient space. (After Raubenheimer and Jones 2006.)

10

12

The Geometry of Nutrition  |  27

show that the locusts from the four treatment groups reached a statistically indistinguishable point in nutrient space, despite the marked dif­ ferences in the food combinations they had available. This demonstrates that food mixing by these insects is geared toward gaining a specific amount and balance of macronutrients, strongly suggesting that the selected point is the protein-carbohydrate intake target. In the above example, the four groups of locusts were entered into the experiment in a similar nutritional state and challenged to select a nutrient intake from different combinations of foods. Raubenheimer and Jones (2006) performed an experiment on German cockroaches that illustrates a very different approach to measuring the position of intake targets. This experiment involved three groups of cockroaches, each of which was put through two phases. In the first phase, the three groups were each restricted to a single food type that had either a high, intermediate, or low protein to carbohydrate balance, and in this way manipulated into one of three nutritional states. In the second phase of the experiment, all three groups were provided with access to all three foods, thus enabling them to reveal their preferred trajectory and position within the nutrient space. If macronutrient balance was not important to these animals, then in phase two of the experiment all three groups would spread their feeding similarly across the three choice foods and in so doing take similar trajectories through nutrient space from their different starting positions. On the other hand, if feeding was geared toward achieving a particular macronutrient target, then the animals would take very different trajectories (to compensate for the difference in their starting positions) that would converge on a common point in nutrient space (the intake target). The data, presented in figure 2.3B, show very clearly that the cockroaches in this experiment took the latter option, behaving just like nutrient-seeking missiles and suggesting that we had identified the intake target.

Rules of Compromise The above experiments show that, when possible, both locusts and cockroaches mix their intake from nutritionally complementary foods to achieve an intake target. How would they respond when ecological or other circumstances prevent them from achieving the target? We have attempted to answer this question by performing experiments in which several groups of insects are all confined for the duration of the experiment to one of a range of foods varying in the protein to carbohydrate balance. As discussed in the theory section above, the points to which the different treatment groups feed on their respective nutritional rails collectively form an intake array that describes the rule adopted for

28  |  Chapter two

Carbohydrate eaten (mg)

A.

B.

200

200

150

150

100

100

50

50

0

0

50

100 150 Protein eaten (mg)

200

0

0

50

100 150 Protein eaten (mg)

Fig. 2.4. Intake arrays for the first three days of the fifth larval stadium of Locusta migratoria (A) and Schistocerca gregaria (B). Points with bidirectional standard error bars represent the intake target selected by insects given a simultaneous choice of two nutritionally complementary foods. Error bars for animals given a single food type are not bidirectional, as these were constrained to vary along the food rails. (Data from Raubenheimer and Simpson 2003.)

trading off the excesses and deficits of the two nutrients—the rule of compromise. The results of such an experiment are shown in figure 2.4A, involving the same age-group of African migratory locusts as featured in figure 2.3A (Raubenheimer and Simpson 2003). The intake array for these insects is arc-shaped, forming a segment of a circle that is centered midway between the origin and the intake target. Geometrically, this rule of compromise results from each group of locusts feeding to the point where their respective nutritional rails pass most closely to the intake target. They are, in other words, minimizing the “as-the-crow-flies” distance between the nutrient intake they achieve and the intake target. This configuration is the Closest Distance rule, to which we were introduced in the context of figure 2.1D. The results of an identical experiment performed on the same developmental stage of another species of locust, the desert locust (Schistocerca gregaria), are shown in figure 2.4B (Raubenheimer and Simpson 2003). The intake array in this case is not arc-shaped, but a negatively sloped line with a gradient of 45 degrees (that is, a slope of -1). Geometrically, this pattern corresponds with feeding to the point on the respective rails

200

The Geometry of Nutrition  |  29

where the distance from the target in the carbohydrate dimension equals the distance in the protein dimension. This is the Equal Distance rule, which we introduced in figure 2.1D. An interesting aspect of the comparison between these two locust species is that the Equal Distance rule adopted by the desert locust involves feeding to points on the respective nutritional rails that are further from the intake target than they would have achieved if they followed the Closest Distance rule. Why might these insects voluntarily accept a greater discrepancy than they need to between their actual nutrient gains and the target gains? We have already addressed this in the most general sense in our theoretical discussion on performance consequences: there are likely to be differences between these species in the function relating macronutrient intake to fitness. The important question, however, is how these functions relate to real biology—in other words, how do desert locusts benefit from adopting the Equal Distance rule or, conversely, what would be the costs of adopting the Closest Distance rule? We postpone discussion of this interesting question to chapter 6 and instead now provide an example illustrating the flexible use of postingestive processing in nutritional regulation.

Postingestive Regulation A protein-carbohydrate nutrient space for the same species of locust we encountered above, the African migratory locust, is presented in figure 2.5A. Plotted in this space are nutrient intakes across the final juvenile stage (the constellation of points furthest from the origin) and also the body compositions of the animals at the end of this period (the cluster of points closest to the origin). The key message is that the growth points clustered tightly within the nutrient space compared with the intake points, which variedly considerably. Indeed, if we exclude from consideration the animals that were given very low- or very high-protein foods (distinguished in the figure as hollow triangles), then the growth points clustered very tightly indeed, suggesting that the region where they converge is a growth target. Since the distance in the nutrient space separating intake and growth points represents the amounts of the nutrients that were eaten but not retained in the body (i.e., voided through defecation, excretion, and respiration), these data show that locusts are able to regulate nutrient utilization postingestively to defend a growth target, even when eating foods that prevent them from reaching the intake target (fig. 2.5B; see chapter 3). It is worth commenting briefly on a point of detail concerning the comparison between figures 2.4A and 2.5. The alert reader might have noticed that the intake arrays in these figures differ, even though they

30  |  Chapter two

Carbohydrate eaten or grown (mg)

A.

700 600 500 400 300 200 100

B.

0

0

40

1:1

Diet 7 7 Diet 7 21 Diet 21 7 Diet 21 21

35 Egested nitrogen (mg)

100 200 300 400 500 600 700 Protein eaten or grown (mg)

30 25

IT

20 15 10 5 0

NT = IT × 0.7 0

10

20

50 30 40 Ingested nitrogen (mg)

60

70

80

Fig. 2.5. (A) Intake and growth points across the fifth stadium of Locusta migratoria fed one of a range of foods differing in the protein to digestible carbohydrate ratio. The open square with bidirectional error bars indicates the intake target. Three groups of animals that differed in their growth response (a central cluster of points, and unusually high and unusually low carbohydrate-derived growth) are matched to intake points using solid circles, open triangles, and open inverted triangles, respectively. (Data from Raubenheimer and Simpson 1993.) (B) The data plotted here show how regulation of protein-derived growth was achieved despite variation between diets in protein intake. Nitrogen in the feces is plotted against nitrogen eaten (a surrogate for protein) in locusts confined to one of four diets differing in the protein to digestible carbohydrate ratio. Up to intakes of 30 mg nitrogen (the intake target, IT, which equates to 180 mg protein), levels of fecal nitrogen were low (70% of ingested nitrogen was retained and 30% was lost, the latter reflecting inevitable wastage). In contrast, excess nitrogen beyond the IT

The Geometry of Nutrition  |  31

were measured for the same species. This is not because of inconsistency in the locusts or poor experimental technique on our behalf, but because of the different timescales over which the two experiments were run: the data in figure 2.4A were measured over a fixed time period—the first 3 days of the fifth larval stadium—whereas the data in figure 2.5 were measured over the full stadium (i.e., up to the molt into adulthood). Since diet influences development rate, the latter data therefore reflect not only the rates of nutrient intake but also differences in the number of days over which the intakes were measured. Specifically, locusts fed imbalanced foods took longer to reach adulthood and so continued to move along their respective rails for several days after the locusts on the more moderate diets had molted into the adult stage and were removed from the experiment. This explains why the outer edges of the intake array have broken away from the Closest Distance configuration seen in figure 2.4A, which is apparent in figure 2.5 only for animals on the more moderate diets. It also highlights the fact that one of the costs borne by locusts feeding on foods that differ from the intake target in their protein to carbohydrate balance is a reduction in development rate and hence a delay in reaching sexual maturity. This might help us to understand why these locusts have evolved to select, and defend, the intake target that they do (fig. 2.3A). In the following section we present results of an experiment designed to investigate the relationships between target selection and fitness costs.

Performance Consequences A powerful aspect of geometric analysis is that it can integrate within a single model target selection, rules of compromise, and performance consequences. This, in turn, allows us to understand the reasons why animals select the target that they do, and why they adopt a particular rule of compromise when they are unable to reach their intake target. Simpson and colleagues (2004) carried out such an experiment on the African cotton leaf worm (Spodoptera littoralis) as a real example to support the above-mentioned theoretical models for describing fitness

was voided by being excreted at a high rate, hence the obvious “broken stick” appearance of the relationship in the graph, representing this abrupt rise in protein excretion. The nutrient target (NT) is the intake target less the 30% of ingested nitrogen inevitably lost. (Data from Zanotto et al. 1993.)

32  |  Chapter two

landscapes (plate 1). In this experiment, 35 experimental groups of cater­ pil­lars were each provided with a diet that differed in its nutritional composition from the diets of other groups. These diets spanned five different protein to carbohydrate ratios, spreading the intakes of the experimental groups across different nutritional rails in the nutrient space as in figure 2.4. To spread the intakes even more broadly across nutrient space, each rail was represented by seven diets that varied in the concentration of cellulose, a bulking agent that is not digested by these caterpillars and fills the gut, thereby limiting the rate at which nutrients can be eaten. These dilutions therefore had the effect of spreading the nutrient intakes of the different experimental groups along their respective nutritional rails. In this way, we were able to manipulate the nutritional states of the insects to span a wide range of positions in the nutrient space and measure the consequences for these animals of occupying different positions within this space. As mentioned previously, nutrient intake affects several aspects of performance. In caterpillars, two important components of evolutionary fitness are the time it takes to get through the juvenile stages to sexual maturity (prolonging development reduces the opportunities for reproducing) and, of course, the probability that the caterpillar will survive to the adult stage to reproduce at all. We chose these as the performance measures to relate to the amounts and balance of nutrients eaten. Rather than model them separately, a “performance index” was constructed that takes into account both performance measures, by multiplying the probability of survival to adulthood and the rate of development (growth / time) through the juvenile stage. The performance values achieved by the caterpillars on the different diets could then be plotted as a fitness landscape superimposed on the nutrient intake space, where peaks represent good performance and troughs poor performance (plate 1). It is evident from plate 1 that there was a clear peak on the fitness landscape, indicating that these caterpillars would achieve maximal performance in terms of growth rate and survival if they secured a nutrient intake ratio of 1.2 parts protein to 1 part carbohydrate. We have therefore identified the position in nutrient space representing the optimal diet (i.e., the intake target), giving rise to the question of whether the caterpillars are able to navigate to this point of their own accord. To test this, caterpillars were provided with nutritionally complementary foods that enabled them to move freely over a wide area in nutrient space (as illustrated for locusts in fig. 2.3A). As predicted, the caterpillars navigated to the point of peak performance, suggesting that the combination of development rate and survival probability has played an important role in the evolution of feeding regulation in this species. By combining evidence from regulatory behavior and performance (fitness), these results provide

The Geometry of Nutrition  |  33

the strongest measure possible for the existence and position of an intake target in S. littoralis caterpillars.

What If Regulatory Behavior and Performance Do Not Align? What if the caterpillars in the above experiment (plate 1) had regulated their protein and carbohydrate intake to a point that was at a lower elevation than the summit of the performance mountain? This would indicate one of two things: either the measure of performance used (the shape of the mountain measured) does not provide a complete representation of Darwinian fitness, or else the animals were behaving maladaptively in a Darwinian sense (Simpson et al. 2004). Regarding the first of these two possibilities, it is becoming increasingly appreciated that different performance measures may yield different landscapes in nutrient intake space. For example, we shall see in chapter 4 that, for flies, life span, lifetime egg production, and the rate of egg production have different summit positions and that, if given a choice, flies mix a diet that maximizes lifetime egg production (Lee et al. 2008a). We will also see in chapter 5 that there are conflicting nutritional demands not only between different life history traits but even between different components of the immune response—hence, some immune components in insects are favored on high-carbohydrate diets and others on high-protein diets. What this means is that the performance landscape that best reflects Darwinian fitness will depend on the animal’s ecological circumstances—its prospects of dying early from predation, its likelihood of being infected with particular diseases, and so on. Measuring performance under laboratory conditions will typically exclude such ecological complexities. Even the most sophisticated survey of fitness in the laboratory will not capture the true fitness landscape if costs such as the timedependent risk of mortality from predation, parasitism, or disease have become “built into” the operation of regulatory mechanisms over evolutionary time. Once all likely measures have been included in a laboratorybased estimate of the shape of the fitness mountain, the distance and direction between the pinnacle of the surface and the point of regulated nutrient intake could even be used as a guide to predict these missing ecological costs (Simpson et al. 2004). The second context where a regulated nutrient intake point may not align with the peak on a performance landscape is when the animals behave maladaptively because their current environment does not match the ancestral environment in which the regulatory systems evolved. One example would be if the regulated point of nutrient intake includes “anticipated” nutrient expenditure that is harmful if not used. Many animals in their natural environments will inevitably spend a substantial amount

34  |  Chapter two

of energy locating food and maintaining body temperature, and this might explain why modern humans and some animals kept under captivity fail to eat less, even though they are not exercising as much as their ancestors would have done. Here it is not that there is a failure to regulate intake, but rather that the intake target is set high because it has evolved to include anticipated energy expenditure, which if not used causes problems (see chapter 10). As discussed above, the same logic can be applied in relation to nutrient retention (the nutrient target)—stored fat might be deleterious if it is retained indefinitely rather than periodically used as an energy source. A more extreme mismatch between evolved regulatory physiology and the current nutritional environment occurs when a changed environment distorts or circumvents the operation of regulatory mechanisms. We introduce one example of this in chapter 10 for modern humans, where we explain how our biological predilection for fat and sugar and our powerful protein appetite lead us astray in the modern nutritional environment.

2.3 Conclusions We opened this chapter with an introduction to the basic concepts underlying the Geometric Framework and then provided some examples where these concepts have been applied in laboratory studies of real animals. The data show that animals have a remarkable ability to regulate their intake of macronutrients, and that they can do so to the point in nutrient space that maximizes fitness benefits. In the next chapter we address the question of how animals achieve this, by examining the behavioral and physiological mechanisms of feeding.

three

Mechanisms of Nutritional Regulation

From what we have seen in the previous chapter, it is clear that animals possess not one appetite system but several, and are able to regulate independently their intakes and utilization of different nutrients to maintain a target diet composition. If environmental circumstances prevent animals from achieving this target, they make compromises between eating too much of some nutrients and too little of others. Nutrient intake represents one part of the budget equation, by convention usually placed on the left-hand side. The right-hand side of the equation is broadly termed “postingestive processing,” and here too there is evidence for regulation of specific nutrients, both by adjusting the efficiency with which different nutrients are digested and absorbed from the gut, and by differentially allocating absorbed nutrients to various bodily functions, including their use as fuel for sustaining activity and life’s varied metabolic processes, tissue repair, growth, nutrient and energy storage, and reproduction. At the far end of the nutrient budget equation—and of the gut—unused nutrients and the waste products of metabolism are defecated or excreted. In this chapter we will consider the two sides of this equation in turn, focusing on how animals are able to regulate their intake and use of multiple nutrients.

3.1 How to Defend an Intake Target In the previous chapter we described some extraordinary feats of regulation, in which animals succeeded in defending an intake target in the face of various dietary manipulations and nutritional challenges. Such feats of regulation involve animals achieving three things: (1) assessing the nutritional quality of available foods, (2) assessing their own nutritional state, and (3) comparing these two to produce appropriate feeding responses. This would be akin to the animal asking three questions: (1) What is the nutritional composition of this food? (2) What nutrients do I need? and (3) How much of this food do I need to eat, if any, to balance my diet? Of

36  |  Chapter three

course, animals don’t actually ask themselves these questions, but they have evolved regulatory mechanisms that provide the answers.

Detecting Nutrients in Foods The simplest means to assess the nutritional composition of foods is to detect different nutrients by tasting them. Not surprisingly, therefore, all organisms, from bacteria to mammals, possess specialized receptors for the detection of key nutrients such as amino acids, sugars, and salts. Animals bear these receptors in several places: on external appendages, such as the tarsi (feet) and mouthpart palps of insects and the long whiskerlike barbels of some fish; within the oral cavity (mouth), such as on the tongue of vertebrates; and lining parts of the alimentary canal (the gut). Together, these receptors supply the central nervous system with information about the nutritional composition of food before, during, and after ingestion (Dethier 1976; Finger 1997; Yarmolinsky et al. 2009). Typically these taste receptors provide an increasingly strong signal as the concentration of the stimulating nutrient in the food increases, up to a point where excessively high concentrations may cause reduced responses. Whether feeding begins, and if so how much is eaten within a meal, are determined by the balance of positive (termed “phagostimulatory”) and negative (“phagodeterrent”) taste inputs arriving at the neural circuitry within the central nervous system that controls feeding. Positive inputs usually come from receptors responding to nutrients, whereas negative inputs include specialized taste receptors that are stimulated by potentially toxic compounds in food, including excessive concentrations of some nutrients. These latter receptors include the bitter receptor of vertebrates (Dong et al. 2009) and the deterrent receptors of herbivorous insects (Schoonhoven et al. 2005). When the positive and negative inputs from these peripheral taste receptors are integrated within the central nervous system, the net outcome defines the “phagostimulatory power” of the food—its capacity to stimulate and maintain feeding behavior. This is a key concept, which we build upon in the next section.

The “Taste Model” of Intake Target Regulation In the evolution of feeding responses, the details of how sensory receptors are tuned to respond to the nutrients in foods, and how the central nervous system converts these sensory responses to behavior, are important targets for natural selection. As discussed in the previous chapter, it is useful to construct models for understanding such complex mechanisms. In this section we describe a model that we have developed for this purpose, which has unimaginatively become known as the “taste model.” Consider an animal that is faced with four foods that are similar in all

Mechanisms of Nutritional Regulation  |  37

respects except one—their concentration of salt (it could be any nutrient, but we will use salt for illustration). One food (A) contains no salt, another (food B) contains the optimal concentration of salt (let us say for now that 2% salt is optimal), a third (food C) contains double the optimal concentration (4% salt), and the fourth (food D) contains four times the optimal salt concentration (8%). What should the animal do? The answer is obvious—it should select food B and avoid the others. This would happen if food B had by far the highest phagostimulatory power (“tasted best”) of the four foods. But what if only foods A and C are present? The animal could still, over time, compose an optimal diet containing 2% salt by eating equal quantities of foods A and C. This is because the average of 0% (food A) and 4% (food C) is 2%. This would happen if foods A and C, although less phagostimulatory than the optimal but unavailable food B, were of equal phagostimulatory power (“tasted equally acceptable”). Finally, what if only foods A and D are present? Again, our hypothetical animal could mix an optimal salt diet, but this time it would have to dilute the high concentration of salt in food D by eating relatively more of salt-free food A. To be precise, it would need to eat three times more of food A than D, which would happen over time if food A had a phagostimulatory power three times that of food D. In this example we have begun to describe how the taste system (defined here to include both peripheral taste receptors and the feeding circuitry within the brain) might fashion an animal’s feeding responses such that it defends an intake target concentration of a nutrient (salt in the example). In figure 3.1A we have taken the example one step further and plotted the mathematical function that defines the relationship between salt concentration and the phagostimulatory power of foods. This equation can be expanded to include more than one nutrient (Simpson and Raubenheimer 1996), but we will stick with one for now. An animal whose taste system can be described by such an equation will show three nutritionally “wise” behavioral responses. First, it will choose to eat predominantly from the optimal food, if it is available. Second, it will distribute its feeding among two or more complementary foods to mix an optimal concentration of nutrients in its diet. Third, if all available foods are suboptimal and noncomplementary in composition, the animal will eat most of the food that is closest to being optimal. So much for theory, but is our model correct? Experiments using locusts suggest that it is. Locusts regulate their intake of salt and macronutrients to an intake target when offered complementary pairs of foods (Trumper and Simpson 1993, 1994). The intake target concentration for the salt mixture was 1.8% dry weight of the food. (Now you can see why our hypothetical example above chose salt as the focal nutrient and 2% as the optimal concentration). The prediction from our taste model is that the phagostimulatory power of foods varying in salt concentration

38  |  Chapter three A. Model

B. Locust 80

80

Percent locusts initiating a meal

Phagostimulatory power of food

100

60 40 20 0 –20 –40 –60

0

1

2 3 4 5 6 Percent salt in food

7

0.5

1.0 1.5 2.0 2.5 3.0 Percent NaCl solution

8

60

40

20

0

0

1

2 3 4 5 6 Percent salt in food

C. Rat 25

Intake (ml)

20 15 10 5 0

0

3.5

Fig. 3.1. (A) A theoretical model shows how the taste system (defined to include both peripheral sense organs and central nervous components) could allow regulation to a target concentration of a nutrient (salt in this case). The model proposes that foods vary in palatability (“phagostimulatory power,” shown on a relative scale from –100 to +100) according to the concentration of a given nutrient, such that the optimal concentration is the most phagostimulatory. (B) Data from experiments on locusts in which the concentration of salt in test foods was varied systematically and the probability of meal initiation recorded as a measure of phagostimulatory power. When locusts were allowed to regulate their salt intake independently of macronutrients, they did so to a target intake of 1.8% salt in the diet (Trumper and Simpson 1993)—which sits at the peak of the curve in B. (C) Equivalent data from rats provided for 1 hour with solutions varying in NaCl concentration. (From Simpson and Raubenheimer 1996.)

7

8

Mechanisms of Nutritional Regulation  |  39

should match the relationship in figure 3.1A. The actual relationship as measured experimentally (Simpson 1994) is shown in figure 3.1B. The two are closely similar. Similarly shaped curves have been reported for sodium in rats (fig. 3.1C) and calcium in a range of animals (Tordoff 2001). A test of a two-nutrient version of the model was carried out by Chambers and colleagues (1997), who provided locusts with two foods: one containing an optimal ratio of protein to carbohydrate, and another that differed increasingly from the optimum. The mathematical model provided a prediction of the extent to which locusts should feed on the suboptimal rather than the optimal food: the predicted and experimental results agreed closely. We have shown that a taste system that functions according to the model in figure 3.1 can enable an animal to mix a diet containing the optimal concentration of one of more nutrients; in other words, it would allow the animal to find and defend the optimal food rail. However, the taste system cannot determine how much of the optimal food needs to be eaten to reach the intake target. Nor can it respond to short-term changes in the intake target, or to environmentally imposed perturbation in the animal’s nutritional state—that is, unless nutrient feedbacks dynamically adjust the shape of the phagostimulation curve. We will return to this point below.

What If Nutrients Are Not Tasted? A Role for Learning Being able to determine the presence and concentrations of nutrients in foods by taste is clearly advantageous, but not all nutrients in food are detected by specialized taste receptors. This is especially the case for micronutrients such as vitamins and trace elements, which are essential to health but required in very small amounts relative to the macronutrients protein, carbohydrate, and fat. If the full nutritional consequences of eating a food are not apparent until after a meal is processed and absorbed from the gut, how can the animal use these postingestive signals to help it reach the intake target? The answer is to learn from the experience of having eaten a food by associating postingestive nutritional consequences with properties of the food, such as its location or its sensory qualities (color, shape, smell, taste, sound, etc.). There are two main types of association between food cues and postingestive consequences: learned aversions and positive learned associations. Food aversion learning was first reported by John Garcia and colleagues (1955) while working in the U.S. Naval Radiological Defense Laboratory in San Francisco. They showed that rats given saccharinsweetened water during a period of gamma irradiation subsequently learned to avoid the taste of saccharin. Conditioned food aversion learning has since been discovered in many different species of animals, from

40  |  Chapter three

slugs to humans, and shown to be induced by toxic effects of eating a food or by nutritional imbalances (Bernays 1993; Tomé 2004; Dukas 2008). An unusual feature of food aversion learning is that an aversive association is made between the flavor of a food (which, in learning parlance, would be termed the conditioned stimulus [CS]) and unpleasant consequences of eating that food, which may not manifest themselves until hours later (equivalent to the unconditioned stimulus [US] in learning theory). In more typical cases of associative learning, the most effective interval between the CS and US is only a few seconds. Pavlov’s dogs would never have learned to associate the ringing of a bell with food if meat had been offered several hours after the bell rang; but if they ate a distinctively flavored food and became ill hours later, they would have learned to avoid that flavor again. Learned positive associations, which include learned preferences for specific flavors, have been demonstrated in a wide range of vertebrates and invertebrates (Sclafani 2000; Touzani and Sclafani 2005; Burke and Waddell 2011; Fujita and Tanimura 2011). In such cases, animals learn to respond positively to previously ineffective cues associated with nutritionally rewarding foods. There are even reports of animals learning cues associated with specific nutrients and only responding when deprived of those nutrients—known as “learned specific appetites” (Booth and Thi­ bault 2000). An example of such an experiment is illustrated in plate 2. Locusts were left for two days in an arena with a high-protein food and a high-carbohydrate food, each associated with one of two colors, green or yellow. During these two days the insects moved regularly between the foods to balance their nutrient intake, at each occasion experiencing the color associated with the food. After two days the locusts were taken out of the arena and rendered either protein- or carbohydrate-deprived for four hours in the absence of the color cues. They were then allowed to choose between the two colors. Locusts that were protein-deprived moved toward the color (yellow or green) previously paired with the high-protein food, whereas those that were carbohydrate-deprived were attracted by the color previously paired with high-carbohydrate food (Raubenheimer and Tucker 1997). Earlier we had used odor cues (spearmint and lemon) in the same experimental design and found evidence for a learned specific appetite for protein but not carbohydrate (Simpson and White 1990). In addition to learning to associate food cues with postingestive consequences, other, more general behavioral responses may help an animal to balance its nutrient intake. One example is known as “neophilia.” Rats normally avoid eating novel foods (they are “neophobic”) but start actively to prefer novel foods (become “neophilic”) when they are deprived of thiamine (vitamin B1). This increases their chance of ingesting thiamine but at the risk of ingesting harmful compounds in a previously un-

Mechanisms of Nutritional Regulation  |  41

tried food (Rogers and Rozin 1966). Such responses are unlikely to result in targetlike regulation of nutrient intake, but they do help maintain the intake of nutrients that are not directly sensed, notably micronutrients such as vitamins, within acceptable bounds—not too high and not too low (Raubenheimer and Simpson 2010). A final point about food learning is that animals may benefit not only from their own experience but also that of others. Social animals in particular have been found to learn from one another about food (Galef and Laland 2005). But not all group-living animals show social learning. Dukas and Simpson (2009) found that locusts would prefer a cinnamonor cocoa-flavored food over a nutritionally equivalent food after only a single meal. However, they did not prefer a novel-flavored food if they had observed another locust eating it, had interacted with an experienced locust, or had eaten an experienced individual and therefore had secondhand experience of the novel food through cannibalism. (We return to cannibalism in chapter 7.)

Assessing Nutritional State For an animal to assess its current nutritional state it needs nutrient and energy sensors within its body. Probably all cells can sense their nutritional state and energy status (Lindsley and Rutter 2004), but coordination of behavior and metabolism at the level of the whole animal is carried out by the brain, in neural and chemical dialogue with other organs, notably the liver, gut, and pancreas in vertebrates and the distributed fat body of insects. An obvious source of nutritional information is the concentrations of nutrients such as glucose, amino acids, free fatty acids, and mineral ions circulating in the blood system (either the closed vascular system of vertebrates or the sluggishly circulating open hemocoel of invertebrates). Circulating nutrients provide an instantaneous measure of nutritional state. This is because their levels reflect the net result of nutrients arriving from the gut (an indication of how recently the animal fed, how much it ate, and the nutritional composition of the food) and those removed from circulation to meet the energy and growth needs of the tissues (Simpson and Raubenheimer 1993a). A complication comes from stored nutrients that are released into the circulatory system during food or nutrient deprivation. These stores include body fat, liver glycogen, and lean tissues such as muscle. Such nutrients liberated during deprivation create a risk of overestimating the body’s nutritional state if only blood nutrient levels are measured: it is ambiguous whether elevated circulating nutrients indicate nutritional repletion or breakdown of stores during deprivation. Hence, a second source of information about nutritional state is needed that indicates the status of reserves—either a measure of their size or some indication that

42  |  Chapter three

they are being used up. The source of such information is often chemical signals released from the tissues themselves. The best-known examples are the hormones leptin and adiponectin, which are released by the fatstoring (adipose) tissue in vertebrates (Zhang et al. 1994; Crespi and Denver 2009; Morton and Schwartz 2011), and insulin and glucagon, which are released from the endocrine pancreas and signal carbohydrate status. To date it is not known whether there is an equivalent hormonal signal for protein, but some of the interleukins (IL 6 and IL 15 for example) that are released by muscle may provide signals of lean status (Febbraio and Pedersen 2002; Pedersen and Febbraio 2008; Quinn 2008). Returning to the level of individual cells in the body, the two bestknown (but not the only) intracellular nutrient-sensing pathways are the protein kinase systems, AMPK (adenosine monophosphate-activated kinase) and TOR (target of rapamycin), which are expressed in a range of tissues in all manner of organisms. The AMPK pathway responds to a decline in circulating levels of glucose, amino acids, and fatty acids, and to energy depletion as indicated by an increased ratio of AMP to ATP. The AMPK pathway triggers a cascade of metabolic effects that are broadly catabolic (energy releasing, breaking down complex molecules into smaller ones) and lead to the release of stored nutrients and the inhibition of growth and reproduction. Acting oppositely, the TOR pathway is stimulated by high levels of ATP and nutrients. Branched-chain amino acids, such as leucine, are powerful stimulators of TOR. This pathway also responds to glucose in a way that is dependent on levels of amino acids, suggesting that it may detect the protein to carbohydrate balance (Simpson and Raubenheimer 2009; see also chapter 4). TOR signaling coordinates a series of broadly anabolic metabolic responses (energy requiring, and involving the synthesis of complex molecules), thus stimulating growth and reproduction. Other nutrient sensor systems include uncharged transfer RNAs that signal deficiency of particular amino acids (Gietzen and Rogers 2006); G-coupled membrane receptors that bind with certain amino acids, small peptides, glucose, fatty acids, sodium ions, calcium ions, or combinations thereof (Wellendorph and Brauner-Osborne 2004; Conigrave and Brown 2007; Chaveroux et al. 2009; Young et al. 2010); and ion-coupled active transporters and cotransporters for glucose and different amino acids (Hundal and Taylor 2009).

Integrating Assessments of Food Nutrients and Nutritional State To produce regulatory feeding behavior, taste and learned cues associated with the nutritional composition of foods must somehow be integrated with systemic signals indicating the body’s nutritional state. Because the

Mechanisms of Nutritional Regulation  |  43

brain coordinates feeding behavior, it is here that nutritional integration must ultimately be coordinated; but nutritional integration is also known to operate more peripherally to direct feeding behavior in some animals. In insects, for example, one of the major sites of nutritional integration is at the mouthpart taste neurons that detect nutrients in food, leaving the brain with the simple task of responding to already integrated, nutrientspecific inputs. Insects possess chemoreceptive neurons within porous cuticular pegs and hairs (called sensilla) located on various body parts, including the antennae and external mouthparts, the preoral cavity, the tarsi (feet), and the ovipositor. Taste sensilla typically have a single terminal pore, through which chemicals on and within foods enter and stimulate the unbranching terminal processes (dendrites) of gustatory neurons within the lumen of the sensillum. Olfactory sensilla, by contrast, are perforated by many pores, through which odorant molecules diffuse and interact with receptors on the highly branched dendrites. Experiments on locusts have shown that concentrations of nutrients and the signaling molecule nitric oxide in the blood directly modulate the responsiveness of gustatory neurons in a nutrient-specific manner (Simpson and Raubenheimer 2000; Newland and Yates 2008). For example, when the insect is deprived of protein but sugar-replete, levels of free amino acids in the blood fall and sugar levels rise (Abisgold and Simpson 1987; Zanotto et al. 1996). These changes in circulating nutrient concentrations act locally to cause the responsiveness of taste receptors to amino acids in food to increase and their responsiveness to stimulation by sugars to fall (fig. 3.2) (Abisgold and Simpson 1987, 1988; Simpson et al. 1991; Simpson and Simpson 1992). The effect is that the insect becomes highly responsive to foods containing amino acids and ignores foods containing sugar: in other words, nutrient-specific feedbacks have altered the phagostimulatory power of foods, resulting in the animal tasting what it needs and eating what it tastes. C. L. Simpson and colleagues (1990) discovered that a suite of amino acids all had to be elevated together in the blood to achieve protein repletion and trigger the associated desensitization of taste receptors to amino acids in the food (fig. 3.3). These amino acids include lysine, alanine, glutamine, valine, phenylalanine, leucine, and serine; removal of any one from a supplementary mixture of amino acids added to the food, or injected into the blood, removed the protein-satiating effect of the rest. The modulation of taste receptor responsiveness by circulating nutrients therefore allows the peripheral taste organs to serve as integrators of information about the quality of food and the animal’s instantaneous nutritional state. This means that a locust can make sophisticated nutritional choices without its brain having to do more than add up incoming

44  |  Chapter three A. Modulation of food selection

Percent locusts feeding for longer than t

100 80

P-pretreated

80

60

60 P food

C food

40

40

20 0

0

2

4

6

8

10

12

14

0

O-pretreated

80

60

60

40

40

20

0

2

4

100

80

C food

20

0 2 4 6 8 10 12 14 Duration of first feeding episode, t (min)

6

8

10

Action potentials per second

0.01 M sucrose 0.01 M amino acids

100 80

P food

0

0 2 4 6 8 10 12 14 Duration of first feeding episode, t (min)

Protein-deficient locust 0.01 M amino acids

0.01 M sucrose

60 Carbohydrate-deficient locust 0.01 M amino acids

40 20 0

0.01 M sucrose P

C Pretreatment diet

PC

14

PC-pretreated

B. Modulation of taste responsiveness 120

12

C food

P food 0

P food

20

C food

100 Percent locusts feeding for longer than t

100

C-pretreated

Mechanisms of Nutritional Regulation  |  45

gustatory nerve impulses. If these positive signals exceed a threshold level, then feeding is triggered; if not, the animal rejects the food and keeps searching for something better. This mechanism is elegantly simple but, as we have seen in chapter 2, highly effective. Learning (see above) augments these direct nutritional regulatory responses, helping the animal to locate and handle nutritionally appropriate foods more efficiently. Dukas and Bernays (2000) conducted an experiment to establish the relative roles of learning and direct nutrient feedbacks on nutrient balancing and growth in a grasshopper. Insects were placed in an arena with two foods, one balanced in its nutritional composition, the other carbohydrate-deficient. Some experimental insects were provided with cues associated with the two foods, including distinct flavors, colors, and spatial locations. Once a food had been allocated a position in the arena, a color, and an added flavor, these remained constant over time, providing the opportunity for the grasshoppers to learn to associate the cues with the nutritional composition of the foods. The remaining grasshoppers had these same cues assigned to the foods randomly twice daily. This confused the animals, preventing them from making dietary associations with the cues. When the cues were unchanging, the grasshoppers rapidly learned not to approach the unbalanced food and ended up spending 99% of their feeding time eating the nutritionally balanced food. When cues were scrambled, grasshoppers continued to approach the two foods with equal likelihood. They, nonetheless,

Fig. 3.2. Nutritional feedbacks onto food selection behavior in locusts. (A) Insects were pretreated for 4 hours on one of four synthetic foods, P (protein-rich), C (carbohydrate-rich), PC (containing both protein and carbohydrate), or O (lacking both macronutrients). The length of the first feeding bout on either P or C foods was strongly affected by nutritional state. C-pretreated insects (protein-deprived) fed significantly longer on P than on C, whereas P-pretreated (carbohydrate-deprived) insects fed longer on C than P. PC-fed locusts (not deprived of either nutrient) ate little of either food, whereas O-pretreated insects (deprived of both) ate both foods with alacrity. (B) The changes in food selection behavior in A are accompanied by changes in the responsiveness of gustatory receptors on the mouthpart palps. P-pretreated locusts (carbohydrate-deprived) have selectively increased responsiveness to stimulation by 0.01 M sucrose, whereas C-pretreated insects (protein-deprived) have elevated responsiveness to a 0.01 M mixture of amino acids. (After Simpson et al. 1991.)

46  |  Chapter three A. Add mixture of amino acids 14%P

28%P

B. Omit individual amino acids 14%P + 14% AA

180 160

E

Food eaten (mg g–1)

E 140

E

E E

120 100

180 80 Food eaten (mg g–1)

160 60

140 120

P(AA) lys ala glu met val phe leu ser rest Diet

100 80 60

Low P

High P Diet

High P(AA)

Fig. 3.3. The nature of the amino acid feedbacks that modulate protein appetite in locusts. (A) Locusts compensated for a dilution of protein in the diet from 28% to 14% (top, first two bars) by increasing consumption over a 12-hour period (bottom, first two bars). When an enriching mixture of 14% free amino acids was added to the 14% protein diet (top, third bar), insects reduced consumption to high-protein levels (bottom, third bar). (B) When individual amino acids were omitted from this enriching mixture, intake significantly increased toward low-protein levels if any one of a suite of amino acids (indicated in gray bars) was absent, even though all other amino acids were present in the mixture. Removal of all other amino acids (“rest”) had no effect—the diet was treated by locusts as if it contained 28% protein (horizontal dashed line). E indicates essential amino acids. (After C. L. Simpson et al. 1990.)

spent 87% of their feeding time eating the balanced food, because when they did contact the carbohydrate-deficient food, they usually rejected it. That these confused locusts managed to ingest the nutritionally balanced food as successfully as they did (87% of the time) is testament to the effectiveness of direct nutrient feedbacks. Nevertheless, the added advan-

Mechanisms of Nutritional Regulation  |  47

tage offered by the opportunity to learn stable food cues provided a 20% higher growth rate compared with that of the confused grasshoppers. To this point we have considered nutritional feedbacks onto taste and learning that operate over a matter of hours. An interesting aside is that the number of taste receptors that develop is itself responsive to the nutritional environment. For example, Rogers and Simpson (1997) found that the number of taste sensilla that develop in adult locusts is influenced by the variety of chemical stimuli experienced over the previous two juvenile stages, whether in the form of nutritionally complementary foods or foods with added flavors. Rearing locusts in an impoverished chemosensory environment, even one that was nutritionally optimal (a single, nutritionally balanced powdered diet), led to fewer chemoreceptors developing on the mouthparts and antennae (Rogers and Simpson 1997). Opstad and colleagues (2004) showed that having more sensilla was associated with more rapid and decisive feeding responses, particularly when locusts were provided with marginally acceptable foods. In mammals the brain is the main site of integration between food composition and nutritional state (Schwartz et al. 2000; Cota et al. 2007), although there is growing evidence that modulation of gustatory responsiveness by nutritional feedbacks may also occur, as described above for insects (Carleton et al. 2010). Taste receptors on the tongue are stimulated by food in the mouth and activate neurons that send their inputs to the hindbrain. Also projecting to the hindbrain are stretch receptor inputs from the upper gastrointestinal tract (GI), nutrient receptor inputs from the GI and liver (carried via the vagus nerve), and neural signals coming from the forebrain. Another source of signals associated with food in the gut is a suite of hormones that are released from the GI and act variously on gut motility, digestive enzyme secretion, the vagus nerve, and directly on the brain. Among these hormones are peptide YY (PYY), glucagonlike peptide-1 (GLP-1), and cholecystokinin (CKK), which are stimulated by food intake and depress feeding behavior (Chaudhri et al. 2008). Secretion of another gut hormone, ghrelin, is inhibited by food intake. Secretion of ghrelin increases with food deprivation and stimulates food intake. Within the forebrain, the hypothalamus plays a central role in integrating nutritional signals, in particular the arcuate nucleus (ARC). The ARC contains two major groups of neurons that control feeding behavior. One of these groups is termed POMC/CART because its neurons coexpress pro-opiomelanocortin (POMC) and cocaine- and amphetamine-related transcript (CART). When activated, these neurons release α-melanocyte-stimulating hormone, which binds to melanocortin-4 receptors and inhibits feeding behavior. The second major population of neurons in ARC is termed NPY/AgRP because when activated they re-

48  |  Chapter three

lease neuropeptide Y (NPY) and agouti-related protein (AgRP), which stimulate feeding behavior. The activity in these two groups of neurons is modulated by the animal’s nutritional state via levels of circulating nutrients and hormones. Signals of nutritional sufficiency, such as elevated levels of amino acids, glucose, fatty acids, insulin, and leptin, inhibit NPY/AgRP neurons and stimulate POMC/CART neurons, as do a low AMP/ATP ratio and low circulating concentrations of ghrelin. Signals of nutritional deficiency (i.e., opposite to the above nutritional and hormonal trends) have the reverse effect, exciting NPY/AgRP neurons and inhibiting POMC/CART neurons. The sensing of these nutritional feedbacks by NPY/AgRP and POMC/ CART neurons in the hypothalamus is mediated by AMPK and TOR signaling pathways (Cota et al. 2007). As we discussed above, these pathways are among the main cellular nutrient sensors. Recruiting them within nerve cells provides an elegant solution to the problem of how to link cellular nutrient-sensing mechanisms to the regulation of nutrient intake and metabolism at the level of the whole organism. Insects too employ neural TOR signaling. Experiments have shown that neuronal TOR signaling is involved in the control of protein intake and nutrient balancing in fruit flies (Ribeiro and Dickson 2010; Vargas et al. 2010). There is cross talk between the arcuate nucleus and hindbrain integrative centers, as well as input from higher brain areas dealing with other sensory modalities and food reward, allowing learned responses to be integrated into the control of feeding behavior (Morton et al. 2006). Quite how specific appetites for different macronutrients are controlled within these hormonal and neural pathways is not yet fully understood (e.g., Potier et al. 2009), but it is worth noting that other forebrain areas such as the anterior piriform cortex have been implicated in regulation of amino acid intake through the activity of uncharged transfer RNAs for amino acids (Gietzen and Rogers 2006).

3.2 Postingestive Regulation Once foods have been ingested, there are two basic options for modifying the rates and ratios at which nutrients are supplied to the tissues. The first option is to modulate the efficiency with which different nutrients are digested and absorbed from the gut. Three possibilities exist here: adjust digestive enzyme secretion and absorptive processes directly; vary the rate at which food passes through the gut; or, over longer time periods, change the shape and size of the gut in response to diet. The second basic option is to modify the efficiency with which nutrients that have been

Mechanisms of Nutritional Regulation  |  49

absorbed from the gut are retained for useful purposes, rather than voided from the body.

The Gut as a Nutrient-Balancing Organ: Regulatory Digestion Until recently, it was believed that the function of the gut is to maximize digestion and absorption of its contents, leaving the job of deciding which nutrients to retain and which to void until after absorption (e.g., Karasov and Diamond 1988; Starck 2005). The prevailing view has been that enzyme secretion is stimulated by the presence of nutrients in the gut and that the higher the concentration of nutrients, the greater the amounts of appropriate digestive enzymes secreted (proteases for protein, lipases for fats, and carbohydrases for carbohydrates). Furthermore, the abundance of the nutrient transporters that absorb digested nutrients across the gut lining has been shown to be positively related to the concentrations of nutrients in the diet (Karasov and Diamond 1983). As a result, the majority of available nutrients are extracted from food in the gut, without wasting energy maintaining unnecessary digestive and absorptive capacity when nutrient levels are low or food is scarce. This view was revised recently by a study on locusts (Clissold et al. 2010). The insects were fed one of four artificial foods for three days. Two of these foods contained a balanced ratio of protein to carbohydrate (1:1 P:C, which is close to the self-selected intake target ratio for locusts; fig. 2.3), but one (pc) was less concentrated than the other (PC) due to the addition of indigestible cellulose. The other two foods were nutritionally unbalanced, containing either a high-protein, low-carbohydrate ratio (Pc) or the converse ratio (pC). Levels of proteases and carbohydrases in the gut were not affected by nutrient concentration, provided that protein and carbohydrate were in balanced proportions in the food (i.e., responses to PC and pc did not differ). However, when protein and carbohydrate were unbalanced in the food, levels of the digestive enzyme for the excessive nutrient were markedly reduced (fig. 3.4). If locusts were fed unbalanced foods and then fed grass, the efficiency with which they extracted protein or carbohydrate from grass reflected the changed balance of digestive enzymes in their guts, indicating that the changes in enzyme levels were meaningful to the animal on a natural diet. These results demonstrate that the gut is not solely a nutrient-maximizing organ: when confined to an unbalanced diet, locusts sacrificed maximal rates of digestion to achieve a better-balanced ratio of absorbed nutrients. But why was there no difference in enzyme levels when protein and carbohydrate were present in near-balanced proportions in the diet but at different total concentrations (PC vs. pc)? As we shall see below, similar patterns were also found for postabsorptive regulatory responses.

50  |  Chapter three A.

B. 30 α-amylase activity (U mL–1)

α-chymotrypsin activity (nmol pNA min–1)

9000

8000

7000

6000

5000

PC pc Pc pC Dietary pretreatment

25

20 10

PC pc Pc pC Dietary pretreatment

Fig. 3.4. Nutrient-specific modulation of digestive enzyme activity in locusts. Insects were pretreated on one of four diets containing high or low concentrations of protein and carbohydrate (both protein and carbohydrate high, PC; both low, pc; high protein, low carbohydrate, Pc; low protein, high carbohydrate, pC). (A) Note how the activity of a protease (chymotrypsin) in the contents of the foregut (the major site of digestion) was down-regulated in locusts on a diet containing excess protein relative to carbohydrate, and (B) how the activity of a carbohydrase (amylase) was lower in insects fed a diet containing excess carbohydrate relative to protein. Dilution of a balanced diet (pc vs. PC) had no effect. (From Clissold et al. 2010.)

The answer appears simple. Provided that nutrients (protein and carbohydrate in this case) are balanced relative to one another in the diet, the intake target can be achieved by adjusting the total amount of food eaten (Raubenheimer and Simpson 1993); but such compensatory feeding cannot correct a dietary imbalance, because every extra mouthful eaten gains extra nutrients in the same unbalanced ratio (chapter 2). Rebalancing an unbalanced diet therefore requires the animal to “jump its dietary rail” and track across nutrient space toward its target state postingestively. As we have seen, this can be achieved, at least in part, by a readjustment of digestive and/or absorptive efficiency. Using natural rather than artificial food, Mayntz and colleagues (2005) reported another, more grisly example of differential digestion of nutrients as a nutrient-balancing strategy. Web-building spiders, Stegodyphus lineatus, were fed for 24 hours on fruit flies that had been specially raised to be either high or low in protein relative to fat. Spiders were then allowed to feed on a protein-poor or protein-rich test fly, after which the remains of the victim’s carcass was analyzed to discover whether differ-

Mechanisms of Nutritional Regulation  |  51

ent amounts of protein had been extracted according to the spider’s feeding history. Spiders do much of their digesting externally rather than in the gut, by injecting digestive enzymes into their victim and sucking up the resulting nutrient soup. Mayntz and his coworkers found that, after feeding, the percentage of nitrogen extracted from the test fly was increased if the spider had previously been fed protein-poor rather than protein-rich flies; moreover, protein-deprived spiders extracted more nitrogen per unit of carbon. These results indicated that protein-deprived spiders sucked out relatively more protein from their prey than did protein-replete spiders. We return to this example in chapter 8 (fig. 8.1C).

The Plastic Gut Changes in digestive enzyme secretion in response to diet are one thing, but the diet can also cause extensive physical remodeling of the gut itself. Perhaps the most extreme example comes from pythons (Secor and Diamond 2000). These sit-and-wait predators feed only infrequently, on prey that can weigh more than they do. After each meal the small intestine of the python doubles in mass, and nutrient transport rates increase up to 20-fold. Once digestion is complete, the gut atrophies and returns to its smaller, less expensive-to-maintain state. Diet-induced remodeling of the gut is found in many other animals too, from grasshoppers to mammals (Yang and Joern 1994; Starck 2005; Naya et al. 2007; Wagner et al. 2009). The most commonly reported effect of diet on gut morphology is that the gut grows larger when the diet contains large quantities of fiber or another bulking agent (Yang and Joern 1994; Starck 2005; Raubenheimer and Bassil 2007). This represents a regulatory response to allow increased food intake to compensate for the dilution of nutrients by indigestible bulk: growing a bigger gut means the animal can eat more and so maintain the rate of supply of nutrients to its tissues. Changes in gut size also accompany altered macronutrient balance in the diet—seen both in locusts (Raubenheimer and Bassil 2007) and in rodents (Sørensen et al. 2010). When Sørensen and colleagues (2010) fed mice on diets that varied in protein to carbohydrate ratio (P:C) at a fixed level of fat and other nutrients, they grew larger stomachs on high P:C diets. Since the stomach is where amino acid–stimulated release of the protease gastrin occurs from specialized cells (G-cells), it was proposed that increasing the mass of the stomach would help deal with high levels of dietary protein. Such a response is “countercompensatory” (Raubenheimer and Bassil 2007), because it appears to offer an increased capacity to digest nutrients that are present in excess. A compensatory response—such as we described above for enzyme release in locusts (Clissold et al. 2010)—would be expected to yield the opposite response, namely, a smaller stomach on high P:C diets.

52  |  Chapter three

The remainder of the digestive system showed the opposite trend, with highest gut masses on the lowest P:C diets. This response might be countercompensatory, facilitating greater absorption of abundant carbohydrates (Karasov and Diamond 1983). It might, however, also be a compensatory response to maximize digestion and absorption of limiting amino acids, a shortage of which limits growth and reproduction, on low P:C diets (Sørensen et al. 2010), especially since the small intestine is a major site of protein digestion by proteases such as chymotrypsin.

Altering Gut Passage Rate as a Compensatory Response Even without making any adjustment to the release of digestive enzymes or gut morphology, an animal could potentially change the rate and balance of nutrients absorbed by changing the rate at which food passes through the gut (Raubenheimer and Simpson 1996, 1998). Such a mechanism could work if different nutrients are digested and absorbed at different rates (fig. 3.5). To illustrate, let us say that an animal eats a food containing nutrients A and B in a 1:2 proportion but its intake target ratio is in the opposite balance, that is, a 2:1 ratio of A to B. If nutrient A is digested and absorbed more quickly than nutrient B—by virtue of its molecular structure, or its greater accessibility because of its location within the physical matrix of the food, or higher levels of the appropriate digestive enzyme in the gut, or whatever—then there will be a point after the meal where the rate of transfer of nutrient A from the gut is greater than that of nutrient B, even though nutrient A is rarer than B in the gut contents. Since nutrient A is digested more quickly than nutrient B, by the time that the digestion of nutrient B is just getting going, nutrient A is already being absorbed from the gut at a high rate. In our hypothetical example shown in figure 3.5, there is a time after the meal, t, when the ratio of nutrients A and B leaving the gut is 2:1—the opposite to their ratio in the food, and the same as the intake target. Were the animal to void its gut contents at this point and eat again, it could provision its tissues with an optimally balanced ratio of nutrients. This would involve wasting large quantities of nutrient B, but if the benefits of a balanced diet outweighed the waste, such a solution would be favored. How might an animal void its gut contents at the appropriate interval to ensure a better-balanced transfer of nutrients from the gut? There are two possibilities—one is simply to eat again and rely on the incoming meal to push the previous meal farther along in the gut; the other is for gut sensors that monitor the ratio of nutrients being absorbed to adjust gut motility and thus food passage rate. The former mechanism is known in animals with a relatively simple, tubelike gut (e.g., locusts; Simpson 1983). Control of motility, and thus transit time, by gut nutrient sensors

Mechanisms of Nutritional Regulation  |  53

Amount of nutrient absorbed

3.0 2.5 2.0

Nutrient B ingested

B

1.5 1.0

Nutrient A ingested

A

0.5 0

t Time since meal

Fig. 3.5. A model showing how different rates of absorption of nutrients in the gut may allow animals to balance their nutrient gain postingestively by adjusting the timing of gut voiding and refeeding. The animal has ingested a meal containing two nutrients, A and B, in the ratio 1:2, but its intake target ratio is 2:1. Note how at time t the ratio of A absorbed to B absorbed is 2:1. Therefore, if the animal were to void its gut contents and feed again at time t it would gain a balanced uptake of nutrients—at the expense of wasting substantial quantities of B. (After Raubenheimer and Simpson 1998.)

is well known in vertebrates and is linked to release of hormones that also cause changes in the release of digestive enzymes (see above). Similarly, in insects there are networks of neuroendocrine cells as well as intrinsic endocrine cells that appear to detect nutrients and release a range of hormonal compounds that modulate gut motility and trigger release of digestive enzymes (Veenstra et al. 2008; Woodring et al. 2009; Clissold et al. 2010). The endocrine cells scattered along the locust gut have projections that simultaneously contact the gut lumen, the blood, and in some cases also the primary urine, and respond to the nutritional state of the animal by releasing modulatory peptides (Zudaire et al. 1998, 2004). Such cells are therefore ideally placed to integrate information about the animal’s nutritional state and gut contents and to coordinate digestive activities accordingly. A similar mechanism may well operate in vertebrates, as seen for example in the gastrin-secreting cells of the stomach, which have calcium–amino acid receptors (CaR) situated both at their apical and basal membranes—the former in contact with the stomach lumen and the latter with the circulatory system (Conigrave and Brown 2006). Just as, in theory at least, an animal might adjust the rate at which food passes through the gut to manipulate the ratio of nutrients entering

54  |  Chapter three

the blood, constraints on the availability of nutrients to digestion may provide an impediment to nutrient balancing. An elegant example here comes from the work of Clissold and colleagues (2006). The Australian plague locust, Chortoicetes terminifera, hatches from eggs laid in the soil and develops following rainfall in an arid environment dominated by two host plants, the annual grass Dactyloctenium radulans (button grass) and the perennial grass Astrebla lappacea (Mitchell grass). These two grasses are of similar nutritional composition, but they differ in their physical properties. Young locusts performed equally well on both grasses, but older nymphs grew faster and bigger when fed button grass than when fed Mitchell grass. Clissold and coworkers (2006) showed that because of differences in the cellular structure of the two grasses, older nymphs with their bigger mandibles extracted less carbohydrate per unit of protein from Mitchell grass than from button grass—and paid the price for this nutritional imbalance. Younger locusts were better able to access the carbohydrate within the cells of Mitchell grass than were older nymphs, and could therefore remain in nutrient balance on both grasses.

Differential Utilization of Nutrients Once Absorbed The next stage at which an animal might rebalance an unbalanced nutrient intake is once nutrients have entered the circulation. There is not much that can be done if too little of a given nutrient has been ingested and absorbed, other than conserve what is there and not waste it; so the main mechanism for rebalancing depends on elimination of nutrients absorbed in excess of requirements. A well-known example of voiding excess absorbed nutrients is when levels of excretory nitrogen rise as the protein requirements for growth and maintenance are exceeded, as shown for locusts by Zanotto and colleagues (1993; see above, fig. 2.5B). This mechanism arises from the breakdown of excess amino acids and their excretion as nitrogenous waste. Locusts have an additional device for dealing with excess ingested protein. As discussed above, C. L. Simpson and colleagues (1990) found that elevated blood levels of eight free amino acids provide a powerful feedback inhibiting food intake. However, when confined to a diet containing a high ratio of protein to carbohydrate, locusts continue to eat somewhat beyond their intake target for protein to gain limiting carbohydrate (Raubenheimer and Simpson 1993; see chapter 2). Zanotto and coworkers (1994) discovered that locusts maintained on such a high-protein, low-carbohydrate diet selectively excreted large amounts of lysine. Because lysine is the most potent of the mixture of amino acids signaling protein repletion (C. L. Simpson et al. 1990), its selective excretion allows a degree of “un-jamming” of the feeding control mechanism and thereby frees locusts to continue eating to gain limiting carbohydrate.

Mechanisms of Nutritional Regulation  |  55

As an aside, it is possible that a similar logic explains why rodents on a high-fat diet show leptin resistance in the arcuate nucleus of the hypothalamus (Münzberg and Myers 2005). Such animals are typically protein-deprived and need to increase total food intake to gain limit-­ ing protein (discussed further in chapter 10). Selective blocking of the appetite-suppressing effects of leptin on the neural circuits controlling feeding would allow continued feeding driven by the protein appetite system. When excess carbohydrate or fat is ingested, some of it is converted to stored body fat, potentially contributing to obesity, but excesses may also appear in the urine (notably as glucose—hence the copious, sweet urine symptomatic of diabetes mellitus) or be “burned off” through a process called “homeostatic waste” (Kleiber 1961), “facultative diet-induced thermogenesis” (fDIT), “adaptive thermogenesis” (Rothwell and Stock 1979), or “wastage respiration” (Zanotto et al. 1997). This process has been most extensively studied in vertebrates, but equivalent mechanisms are ancient, occurring in invertebrates, plants, and bacteria (Fleury and Sanchis 1999; Stock 1999). In mammals, there is a tissue type specialized for fDIT, brown adipose tissue, which is typically located in the thorax and abdomen (Rothwell and Stock 1979; Cypess et al. 2009), although it also occurs in organs such as the brain, gut, liver, heart, and muscles. Facultative DIT is under neural and hormonal control and involves energy-dissipating biochemical pathways, such as futile substrate cycles and mitochondrial uncoupling proteins (Silva 2006). These processes literally burn off excess nutrients without generating ATP but in the process produce heat. As a result, the animal can ingest excess energy to gain limiting protein without gaining commensurately in fat mass. Michael Stock (1999) considered that fDIT evolved “as a mechanism for enriching nutrient-poor diets by disposing of the excess non-essential energy.” Such regulatory responses are especially associated with low-protein diets relative to the animal’s optimal nutrient requirements, and also with animals that have evolved to feed on low-protein foods. For example, the mechanisms for fDIT are highly developed in nectar and fruit eaters such as fruit bats and marmosets (Stock 1999). However, a low capacity to dissipate the heat generated by fDIT can limit the effectiveness of this mechanism for diet balancing—especially in large animals, which have a low surface area to volume ratio, and at high ambient temperatures (Speakman and Króll 2010). Zanotto and colleagues (1997) showed that locusts use wastage respiration to help rebalance their diet postabsorptively. Locusts were fed one of four diets, similar in composition to those discussed above in the study of Clissold and coworkers (2010) on digestive enzyme secretion: two diets were balanced in protein to carbohydrate ratio (1:1 P:C) but were either concentrated (PC) or diluted (pc) using cellulose; the other two

56  |  Chapter three

diets were unbalanced, containing either a low-protein, high-carbohydrate ratio (pC) or the opposite ratio (Pc). The rate of respiration (oxygen consumption and carbon dioxide production) was measured in a flowthrough respirometer. Locusts respired at a substantially greater rate on the pC than on the Pc diet, but dietary dilution had little effect if the diet contained a balanced protein to carbohydrate ratio. Thus, just as described above for the study on digestive enzyme secretion in the gut (Clissold et al. 2010), the compensatory response (in this case an increased respiration rate to void excess carbohydrate) was only shown when insects were in a state of nutritional imbalance. When nutrients were in balance but diluted in the diet, the response was largely behavioral: to eat more or less food.

3.3 Conclusions To regulate the balance of nutrients eaten, an animal needs to assess the composition of available foods in relation to its nutritional requirements. Some key nutrients are detected directly by taste organs, which are located on external appendages in some animals, within the mouth, and along the gastrointestinal tract. The presence or absence of other nutrients is “inferred” by learning to associate features of a food with the consequences of ingesting it. Nutritional state is assessed through systemic nutrient-sensing mechanisms and hormonal feedbacks from body reserves. Integration of information about food composition and nutritional state occurs both at the periphery, by nutrient-specific modulation of taste receptors, and more centrally as signals from systemic and peripheral sources converge onto the neural circuits that control feeding behavior. Learning also plays an important role. Postingestive regulatory responses can assist in rebalancing an imbalanced nutrient intake. The gastrointestinal tract sits at the interface between the food and the internal milieu and is now appreciated to be a key site of regulation, being plastic in both structure and function in response to nutritional state. Once digested and absorbed across the gut, nutrient supplies can be further rebalanced by differentially voiding excess nutrients and conserving nutrients that are in limited supply. During these first three chapters we have set in place the conceptual and mechanistic frameworks for the rest of the book. In the next chapter we turn to some of the consequences of nutrition for the individual, by considering relationships between three of the great biological imperatives: sex, death, and food.

four

Less Food, Less Sex, Live Longer?

The previous chapters have shown that animals have an intricately coordinated suite of regulatory mechanisms for gaining the required amounts and balance of nutrients. However, we have also shown that no matter how adept the regulatory mechanisms, the intake target can only be reached in environments containing nutritionally balanced and/or nutritionally complementary foods. When the intake target cannot be reached, the job of the regulatory mechanisms controlling feeding behavior is to achieve the best balance between overeating some nutrients and undereating others—the rule of compromise. Thereafter, postingestive mechanisms help to rebalance an unbalanced supply of nutrients to the tissues. In the logic of nutritional ecology, it is critical to learn not only how animals respond in different nutritional environments but also why. To do so, we need to perform cost-benefit analyses of the nutritional responses of animals. Our aim in this chapter is to show how this can be done using the Geometric Framework. As we will show, such analyses can provide a window not only into the evolution of nutritional regulatory strategies but also into other fundamental issues in biology. The example we use is the intriguing relationships between nutrition, life span, and sex. In 1935, Clive McCay and his colleagues at Cornell University published the results of a study in which they restricted the amount of food given to laboratory rats (McCay et al. 1935). Their aim was to use rats as a model for understanding the effects of malnutrition and growth retardation in humans. Remarkably, the food-deprived animals lived longer than rats with unrestricted access to food. Since that seminal publication, the view that dietary restriction without severe malnutrition prolongs life has become the big idea in research on aging (Weindruch and Walford 1988; Masoro 2005; Everitt et al. 2010). The list of species that live longer when modestly food-deprived is now extensive: yeasts, nematode worms, fruit flies, rodents, monkeys, and many more. Not surprisingly, there is huge public interest in the possibility that dietary restriction could

58  |  Chapter four

delay aging and extend life in humans [despite a dearth of experimental evidence (Heilbronn and Ravussin 2003; Prentice 2005b; Fontana 2006)]: try typing “calorie restriction lifestyle” into your search engine and enjoy some of the more than 400,000 entries that pop up. But what does dietary restriction actually mean? The most widely accepted view is that eating fewer calories (called “caloric restriction” or “calorie restriction”) is the reason for living longer, whatever the source of those fewer calories might be, whether protein, carbohydrate, or fat (Weindruch and Walford 1988; Masoro 2005; Everitt et al. 2010). Some have suggested, however, that calories may not hold the key to a long life; rather, the reduced intake of specific nutrients such as proteins and certain amino acids may be responsible (e.g., Ross 1961; Zimmerman et al. 2003; Mair et al. 2005). Separating the effects of calories and nutrients was important for the field of aging research but not easy to do. Geometric designs offered a possible solution but required a large number of dietary treatments, spanning numerous ratios and concentrations of nutrients (Simpson and Raubenheimer 2007; Archer et al. 2009). Lee and colleagues (2008a) set out to try such an experiment on one of the best-known model systems for aging research, the fruit fly Drosophila melanogaster. The experiment was inspired by an earlier study (Mair et al. 2005), which had shown, using four dietary treatments, that life span was affected more by the yeast content of the diet (yeast being the protein source) than by its caloric density. Fruit flies offer two major advantages for a geometric nutritional study, and one substantial difficulty. On the positive side, flies gain the majority of their calories from two macronutrient groups, carbohydrate and protein, with fat being relatively insignificant. This means that only these two macronutrient dimensions need to be considered when designing experiments, at least in the first instance. A second advantage is that fruit flies don’t live very long: a fly equivalent of Methuselah lives for less than three months. Counted against these advantages is the fact that flies are very small and suck their food through a proboscis, which makes measuring how much they eat a considerable challenge. Measuring intake is important, however, since it cannot be assumed that manipulating the concentration of nutrients in the diet equates directly to similar changes in nutrient intake (Simpson and Raubenheimer 2007). This is because animals typically compensate to some degree by eating more or less food to stabilize nutrient intake (see chapter 2). Prior to 2008, nobody had measured lifetime food intake in fruit flies. Lee and colleagues miniaturized a technique for measuring intake that stemmed back to studies on much larger blowflies (Gelperin and Dethier 1967; Simpson et al. 1989). They presented foods in liquid form in tiny (5 millionths of a liter) glass capillary tubes and measured the change in

Less Food, Less Sex, Live Longer?  |  59

the position of the meniscus as flies lapped the fluid from the end. The same technique was developed independently by Ja and colleagues (2007) and ingeniously termed the CAFE (for “capillary feeding”) assay. Using the CAFE technique, Lee and coworkers offered more than 1,000 mated female fruit flies one of 28 diets varying in the ratio and concentration of yeast to sugar and measured intake and egg production over the entire life span of each fly. Since the intake of protein and carbohydrate, which are the major energy-yielding nutrients for Drosophila, could be calculated from the volume ingested and the composition of the diet, it was possible to associate a given intake of nutrients with the consequences for life span and egg production. This relationship was examined by plotting response surfaces (see chapter 2) on top of the array of protein and carbohydrate intakes (plate 3A). These landscapes are colored dark red at their summit and change progressively to orange, yellow, and green before descending to deep blue at the foothills. Flies lived longest on a diet containing a 1:16 ratio of protein to carbohydrate and lived progressively less long as the ratio of protein to carbohydrate either decreased to zero or increased beyond 1:16. Because protein and carbohydrate yield similar numbers of calories per amount eaten, all points lying along a line with a slope of –1 on the intake graph are equal in the numbers of calories eaten (i.e., are “isocaloric”). Hence, 0.2 mg of protein contains the same number of calories as 0.2 mg of carbohydrate, as does 0.1 mg of protein plus 0.1 mg of carbohydrate (see the dashed line in plate 3A). If life span corresponded with the number of calories eaten, as predicted by the calorie restriction hypothesis, the life span contours would run parallel with these isocaloric lines. This was not the case; rather, the contours of the longevity landscape run almost at right angles to these lines. This can be seen by following the dashed isocaloric line from left to right on plate 3A and noting how life span falls progressively as the ratio of protein to carbohydrate in the diet increases. The data therefore prove that caloric restriction cannot explain the variation in life span; instead the balance of carbohydrate to protein ingested correlated most strongly with longevity. Lee and colleagues compiled another longevity landscape based on previously published studies on Drosophila. Each of these studies had involved flies housed in groups in cages and fed one of a limited number of diets, presented as nutrients set in agar jelly. Despite procedural differences across the various studies, the shape of the landscape aligned closely with that in plate 3A. Capillary-fed flies lived less long than flies housed in groups with agar-based diets, but the pattern of life span in relation to nutrient intake was the same: the ratio of protein to carbohydrate was the key factor. To confirm these conclusions, Lee and coworkers then conducted a more traditional cage trial in which groups of flies were offered agar-based diets and intake was not measured. These flies lived lon-

60  |  Chapter four

ger than their capillary-fed counterparts, but the pattern of living less long as the dietary ratio of protein to carbohydrate increased was the same (Lee et al. 2008a). A parallel experiment was conducted by Fanson and colleagues (2009) on the Queensland fruit fly, Bactrocera tryoni—a major pest of fruit crops in Australia and a member of a different family of flies from Drosophila. As can be seen in plate 3B, the results were very similar to those for Drosophila: once again, dietary protein to carbohydrate ratio and not energy intake was strongly associated with life span. In the studies of Lee and colleagues (2008a), Fanson and colleagues (2009), and others that have reached similar conclusions (e.g., Carey et al. 2008; Skorupa et al. 2008; Ja et al. 2009; Le Rohellec and Le Bourg 2009; Vigne and Frelin 2010), the major source of dietary protein was yeast. But yeast is a complex ingredient, containing micronutrients and other chemicals in addition to protein and carbohydrate. This leaves open the possibility that something correlated with protein in yeast, rather than protein itself and its constituent amino acid building blocks, was the active ingredient. Two lines of evidence indicate that protein really is primarily responsible, however. The first comes from a study by Maklakov and colleagues (2008) on adult field crickets, Teleogryllus commodus. They fed male and female crickets one of 24 chemically defined diets (i.e., made from pure nutrient sources rather than complex ingredients such as yeast) varying in protein and carbohydrate content (plate 3C and D). Just as in flies, crickets died progressively earlier as the ratio of protein to carbohydrate ingested increased beyond an optimal level. Male but not female crickets had a reduced life span when eating large quantities of low-protein, highcarbohydrate diets, which was associated with their greater propensity to lay down excess body fat on such diets (a point that we consider further below). The life-shortening consequences of high dietary protein to carbohydrate ratios can extend beyond the individual, as seen in recent experiments on social insects. Dussutour and Simpson (2009) used chemically defined foods to show that ant colonies died sooner at high protein to carbohydrate ratios (see chapter 7), as did Cook and colleagues (2010) in another ant species and Pirk and colleagues (2010) in honeybees. The second line of evidence in support of protein and its amino acid constituents causing the life span effect in flies, rather than some correlated component of yeast, comes from recent experiments on Drosophila in which scientists replaced yeast with mixtures of amino acids. Troen and colleagues (2007), inspired by reports of a link between the amino acid methionine and longevity in rodents (Orentreich et al. 1993; Zimmerman et al. 2003; Miller et al. 2005), developed four chemically defined

Less Food, Less Sex, Live Longer?  |  61

diets in which methionine and glucose were varied. Small but significant effects of dietary methionine on life span were reported. Subsequently, Fanson and Taylor (2011) developed a chemically defined diet for Queensland fruit flies and replicated the findings of Fanson and colleagues (2009) (plate 3B); therefore the life-shortening effect of an elevated yeast to sugar ratio in the diet was because of protein, not some unidentified components in yeast. Grandison and colleagues (2009) used a different strategy, in which they added supplementary mixtures of amino acids to yeast-based diets. They used two baseline dietary regimes, one that they called “full feeding” (FF) and the other “dietary restriction” (DR). The FF diet had a higher protein to carbohydrate ratio (1:1.3) than the DR diet (1:1.9); hence, as expected from plate 3A, flies lived longer on DR than on FF. Grandison and coworkers added various mixtures of free amino acids to the DR diet and measured the effects on longevity and egg production. We will concentrate on life span here and come back to reproduction below. When a complete mixture of 20 amino acids was added to the DR diet, flies lived less long; that is, the supplement mimicked the effect of having the extra yeast in the FF diet. If a mixture of only the 10 essential amino acids was added to DR (defined as those amino acids that the fly cannot synthesize itself), then the flies also lived less long, whereas a supplement of only the 10 nonessential amino acids had no such lifeshortening effect. Therefore, the active constituent (or constituents) in shortening life span must lie among the essential amino acids. However, if individual essential amino acids were added to the DR diet, none alone caused a reduction in longevity, showing that some combination of essential amino acids in the diet causes decreased longevity. When Grandison and coworkers omitted only methionine from an enriching mixture of essential amino acids, flies lived longer; but if only tryptophan was omitted, then there was no extension of life span. The inescapable conclusion from the results of Grandison and colleagues (2009) is that methionine is among a suite of amino acids, which does not include tryptophan but whose other members remain unknown, all of which have to be elevated in the diet to trigger life-shortening effects. Although the nature of the other amino acids in the suite remains unknown, there are obvious parallels in the results of Grandison and colleagues (2009) to the experiments by C. L. Simpson and colleagues (1990; see chapter 3, fig. 3.3), who discovered that protein satiety in locusts required that a suite of eight amino acids all be elevated simultaneously in the diet. This suite included methionine and four other essential amino acids (lysine, valine, phenylalanine, and leucine). The one apparent difference with the fly longevity results was that three nonessential amino acids (alanine, glutamine, and serine) were also in the suite of

62  |  Chapter four

eight for locusts. Omission of only one of these eight amino acids from an otherwise complete supplementary mix rendered a diet “low protein” so far as the animal was concerned. Signaling elevated protein status, whether to induce protein satiety in locusts or to trigger pathways involved in shortening life span in flies, therefore requires a specific mixture of amino acids. Taken together, the results from insects provide overwhelming evidence that caloric restriction is not responsible for life span extension. Instead, the ratio of protein to carbohydrate in the diet is crucial, with the protein component of the response mediated by a mixture of key amino acids, which includes, but is not exclusively, methionine. An important message from the insect results is that experiments in which single amino acids are manipulated in the diet without taking account of interactions with other amino acids (or with other macronutrients, notably carbohydrate) are at risk of being misinterpreted—a message that applies to studies on other animals too. What about mammals? Although it is widely held that caloric restriction, not specific nutrient effects, is responsible for life span extension in mammals (Weindruch and Walford 1988; Masoro 2005; Everitt et al. 2010), no experiment to date has contained sufficient dietary treatments to disentangle calories from specific nutrients (Simpson and Raubenheimer 2007). There have been numerous reports, stemming back to early work by Ross (1961), that protein restriction, and restriction of methionine in particular, extends life span in rodents (Orentreich et al. 1993; Zimmerman et al. 2003; Miller et al. 2005; Ayala et al. 2007; Sun et al. 2009), so it is at least plausible that the response of mammals—including humans—is similar to that of insects. Spurred on by the need for a geometric analysis of aging in mammals, we have embarked upon just such a study in mice with David Le Couteur at the ANZAC Research Institute in the University of Sydney. A full design for rodents has required expanding from two to three macronutrient dimensions with the inclusion of dietary lipid in addition to protein and carbohydrate. At the time of writing, the 30-diet experiment is still underway, but the data are already proving to be instructive.

4.1 How Does Macronutrient Balance Affect Life Span? We have seen that eating excess protein relative to nonprotein energy shortens life span, at least in insects and perhaps also in mammals. The mechanisms causing this effect are not yet understood, but there are some tantalizing candidates. These include altered production of radical oxygen species (“free radicals”) with associated damage to DNA and cellular pro-

Less Food, Less Sex, Live Longer?  |  63

teins (Sanz et al. 2004; Ayala et al. 2007); toxic effects of nitrogenous breakdown products arising when protein is used instead of carbohydrate or fat as an energy source; changes in immune function and alteration in the capacity to deal with other dietary toxins (as we discuss in chapter 5); and perhaps even changes in the entrainment of circadian rhythms (Hirao et al. 2009). However, it is becoming increasingly apparent that the central coordinators of the effect of macronutrient balance on life span are the nutrient-signaling pathways that we introduced in chapter 3. These pathways are shared by a diversity of organisms from yeasts to humans and include the insulin/insulin-like growth factor (IGF), TOR, and AMPK pathways (Kapahi et al. 2010; Kenyon 2010; Katewa and Kapahi 2011; Mair et al. 2011). It is not only aging that is affected by these pathways; they are emerging at the heart of multiple life-threatening disease processes, including eating disorders such as anorexia and cachexia (a wasting condition common in cancer patients), obesity, cancer, type 2 diabetes, cardiovascular disease, and other metabolic disorders (fig. 4.1). What is needed next are biochemical and molecular genetic studies in which gene expression patterns and metabolic responses are mapped as surfaces onto nutrient intake arrays, as has been done for major life history variables such a life span and fecundity (plate 3). Such studies will help unite nutrition, aging, and their affiliated diseases within a single explanatory framework, spanning genes to behavior. Increasing the ratio of protein to nonprotein energy in the diet decreases life span, but if this ratio falls too far there is an increased risk of an early death associated with obesity. We will address this issue in detail in chapter 10, but it warrants some discussion here. The reason why the risk of obesity increases as the dietary ratio of protein to nonprotein energy falls below the intake target ratio is that many animals, especially herbivores and omnivores (including humans, as we shall see in chapter 10), regulate their intake of protein more strongly than that of carbohydrate and fat. Consequently, when confined to diets that are high in the proportion of fat and/or carbohydrate relative to protein, animals overeat to gain the target protein intake. Unless these excess calories from fat and carbohydrate are voided by increased activity levels or the up-regulation of thermogenic (heat-generating) mechanisms (see chapter 3), the animal becomes obese and prone to various metabolic disorders. As we discuss in chapter 6, the propensity to store excess calories as body fat, rather than burn them off, varies among species, populations, individuals, and sexes, and can be shown to shift across generations in response to a change in the nutritional environment (e.g., Warbrick-Smith et al. 2006). An example of how individuals of the same species differ can be seen in the comparison of male and female field crickets shown in plate 3C and D; other examples are provided in chapter 6.

64  |  Chapter four AMPK activity

High P:C diet

High P:C diet

Levels of amino acids

Levels of amino acids

Leptin; insulin/IGF; etc.

Stress factors; sirtuins; etc.

Nutrients High aa:glu

Insulin resistance; autophagy and repair inhibited

Low P:C diet

Levels of glucose

Low P:C diet

Levels of glucose

TOR activity

Low aa:glu

TOR

AMPK

Eat less

Anabolic responses Protein synthesis, lipogenesis, cell proliferation, growth, reproduction

Insulin sensitivity; autophagy and repair promoted

Eat more

Catabolic responses Cell cycle arrest, inhibition of growth and reproduction, lipolysis, proteolysis

Vicious cycle to anorexia

Vicious cycle to obesity Live longer

Pathological increase in lean signalling (e.g. tumour release of IL6)

Overeat on low %P diets

Obesity and insulin resistance

Vicious cycle to anorexia or cachexia

Die early

Lipolysis, elevated FA, lean muscle breakdown, enhanced hepatic gluconeogenesis Depleted muscle mass and amino acid pool; reduced lean signal (IL-15?); low aa:glu; high AMPK

Less Food, Less Sex, Live Longer?  |  65

A central theme of this book is that nutritional imbalance comes at a cost. When considering aging and life span, most diet restriction experiments use two treatment regimes: food-restricted and unrestricted. If animals live longer when food-restricted, the usual interpretation is that there are benefits to eating less. But there is an alternative explanation: food-restricted animals don’t live longer because of the benefits of restriction, but, rather, the other group dies sooner because of the damage caused by unrestricted access to food (Raubenheimer et al. 2005). The likelihood of this latter alternative will depend on how nutritionally balanced the diet is in relation to the changing requirements of the animal across its lifetime (Simpson and Raubenheimer 2007, 2009). The majority of diet restriction studies maintain animals in captivity on a single food type; yet nutritional requirements change over time and cannot be met by a single food at all stages of an animal’s life (chapter 6). The ideal diet for a juvenile will differ from that of a mature individual, which will differ again for the elderly. Because the required balance of nutrients changes over time, yet the diet is fixed, individuals in a typical experiment will constantly be making compromises between eating too much of some nutrients and too little of others—with associated costs. The accumulated damage resulting from the nutrient excesses may well be eased by modest diet restriction. Results from studies on calorically restricted monkeys (Colman et al. 2009) are consistent with the beneficial effects of diet restriction resulting

Fig. 4.1. Schematic summarizing ways in which dietary macronutrient balance might affect life span via the TOR and AMPK signaling pathways. It is hypothesized that TOR and AMPK respond not only to the concentration of circulating nutrients but also to nutrient balance, notably the ratio of amino acids (especially branched-chain amino acids such a leucine) to glucose (aa:glu). Hypothetical response surfaces for TOR and AMPK are shown at the top, with the response level rising from light gray to black. The vicious cycle to obesity is shown, in which chronic exposure to a diet low in protein and high in nonprotein energy (carbohydrates and fat) can drive overconsumption, metabolic disorders, and shortened life span unless excess ingested energy is dissipated (see chapter 10). Otherwise, low-percentage protein diets are life extending via the normal actions of AMPK, whereas high-percentage protein diets shorten life span and encourage aging via the TOR pathway. Additionally, it is hypothesized that a pathological increase in signals associated with protein repletion (e.g., release of IL 6 by tumors) may inhibit food intake and drive development of anorexia and cachexia. (After Simpson and Raubenheimer 2009.)

66  |  Chapter four

from the costs of excess in the unrestricted controls, rather than the intrinsic benefits of eating less. Monkeys either had restricted or unlimited access to a single food, monkey chow, for 20 years. The major reasons why restricted animals lived longer was a reduction in the incidence of diabetes, cancer, and cardiovascular disease relative to control animals. These are typical diseases of nutrient excess and imbalance. It follows from the above logic that the closer a diet matches the changing needs of an individual, the less beneficial will be the effects of diet restriction. In the extreme, for a perfectly balanced diet there will be no benefits, only costs, to diet restriction (see chapter 2). This might explain why a recent study in rodents saw no change in life span in response to food restriction when animals were confined to a food that was “optimized for health and longevity benefits” (Smith et al. 2010).

4.2 Less Sex, Live Longer? Living long enough to reproduce is the raison d’être of all organisms, so it is no surprise that when resources are scarce many organisms shut down reproduction and wait it out until conditions improve and the prospects for reproduction are brighter (Zera and Harshman 2001). Since “waiting it out” may lead to organisms living longer than if they had reproduced earlier in life—sometimes tens or hundreds of times longer in the case of organisms that enter a state of suspended animation (Withers and Cooper 2010)—the idea arose that reproduction and aging trade off against each other by competing for resources. Hence, limited resources are allocated with highest priority to maintaining and repairing the organism’s body (its “soma”), thereby improving the chances that it will live long enough to experience better conditions and reproduce (Kirkwood 2005). The obverse of this trade-off hypothesis is that when there are sufficient resources to support reproduction, somatic maintenance is allocated resources sufficient only to survive the reproductive period—so, not only does inhibition of reproduction cause delayed aging, but reproduction effectively shortens life span. A further variant of this idea is that reproduction doesn’t just compete with somatic maintenance for resources; it produces its own damaging side effects that shorten life span (Tatar 2007; O’Brien et al. 2008). Under all these various manifestations of the trade-off hypothesis, maximal life span and maximal reproductive output are mutually exclusive—you can have one or the other, but not both. We will return to the issue of mutual exclusiveness later, but first we will consider what is meant by “resources.” When considering the tradeoffs between reproduction and aging, “resources” and “energy” are usu-

Less Food, Less Sex, Live Longer?  |  67

ally used as synonyms. However, as we have discussed above, energy alone cannot explain aging and life span: so how do life span and reproduction relate to one another in a more complex representation of the nutritional landscape? To answer that question, we need to return to the experiments illustrated in plate 3. As discussed above, Lee and colleagues (2008a) found that mated female Drosophila lived longest on a diet containing a 1:16 ratio of protein to carbohydrate (P:C) (plate 3A). The same flies, however, laid most eggs across their life span when confined to a more protein-rich diet, 1:4 P:C. An increase in P:C beyond 1:4 resulted in a drop in lifetime egg production. These results suggest two important conclusions. First, the number of calories eaten cannot explain the relationships between diet, longevity, and reproduction—instead, these relationships can only be understood by considering nutrient balance. Second, because the performance landscapes for life span and lifetime egg production have different shapes and their peaks lie in different locations, flies cannot maximize both life span and lifetime egg production on a single diet. An interesting question is which peak do the flies choose to climb if given free choice? To address this, Lee and coworkers (2008a) offered flies one of nine complementary food pairs in the form of separate yeast and sugar solutions differing in concentration. The flies converged upon a diet comprising 1:4 P:C, thereby maximizing lifetime egg production and paying the price of a diminished life span. Fanson and colleagues (2009) found something very similar in their study of Queensland fruit flies. These insects lived longest on P:C 1:21, laid most eggs on P:C 1:3, and selected P:C 1:3 when offered a choice (plate 3B). Maklakov and coworkers’ (2008) experiment on field crickets added a new dimension by considering the reproductive efforts of both females and males. A major investment in reproduction by male crickets is their calling to attract females. Males sing by rubbing their wing covers together. Not only is singing energetically expensive, but in addition to luring females it may attract the unwelcome attention of predators. So for females, Maklakov and colleagues counted the number of eggs laid, and for males they recorded the amount of time spent singing throughout their lives, on 24 diets varying in protein and carbohydrate content. Both sexes of cricket lived longest on low P:C diets (plate 3C and D), and, like flies, female crickets laid most eggs on a more protein-rich diet (P:C 1:1). Males, however, sang most on low P:C diets. For males, the response surfaces for life span and singing were similar, but female crickets, like female flies, were faced with a quandary: they could not live longest and lay most eggs on the same diet. When offered a choice, both sexes selected a diet composition that neither maximized life span nor reproductive effort. Maklakov and colleagues argued that this represented a compro-

68  |  Chapter four

mise between the reproductive optima of males and females, with neither sex able to achieve its own optimum because of constraints imposed by a shared nutritional regulatory biology. Sex-specific regulation of nutrient intake to meet differing reproductive needs is seen in other insects, including flies (Dethier 1976; Barton Browne 1995; Ribeiro and Dickson 2010, Vargas et al. 2010), but, if this interpretation is correct, not in crickets. South and colleagues (2011) recently used nutritional geometry to show that nutrient balance plays another role in defining the battle of the sexes, this time in cockroaches. These authors manipulated the intake of proteins and carbohydrates by male cockroaches, Nauphoeta cinerea, and related this to their production of sex pheromones, their success in being chosen as mates by females, and their status among competing males. Carbohydrate (but not protein) intake affected sex pheromone production and male attractiveness, but did not affect male dominance status. When males were given a choice between complementary foods, they selected a higher carbohydrate intake, thereby smelling more attractive to females. Maklakov and colleagues’ (2008) experiment on crickets provided another opportunity: a chance to test evolutionary theories about the relationship between aging and reproductive senescence by measuring how male and female reproductive effort changed throughout their life span on different diets (Maklakov et al. 2009). The trade-off hypothesis for aging versus reproduction that we discussed at the beginning of this section (and related evolutionary theories of aging) predicts that reproductive effort will decline with advancing age (Hughes and Reynolds 2005). In contrast, life history theory predicts that animals should devote an increasing proportion of their resources to reproduction as they approach the end of their lives—they should go out with a bang, as it were (Williams 1966). Males are predicted by sexual selection theory to continue advertising their sexual wares well into later life especially vigorously because, compared with females, reproduction is relatively cheap (Kokko 1997; Graves 2007)—what we might call the “go out with a smile on your face” hypothesis. Maklakov and coworkers (2009) observed that females reached their egg-laying peak early in life and then their reproductive output declined toward death—as predicted by theories of aging. Males, however, continued to increase their singing effort throughout their lives, reducing calling rates only at ages unlikely to be reached in the wild (fig. 4.2)—they died ever hopeful. Finally, let us return to the matter of the mutual exclusiveness of reproduction and somatic maintenance. As seen in plate 3, the peaks for maximal reproduction and longevity occupy different regions in nutritional space for female insects (although they come close to aligning for male field crickets), such that females cannot achieve maximal fecundity and

Less Food, Less Sex, Live Longer?  |  69 A. Female crickets 80

B. Male crickets

Singing (seconds per night)

Eggs laid per week

60

40

20

0

800

Max. in wild

0

20 40 60 Age (days)

80

Max. in wild

600

400

200

0

0

20

40 60 80 Age (days)

100

120

Fig. 4.2. The relationship between life span and age-dependent female egg laying (A) or male singing (B) in field crickets. Reproductive aging occurs well within the natural life span (dashed vertical arrow) of females. In contrast, males continue to sing well into old age, decreasing their calling only after the maximal life span recorded in the wild. (After Maklakov et al. 2009; courtesy of John Wiley and Sons.)

life span on the same diet. The trade-off theory proposes that longevity and reproductive output cannot both be maximized simultaneously, either because they compete for the same resources, or there are direct costs to reproduction that shorten life span. Is this true? Grandison and colleagues (2009) have shown that, for flies at least, the answer is no. As we discussed above, when fed a low P:C diet (what Grandison and coworkers called the diet restriction condition, DR), flies laid fewer eggs and lived longer than if fed a higher P:C diet (what they termed the full feeding condition, FF). When a mixture of essential amino acids was added to the restricted diet, flies increased their egg production but lived less long—as expected from the trade-off hypothesis. However, when the restricted diet was augmented with only methionine, flies both lived long and had high egg production. As we considered above, methionine addition alone was not enough to cause life span to shift (because other essential amino acids need also to be elevated in the diet to trigger the lifeshortening effects of a high P:C diet); but methionine enrichment of the DR condition was sufficient to increase fecundity to that in the FF condition. So maximal longevity and fecundity can coexist under the appropriate nutritional conditions, which means that the trade-off hypothesis as

70  |  Chapter four

usually formulated is not correct, nor is the variant hypothesis that there are direct costs of reproduction that shorten life span (see also Flatt 2011; Tatar 2011).

4.3 Conclusions Dietary restriction without malnutrition is considered to be a universal mechanism for prolonging life span. It is generally believed that the benefits of dietary restriction arise from eating fewer calories. However, GF experiments on insects in which the effects of macronutrients have been separated indicate that, rather than calories, a key determinant of the relationship between diet and longevity is the balance of protein to nonprotein (fat and/or carbohydrate) energy in the diet. Whether the same is true for mammals remains to be seen, but existing data indicate that it may well be. As we shall see in following chapters, the ratio between protein and nonprotein energy intake affects not only life span but also total energy intake, metabolism, immunity, and the likelihood of developing obesity and associated metabolic disorders. Among various possible mechanisms linking macronutrient balance to life span, the interaction between the TOR and AMPK signaling pathways is emerging as a central coordinator. The nutrient signals that activate these pathways remain to be elucidated, but it is likely that a mixture of amino acids must be elevated in the circulation to produce protein satiety and to activate parallel metabolic pathways that are implicated in aging. Finally, the presumption in much of life history theory that life span and reproduction trade off against each other for limiting resources (usually considered to be energy) is shown to be too simplistic. These two life-history variables certainly have differing nutritional optima, but they can be dissociated and do not inevitably trade off. Reproductive senescence and aging may proceed at different rates in males and females, as predicted by sexual selection theory. In the next chapter we show that it is not only aging and reproduction that have differently shaped response surfaces when mapped onto nutrient intake arrays, but so too do the physiological systems that respond to toxins and disease.

five

Beyond Nutrients

To this point in the book we have focused our geometric analyses on the macronutrients—protein, carbohydrates, and fat. This is for a reason: as the data show, macronutrients can explain a good deal of the variation in the behavioral, physiological, and performance responses of animals. Macronutrients are, however, not the only important nutritional components of foods: the constituent molecules in macronutrients (amino acids and fatty acids, for example) and micronutrients such as vitamins and minerals also play a critical role in an animal’s nutritional strategies and physiology, as do other components of foods that would not normally be considered nutrients. Examples of the latter include refractory fiber (e.g., lignin from plants and chitin from some animal foods) and a range of secondary metabolites that play various roles from toxic defenses to essential antioxidants and even antibiotic components that act as medicines (Huffman 2001, 2003; Villalba and Provenza 2007). Our aim in this chapter is to demonstrate that the conventional categorization of food components into “macronutrient,” “micronutrient,” “toxin,” “medicine,” and so on works well from a distance, but on greater magnification the boundaries between these categories blur (Raubenheimer and Simpson 2009). When viewed through a geometric lens, however, a new structure falls into focus, which emphasizes not the chemical identity of the food component but the target-like perspective of optimal intakes. We will structure our argument around three interlinked themes: (1) the distinction between “nutrient” and “toxin” is fuzzy and sometimes imaginary; (2) the phenomenon of “self-medication” in nonhuman animals can involve compounds that are conventionally classified either as nutrients or natural “medicines”; and (3) even when a compelling case can be made for distinguishing a “toxin” from a “nutrient,” the biological impacts of the toxin depend on the levels of nutrients in the food relative to the intake target for those nutrients.

72  |  Chapter five

5.1 The Distinction between Nutrients and Toxins Conventionally, those chemical compounds in food that enhance health and fitness and elicit appetitive responses (e.g., amino acids, fatty acids, sugars) are categorized as “nutrients,” while those that are deleterious and trigger aversive and defensive responses (e.g., alkaloids, polyphenolics, terpenoids) are classified as “toxins.” Although such classifications seem intuitive and are for many purposes operationally useful, attaching functional labels to compounds in this way risks obscuring the dynamic, context-dependent nature of biology. In fact, “toxin” versus “nutrient” is a loose distinction, and for many purposes a more helpful dichotomy is between the adjectival versions “toxic” versus “nutritious” or, more generally, “deleterious” versus “beneficial” (Berenbaum 1995; Raubenheimer and Simpson 2009). We will use two well-established phenomena to illustrate this point: hormesis and Bertrand’s rule, both of which are modern restatements of the ancient wisdom that “only the dose makes the poison” (attributed to Paracelsus 1538—Stumpf 2006).

Hormesis Hormesis is a concept developed in toxicology, in which the effects on biological systems (cells, tissues, organs, organisms, populations) of exposure to a substance are reversed with increasing concentration (Calabrese and Baldwin 2003). Hormetic dose-response relationships can take two forms. The most common of these is the inverted U shape, in which low doses of a substance stimulate and high doses inhibit beneficial biological responses (e.g., growth, fecundity, or longevity) (fig. 5.1A). A second form is the J-shaped curve, where low doses reduce and high doses exacerbate a deleterious response (e.g., tumor formation, mortality, or growth suppression) (fig. 5.1B). Hormetic dose-response curves are different from other dose-response relationships, such as the linear threshold (LT) and linear nonthreshold (LNT) models (fig. 5.1C and D). Traditionally LT and particularly LNT models have formed the basis of dose-response thinking in toxicology (Hayes 2007). Over recent decades, however, it has become apparent that hormetic responses are common, having been observed for a wide range of chemicals, organisms, and biological responses (Calabrese et al. 1999; Calabrese 2005). Indeed, it has been proposed that hormetic responses can be predicted by evolutionary theory (Forbes 2000; Parsons 2001). The acceptance of hormesis in mainstream toxicology and biomedical sciences has, however, been surprisingly slow given its long history (see, for example, Calabrese and Baldwin 2000) and strong empirical foundations (Calabrese and Baldwin

Beyond Nutrients  |  73

D.

Dose

Response

Dose

Response

C.

Response

B.

Response

A.

Dose

Dose

Fig. 5.1. The hormetic inverted U (A) and J-shaped (B) dose-response curves, contrasted with the nonhormetic linear threshold (C) and linear nonthreshold (D) curves. In each case the dashed horizontal line represents the reference (e.g., control) response. (From Raubenheimer and Simpson 2009.)

2001). The reasons for this are complex, but two factors that have likely played a role (Calabrese 2005) are a failure to consider in experiments the full dose-response curve and the historical confusion of hormesis with the scientifically questionable concept of homeopathy (Park 2000). Furthermore, similar dose-response relationships have been described in several subdisciplines of toxicology and biomedical science, but poor communication among these fields has led to a proliferation of terminology and lack of appreciation for the generality of this phenomenon. Recently, this literature has been unified under the concept of hormesis (Calabrese et al. 2007). We have argued (Raubenheimer and Simpson 2009), however, that this unification will remain incomplete until it encompasses parallel developments in the nutritional sciences, the subject to which we now turn.

Bertrand’s Rule Although some researchers in the area of hormesis have recognized similarities between toxin and nutrient dose-response curves (Luckey and Stone 1960; Hayes 2007), there have been separate developments in the nutritional sciences that are directly relevant. These developments stem

74  |  Chapter five Optimum

Function

100%

Deficit costs

Excess costs Dose

Fig. 5.2. Bertrand’s rule: at low doses of a nutrient, increased intake is associated with increasing benefits, but beyond an optimal intake any further increase results in health costs. (From Raubenheimer and Simpson 2009; modified from Mertz 1981.)

back to the French scientist Gabriel Bertrand, who in 1912 established a rule concerning the dose-response curve for mineral nutrients. According to Bertrand’s rule (fig. 5.2), . . . a function for which a nutrient is essential is very low or absent in a theoretical, absolute deficiency, and increases with increasing exposure to the essential nutrient. This increase is followed by a plateau representing the maintenance of optimal function through homeostatic regulation, and a decline of the function toward zero as the regulatory mechanisms are overcome by increasing concentrations that become toxic. (Mertz 1981) Bertrand’s rule is believed to apply to all essential micronutrients, with the detailed shape of the curve depending on the nutrient in question and the biological context (Mertz 1981). In contrast to micronutrients, it was generally assumed that Bertrand’s rule does not apply to macronutrients, despite evidence to the contrary—not least of which is the starvation (costs of energy shortage)–obesity (costs of energy excesses) spectrum in modern humans (see fig. 10.1). Indeed, the concepts (and reality) of targets and rules of compromise in the Geometric Framework are predicated on there being costs to excesses as well as deficits of macro- and micronutrients. Of course, there is no question that the first phase of the curve would apply for macronutrients (increasing benefit with amelioration of a deficit), but the second (plateau) and third (toxic) phases are, for several reasons, less generally agreed upon (Raubenheimer and Simpson 2009). First, the strong emphasis in nutritional biology on energy being a primary limiting nutritional currency has tended to cast doubt on the possibility that

Beyond Nutrients  |  75

an animal could ingest excessive levels of useful energy-yielding macronutrients. Second, it is difficult to induce animals to overingest macronutrients in a way that enables the health costs of those excesses to be disentangled from the impacts of concomitant deficits of other nutrients (Simpson et al. 2004; Raubenheimer et al. 2005). The reason for this is that, as demonstrated in previous chapters, the regulatory mechanisms of animals will usually resist large excesses of macronutrients—which in itself suggests that surpluses, like deficits, are costly (Raubenheimer and Simpson 1997). As a result, food intake (and hence intake of micronutrients) is reduced when there are high compared with lower concentrations of macronutrients. An illustration of this is given in chapter 9, where we discuss the design of supplementary feeds for the critically endangered kakapo parrot. One way to get around the problem of confounding the effects of excesses of specific nutrients with deficits of others is to conduct the kinds of experiments we described in the previous chapter, where nutrient intakes are spread across a broad area of macronutrient space and response surfaces are used to assess the performance consequences of different nutrient intakes. Comparisons can then be made among selected groups within the experiment to determine the effects on performance of specific macronutrient excesses and deficiencies. Using the data we discussed in chapter 2 for the caterpillar Spodoptera littoralis, we took this approach to isolate performance costs that were due specifically to carbohydrate excesses, independent of other nutrients (fig. 5.3) (Raubenheimer et al. 2005). Three foods were involved in this comparison, all containing a carbohydrate to protein balance of 5:1, which was shown using both selfselection treatments and performance measures to be carbohydratebiased relative to the optimal balance (1:1.2) for these caterpillars (Simpson et al. 2004). The foods differed, however, in the concentration of the macronutrient mix, comprising either 42% (35% C + 7% P), 63% (52.5% C + 10.5% P), or 84% (70% C + 14% P) by dry mass of the two nutrients. Each of the four experimental groups comprised a pair of diet blocks with differing average macronutrient density as follows: 42% and 42% (labeled L in fig. 5.3), 63% and 42% (ML), 84% and 42% (MH), or 84% and 63% (H). By varying the food combinations in this way, we were able to coax the different experimental groups to ingest amounts of nutrients that enabled us to disentangle the effects of carbohydrate surpluses from protein deficits and micronutrient intakes (see Raubenheimer et al. 2005 for further details). Our results (fig. 5.3) showed that caterpillar performance declined with increasing macronutrient concentration of the experimental treatments (fig. 5.3A). This effect could not be explained by micronutrient intake, because intake of micronutrients didn’t differ systematically across treat-

76  |  Chapter five A.

B. 40 Micronutrient intake (mg)

40

Performance

30

20

10

0

L

ML

MH

10

L

ML

MH

H

D. 500 Macronutrient intake (mg)

500

Cellulose intake (mg)

20

0

H

C.

30

400 300 200 100 0

L ML MH H Macronutrient concentration

400

Protein Carbohydrate

300 200 Pt Ct

100 0

L ML MH H Macronutrient concentration

Fig. 5.3. Effects of dietary macronutrient concentration on (A) performance (development rate × survival) (B) micronutrient intake (C) cellulose intake, and (D) macronutrient intake by sixth-instar Spodoptera littoralis caterpillars. In (D), the solid horizontal line labeled Pt represents the level of protein intake that was self-selected by caterpillars over the same developmental period and provided maximum performance (i.e., the target intake), while the dashed horizontal line labeled Ct represents the target intake for carbohydrate. Symbols for nutrient concentrations are L = lowest, ML = medium low, MH = medium high, H = highest. (From Raubenheimer et al. 2005.)

ments (fig. 5.3B). Excessive cellulose intake can affect the performance of caterpillars through its influence on digestion and absorption (e.g., Lee et al. 2004a), but this cannot explain our results, because performance was better at high levels of cellulose intake (fig. 5.3C). The effect must therefore have been due to macronutrient intake. Comparisons among the ex-

Beyond Nutrients  |  77

perimental groups showed that both carbohydrate and protein intake increased with increasing macronutrient density of the treatment pairings (fig. 5.3D). There was, however, a key difference between the trends for carbohydrate and protein: the increase in carbohydrate intake was over and above the required (target) intake, and thus represented an increasing excess, whereas protein intake was below the target level, and therefore the increased protein intake on macronutrient-denser foods represented a diminishing protein deficit. Since reducing a protein deficit is beneficial, the observed reduction in performance could only be due to excessive carbohydrate intake. This analysis shows that carbohydrate excesses are costly for caterpillars, thus demonstrating that Bertrand’s rule does, indeed, apply to a macronutrient. We suspect that Bertrand’s rule is at least as prevalent in nutrition as is hormesis in toxicology (see also Boersma and Elser 2006; Zehnder and Hunter 2009).

Toxin or Toxic? The striking similarity between the inverted-U-shaped hormetic doseresponse curve (fig. 5.1A) and Bertrand’s rule (fig. 5.2) suggests that these should be regarded as related phenomena. In both cases the ingestion of a substance can have positive or negative effects depending on the dose, and from a functional (hence evolutionary) perspective it makes no difference to the animal whether science has labeled these substances “toxin” or “nutrient.” An alternative and for some purposes more productive distinction would be between “beneficial,” “neutral,” and “toxic” regions of the dose-response curve, regardless of whether that curve pertains to components of foods that have traditionally been the focus of nutrition or to substances that have traditionally been the focus of toxicology. Many definitions of the word “nutrient” stress the advantageous effects of ingesting these substances, as well as the mechanistic basis for these effects—for example, “any substance that can be metabolized by an animal to give energy and build tissue” (http://wordnetweb.princeton. edu/). Toxins, on the other hand, are defined as substances that disrupt life-sustaining processes. Although these definitions can be useful in some contexts, there are many cases that are not adequately covered by either of them. Take, for example, ethanol. This alcohol is widely regarded as a toxin to humans, yet in some populations it can comprise 5% of the energy budget and in some individuals considerably more (Prentice 2005a). Syrian golden hamsters achieve even greater feats of dipsomania and can derive up to a third of their energy budget from ethanol. In choice assays these animals select ethanol over water, even at concentrations as high as 45% (DiBattista and Joachim 1998). Another example concerns plant

78  |  Chapter five

secondary metabolites that are toxic to some animals but are beneficial to others (e.g., Bernays and Woodhead 1982; Slansky 1992). There are, likewise, plant primary metabolites that are nutritious for some herbivores but toxic for others—for example, some sterols (see Behmer and Nes 2003). These awkward cases demonstrate that the chemical identity of a compound is not always a reliable indicator of whether it is a “toxin” or a “nutrient”—the relationship between the levels ingested and the biological costs and benefits for the animal must be taken into account. It has, however, been argued by toxicologists that it can also be problematic to think about substances in terms of these dose-response relationships. One problem that has been raised is that it can be difficult to attribute biological benefit to a specific response that might result from ingesting a given amount of a substance (Calabrese and Baldwin 2002). However, the same applies to nutrition, and yet in that field the question of biological benefit is handled using a number of approaches, including geometric analysis (chapter 2). It has also been pointed out that although individual components of fitness, such as individual life-history traits, might respond hormetically to stressors (i.e., depending on the dose they might respond positively or negatively), evolutionary theory does not predict that fitness as a whole would respond in this way (Forbes 2000). The argument here is that the complex web of trade-offs that exists among traits means that a given dose of a particular substance might, for example, enhance the immune system but at the same time create disadvantages in other respects—for example, reduce fertility. However, as we have seen in earlier chapters, a similar web of trade-offs exists in nutrition (e.g., see the discussion in chapter 4 on the trade-off between life span and reproduction), and again these can be dealt with using a suitable approach such as geometric analysis. If toxins can be nutritious and nutrients toxic, it seems safe to predict that animals should evolve foraging and feeding responses that reflect the relationships between dose and consequences—that is, they should specifically target substances at doses that are beneficial and avoid toxic doses (Raubenheimer and Simpson 2009). As shown in chapters 2 and 4, there is good evidence that animals do indeed regulate their intake of foods so as to avoid ingesting both excesses and deficits of specific nutrients, and in many cases a good deal is known about the mechanisms involved (chapter 3). A recent literature shows that animals likewise regulate their intake to avoid ingesting harmful levels of toxins (Torregrosa and Dearing 2009). An important question is whether animals also select foods specifically to ingest toxins at levels that are beneficial. The interesting phenomenon of “self-medication”—to which we turn in the following section—suggests that they do.

Beyond Nutrients  |  79

5.2 Self-medication and Ecological Immunology: The Distinction between Nutrients and Medicines Self-medication, or zoopharmacognosy (Rodriguez and Wrangham 1993), is the phenomenon in which animals use plant secondary metabolites or other nonnutritional substances to prevent or treat disease (Huffman 2003). This is common practice among traditional human societies (Johns 1990; Huffman 2001). Dan Janzen raised the possibility that the same might be the case in nonhuman animals. Janzen (1978) compiled a list of examples of food selection by animals in the wild that could not readily be explained on the basis of nutrient requirements or toxin avoidance, and suggested that animals might target these foods specifically for their medicinal properties—for example, as laxatives, antibiotics, antihelminthics, or antidotes for previously ingested toxins. Several studies have since suggested that this is indeed the case for animals as diverse as chimpanzees, leopards, bears, geese, dogs, and sheep (Villalba and Provenza 2007). The most extensive evidence for self-medication by animals in the wild comes from studies of chimpanzees. These apes are known to target several plant species containing compounds that at the levels ingested have medicinal properties, including the antibiotic methoxysporalen in Ficus exasperata (Rodriguez and Wrangham 1993), the antimalarial liminoids in Trichilia rubescens (Krief et al. 2004), and the antihelminthic sesquiterpene lactones in the pith of Veronia amygdalena. The evidence suggests that chimpanzees target at least one of these species, V. amygdalena, specifically for its medicinal properties (Huffman 2001, 2003). Despite its year-round availability, this plant is eaten mainly during the rainy months, when infection by nematodes is at its peak, and apparently it is targeted specifically by chimpanzees that show signs of sickness. Observations have suggested that the consumption of V. amygdalena is associated with a dramatic reduction in parasite load and a recovery of health (Huffman 2001). Given the difficulties of performing experimental work with free-ranging primates, evidence for self-medication in chimpanzees has remained largely correlative and observational (Lozano 1998; Hutchings et al. 2003). In contrast, research on livestock has provided strong experimental evidence for self-medication in nonhumans, particularly in the context of reducing the negative impacts of ingested toxins. Provenza and colleagues (2000) showed that sheep that are fed foods high in tannins selectively eat polyethylene glycol (PEG), which reduces the adverse effects of these plant secondary compounds. Sheep also selectively eat sodium bicarbonate and sodium bentonite to offset the acidic consequences of eating some grains, and dicalcium phosphate to ameliorate the effects of

80  |  Chapter five

ingesting oxalic acid (Villalba et al. 2006). Experiments have demonstrated that associative learning (see chapter 3) forms an important part of the mechanisms involved in self-medication by sheep. Insects have also been show to self-medicate. Caterpillars of the tiger moths Grammia incorrupta and Estigmene acraea defend themselves against insect parasitoids by sequestering pyrrolizidine alkaloids (PA) from their food plants. These caterpillars have specialist taste receptor cells (chapter 3) that detect PA even at extremely low levels and stimulate feeding. Bernays and Singer (2005) showed that the PA receptors on the mouthparts of parasitized caterpillars responded more vigorously than receptors of unparasitized controls. Based on previous work demonstrating that increased responsiveness of the taste receptors to PA stimulates feeding, Bernays and Singer concluded that these caterpillars have a mechanism for increasing the intake of protective chemicals as a response to parasitization. Singer and colleagues (2009) extended this work for G. incorrupta, and set it in the context of a framework for rigorously detecting cases of therapeutic self-medication. They made the important argument that three criteria should be satisfied to verify cases of therapeutic self-medication: (1) the behavior should improve the fitness of animals infected with parasites or pathogens; (2) it should decrease fitness in uninfected animals; and (3) the behavior should be triggered by infection. They found that all three criteria were satisfied for the ingestion of PA by G. incorrupta. Although much remains to be done in this field (Lozano 1998; Hutchings et al. 2003), the research discussed above shows that poisoned or parasitized animals can specifically select foods containing nonnutrient compounds that help them to recover. Might they also select diets with a nutritional composition that neutralizes toxins or reduces parasite load and associated disease? In addressing this question for parasitic infection, Hutchings and colleagues (2003) discussed several experiments that have shown a link between diet and immune responses. For example, studies by Cosgrove and Niezen (2000) and by Hutchings and coworkers (2000) have shown that sheep infected with gastrointestinal parasites select a diet relatively high in nitrogen-rich clover. However, as Hutchings and colleagues (2003) point out, for these observations to be categorized as self-medication, it needs to be shown that the shift in diet by infected animals increases resistance to parasites—that is, that criterion (1) of Singer and colleagues (2009) is met. Lee and colleagues (2006b) produced such evidence for the caterpillar Spodoptera littoralis exposed to a highly virulent viral pathogen, nucleopolyhedrovirus (fig. 5.4). When infected caterpillars were confined to one of five foods varying in their balance of protein to carbohydrate, resistance to pathogen attack and constitutive immunity were greater as

Beyond Nutrients  |  81 8 7

Control

Performance

6 5 4 3

Viruschallenged

2 1

35:7

28:14 21:21 14:28 7:35 % protein:% carbohydrate in diet

Fig. 5.4. Performance of Spodoptera littoralis caterpillars challenged with nucleopolyhedrovirus and control caterpillars when confined to one of five diets varying in protein and carbohydrate content. Performance was calculated as the proportion of insects surviving × their average growth (mg) per day. Each descending arrow indicates the pathogen-induced performance loss of insects feeding on each P:C diet. (From Lee et al. 2006b.)

the proportion of protein in the diet increased, perhaps reflecting the relatively high protein costs of resistance (Lee et al. 2006b, 2008b). A similar response was reported earlier by Peck and colleagues (1992), who found that after being inoculated with Salmonella typhimurium mice survived better on diets containing a higher ratio of protein to carbohydrate. In the research (2006b) of Lee and coworkers, uninfected caterpillars performed best on a diet with lower protein concentration than that which supported maximal performance in infected individuals (see also plate 1). When allowed to self-select their diet, caterpillars surviving infection increased their relative intake of protein compared with controls and with caterpillars that died of infection. This experiment demonstrates that these caterpillars are able to combat viral infection by modulating the macronutrient composition of their diet, in the same way that sheep are able to neutralize the effects of toxins, and chimpanzees and caterpillars are able to fight parasitic infection, by supplementing their diet with nonnutritional medicines. Furthermore, the experiment was performed within a framework that satisfies the three criteria required by Singer and colleagues (2009) as definitive evidence for therapeutic self-medication: a diet high in protein increased performance in infected animals, reduced performance in uninfected controls, and was selected only by infected animals.

82  |  Chapter five

Povey and colleagues (2009) extended the study of Lee and colleagues (2006b) to another pathogen, the bacterium Bacillus subtilis, with similar results. Infected caterpillars survived better on a high-protein diet, and larvae injected with a sublethal dose of bacteria selectively increased their intake of protein if offered a choice of foods. Povey and coworkers (2009) also found that components of the immune system of the caterpillars interacted with one another in relation to diet. Elevation of some immune components was associated with a decline in others. The traditional view of such immune trade-offs (as is also the case for other life history traits; see chapter 4) is that different components compete for a limiting pool of resources, typically energy. However, as we saw in chapter 4 for life span and reproduction, another possibility is that, rather than competing for a single limiting resource, there is something akin to internal “niche partitioning” among different elements within the immune system, with each component having a different nutritional optimum. This idea was explored by Cotter and colleagues (2010) in experiments on caterpillars in which individuals were restricted to one of 20 diets varying in the quantity and ratio of protein and carbohydrate. Some caterpillars were immune-challenged by being pricked with a needle dipped in a solution of dead cells of the bacterium Micrococcus lysodeikticus, whereas others were not challenged. Results indicated that immune traits do in fact exist within an organismal “ecology,” in which each trait has its own specific nutritional requirements. This was shown by plotting response surfaces for different immune traits onto nutrient intake arrays (plate 4). No one diet simultaneously optimizes all immune components, but the opportunity exists for the animal to adjust its food selection behavior to attain a nutritional state that supports the balance of immune responses that best resists infection by a particular pathogen. These and other examples that explore ecological and evolutionary aspects of immunity have shaped the emergence of a new research field, ecological immunology (Sheldon and Verhulst 1996; Owens and Wilson 1999; Rolff and Siva-Jothy 2003; Wilson 2005; Hawley and Altizer 2011). Central to this field is the question of how foods and their nutrient and nonnutrient constituents affect the ability of animals to mount immune responses against pathogens and parasites. The other side to ecological immunology, from a nutritional perspective, is that parasites and pathogens rely on the host for provision of resources. Because these organisms may not share the same nutritional requirements as their hosts, there are possibilities for resource competition and manipulation between the different parties (Smith and Holt 1996; Ponton et al. 2011a, b). The complexity of the nutritional interactions between hosts and pathogens is made even greater because animals play host to entire communities of commensal and symbiotic microorganisms,

Beyond Nutrients  |  83 Feeding behavior

Host diet

Microbial community in the gut

Host immune function Host nutritional state

Pathogen/s

Host fitness

Fig. 5.5. The network of interactions between nutrition and immunity. Diet affects host nutritional state and immune status, both of which interact with microbial symbionts, commensals, and pathogens to affect the fitness of all partners. Because nutrient feedbacks modulate host feeding behavior, the potential exists for the host to adjust its diet to optimize its microbial interactions and increase resistance to infection. Alternatively, parasites and pathogens might subvert host feeding behavior to their nutritional advantage. (From Ponton et al. 2011b.)

which receive their nutrition from the host and in turn contribute essential nutrients (fig. 5.5; Topping and Clifton 2001; Douglas 2010). The human gut microbiota alone comprises about 1014 microorganisms, with a collective genome (“microbiome”) containing 100 times as many genes as their host (Gill et al. 2006). Gut microorganisms have substantial effects on immune and inflammatory responses (Wen et al. 2008; Mazmanian et al. 2008; Maslowski et al. 2009; Ryu et al. 2010), and in mammals, disturbances of the gut microbiota have been implicated in obesity, type 1 diabetes, and various cancers (Topping and Clifton 2001; Sekirov et al. 2010). Moreover, the diet of the host has a strong effect on the gut microbiota (Turnbaugh et al. 2009; De Filippo et al. 2010; Kau et al. 2011), both by serving as a vector for microorganisms and by affecting the physical, chemical, and structural properties of the gut (Flint et al. 2007; Ley et al. 2008; Duncan et al. 2008). These interactions are ripe for geometric analysis (Ponton et al. 2011b). We conclude from this section that animals can ameliorate the impacts of toxins, parasites, and pathogens by flexibly modulating their choice of diet. From a functional viewpoint, whether the targeted medication is nutritional or nonnutritional is a detail that, for many purposes, is irrelevant. A similar conclusion was reached by Villalba and Provenza (2007),

84  |  Chapter five

who pointed out that in this respect both nutrient and nonnutrient components of food can be represented in geometric models of nutrition as axes with target (optimal) coordinates > 0 (see also Simpson and Raubenheimer 1999; Raubenheimer et al. 2009). In the following section we provide an example demonstrating another approach to the analysis of nutrient-toxin interactions using nutritional geometry.

5.3 Toxins and Nutrients Interact We have argued above that for many purposes nutrient and nonnutrient components cannot be simply distinguished, because both can be nutritious, medicinal, or toxic. On the other hand, there clearly are cases in which nutrients are nutritious and toxins are categorically harmful—as is true for many coevolved defensive compounds (Sotka et al. 2009). In this section we show that even when toxins can be distinguished categorically in this way, it is often the case that their effects can only be understood in the context of the background nutritional milieu. The modes of interaction of nutrients and toxins are diverse, involving intake, digestion, and absorption, as well as postabsorptive effects (Slansky 1992; Sotka et al. 2009). In relation to intake, an animal’s current nutritional state (hence recent feeding history) can exert a powerful influence on its willingness to ingest toxins (e.g., Cronin and Hay 1996). In addition to the animal’s recent feeding history, the nutrients co-occurring in foods with toxins can also influence the amounts of toxins ingested. For example, Slansky and Wheeler (1992) showed that a compensatory response to dilution of the overall nutrient content of a diet led the velvet bean caterpillar (Anticarsia gemmatalis) to ingest toxic levels of caffeine, whereas other studies have demonstrated a specific effect of macronutrient balance on the intake of toxins (Raubenheimer 1992; Simpson and Raubenheimer 2001; see below). The water content of food is also known to affect the deterrency of toxins directly, rather than via its effects as a diluent of dietary nutrients (Glendinning and Slanksy 1994). Many examples are known in which nonnutrient components of plants interact with nutrients to influence the digestion and absorption of foods. Perhaps the best-known example concerns polyphenolics, which bind to dietary proteins and other macromolecules in the guts of animals and thereby lower the digestibility and availability of these nutrients (e.g., Tugwell and Branch 1992; Targett and Arnold 2001; Bennick 2002). The nutrient content of foods, and the subsequent impact of those nutrients on an animal’s nutritional state, can similarly influence the way that animals handle toxins after they have been absorbed through the gut wall.

Beyond Nutrients  |  85

One example concerns the action of xenobiotic-metabolizing cytochrome P450 enzymes, which are sensitive to dietary levels of various nutrients. In general, nutritional deficiencies result in reduced rates of cytochrome P450-mediated xenobiotic metabolism, although in some circumstances (e.g., thiamine deficiency) the activity of these enzymes might increase (Yang et al. 1992). Our application of the Geometric Framework to a study of nutrienttoxin interactions shows how complex and important these interactions can be (Simpson and Raubenheimer 2001). Juveniles of Locusta migratoria were confined to one of five diets varying in protein to carbohydrate ratio (P:C) and containing either 0, 3.3, 6.7, or 10% tannic acid (TA). The striking result was that locusts were impervious to TA, even at the highest level of 10%, provided that their diet was near balanced in its P:C relative to the intake target. As the diet became either protein- or carbohydratebiased, locusts were progressively more susceptible to TA (fig. 5.6); however, the reason why nutritional imbalance affected susceptibility to TA differed according to the direction of nutritional imbalance. As the diet became more carbohydrate-biased, insects suffered because of the increasingly powerful antifeedant effects of TA. Tannic acid mixed in a low P:C food caused locusts to eat so little that they starved to death. By contrast, food intake was not affected on high P:C diets, but instead TA had toxic effects postingestively. Behmer and colleagues (2002) next explored the interactive effects of tannic acid and macronutrients in a more complex environment, in which two nutritionally complementary foods (one high in carbohydrate, the other high in protein) containing TA were provided along with a third, TA-free food. Providing that the TA-free food was carbohydratebiased, locusts were able to maintain their macronutrient intake target by mixing this food with the TA-containing high-protein food. However, they abandoned intake target regulation if the TA-free food was proteinbiased, because of their aversion to TA-containing high-carbohydrate foods (fig. 5.7). These results (Simpson and Raubenheimer 2001; Behmer et al. 2002) put a different complexion on the evolution of plant phenolics such as tannic acid as defenses against herbivory. The prevailing view was that, because phenolics are carbon-based and do not require nitrogen for their synthesis (unlike other plant secondary metabolites such as alkaloids), plants with an excess of carbon (e.g., due to high sun exposure) and a shortage of nitrogen manufacture them as protective chemicals (Coley et al. 1985). Our results showed that tannic acid only serves as a feeding deterrent when present in foods containing an excess of carbohydrate relative to protein.

86  |  Chapter five

Percent survival

100

Diet 7:35

80 60 40 20

Percent survival

0 100

Diet 14:28

80 60 40 20

Percent survival

0 100

Diet 21:21

80 60

Near optimal P:C ratio – resistant to tannic acid in diet

40 20

Percent survival

0 100

Diet 28:14

80 60 40 20 0 100

Percent survival

Lower than optimal P:C – TA feeding deterrent

Diet 35:7

80

Greater than optimal P:C – TA post-ingestive toxin

60 40 20 0

0

10 3.3 6.7 Percent tannic acid in diet

Fig. 5.6. Interaction between nutritional balance and tannic acid in the diet of juvenile locusts (Locusta migratoria). The intake target for protein and carbohydrate is near to 21:21 (21% protein, 21% carbohydrate). Note that locusts were essentially resistant to the presence of tannic acid in a near optimally balanced diet, but that survival was reduced by addition of tannic acid to the food as the ratio of protein to carbohydrate shifted away from the intake target. (From Simpson and Raubenheimer 2001.)

Beyond Nutrients  |  87

Carbohydrate eaten (mg)

450

C+

375

1

2 3

350 225

1

2

150

5 P+

4

75 0

4 3 5

0

75

150 225 300 Protein eaten (mg)

375

450

Fig. 5.7. Interaction between nutritional balance and tannic acid (TA) when locusts could choose between three foods. Two of these foods contained TA: C+, which was rich in carbohydrate, and P+, which was rich in protein. The third food was one of five that varied in the ratio of protein to carbohydrate (1–5) but lacked tannic acid. When the TA-free food was carbohydrate-biased (foods 1–3), locusts regulated macronutrient intake to a target (indicated by the ellipse) by mixing this food with P+. However, if the TA-free food was protein-biased, locusts abandoned regulation to the intake target and ingested a more protein-biased diet because of their aversion to TA-containing, high-carbohydrate foods. (After Behmer et al. 2002.)

5.4 Conclusions We have shown in this chapter that “toxins,” “medicines,” and “nutrients” are closely—sometimes inseparably—interrelated. This is because the consequences for animals of ingesting various compounds are massively contingent, depending not just on the chemical structure of the compound but also on, among other things, the organism (e.g., taxon, developmental stage, nutritional state, health), the dose, the balance of other compounds in the food, pathogen challenges, host detoxification and immune responses, and multifarious interactions among these factors. The Geometric Framework is well suited to disentangling these complex relationships, because it provides a method for interrelating within a single model diverse dietary components and animal attributes, including optimal requirements and performance responses. In the next chapter we show how these relationships are dynamic, and how animals adapt over different timescales to changing environments.

six

Moving Targets

In earlier chapters we have shown that intake targets, rules of compromise, and postingestive responses are fundamental, interrelated traits, against which the adequacy of diets and nutritional environments can be calibrated and the consequences of nutritional imbalance assessed. For clarity, we have treated intake and growth targets as static points integrated across a particular period in the life of an animal. In reality they are, of course, not static but rather trajectories that move in time. In the short term, the requirements of the animal change as environmental circumstances impose differing demands for nutrients and energy. At a somewhat longer timescale, targets move as the animal passes through the various stages of its life, from early growth and development to maturity, reproduction, and senescence. On an even longer timescale, nutritional traits are subject to natural selection and move as species evolve to exploit new or changing nutritional environments and to adopt differing life-history strategies. Presaging such evolutionary change in gene frequencies within populations are epigenetic effects, whereby the nutritional experiences of parents influence the behavior and metabolism of their offspring without requiring changes in gene frequencies.

6.1 Moving Targets in the Short Term An animal’s nutritional requirements will depend on the physiological demands placed on it, and the intake target should move accordingly. To illustrate this, figure 6.1 shows the case of an adult locust placed upon a flight mill (Raubenheimer and Simpson 1997). Locusts were fixed by a wire to the top of their thorax and allowed to fly around and around in circles. They are prodigious migrants and can fly in this manner for hours on end. Locusts were removed after flying for 20 minutes or after two hours, then placed in a container and allowed to select a protein and carbohydrate intake from two complementary foods over the following 24 hours. Note how the point of nutrient intake shifted with the time

Moving Targets  |  89

Carbohydrate eaten (mg)

45 Flown 120 min

40

35 Flown 20 min Unflown

30

25

20 20

25

30 35 Protein eaten (mg)

40

45

Fig. 6.1. The intake target moves in response to exercise. Adult locusts were flown on a flight mill for different periods and allowed to select their diet from two complementary foods. The bivariate mean (+ standard error) intakes selected by locusts flown for 20 minutes, 120 minutes, and unflown controls show that carbohydrate intake increased more than protein intake after prolonged flight. This reflects the fact that dietary carbohydrate provides the main flight fuel. Dashed lines indicate nutritional rails of the two complementary foods. (From Raubenheimer and Simpson 1997.)

spent flying. Notably, locusts that had flown for two hours selected significantly more carbohydrate but not significantly more protein than those that had flown for 20 minutes, or not flown at all. This result reflects the fact that dietary carbohydrate provides the primary source of fuel for flight in locusts. Another major influence on nutritional requirements is environmental temperature. Nutrient intake varies as a function of body temperature, as do metabolic rate and the efficiency with which ingested nutrients are utilized (e.g., Angilletta 2009; Miller et al. 2009; Coggan et al. 2011). In warm-blooded animals the target mixture of nutrients can vary as a function of the energy demands for thermoregulation, as seen in the increased consumption of carbohydrate but not protein in weanling rats kept at 8ºC rather than 23ºC (Musten et al. 1974; Simpson and Raubenheimer 1997). As we discussed in chapter 5, yet another change in environmental circumstances that will shift the intake target is the need for animals to meet the nutritional demands associated with fighting pathogens and mitigating the effects of toxins.

90  |  Chapter six

Changes in demand for nutrients imposed by activity, temperature, pathogens, toxins, or other short-term influences will act via the suite of mechanisms introduced in chapter 3, but over what time periods are these changes detected and tracked? In theory an animal with appropriate complementary foods available could adjust its feeding behavior and food choices mouthful by mouthful to track its moving intake target with very high resolution; but each time the animal shifts between foods, it risks attracting unwarranted attention from natural enemies and wastes both time and energy (Houston and Sumida 1985). At the other extreme, an animal might switch just once, maintaining fidelity to one food until it reaches the point where it must swap foods if it is to attain its target integrated over some longer period. However, by following this strategy the animal will suffer the cost of increased time in a state of nutritional imbalance (far from the target trajectory)—in chapter 4 we discussed the possibility that accumulated effects of nutritional imbalance play a role in shortening the life span of animals. These scenarios are illustrated in figure 6.2, with the intake target trajectory shown tracing a smooth curve through nutrient space. In one case (A) the animal makes only a single food switch to attain the final end position of the target (the bull’s-eye symbol), but the area between its actual intake trajectory and the target trajectory is large. In contrast, the animal represented in track B switches foods on five occasions and therefore tracks the target trajectory closely. The optimal switching interval for a real animal under natural conditions should reflect the relative costs of staying on an imbalanced food versus those of moving to find a complementary food (Houston et al. 2011). The costs of moving can include failing to locate a nutritionally complementary food (discussed further below), wasting time searching that could be better spent finding mates or on other fitness-enhancing activities, encountering predators along the way, or succumbing to inclement conditions. Costs of remaining on an unbalanced food include not only eating deficient amounts of some nutrients and surpluses of others but also missing out on better-balanced foods located elsewhere, inducing protective responses from the food (especially an issue for herbivores, in which feeding can induce the production of toxins), running out of food, and attracting predators and parasites by continuing to occupy the same location (Chambers et al. 1998; Simpson et al. 2010). Chambers and colleagues (1995) found that locusts switched between two complementary foods differing in protein and carbohydrate content on average every four hours, maintaining fidelity to one food type for runs of around 4–5 meals in a row before shifting to the other food. Earlier it had been found that nutritional feedbacks operate after one hour but reach full strength after four hours (Simpson et al. 1988, 1990). It seems, then, that locusts integrate their state in relation to their intake target over

Moving Targets  |  91 Food 1 Intake target trajectory 5 3

B

A

4 Food 2 1

1

2

Fig. 6.2. Two routes to achieve the moving intake target when selecting between complementary foods. The trajectory through nutrient space of the intake target over time is shown by the gray curve. In case (A), a hypothetical forager has switched foods only once to reach the final target position, but has not tracked the moving trajectory. In case (B), the forager has switched foods on five occasions, thereby tracking the moving target closely.

periods of about four hours, and this determines their food-switching frequency. Humans, by contrast, appear to integrate macronutrient feedbacks over a period of 1–2 days (de Castro 2000; Weigle et al. 2005; Gosby et al. 2011; see chapter 10), which, intriguingly, represents a similar number of meals as for locusts (4–5). Forager ants also take a day or so to begin collecting less of a concentrated sugar diet and more of a diluted diet to achieve nutritional homeostasis at the colony level (Dussutour and Simpson 2008; see chapter 7).

6.2 Moving Targets in Developmental Time As animals grow, develop, reproduce, and grow old, their nutritional requirements change, in both the amount and the blend of nutrients needed. We discussed an extreme example of this in chapter 1, where an animal in the juvenile (caterpillar) stage eats leaves, and in the adult (butterfly) stage eats nectar. In that case the developmental change is obvious, because it involves two very different foods. However, equivalent changes take place in most if not all animals, but they are often more subtle, involving not a shift between food types but a change in the relative pro-

92  |  Chapter six 80 Maintenance

60 50

Growth phase

Carbohydrate eaten (mg)

70

40 30 20 10 0

0

10

20

30 40 50 60 Protein eaten (mg)

70

80

Fig. 6.3. The intake target moves during development. The points show successive daily intakes of protein and carbohydrate selected by locusts over the first two weeks of adulthood. (From Simpson and Raubenheimer 1993b; after data from Chyb and Simpson 1990.)

portions of the same food types in the diet. For example, in figure 6.3 we have plotted the changing daily intakes of protein and carbohydrate from two nutritionally complementary synthetic foods for young adult locusts (Chyb and Simpson 1990). During the first days after the adult molt, the insect grows and lays down new muscle and fat in readiness for longdistance migratory flight, after which it enters a period of maintenance leading up to reproductive maturity. The daily intakes of protein and carbohydrate track these patterns of growth and maintenance. Another example is shown in figure 6.4, where the intakes of protein and carbohydrate are plotted for rats during different life stages. Young rat pups need protein, carbohydrate, and fat to grow, but once they reach puberty their intake of protein declines, and carbohydrate and fat dominate the diet (fig. 6.4A). Male and female rats have different intake trajectories, with males selecting more protein as they grow, presumably to sustain their greater size and muscle mass. When females reach maturity, their macronutrient intake varies with the reproductive cycle (fig. 6.4B). During pregnancy they select more protein than before mating, but not more carbohydrate or fat (the latter is not shown here—see Simpson and Raubenheimer 1997). When suckling the resulting pups, mothers increase their intake of protein and fat but not carbohydrate. After the pups are weaned, the mother reduces her protein intake to premating levels but

Moving Targets  |  93 A. 14

Carbohydrate eaten (g)

12 10 8

Females

6

Males

4 2 0

0

2

4

6 8 10 Protein eaten (g)

12

14

B. 5

Carbohydrate eaten (g)

4 Lactating 3

Premated Post-weaning

Gestating

2

1

0

0

1

2 3 Protein eaten (g)

4

5

Fig. 6.4. Developmental changes in daily protein and carbohydrate intake in rats during (A) successive weeks from weaning until maturity, and (B) different periods of the reproductive cycle in females. (From Simpson and Raubenheimer 2007; after data from [A] Leibowitz et al. 1991; [B] Richter and Barelare 1938.)

94  |  Chapter six

sustains high fat intake to restore body reserves that have been depleted during reproduction. Compared with rodents, less is understood about how the intake target for macronutrients (as distinct from energy requirements) changes throughout life in humans, or about the consequences of failing to track this moving target (we return to this in chapter 10). There has, however, been substantial work on the nature and evolution of our peculiar life history (Hawkes and Paine 2006; Wells 2010; Crespi 2011). Human breast milk, like that of other primates, is relatively dilute, and relatively low in protein and fat and high in sugar (lactose), compared with that of nonprimate mammals (ca. 7% of total energy content is protein, 43% carbohydrate, and 50% fat) (Jenness 1979; Sellen 2007). This composition has been associated with the slow growth rates and extended development times of primates (Oftedal 1991). Since a suckling infant is unable to regulate the composition of its diet—only, to some extent, the amount ingested—it might be expected that natural selection has ensured that the mother produces milk that is balanced with respect to the infant’s age-specific needs (Trott et al. 2003). Although there are many dimensions in which nutrient balance is no doubt important, there is reason to expect from the numerous examples presented throughout this book that the balance of protein in respect to other macronutrients will be among them. Studies of primates suggest that this is, indeed, an aspect of the composition of milk that is tightly regulated (Power et al. 2008; Raubenheimer 2011). Humans are unique among mammals in weaning very early and introducing nutritionally rich complementary foods to the diet before suckling ends (Sellen 2007). Bottled formula feeding (which, unlike breast milk, represents a fixed nutrient rail that typically lacks immune components) and reliance on manufactured baby foods potentially threaten infant health prospects (Sellen 2007) and require a better understanding of the infant intake target and rules of compromise when restricted to suboptimal diets. After weaning, human children are unique among primates in not being able to forage independently. Rather, they are reliant on parental feeding—which both sets up possibilities for nutritional misdirection and provides opportunities for early dietary interventions for improving lifetime health (Gluckman et al. 2009). Following a period of slow growth throughout childhood, humans undergo a rapid pubertal growth spurt, with associated changes in nutritional requirements. Then, as adults, the nutritional demands of reproduction must be met (with consequences for the nutritional well-being of the next generation—see below), after which there are changes associated with older age and senescence (Bogin 2006; Le Couteur and Simpson 2011). Measuring the human intake target throughout the life course remains one of the major challenges in nutrition.

Moving Targets  |  95

There are two basic mechanisms for tracking developmental shifts in the intake target. Barton Browne (1995) termed these “demand-mediated” and “non-demand-mediated” mechanisms. The former are those mechanisms that we have discussed in chapter 3, involving dynamic responses to changing nutritional needs. Hence, the animal experiences a developmental change in patterns of metabolism and growth (induced, for example, by environmental or hormonal signals), and the resulting shift in demand for nutrients causes an appropriate change in feeding behavior. Alternatively, the developmental switch may itself cause the shift in diet that achieves a new intake target without involving nutrient feedbacks, as seen, for example, in female blowflies and mosquitoes, which upon mating start to develop eggs and become responsive to odors from carrion and live animal hosts, respectively (Barton Browne 1995). Attraction to these resources results in ingestion of high-protein food (rotting flesh or fresh blood) that supports egg development. Recent studies on the fruit fly Drosophila melanogaster have begun to provide molecular insights into the relationship between demandmediated and non-demand-mediated mechanisms associated with female reproduction. Ribeiro and Dickson (2010) found that the rapidly induced preference for protein-rich yeast in female flies after mating results not from depletion of protein reserves as eggs are developed, but as a direct result of a sex peptide that is introduced with the male’s seminal fluid during mating and stimulates special sensory neurons in the female’s reproductive tract. An additional demand-driven mechanism then modulates how much yeast is eaten, involving TOR/S6 kinase and serotonin signaling pathways in the central nervous system (Ribeiro and Dickson 2010; Vargas et al. 2010; see chapter 3).

6.3 From Parents to Offspring—Epigenetics Some changes in the nutritional environment are sufficiently persistent that intake targets, rules of compromise, and postingestive responses need to shift across generations. The main mechanism allowing the phenotype to track changing environments over generations is, of course, natural selection acting via changing gene frequencies within populations. However, it has become increasingly appreciated that effects of the nutritional environment experienced by one generation can be passed to subsequent generations without requiring changes in gene frequencies. For example, dietary-induced obesity in mothers and even grandmothers has a direct influence over the metabolism and risk of obesity in offspring (Barker 1998; McMillen and Robinson 2005; Gluckman and Hanson 2006a). Obesity in fathers can, it seems, also affect metabolic function in off-

96  |  Chapter six

spring (Ng et al. 2010). That nutritional consequences in one generation can reverberate and become amplified across subsequent generations has profound health and socioeconomic implications (Gluckman et al. 2009). The search for controlling mechanisms of epigenetic effects is well underway. A major focus of research has been on inherited patterns of gene activation, involving processes such as gene methylation, histone modification, and involvement of micro-RNAs, but other pathways are also possible, including changes in milk production by lactating mothers and cultural influences. Another focus of interest is the extent to which epigenetic influences represent adaptive developmental responses, or are simply unavoidable changes that do not improve or might even reduce the fitness of the offspring. Responses that are specifically adaptive have been termed “predictive adaptive responses” to denote the fact that the developmental trajectory is cued to take a course that is adaptive (bene­ ficial) with respect to the predicted state of the future environment. For example, if the mother’s diet has low protein content, this might serve as a cue to a developing fetus that it is likely to be born into a proteindeficient environment, triggering the development of behavioral and metabolic responses that are appropriate for handling a protein-deficient diet. As discussed further in chapter 10, such responses can, however, lead to problems when there is a mismatch between the predicted future environment and that actually encountered (Gluckman and Hanson 2004, 2006b)—for example if a fetus that has developed a phenotype suitable for a low-protein environment subsequently finds itself in a high-protein environment. Researchers have begun using the Geometric Framework to investigate the extent to which rats develop phenotypes that are specifically adapted to the nutritional conditions experienced by their mothers (Coveny 2009; Coveny et al. 2010). These experiments have involved several groups of rats whose mothers were transferred at the onset of puberty from a standard, grain-based diet to one of several synthetic diets of differing macronutrient composition. Each group of rats was then maintained on the new diet over several generations. The first generation of rats that had been exposed to a high-carbohydrate diet throughout their lives were small with a high proportion of body fat. However, the offspring of these rats had larger lean body mass and lower fat density than their parents, even though they had experienced rearing conditions (including diet) identical to those of their parents. This shows that rats whose parents were raised on a grain-based diet are more susceptible to high-carbohydrate synthetic diets than genetically similar rats whose mothers were raised on the high-carbohydrate diet. Since the only environmental factor that differed between the two groups of rats was the diet of the mothers, this provides evidence for rapid, transgenerational

Moving Targets  |  97

but nongenetic adaptation to nutritionally altered environments. Experiments on locusts (discussed below) have shown that rules of compromise can shift in response to the maternal environment (Simpson et al. 2002). Although still in its infancy, we believe that the use of GF experimental designs has great potential for the study of nutritional epigenetics.

6.4 Evolving Targets

Performance (percent survival, final μg dry wt.)

As organisms evolve to exploit different nutritional niches or to track changing nutritional environments, their intake targets will evolve to match the diet. As a result, the optimal diet composition will differ substantially among organisms. The optimal diet for an aphid that feeds on sugary plant sap low in nitrogen is not the same as the optimal diet for a locust that eats whole plant tissues (fig. 6.5) or for a carnivore that eats predominantly meat. We elaborate on this point in chapter 8, but for now we will show how comparative data on intake targets from many species can be used to explore the association between an animal’s diet, its life history, and its evolutionary past.

100 Locust

90

Aphid

80 70 60 50 40 30 20

0

0.2

0.4

0.6 0.8 1.0 1.2 C:P ratio (radians)

1.4

1.6

1.8

Fig. 6.5. Performance in relation to the ratio of carbohydrate to protein (or amino acids) (plotted in radians) in the diets of juvenile migratory locusts and pea aphids. Both the optimal composition for performance (survival in locusts and growth in aphids) and the range of diet compositions tolerated reflect the natural foods of the two species. (Data from Raubenheimer and Simpson 1993; Abisgold et al. 1994.)

98  |  Chapter six Without symbionts Locusta migratoria Schistocerca gregaria Tribolium confusum

100

Blattella germanica Periplaneta americana

aphids wood and grain beetles Supella longipalpa

With symbionts

Digestible carbohydrate

80

60

Spodoptera littoralis Plodia interpunctella

40

Exeristes roborator Helicoverpa zea

20

0

0

20

40 60 Protein/amino acids

80

Phormia regina 100

Fig. 6.6. Selection of data from a comparative analysis of intake target ratios for protein and carbohydrate (P:C) in 117 species of insect. Dashed target rails indicate species with complete metamorphosis (endopterygote development), while unbroken rails indicate species with incomplete metamorphosis (exopterygote development). Irrespective of developmental strategy or taxon, species with endosymbiotic bacteria had a lower P:C than those without. (From Raubenheimer and Simpson 1999; after data from Simpson and Raubenheimer 1993b.)

Our example concerns an analysis in which the optimal protein to carbohydrate ratios (P:C) in the larval diets of 117 insect species were estimated from a combination of GF experiments and a published compendium of diet recipes used for rearing insects in the laboratory (Simpson and Raubenheimer 1993b). These ratios were superimposed on a phylogenetic tree of the insect groups, to control for shared ancestry when associating diet composition with features of life history. A subset of the 117 species is plotted in figure 6.6. The strongest pattern to emerge from the comparative analysis was the association between a reduced P:C ratio

Moving Targets  |  99

and the acquisition of symbiotic microorganisms (endosymbionts) living within specialized body cells in the host insect. Species with endosymbionts have a capacity to thrive on nitrogen-poor diets, such as phloem sap (aphids), dead wood (borer beetles), and detritus (cockroaches). Indeed, such insects have evolved a mutual association with symbionts precisely because they provide the host with a novel metabolic capability—the upgrading of dietary nitrogen (Douglas 2010). Possessing symbionts allows the host to exploit low-protein resources that are inadequate for other species. The corollary of acquiring this new metabolic capacity is that diets with higher levels of protein, although optimal for other species, may become nutritionally imbalanced for species with endosymbionts. This is clearly seen in figure 6.5 for locusts and aphids.

6.5 Evolving Rules of Compromise: Nutrient Specialists and Generalists Not only is the intake target dynamic over physiological, developmental, and evolutionary timescales, but so too are rules of compromise. We gave an example in chapter 2 (fig. 2.4) when comparing nymphs of the mi­ gratory locust, Locusta migratoria, and the desert locust, Schistocerca gregaria. If offered nutritionally complementary foods, both species regulated to the same intake target ratio of protein to carbohydrate (P:C), but when confined to imbalanced diets, Schistocerca nymphs ate more than did nymphs of Locusta. The intake array of Schistocerca was near linear, while that of Locusta was convex (arc-shaped) (fig. 2.4; replotted in fig. 6.7). As a result of eating more of the nutritionally imbalanced foods, Schistocerca ingested greater excesses relative to the target of the more abundant macronutrient in the diets and a smaller deficit of the limiting nutrient. Locusta, in contrast, was prepared to suffer a more substantial deficit in the deficient nutrient to avoid overconsuming the excess nutrient (Raubenheimer and Simpson 2003). Over time during development, the intake arrays in both species opened out as locusts on the more balanced diets molted to become adults while those on imbalanced diets continued eating. This was especially so on low-protein diets, where feeding continued until a threshold level of protein intake was attained, allowing molting to proceed (see fig. 2.5A; Raubenheimer and Simpson 1993). Why should one species of locust be more willing to overeat imbalanced diets than the other? One obvious difference between the two species is that Locusta is a host-plant specialist, restricted in its range of food-plants to species of grasses, whereas Schistocerca is notoriously

100  |  Chapter six Nutrient generalists

Transitional forms

Nutrient specialists 200

Locusta

150 100 50 0

Carbohydrate eaten (mg)

200

Carbohydrate eaten (mg)

200

150

150

100

100

100

50

50

50

200

0

50

100

150

200

0

0

50

100

200

Spodoptera littoralis

0 80

150

60

100

40

50

20

0 200

Carbohydrate eaten (mg)

200

Schistocerca: transiens

150

0

0

50

100

150

0

200

Heliothis virescens

200

150

150

100

100

50

50

0 40

Carbohydrate eaten (mg)

Schistocerca: gregarious

0

50

100

150

200

Blattella germanica

0

Hv x Hs hybrid

150

50

100

150

200

Schistocerca: solitarious

0

50

100

150

200

Spodoptera exempta

0

20

40

60

80

Heliothis subflexa

100 50

0

50

100

150

200

0

Protein eaten (mg)

1.5

30

0

0

50

100

150

Plutella xylostella

1.0

20 0.5

10 0

0

10

20

30

Protein eaten (mg)

40

0

0

0.5

1.0

1.5

Protein eaten (mg)

Moving Targets  |  101

catholic in its tastes, readily consuming a wide range of plants. Bearing this ecological difference in mind, a possible explanation presents itself from the ideas outlined above regarding the costs of staying versus leaving an imbalanced food, as follows. Having come across a nutritionally imbalanced food, a foraging animal must decide how much of that food to eat before moving on. By eating a lot rather than a little, the animal will gain the benefit of getting closer to the intake target level for the limiting nutrient, but will pay the price of simultaneously ingesting too much of the excess nutrient. If the animal happens to find a complementary imbalanced food in the near future, however, the potentially costly excess from the previous food could be turned into a greater total intake of a balanced mix of nutrients. Accordingly, the extent to which it is worth overconsuming an imbalanced food will depend on the probability of encountering a complementary food (the “nutritional antidote”) in the future (Raubenheimer and Simpson 1999). This probability would be expected to be higher for a food-plant generalist like Schistocerca than for a food-plant specialist such as Locusta, because a host-plant generalist is likely to experience a greater range of food compositions and greater nutritional variability (heterogeneity) than a host-plant specialist. (It is not necessarily true that nutritional heterogeneity is greater between the many food-plant species eaten by a food-plant generalist than it is within the restricted diet of a specialist, but it seems likely to be the case.) Nutritional heterogeneity provides a plausible, or at least a possible, explanation for why Schistocerca should adopt a rule of compromise different from that of Locusta, but these are different species from different branches within the family tree of grasshoppers and locusts, so there might be other reasons why they differ. Fortunately, Schistocerca offered a fascinating opportunity to test our hypothesis further. When reared under crowded conditions, Schistocerca nymphs develop into the brightly colored, actively aggregating, highly mobile “gregari-

Fig. 6.7. Intake arrays recorded during the early part of a larval stadium in a selection of four nutrient generalists, five nutrient specialists, and two transitional forms—intermediate-phase desert locusts and a hybrid between two species of Heliothis. (Data from Raubenheimer and Simpson 2003 [Locusta vs. gregarious Schistocerca]; Simpson et al. 2002 [Schistocerca phases]; Lee et al. 2002, 2003, 2004b [Spodoptera]; Lee et al. 2006a [Heliothis]; Raubenheimer and Jones 2006 [Blattella]; and Warbrick-Smith et al. 2009 [Plutella].)

102  |  Chapter six

ous” phase, which forms mass-migrating, marching bands of juveniles and then winged swarms of adults (see chapter 7). If reared alone, however, nymphs develop into the “solitarious” form, which has green or brown camouflage coloration, avoids other locusts, and is much more sedentary in its behavior (Simpson et al. 1999; Simpson and Sword 2008; Pener and Simpson 2009). Locusts have the potential to become either form, depending on their experience of crowding: they are, in effect, two animals packed within the same genome. Crowding is detected by sense organs on the hind legs and head, which set in train the process of phase transition by triggering chemical modulation of neural circuits controlling behavior (Simpson et al. 2001; Anstey et al. 2009; Pener and Simpson 2009). Under natural conditions in North Africa, solitarious locusts live in habitats that are dominated by a small number of host plants. When combined with their sedentary nature, this means that solitarious locusts should encounter lower nutritional heterogeneity than their much more mobile, dietary generalist, gregarious counterparts (Simpson et al. 2002). These differences between the solitarious and gregarious forms of Schistocerca sets up two alternative predictions. If evolutionary relatedness (i.e., phylogeny) is the main determinant of the rule of compromise, then solitarious-phase Schistocerca should show the same linear intake array as their gregarious-phase counterparts, but different from Locusta, since solitarious and gregarious Schistocerca share the same genotype. However, if nutritional ecology matters more than genetic relatedness, then solitarious Schistocerca should demonstrate the arc-shaped array seen in Locusta, because both are nutritional specialists. As can be seen in figure 6.7, the intake array of solitarious-phase Schistocerca was strongly arc-shaped, suggesting that rules of compromise are, indeed, sensitive to the nutritional ecology of these locusts. When locusts were in transition toward the solitarious condition, having been reared in isolation from eggs laid by crowd-reared parents, they developed an intermediateshaped intake array (fig. 6.7), providing evidence for epigenetic inheritance of the rule of compromise as for other phase-related traits (Pener and Simpson 2009; Miller et al. 2008). Further comparisons were made by Lee and colleagues (2002) in a series of experiments on species of host-plant specialist and generalist caterpillars. The extreme host-plant generalist Spodoptera littoralis showed a linear intake array, whereas that for the grass-feeding specialist Spodoptera exempta was arc-shaped (Lee et al. 2003; 2004b). The array for the generalist Heliothis virescens was linear, but that for the specialist feeder Heliothis subflexa, which only eats plants in the genus Physalis, was arc-shaped (Lee et al. 2006a). When these two Heliothis species were cross-bred, the hybrid caterpillars showed the same protein-biased intake target as their generalist H. virescens fathers, but an intake array more like that of their specialist H. subflexa mothers (fig. 6.7; Lee et al. 2006a).

Moving Targets  |  103

Another example of an extreme food-generalist is the cockroach Blattella germanica, which was found to have a distinctly linear intake array (fig. 6.7; Raubenheimer and Jones 2006). The fact that food specialists have arc-shaped intake arrays, whereas generalists have more linear arrays, indicates that the costs of ingesting excess nutrients are greater for specialists than for generalists (fig. 6.8; see also fig. 2.2; Simpson et al. 2004). We have argued above that these greater costs relate, in part at least, to the lesser likelihood of a specialist finding an antidote for the excess in the form of a complementary food. A compatible prediction is that generalists should be better able to deal with excesses postingestively. This was confirmed in the studies of locusts and caterpillars discussed above (Simpson et al. 2002; Raubenheimer and Simpson 2003; Lee et al. 2002, 2003, 2004b, 2006a). In each instance, nutrient utilization budgets indicated that generalists were better able than specialists to use excess ingested protein on high-protein, low-carbohydrate diets for energy metabolism, thereby simultaneously reducing the cost of surplus ingested protein and offsetting the carbohydrate deficit (see Lee et al. 2003). Warbrick-Smith and colleagues (2009) explored the correlates of nutritional specialism one step further in a study on caterpillars of the diamondback moth, Plutella xylostella, a tiny but economically important pest on cabbages and other brassica crops. A colony of Plutella was used that had been reared in the laboratory for at least 350 generations, during which the moths and their larvae had been exposed to a constant nutritional environment in the form of a synthetic food medium. Having started as food-plant specialists in the wild, and then having been maintained for hundreds of generations on a single diet, these insects should represent extreme dietary specialists that would be finely attuned to the composition of their ancestral rearing diet (fig. 6.8). Results suggested that this was indeed the case. Caterpillars regulated their intake to the nutritional composition of the long-standing laboratory diet (26% protein, 26% carbohydrate) when offered pairs of complementary foods, and performed by far the best on fixed diets formulated with this composition. A relatively small shift in diet composition, to 33% protein and 19% carbohydrate or the reverse ratio, resulted in a precipitous decline in survival rate. Those larvae that survived took longer to develop and grew to a smaller size. Caterpillars were also unable to compensate for dilution of the diet, as might be predicted from not having experienced much variation in dietary concentration for hundreds of generations. Although these caterpillars at first showed some evidence of an arcshaped intake array during the final larval stage (fig. 6.7), especially on high-protein diets, the arc soon straightened. As mentioned above and in chapter 2, a straightening and then opening out of the array is expected over time because of the effect of diet on the rate of development. Plutella

104  |  Chapter six A. Nutrient generalist

B. Nutrient specialist

a

a

b

Carbohydrate eaten

Carbohydrate eaten

b

Protein eaten

Protein eaten

C. Nutrient balance

D. Nutrient dilution

Performance

Performance

generalist generalist

specialist

specialist Low

P:C ratio in diet

High

Low

Concentration of diet

High

Fig. 6.8. Schematic summarizing the predicted differences between nutrient generalist and specialist feeders. For illustrative purposes, we have assumed that both the generalist (A) and the specialist (B) have the same intake target (bull’s-eye symbol). The gray isoclines define the fitness landscape for each feeding type, which reaches a maximum at the intake target. The generalist has a tilted elliptical fitness surface, whereas the specialist’s fitness surface is more restricted and circular (see chapter 2 and Simpson et al. 2004 for theoretical basis). Compared with the specialist, the generalist is less susceptible to variation in P:C ratio in the diet. This can also be seen in (C), where performance is shown along a cross-section taken at dashed line (a) in panels (A) and (B). The generalist is better able to tolerate dilution of dietary nutrients, as shown in (D), where performance is shown along a cross-section taken at dashed line (b) in panels (A) and (B). The gray points on each diet rail in panels (A) and (B) indicate the point of maximum fitness available, and thus the locally optimal intake point, when restricted to a given food rail. Col-

Moving Targets  |  105

is very small (weighing only 5 mg when adult) and develops rapidly, so straightening might be expected to occur after only a day or so. WarbrickSmith and coworkers (2009) suggested another explanation for the absence of a strongly arc-shaped array, which was that selection against overeating imbalanced diets had been relaxed during the hundreds of generations of rearing on a fixed diet composition. This would help explain why the insects were so susceptible to a shift away from the ancestral diet composition: not having experienced imbalanced diets, they were ill equipped to recognize or deal with them either pre- or postingestively.

6.6 Evolving Postingestive Responses Warbrick-Smith and colleagues (2006) set about to discover whether Plutella could adapt over generations to a shift in the mean composition of the diet from the ancestral laboratory diet. We reared multiple lines of Plutella for a total of eight generations on either a carbohydrate-rich diet or a protein-rich diet. The carbohydrate-rich diet comprised either chemically defined artificial food (12% protein and 40% carbohydrate) or a high-starch mutant of the plant Arabidopsis (a small mustard cress, beloved of plant geneticists). Over the eight generations, caterpillars progressively developed the ability to eat excess carbohydrate without laying it down as body fat (fig. 6.9), which is strong evidence that storing excess fat has fitness costs. Other replicate lines of caterpillars were reared for eight generations in a protein-rich, carbohydrate-scarce environment, comprising either artificial diet (42% protein, 7% carbohydrate) or a low-starch Arabidopsis mutant. In contrast to insects reared under a high-carbohydrate environment, these caterpillars developed an increased propensity to store ingested carbohydrate as fat. Female moths also developed a preference for laying their eggs on the low-starch plant, whereas those selected on the high-starch Arabidopsis mutant showed no preference. These results provide an example of how postingestive responses can change over generations in response to a shift in the nutritional environment. By becoming less prone to laying down body fat in a high-carbo-

lectively, the array of such points defines the extent to which animals tolerate ingesting excesses and deficits of nutrients on suboptimal diets (those that do not intersect the intake target). The intake array for the generalist is straighter (less arc-shaped) than that for the specialist. (From Warbrick-Smith et al. 2009.)

106  |  Chapter six Ancestral regime for 350 generations (p26:C26)

Pupal lipid content (mg)

0.9

0.9

High protein regime (P45:C7) G8

0.8 0.7

G8

0.7

G4

G1

0.6

0.6 G1

0.5

0.5

0.4

0.4

0.3

0.3

0.2

High carbohydrate regime (P12:C40)

0.8

0

1

2 3 4 5 6 Carbohydrate eaten (mg)

7

0.2

G4

0

1

2 3 4 5 6 Carbohydrate eaten (mg)

Fig. 6.9. Changes in the efficiency with which ingested carbohydrate was converted to body fat in diamondback moth caterpillars reared for eight generations under either a high-protein or a high-carbohydrate regime. At generations 1, 4, and 8, caterpillars were removed and reared in a common environment with nutritionally complementary foods for the first three larval stages, then exposed for the final larval stadium to one of five synthetic diets containing different proportions of protein and carbohydrate (% protein to % carbohydrate = 47:5, 40:12, 29:23, 19:33, or 12:40). Over the eight generations, caterpillars from the highprotein environment progressively laid down more body fat per intake of carbohydrate (upward gray arrow), while the opposite occurred under high-carbohydrate environments (downward gray arrow). (Data from Warbrick-Smith et al. 2006.)

hydrate world, caterpillars could minimize the costs of obesity—but at the risk of starving for want of energy reserves if food became scarce. By contrast, in a low-carbohydrate world, retaining and storing carbohydrates as fat was favored; but should such a phenotype be placed into a high-carbohydrate world, it would be at risk of obesity. It was not known whether these changes in physiology were the result of genetic selection or accumulating epigenetic effects (see above). Either way, the implications are clear for humans in the modern nutritional environment—a topic to which we return in chapter 10.

6.7 Conclusions Intake targets, rules of compromise, and postingestive growth and metabolic responses are dynamic across several timescales. Intake targets move in physiological time in response to the demands of activity, environmental temperature, and the need to deal with ingested toxins or at-

7

Moving Targets  |  107

tack by pathogens. Mechanisms to track moving targets need to operate across time periods that are not so short that the animal is driven constantly to switch between foods, or so long that it fails to track the target sufficiently closely. At a longer timescale, targets track the changing needs of different life stages. A combination of demand-driven and programmed (non-demand-driven) mechanisms allows animals to match intake to these changing requirements. Nutritional traits are also dynamic across generations, tracking persisting environmental changes and allowing organisms to adapt to new nutritional niches. Genetic selection and epigenetic influences contribute to such cross-generational shifts. Comparative analysis offers a means to reconstruct the evolutionary history of nutritional traits and the ecological and life history forces that have shaped nutritional strategies. In the next chapter we move beyond the individual and consider how the nutritional responses of individuals translate to groups and societies. As we shall see, the consequences of an individual’s nutritional state for the group may be profound.

seven

From Individuals to Populations and Societies

So far, we have viewed the world of nutrition from the perspective of individual animals. But individuals interact with one another—in family groups, aggregations, swarms, and societies—and many of these interactions involve nutrition. Can an individual’s nutritional state influence how a group behaves? We are beginning to appreciate that the answer to this question is yes. And as we shall see, sometimes the consequences are gruesome. Animal groups often appear to act as if they possess a single mind: observe, for example, how exquisitely flocks of birds and schools of fish avoid the attacks of predators, how vast numbers of wildebeest migrate each year across the plains of East Africa, or how the individual members of an ant nest work together to provision the colony with food. Whereas it might appear that such groups have collective intelligence, we now know that complex, coordinated behavioral patterns can arise from simple local interactions between individuals (Sumpter 2006, 2010). By interacting locally with no knowledge of the behavior of others except their close neighbors, individuals within a group can switch their collective behavior in an instant from disorder to order, or between one coherent pattern and another—a “phase transition” as acute as when water turns to ice. For example, when a certain density is reached, disordered aggregations of locusts will suddenly start to march together as a coordinated whole. The marching army can number billions of locusts, yet is seemingly coordinated through each insect following the simple rule of aligning with its moving neighbors within a radius of 14 cm and being insensitive to other locusts beyond that distance (Buhl et al. 2006, 2011). Swimming fish also align with their neighbors, and small adjustments in the alignment radius cause fish schools to shift suddenly from comprising a stationary aggregation, to a spirally rotating mass, to a group moving in a single direction with all fish aligned in parallel (Couzin et al. 2002). Subtle changes in the ways in which individuals interact can therefore have profound effects at the collective level. We also know from earlier

From Individuals to Societies  |  109

chapters that nutritional state influences the behavior of individual animals. We will now put these two pieces together.

7.1 Cannibal Mormon Crickets The Mormon pioneers planted crops in the spring of 1848, after suffering great hunger in their first winter in the Salt Lake Valley. As the crops ripened, hordes of devouring crickets descended upon them from the foothills east of the valley. The Saints fought them with fire, clubs and water. As they despaired of saving the next winter’s food, their prayers of deliverance from almost sure starvation were answered when thousands of seagulls came to feed on the crickets. The Sea Gull Monument commemorates this modern-day miracle. The sea gull is now the Utah State bird. Inscription on the Sea Gull Monument, Temple Square, Salt Lake City, Utah, USA

Welcome to the Mormon cricket, Anabrus simplex. Just as the Mormon pioneers experienced, during late spring and early summer in western North America one regularly sees millions of these flightless insects ganging together in huge marching bands that extend for up to 10 km in length and travel 2 km per day. Both theory (Reynolds et al. 2009) and experimental evidence (Sword et al. 2005) can explain why it is better for an individual cricket to join a superaggregation than to go it alone (the short answer is to avoid predation), but why do they form marching bands? An obvious answer to this question would be to find more food, having depleted the supply locally. However, even though Mormon crickets are omnivores with highly catholic food tastes, they do not strip the habitat bare as they go. In fact, it is often difficult to tell that a band has just marched through an area. If some nutritional resource is limiting, then it must be more specific than edible food. Clues for what the limiting resource might be come from watching Mormon crickets in the wild. They avidly consume flowers and seed heads, the leaves of legumes, carrion, animal feces, urine-soaked soil—and one another. Their cannibalism is renowned, and can cause unexpected problems for state highway departments. As crickets cross a road and get run over, other crickets stop to eat them and themselves get run over, and so on until there is a hazardous slippery slurry across the road and snow plows have to be deployed. Two obvious nutrient groups emerged from these observations as candidates for the limiting resource: proteins and mineral salts. To test

110  |  Chapter seven

whether protein is preferentially ingested, dishes of powdered, chemically defined diets differing in protein and carbohydrate content (as used in other experiments on locusts but never experienced by field populations of Mormon crickets) were placed in front of marching bands of crickets (Simpson et al. 2006). The insects streamed over the dishes as the band marched by, stopping to feed far more frequently when the foods contained protein than when they contained carbohydrate (fig. 7.1A). When the experiments were repeated using pads of cotton wool soaked in different concentrations of salt solution, crickets stopped to drink from solutions equivalent in concentration to their own body fluids and avoided either pure water or more concentrated salt solutions (fig. 7.1B). These results indicated that Mormon crickets in the marching bands were selectively feeding for protein and salt. When crickets were taken from the band and housed with access to plentiful amounts of proteinrich and carbohydrate-rich foods in a simple Geometric Framework experiment, they initially ate predominantly protein-rich food, but over the next day they mixed a diet that was much more typical of other crickets and grasshoppers, containing protein and carbohydrate in near equal proportions (fig. 7.1C). In other words, the crickets in the marching bands were not obligatorily protein-seeking, but rather they were selectively deprived of protein as a result of local environmental conditions. Whereas the environment may have let them down, there is one highly abundant source of lightly salted protein in the midst of a cricket horde: other Mormon crickets. So, is the reason that cricket bands migrate because they are on a forced march for protein and salt, with each insect chasing the moving meal in front and escaping the cannibals behind? This macabre hypothesis makes two predictions: (1) if crickets become less mobile, they should stand a higher chance of getting cannibalized; and (2) satiating crickets with protein and salt should reduce both cannibalism and marching. In support of the first prediction, reducing a cricket’s capacity to avoid or fend off attacks with its strong, kicking hind legs greatly increased its vulnerability to cannibalism. (We will spare the details of how this was achieved, but think “hot glue gun and fishing line.”) The second prediction was also borne out (fig. 7.2A–C). Providing crickets with proteinrich food or salt solution for 4 hours caused them to ignore or only nibble at a victim thrown into the arena. In contrast, crickets that had been given access to carbohydrate-rich food consumed the majority of the victim, and in many cases ate an entire cricket of the same size as themselves in a single meal, which is probably the record meal size for any animal that chews its food. Not only did the nutritional quality of their diet affect Mormon crickets’ cannibalistic tendencies; if crickets were confined alone with protein-rich food, they spent only half as much time walking

From Individuals to Societies  |  111 A. Mean percent insects at dish

40

42% protein 21% protein

30

0% protein

20

10

0

B.

PC

O

Carbohydrate eaten (g)

0.10

30

20

10

0

C Diet

C. 40

Mean percent insects at dish

P

Water

0

0.5 1.0 1.5 NaCl concentration (M)

2.0

0.08 0.06 Day 2

Day 1

0.04 0.02 0

0

0.02 0.04 0.06 0.08 0.10 0.12 Protein eaten (g)

Fig. 7.1. Mormon crickets on a forced march for protein and salt, driven by cannibalism. (A) When Petri dishes containing chemically defined synthetic foods were placed in front of a marching band, crickets selectively stopped and fed from diets containing protein (P, PC). Food P contained 42% protein and no digestible carbohydrate; PC contained 21% protein and 21% carbohydrate; C contained 42% carbohydrate and no protein; O contained neither macronutrient, these being replaced by additional cellulose to maintain the same concentration of micronutrients in all four foods. (B) Crickets were also highly responsive to dilute salt solutions, with 0.125 M NaCl being most stimulating—the concentration of salt in cricket blood. (C) The preference for protein shown in (A) was not due to crickets being obligate high-protein feeders; when confined with P and C foods for two days, they initially selected a high-protein diet but then reverted to a slightly carbohydratebiased diet typical of other crickets and grasshoppers. (From Simpson et al. 2006, copyright 2006 National Academy of Sciences, U.S.A.)

112  |  Chapter seven A.

Number of cases where cannibalism exceeded value on x-axis

B.

C. 20 15

Protein replete: PC P

10

Protein deprived: C O

50 40

Salt replete 20

5 0

Salt deprived

30

10 0

20

40 60 Percent eaten

80

100

0

0

20

40 60 Percent eaten

80

100

Fig. 7.2. Allowing Mormon crickets access to protein (foods P or PC) for 4 hours reduced their cannibalistic tendency (panels A and B) when offered a victim, relative to crickets with access to foods lacking protein for the same period (foods C, O). (C) Allowing crickets access to a salt solution for 4 hours also reduced their cannibalistic tendencies compared with crickets given plain water. (From Simpson et al. 2006, copyright 2006 National Academy of Sciences, U.S.A.)

around the arena as did others kept with only carbohydrate-rich food (Simpson et al. 2006). The conclusions from these experiments and others by Sword and coworkers (2005) are clear: Mormon crickets form aggregations at high population levels to reduce the risk of predation, but having done so they locally deplete the supply of key nutrients such as protein. Mass migration at the group level emerges as the indirect effect of individual nutritional responses, mediated via cannibalistic interactions between indi-

From Individuals to Societies  |  113

viduals: each protein- and salt-hungry cricket is “pulled” by the moving meal ahead, and at the same time “pushed” by the cannibal behind. As a result, there is a twofold incentive to move, and a Mormon cricket band travels 2 km each day rather than only tens of meters when the crickets are in low-density populations (Lorch et al. 2005). The consequence of migration is a greatly increased chance of encountering fresh nutritional resources; but meanwhile, to use Greg Sword’s phrase, the crickets are “travelling with their lunch” (Hansen et al. 2011).

7.2 Locusts Are Cannibals Too The experiments on Mormon crickets made sense of a frustrating experience in 1982. As part of a study to localize stretch receptors on the gastrointestinal tract responsible for regulating how much food is eaten in a meal (Simpson 1983), the ventral nerve cord was painstakingly sectioned in a large number of locusts to remove sensory feedback from the hindgut. The operated insects were left together to recuperate overnight. Next morning, the majority of them had no abdomen beyond the site of the lesion. There was even a daisy chain of insects, each munching the numbed abdomen of the animal in front. Flightless juvenile locusts also, like Mormon crickets, famously form massive marching bands. Buhl and colleagues (2006) used a combination of laboratory and field experiments and computer simulations, based on mathematical models taken from statistical physics called self-propelled particles models, to show that collective movement in locusts arises because individuals align with their moving neighbors. Interactions between individuals within a distance of only a hand span result in the cohesive movement of millions or even billions of insects in bands extending kilometers (Buhl et al. 2011). Mormon crickets align with their neighbors to avoid cannibalism—is the same true of locusts? The sorry outcome from the nerve-cutting experiments in 1983 suggested so, and offered a way of testing the hypothesis. Bazazi and coworkers (2008) introduced desert locust nymphs with their ventral nerve cords cut into the “Mexican hat” marching arena—a device that allows locusts to march endlessly round and round in circles (Buhl et al. 2006). When individuals were alone, levels of activity were similar between operated-on and control locusts, but in crowds marching was absent when locusts had no sensation from their rear end. Detailed video analysis showed that these locusts behaved normally in all but one respect: they did not react when touched from behind. The consequences were that they failed to form marching bands—and they had the tips of their abdomens cannibalized. Painting the rear part of the insects’ compound eyes—effectively applying tight-fitting blinkers—also reduced

114  |  Chapter seven

marching, because locusts were oblivious to the sight of neighbors approaching from behind. Bazazi and coworkers (2011) next manipulated the nutritional state of individual locusts using the same types of synthetic diet as used on Mormon crickets in the field and showed that protein but not carbohydrate deficiency promoted high levels of marching in groups. Locusts too are traveling with their lunch.

7.3 Communal Nutrition in Ants In locusts and Mormon crickets, the ultimate in noncooperative individual nutritional interactions, cannibalism, produces an extraordinary and potentially devastating collective response. Colonies of social insects, on the other hand, are far more collegial: interactions among individuals somehow coordinate the activities of the entire colony so that it acts as a nutrient-acquiring, distributing, and digesting “superorganism” (Hölldobler and Wilson 2009), which is able to meet the disparate needs of all its members. Indeed, it may well be that nutrition has played an important role in the evolution of advanced sociality among animals, including the ants, bees, and wasps (Hunt and Nalepa 1994; Kamakura 2011). As we saw in chapter 3, nutritional regulation in an individual animal involves three components: (1) assessing the nutritional quality of available foods, using sensory organs and past experience; (2) assessing current nutritional state in relation to the optimal state, using circulating metabolites and hormones; and (3) integrating these two in the brain and peripheral organs to produce regulatory behavioral and physiological responses. The complication for social animals such as ants is that a minority of individuals collect the food for the colony, and these foragers have nutritional requirements very different from those of other colony members (Cassill and Tschinkel 1999; Behmer 2009). In particular, the egglaying queen and the legless, grublike larval ants have a much higher need for protein than do workers, but they do not contribute to food collection. A forager ant who responded only to her own needs would collect carbohydrate-rich nectars and exudates in small amounts, since she is already fully grown. If all foragers behaved in this manner, the brood, queen, nurse ants, and other nonforaging workers would soon starve to death. How then does the entire colony communicate its complex needs to the forager ants? To answer this question, Dussutour and Simpson (2008) set about posing nutritional conundrums for an Australian species, the greenheaded ant (Rhytidoponera metallica). In a first series of experiments, dozens of colonies were established in the laboratory and provided for 5-day periods with one of three concentrations of a sugar solution (di-

From Individuals to Societies  |  115

lute, medium, and concentrated). How many ants arrived at the food, how much each ant collected to take back to the nest, and how much in total was collected over time were then measured. For the first day of each experimental period, ants consumed the most sugar, and recruited the largest numbers of other foragers, in colonies given only the concentrated solution, as would be expected were ants maximizing their rate of sugar acquisition. However, over successive days, the numbers of ants recruited and the volumes collected by each ant reversed, such that the greatest amount was consumed by colonies with the dilute solution and least by those with the concentrated solution. This result proved that the ants were compensating for the dilution to maintain the supply of sugar to the colony at a target level: they were regulating, not maximizing, sugar intake. How did the foragers “know” when they needed to collect more of the dilute solutions or less of the concentrated solution to maintain sugar supply to the colony? An important clue came when we noticed that ants became better at regulating their sugar intake when larvae appeared in the nest as the weeks progressed. Were these larvae, perhaps, the source of nutritional feedback? We artificially doubled the numbers of larvae in some colonies and doubled the numbers of workers in others. Ants regulated sugar collection more precisely when the number of larvae was increased than when the number of adult workers was doubled, indicating that larvae were indeed a source of nutritional feedback from the colony to the workers. The feedback from larvae to foragers most probably comes via a cascade of interactions. This is because in most ant species foragers pass their collected food to reserve workers, who in turn give it to nurse ants, who then feed it to the larvae (Sorensen et al. 1981, 1985; Cassill and Tschinkel 1999). A simple scenario would be as follows: A worker collects a droplet of low-concentration sugar solution and returns to the nest. A reserve worker takes the droplet without delay, keeps some for her own needs, and passes the rest to a nurse, who keeps her share and immediately feeds what’s left to a hungry, vigorously begging larva. Having had its food load accepted promptly and enthusiastically, the forager hurries back along the foraging trail to collect more. If, in contrast, a forager returned with a highly concentrated sugar droplet and had difficulty offloading her burden because reserve workers, nurses, and larvae were sugar-replete, then the forager would be less likely to return to that food source. As a result, across the colony fewer foragers would be recruited to a more concentrated source, and less food would be collected. The next question to be addressed was whether such feedbacks emanating from larvae and nest-bound ants could allow ant colonies to regulate the amounts and balance of multiple nutrients (Dussutour and Simp-

116  |  Chapter seven Concentration 300g L–1 200g L–1 100g L–1

Carbohydrate collected (mg per ant)

Colony Without larvae With larvae 1.0

1.0

1:3

0.8

0.8

0.6

0.6 2:1

0.4 0.2 0

1:2

0.4 3:1

0.2

0

0.2 0.4 0.6 0.8 Protein collected (mg per ant)

1.0

0

0

0.2 0.4 0.6 0.8 Protein collected (mg per ant)

1.0

Fig. 7.3. Green-headed ant colonies regulate the amount of protein (P) and carbohydrate (C) collected to an intake target. Colonies with or without larvae adjusted their collection from two nutritionally complementary agar-based foods: either 1:3 P:C versus 2:1 P:C (left panel) or 1:2 P:C versus 3:1 (right panel). These foods were offered at one of three total nutrient concentrations (100, 200, or 300 g L–1). Crosses and open circles indicate colony replicates, and filled squares and circles are means across colonies. As well as maintaining a target P:C ratio, ants maintained total nutrient amounts collected by adjusting the volume of food collected according to the concentration of nutrients present. Note how colonies without larvae selected a lower P:C diet than did those with larvae, and that whereas colonies with larvae maintained the amount of nutrients collected in the face of a threefold dietary dilution (involving bringing back three times more food to the colony), colonies without larvae did not maintain nutrient intake at the highest dilution (the cluster of points for 100 g L–1 falls well below intakes for the two higher concentrations). (From Dussutour and Simpson 2009, photo © Dr G. A. Miller.)

son 2009). We challenged colonies of green-headed ants with or without larvae to maintain intake of both protein and sugar by giving them choices between two foods varying in the ratio and concentration of protein and carbohydrate. Colonies with larvae adjusted the amounts of each food collected to maintain the ratio and amounts of protein and

From Individuals to Societies  |  117

sugar collected remarkably constant (fig. 7.3). Colonies without larvae differed in three respects: (1) they regulated to a lower protein to carbohydrate ratio; (2) they collected lesser total quantities of nutrients; and (3) they were unable to respond as effectively to dilution of nutrients in the food (fig. 7.3). Hence, the presence of fast-growing, energy- and protein-demanding larvae changed both the quantity and blend of nutrients required and determined the effectiveness of nutritional regulation. These results showed one further remarkable collective feature: an ant colony is not only a nutrient-balancing superorganism at the level of food collection—it is also a collective nutrient-processing, storage, and waste disposal device—a collective mouth, gut, and anus, as it were. As we have seen in chapter 3, when the nutritional composition of the food an individual has ingested is imbalanced with respect to requirements, nutritional balance can be restored postingestively by retaining limiting nutrients and getting rid of those in excess. Our experimental ant colonies behaved collectively in a similar way. When confined to one of five foods containing a fixed ratio of protein to carbohydrate, ants did not consume all the food collected—rather they stored food as pellets in the nest and then threw them onto a waste dump outside. Hardly any pellets were discarded in colonies fed the most carbohydrate-rich, protein-poor diet, but the size of the waste dump got larger as the ratio of protein to carbohydrate in the diet increased. The ants were extracting the limiting sugar from the high-protein diets and rejecting the excess protein residue as pellets. Intriguingly, colonies with larvae were more effective at getting rid of excess protein than colonies without larvae. Therefore, not only did larvae provide nutritional feedbacks that influenced nutrient collection; they also affected nutrient extraction (Dussutour and Simpson 2009). One final twist was that the presence of larvae helped offset the costs of processing high-protein diets, whereby workers died sooner on highprotein, low-carbohydrate diets than on low-protein, high-carbohydrate diets—a result that, intriguingly, mirrors the effects of high-protein diets on life span in other animals as discussed in chapter 4. Cook and coworkers (Cook and Behmer 2010; Cook et al. 2010) have since reported a similar finding in fire ants in their GF study of collective nutrition in that species and also in a field study on ants in Costa Rica.

7.4 The Blob Individual animals and social-insect colonies possess some striking parallels as nutritional regulatory systems. Both have specialized components to deal with nutrient supply and demand: brains and peripheral organs in

118  |  Chapter seven

the case of animals, foragers and brood in ant colonies. As we have seen, there are even collective equivalents of mouths, guts, and anuses. But some of the most ecologically significant organisms on earth do not fit this mold: molds and slime molds, in fact. Fungi and slime molds are essential to the proper functioning of soils, for the breakdown of organic matter, and ultimately for carbon sequestration and the cycling of nutrients within the world’s ecosystems. These are important roles for such unprepossessing organisms. Whereas most people are aware of fungi in their various manifestations, the acellular slime molds are especially strange and little-known organisms. They are single, multinucleate-celled amoebae that form moving blobs called plasmodia, which can cover an area of several hundred square centimeters. Happily, this real-life counterpart of the alien monster featured in The Blob only terrorizes communities of bacteria and other microorganisms. In forests, for example, it streams over and among the leaf litter, sending out armlike projections (“pseudopods”) that engulf microorganisms, fungi, and dead organic matter and extract their nutrients. Plasmodia can travel several centimeters per hour as they explore their environment. Recent experiments have shown that a slime mold plasmodium can find the shortest route through a maze, anticipate the timing of regular events, and design an efficient and fault-tolerant network of connections between multiple food sources that closely mimics the design of the Tokyo subway system (Tero et al. 2010). The really interesting point is that plasmodia can do all this without possessing a brain or central coordinating system—they are the ultimate distributed problem-solvers. Dussutour and colleagues (2010) set out to a see whether the acellular slime mold Physarum polycephalum is equally clever at solving complex nutritional challenges. The first step was to establish whether slime molds performed differently on diets containing different ratios and concentrations of protein and carbohydrate. A small piece of slime mold was placed in the center of a Petri dish filled with one of 35 diets set in agar gel; then the size, shape, mass, and position of the blob was measured 60 hours later (plate 5A). The more dilute the nutritional medium, the greater the area covered by the slime mold. This was the slime mold equivalent of more ants being recruited to diluted food sources in Dussutour and Simpson’s ant experiments—the blob’s way of compensating for the lower concentration of nutrients in the food medium. The second major finding was that slime molds increased their mass most rapidly on a diet comprising a 2:1 ratio of protein to carbohydrate. This was evidently their optimal diet composition. Next, fragments of slime mold were offered a choice between two pieces of food placed upon a nutrient-free agar plate. There were six different pairwise food combinations. In every case, the slime molds grew to

From Individuals to Societies  |  119

distribute themselves between the two pieces of food in the same ratio and area needed to obtain the optimal diet mixture of close to 2:1 protein to carbohydrate (plate 5B). In the pièce de résistance, slime mold fragments were offered a clock face of 11 different food ratios (from 9:1 to 1:9 protein to carbohydrate). The blob grew outward from the center, migrated around the Petri dish, and settled on precisely the optimal dietary ratio (plate 5C). We don’t yet know how slime molds are able to mix an optimal diet in a complex nutritional environment, but there is no doubt that they can do so even without a central coordinating center. It seems likely that local nutrient sensing, growth, and movement responses are involved, with the entire system self-organizing to produce coherent, adaptive behavior.

7.5 Conclusions Mormon crickets, ants, and slime molds illustrate how local nutritional interactions can fashion cohesive, group-level behavioral patterns. In Mormon crickets and locusts, order emerges from a base nutritional response: attempt to eat your neighbor. The risk of leaving the band outweighs the costs of remaining: “Living and moving together in a migratory band is a compromise that makes the best of a seemingly very bad situation” (Simpson et al. 2006). While it is true that one consequence of being driven on a cannibalistic forced march is an increased probability of finding better conditions elsewhere, it seems vanishingly unlikely that this has been selected as an emergent property at the level of the migratory band. In contrast, the emergent outcomes arising from nutritional interactions within ant nests and slime mold plasmodia clearly do enhance the evolutionary fitness of the collective entities and their component parts. It is therefore valid to use the term “superorganism” for such systems (Hölldobler and Wilson 2009). In the next chapter we will see how nutritional interactions may shape not just the behavior of groups and populations but, furthermore, cascade throughout networks of interactions among assemblages of species, fashioning the way in which ecosystems function and respond to changing environmental conditions.

eight

How Does Nutrition Structure Ecosystems?

Ecology is the study of the factors that determine the distribution and abundances of organisms. We have had a taste in the previous chapter of how the requirements for a nutritionally balanced diet can influence the first of these issues, by driving large-scale shifts in the distribution of locusts, Mormon crickets, and (on their own modest scale) slime molds. In this chapter we will expand our focus to consider also population sizes and interactions between species. The question of how the properties of ecosystems are influenced by the nutritional biology of the component organisms—their foraging, food choices, and functional responses to nutrition—has long been a focus of community ecologists. And yet, as summed up in a recent article (Beckerman et al. 2010), progress has been modest: It is now well established that animals adaptively choose their diets, shift their habitat use and alter their morphology in response to temporal and spatial variation in resource availability, as well as the presence of predators and parasites. Optimal foraging remains the central theory linking resource and consumer traits to patterns of resource selection. Yet, even after 40 years of development, there are precious few advances towards truly synthesizing the connections between individuals, populations and large interconnected food webs. . . . We suspect the reason that progress has been slow in this area is that the central link referred to in this quote, optimal foraging theory, does not characterize the nutritional biology of animals very well (Raubenheimer et al. 2009). Much research in the field of optimal foraging theory has been based on the assumption that the foraging behavior and food choices of animals are geared toward maximizing the gain of energy in as little time as possible. Although there have been important efforts to incorporate more than one nutrient into models of optimal foraging theory (e.g., Hengeveld et al. 2009; Houston et al. 2011), food components other than energy are typically considered to be important only insofar as they form impediments that constrain the gain of energy. However, as should be

How Does Nutrition Structure Ecosystems?  |  121

clear from previous chapters, there is now a respectable body of incontrovertible evidence that animals respond not to the energy content of foods per se, but to the amounts and balance of specific nutrients and nonnutritional components. We believe that substituting this multidimensional view for the energy-centered link will help to stimulate progress in the understanding of how individual traits influence ecological communities. In chapters 2 and 4 we showed how the success of individual organisms can be related to their intake of multiple nutrients by mapping measures of performance (life span, reproductive output, immunity to disease, and so on) as landscapes onto nutrient intake arrays. In chapter 7 we showed how nutritional responses of individuals can in turn fashion the behavior of groups and societies. We now take this one step further to consider the consequences of individual nutrition for populations and the assemblages of species that comprise ecological communities. We will argue that trophic dynamics are in large part an emergent property of regulation of nutrient balance, and that such regulation takes place at all trophic levels.

8.1 From Individual Fitness to Population Growth Rates If the nutritional responses of individual organisms represent those of the species, it ought to be possible to use performance landscapes measured from a representative sample of individuals to model the success of an entire population of that species. Ecologists measure the success of a population in terms of its population growth rate (pgr). By definition, a population will increase in an environment when pgr is greater than zero. Population responses have been described in geometric terms similar to those we have used for individual organisms, most famously by David Tilman (1982), who classified population growth rate responses as surfaces in a two-resource graphical space. Interestingly, Tilman, who worked principally on plant communities considered the cases where excessive consumption results in reduced population growth rate to be uncommon (Tilman 1982, p. 20). While it is undoubtedly the case that population growth rates can be defined within the same multidimensional nutritional spaces used to describe the responses of individual organisms, the key question is whether performance landscapes derived from individuals, such as those in plate 3, can be directly translated into pgr landscapes: can we simply multiply the valves on isoclines for individual reproductive fitness by the number of reproductive individuals in the population to get pgr isoclines? The answer is no—or at least not always. The reason is that populations are not the simple sum of their individual parts: members within a population interact with one another and with the rest of their environment.

122  |  Chapter eight

This was illustrated in chapter 7: it would not be possible to predict the growth rate of an ant colony from the individual nutritional responses of workers or larvae, nor to predict mass migration from the responses of single locusts or Mormon crickets taken from a migratory band, even though individuals within a band are all essentially the same. Rather, local interactions between individuals cause unexpected patterns to emerge at the level of groups and populations. Similarly, transitions in organizational complexity are expected to arise as a result of interactions between organisms of different species (Levin 1992; Peters et al. 2007). When these interactions involve nutrition they are termed “trophic interactions.” Trophic interactions with other species can have profound effects on a population’s growth rate, and may result in very different responses to diet composition than would be predicted from the study of individual organisms. For example, we know that improving the nutritional quality of host plants by adding fertilizers can improve the performance of individual insect herbivores, but may have exactly the opposite effect at the population level because of effects on the nutrition of predators and other natural enemies (Kytö et al. 1996). Similarly, a change for the worse in diet composition at the individual level could result in an increase in population growth rate. Let’s say that a change in plant quality results in herbivores having lowered body nitrogen content. This change in herbivore body composition will change the diet composition of the predators that eat the herbivore. These predators will have their own intake targets and nutritional priorities—which, as we shall see below, are not the same as those of their prey. If the shift in herbivore body composition limits the success of their predators, then the resulting reduction in predation pressure may outweigh the cost to individual herbivores of a lower-quality diet and lead to an increase in herbivore population growth rate (Raubenheimer and Simpson 2004). Two important conclusions can be drawn from the above discussion. First, the ecological consequences of nutrition involve not only the direct effects on the individual organism but also the direct and indirect inter­ actions occurring among individuals, both within and between species. Second, we cannot understand the nature and consequences of such interactions unless we know the nutritional needs, priorities, and regulatory capacities of the various interacting organisms. We next elaborate on each of these in turn.

8.2 Interactions among Organisms and the Environment The most direct nutritional interaction of all is when one organism attempts to eat another. The outcome that is most commonly considered in classical community ecology is that the predator succeeds in killing and

How Does Nutrition Structure Ecosystems?  |  123

eating the victim, and returning its unused components to the environment in the form of feces. But the interaction might be ecologically significant even if the prey animal is only wounded or escapes unharmed. It might, for example, change its subsequent behavior to avoid the risk of further attack, or the interaction might result in the induction of defensive chemicals and immune responses to prevent or counteract future threats (Karban and Baldwin 1997; chapter 5). Being removed from the environment as a living organism and returned as feces—or being changed in behavior, chemical composition, and defensive armory—may resonate throughout the ecosystem in unexpected ways (Simpson et al. 2010). This is illustrated by an interesting experiment that we introduced in chapter 1: the study by Hawlena and Schmitz (2010) of the North American grasshopper Melanoplus femurrubrum. This herbivore selects a diet of grasses and herbs and is preyed upon by hunting spiders, such as Pisuarina mira. Hawlena and Schmitz housed grasshoppers in the presence of spiders, which were kept in clear plastic cylinders where they were visible but unable to attack the grasshoppers. Compared with controls that were not exposed to spiders, the grasshoppers selected a diet containing 40% more carbohydrate, whereas protein intake was not affected by the presence of spiders. This shift in the intake target toward a higher carbohydrate to protein ratio was due to grasshoppers having an increased metabolic rate when exposed to spiders: the “fear effect.” There was also a change in the chemical composition of the grasshopper’ feces, which were lower in carbon to nitrogen ratio (C:N) when spiders were around, reflecting the higher metabolic need for carbohydrate. Finally, the body composition of spider-exposed grasshoppers differed, being higher in C:N than controls. Hawlena and Schmitz calculated that the change in the grasshoppers’ intake target when at risk of spider predation would shift the structure of the natural host plant community as a result of selective foraging by fearful grasshoppers. The changes in body and fecal chemistry and plant community structure would, in turn, affect the chemical composition of the pool of detritus in the ecosystem. We already have seen in the previous chapter evidence of how one important decomposer, the slime mold, responds to nutrient balance, and this would have knock-on effects for soil ecology and plant growth. Hawlena and Schmitz (2010) concluded that their study “leads to the testable hypothesis that trophic dynamics are neither bottom-up- nor top-down-regulated, but instead may be an emergent property of herbivore regulation of nutrient balance that emanates from the middle of the food chain.” We agree with this interesting point, but would argue that the message is broader than this: trophic dynamics might be an emergent property of regulation of nutrient balance at all trophic levels.

124  |  Chapter eight

The example of the response of grasshoppers to the risk of spider predation affecting the species composition of grassland plant communities and the chemistry of the detritus pool illustrates another key feature of nutritional interactions: organisms both respond to their environment and modify it; this is what is known as the “biotic interaction milieu” (McGill et al. 2006). They may even “engineer” their environment to better meet their nutritional needs (Odling-Smee et al. 2003). Human agriculture (chapter 10), beaver dams, complex plant galls induced by insect herbivores, and the tending of fungus gardens by leafcutter ants are all examples. The production of foraging trails is another, such as when large herbivores trample and graze the vegetation and produce physical trails in the environment, or when ant foragers lay down chemical trails to food. Such trails act almost like a form of collective memory spanning the landscape. Well-worn trails leading to profitable feeding sites continue to be followed, while trails leading to depleted feeding sites fall into disuse (Helbing et al. 1997; Couzin and Krause 2003). So far, our examples of trophic interactions have operated over ecological timescales, but they may also act over longer time frames. For example, any characteristic that influences an organism’s susceptibility to being consumed is likely to be under strong genetic selection, which, over evolutionary time, will allow such genetic influences to affect other trophic levels (Shuster et al. 2006). Traits that are heritable can result in patchy nutritional environments at multiple scales. For instance, foliar concentrations of plant secondary metabolites that are toxic to mammals are genetically determined in Eucalyptus species. Because of the spatial distribution of different genotypes of trees, herbivorous marsupials such as possums and marsupial gliders must move over distances greater than 40 m to encounter significant variation in plant secondary chemistry within a eucalypt forest (Andrew et al. 2007).

8.3 Do Predators Regulate Nutrient Intake? Understanding nutritional interactions and their effects requires that we know the differing nutritional requirements, priorities, and regulatory abilities of the various interacting organisms. So far in this book we have shown, using GF designs, that herbivores and omnivores, including species of insects, birds, and mammals, regulate their intake of macronutrients and some micronutrients and make postingestive adjustments to help attain the optimal balance of nutrients to meet their various requirements. In contrast, as we already have mentioned in chapter 1, the prevailing view until recently was that predators (carnivores) do not need to practice nutrient balancing and are unlikely to possess mechanisms to do so.

How Does Nutrition Structure Ecosystems?  |  125

One reason for this view is that animal prey are commonly thought to be of high nutritional quality (by which it is usually meant that they are similar in composition to predator tissues—we will return to this assumption in the next section) and to vary little in composition, as compared with the much wider range of generally less nutritious food compositions eaten by herbivores and omnivores (Galef 1996). If their foods are, indeed, constant and well balanced, then it is true that there is no need for predators to have separate appetites for different nutrients (Rau­ ben­heimer and Simpson 2010). Under such circumstances, the only decision for the animal to make is how much to eat, and for this, foods need only be represented by a single quality: any quality will do—mass, energy, or selenium content would all serve equally well as regulated variables. Accordingly, most models have assumed that the rate of energy acquisition is what determines predator foraging decisions (Stephens and Krebs 1986). The second reason for believing that predators do not exhibit nutrient balancing is that they are thought to be more limited by the availability and challenges of capturing prey than by their nutritional quality (Westoby 1978; Fryxell and Lundberg 1997)—in other words, predators can’t be fussy and must take what they can get. Recent work suggests that both of these assumptions about predators may be incorrect. First, prey animals do vary in composition. Predators that feed on arthropods may experience variation in prey nitrogen content ranging from less than 6% to 13% (Fagan et al. 2002; Denno and Fagan 2003). The fat content of arthropods can vary from less than 5% to more nearly 60% (Barker et al. 1998; Hahn 2005; Raubenheimer et al. 2007), as can fat content in birds and rodents, depending on such factors as whether they are starved or well fed, resident or migrating, about to enter hibernation or done with it, having just emerged after winter (Mc­ Landress and Raveling 1981; Ankney 1984; Batzli and Esseks 1992; Voltura 1997). That a change in prey composition may really matter to predators is nicely illustrated by declines in populations of predatory marine birds and sea lions in the Gulf of Alaska. The structure of fish communities in this region changed as a result of the climate shifting in the North Pacific Ocean during the 1970s, causing predators to rely more heavily on low-fat fish species and suffer accordingly (Rosen and Trites 2000; Ro­mano et al. 2006). Why fat should be particularly limiting to carnivores will become clear in the next section on food webs. But even if prey were invariant in composition, this of itself does not preclude the need for nutritional regulation and nutrient balancing. As discussed in chapter 6, the intake target is not fixed—the optimal blend of nutrients in the diet changes throughout an animal’s life and with environmental conditions. Requirements for protein and fat will shift markedly as a result of the changing demands of growth, health status, re­

126  |  Chapter eight

production, migration, hibernation, and so forth. Tracking these changing requirements in relation to a fixed food composition requires nutrientspecific feedbacks to operate, acting both on feeding behavior and postingestive processing. As for predators being chronically food-deprived, it seems that this is not invariably the case (Herbers 1981; Jeschke 2007). Even when food is scarce, this does not mean that nutritional regulation and nutrient balancing are unimportant. Fat deposits are the first reserves to be depleted when an animal is starved. Starved predators would benefit from selectively choosing high-fat prey to replenish these reserves, which would require that they be able to regulate their intake of fat separately from other nutrients (Raubenheimer et al. 2007). Equally, when approaching or emerging from an energy-demanding period such as migration, hibernation, or reproduction, it may be advantageous to select high-fat food items. It would seem, then, that the reasons for presuming that predators do not demonstrate nutrient balancing are not compelling—indeed, we might expect that predators do have this capacity, at least for fat and protein. To find out, Mayntz and colleagues (2005) studied three species of invertebrate predator with different prey-capture strategies: an active hunter (the ground beetle Agonum dorsale, now renamed by taxonomists Anchomenus dorsalis), an ambush hunter (the wolf spider Pardosa prativaga), and a trap builder (the web-building desert spider Stegodyphyus lineatus). These predators were restricted for a period to a diet of Drosophila flies that were either high or low in the ratio of protein to body fat. These flies were generated by rearing the larvae on foods of different nutritional composition. After the period feeding on the manipulated flies, the predators were tested to discover whether they could redress the nutritional imbalance by selective feeding. All three predators demonstrated regulatory feeding behavior in response to their nutritional state. The mechanism of regulation employed varied according to the predator’s foraging strategy (fig. 8.1). The highly mobile ground beetles selected different proportions of fat-rich and protein-rich foods when offered a choice (fig. 8.1A); wolf spiders balanced their intake of macronutrients by eating more or less of a single fly type (fig. 8.1B); and web-building spiders differentially extracted the deficient nutrients from flies, digesting and sucking out nutrients in proportion to their requirements (fig. 8.1C). Having established that invertebrate predators can respond to a recently imposed change in nutritional state, Raubenheimer and colleagues (2007) initiated a further study on the ground beetle A. dorsale to discover whether longer-term, seasonal changes in nutritional state were also accompanied by nutrient-specific foraging. During winter, adult A. dorsale hibernate belowground. While in hibernation the beetles deplete their

How Does Nutrition Structure Ecosystems?  |  127 A. Ground beetle

Food eaten (mg)

2.5

B. Wolf spider 2.0

Pre-fed protein-rich food Pre-fed lipid-rich food Prey eaten (mg)

3.0

2.0 1.5 1.0

Pre-fed protein-rich prey Pre-fed lipid-rich prey

1.5

1.0

0.5

0.5 0

Feeding on lipid-rich food

Feeding on protein-rich food

0

Feeding on lipid-rich flies

Feeding on protein-rich flies

C. Web spider 15

Percent nitrogen in fly

14

Pre-fed lipid-rich prey Pre-fed protein-rich prey

13 12 11 10 9 8 7

Live flies

Remnants

Fig. 8.1. Nutritional regulatory responses shown by three invertebrate predators with different hunting strategies. (A) Ground beetles, which are highly mobile predators, were pretreated for 48 hours on semisynthetic foods that were either high in protein and low in fat (protein-rich) or low in protein and high in fat (lipid-rich), then offered a choice of protein-rich or lipid-rich foods for a further 48 hours. They ingested most of the food containing the previously limiting nutrient. (B) Wolf spiders (ambush predators) were pretreated for 72 hours on Drosophila that had been reared on different larval media to produce protein-rich or lipid-rich flies. When subsequently offered protein- or lipid-rich flies for 24 hours they ingested more of the flies that contained the macronutrient that was limiting in the pretreatment diet. (C) Web-building spiders were offered a fly after a period of 24 hours during which they had been fed either a protein-rich or a lipid-rich fly. When the remnants of the test victim were analyzed after the spider had finished feeding, it was found that more nitrogen had been extracted by spiders that were protein-deprived, having previously been fed on a lipid-rich fly. (After Mayntz et al. 2005, reprinted with permission from AAAS.)

128  |  Chapter eight

fat reserves, and when they emerge in spring they need both to replenish these reserves and to ingest sufficient protein for their reproductive needs. To establish the priority given to replenishing fat versus gaining protein, field-collected beetles that had just emerged from hibernation were offered a choice of nutritionally complementary foods (one a high-protein, lowfat food; the other a low-protein, high-fat food). Over the first 2 days the beetles prioritized fat ingestion, eating large quantities of the high-fat food. As a consequence, their body fat reserves increased more than threefold over 2 days, while protein content increased only marginally. This indicated that the main priority of beetles emerging from hibernation is to replenish fat, and that they are able to fill their reserves with remarkable alacrity. Over subsequent days beetles greatly reduced their food intake, especially of the high-fat food, and shifted toward mixing a protein-rich diet. Other groups of beetles were confined to one of five diets differing in their protein to fat ratio (P:F) (fig. 8.2B). The main feature of the nutrient intake array across those five diets was a progressively developing pattern for beetles to maintain intake of fat more constant than that of protein. Hence, when dietary P:F was higher than that chosen by insects able to select between complementary foods, beetles ate more protein to gain limiting fat—although this overconsumption of protein was insufficient to prevent a lower rate of gain in body fat than that which occurred in selfselecting beetles. In contrast, when confined to low-P:F diets, beetles did not substantially overeat fat to gain limiting protein, ending up with lower body protein content but the same fat content as self-selecting beetles. These data on ground beetles were intriguing in one key respect: the prioritization of fat above protein intake is the opposite of the pattern seen in herbivores and omnivores, which usually prioritize protein intake above fat and carbohydrate (various examples are scattered throughout the book, although as we shall see in chapter 9 mountain gorillas are an exception). Is the ground beetle typical of all predators, or is it anomalous? Mayntz and colleagues (2009) addressed this question using a very different predator—a mammalian carnivore, the mink (Mustela vison). Mink were offered one of five pairings of foods varying in protein to fat ratio (P:F). In four of these pairings the mink converged tightly onto the same nutrient intake trajectory, proving that they have separate regulatory mechanisms for protein and fat. The fifth food pairing was such that it did not allow them to reach this regulated intake point, and in that case the mink fed overwhelmingly from the food closest in composition to the target. When mink were confined to one of five diets varying in P:F, the intake array was similar to that of ground beetles: fat intake was clearly prioritized over protein intake when mink were unable to reach their intake target (fig. 8.2A).

How Does Nutrition Structure Ecosystems?  |  129 A. Mink

B. Ground beetle

500

10

IT

300 200 100 0

8 Lipid eaten (mg)

Lipid eaten (g)

400

6 IT 4 2

0

100

200 300 400 Protein eaten (g)

500

0.2 0.4 0.6 0.8 Protein eaten (mg)

1.0

0

0

2

4 6 8 Protein eaten (mg)

C. Spider 1.0

Lipid eaten (mg)

0.8 0.6 0.4 0.2 0

0

Fig. 8.2. Intake arrays for fat and protein for three carnivores: the mink, a ground beetle, and a wolf spider. Note how all three species prioritized intake of fat over that of protein, as indicated by the shallow slope of the array. Hence, they tolerated substantial overconsumption of protein to gain limiting fat on diets high in protein and low in fat. The two species in which it was tested (A, B) showed strong regulation to an intake target trajectory when offered a choice of complementary foods, indicating that they have separate appetites for protein and fat. The target rails for these species are indicated as unbroken lines labeled IT, whereas the nutrient balance in the no-choice diets used to generate the intake arrays are shown as dashed lines. The array for the mink was over an 11-day period (after Mayntz et al. 2009), that for the beetles was over 10 days (after Raubenheimer et al. 2007), and for the spider, over the second stadium (an average of 13 days, which was the same across diets) (after Jensen et al. 2012, copyright 2011, with permission from Elsevier).

10

130  |  Chapter eight

And mink and the ground beetle are not alone: other examples of predators regulating and balancing intake of protein and fat in similar ways came from studies on wolf spiders (Jensen et al. 2011; fig. 8.2C), predatory fish (Sánchez-Vázquez et al. 1999; Rubio et al. 2003), and domestic cats (Hewson-Hughes et al. 2011). We will consider the last two of these in chapter 9. In short, predators, like herbivores and omnivores, regulate their intake of macronutrients. Their intake target is high in protein, reflecting their diet, but they prioritize fat intake when forced to trade off protein against fat on nutritionally unbalanced diets, implying that fat limitation incurs greater costs than protein excesses. (As an aside here, predatory fish appear to be more constrained than terrestrial predators in their ability to overeat protein to gain limiting fat because excess nitrogen is excreted as toxic ammonia via the gills, which is toxic and impairs gill function.) Jensen and colleagues (2012) confirmed and measured these costs for ground beetles in an experiment in which mated female beetles were confined to one of 25 diets varying in protein and fat content (plate 6). The number of eggs produced by beetles was measured and related to nutrient intake. When beetles were allowed to compose their own diet from complementary foods, they selected a lipid to protein ratio that corresponded to maximal egg production—the highest elevation on the egg production landscape. When beetles were restricted to a diet with an imbalanced ratio of lipid to protein, they fed to reach the greatest local elevation on the fecundity landscape. Beetles suffered a particularly large loss of egg production when unable to reach an adequate lipid intake. Why is there such a marked shift in the prioritization of protein relative to nonprotein energy (carbohydrate and fat) between herbivores and omnivores, on the one hand, and carnivores, on the other? A likely answer becomes apparent when we consider the geometry of food webs.

8.4 The Nutritional Geometry of Food Webs The combination of all trophic interactions within an ecosystem comprises the food web. Typically, food webs contain a maximum of four to six trophic levels. Give or take a level, these begin with plants and algae (level 1 producers), which are consumed by herbivores (primary consumers, level 2), which are eaten by primary carnivores (secondary consumers, level 3), which are eaten by secondary carnivores (tertiary consumers, level 4), which are consumed by apex predators, which are not eaten by anything else (level 5). This classification is simplistic, to the extent that trophic interactions do not form simple chains but, rather, intricate webs. Animals do not necessarily (or often) restrict their feeding to prey in the trophic level immediately below; they may feed across several levels, and

How Does Nutrition Structure Ecosystems?  |  131

even on others within their own trophic level. Omnivores, for example, include both plants and animals in their diet. “Intra-guild predators” (Polis et al. 1989) are predators that “compete for a shared prey population [and] also engage in predator–prey interactions with each other” (Rosenheim 2007). A special case of intraguild predation is cannibalism, where the victim is of the same species as the consumer (chapter 7; Elgar and Crespi 1992). Most theories for explaining food web dynamics have been based on fluxes of mass and energy between consumed and consumers. We have seen several times already in this book that unidimensional models are insufficient to capture the essential features of nutritional systems, and we suspect the same is true for trophic ecology. Some food web models have, however, considered fluxes of multiple nutrients or chemical elements (e.g., DeAngelis et al. 1989; Sterner et al. 1996; Sterner and Elser 2002). Sterner, Elser, and colleagues have developed a body of theory known as Ecological Stoichiometry, which has had considerable success at explaining patterns of species interaction and population growth, particularly within aquatic ecosystems (Sterner and Elser 2002; Hall 2009). Sterner and coworkers (1996) proposed that food webs can be represented as “tropochemical diagrams,” which are essentially nutrient (or chemical element) spaces in which species are represented as points located according to their body compositions, with arrows joining consumers and consumed. We would argue that an important aspect of the animal that needs to be represented in models of food webs is the concept of nutritional targets—intake, nutrient, and growth targets. These are fundamental in ecological interactions because, as we have seen throughout this book, targets provide a good deal of predictability about the nutritional and foraging decisions of animals. We believe that incorporating the target concept into models of ecological communities can, similarly, help to predict and understand the fluxes of nutrients in ecosystems. To illustrate, we now provide a model that generates a hypothesis about an important structural feature of ecosystems, the fact that the number of trophic levels is limited. The intake targets of consumers include nutrients required for respiration and inevitable wastage as well as those that end up in body tissues and determine body composition (chapter 2). Simply put, and contrary to the popular adage, animals are not what they eat. Let’s consider a striking illustration: aphids that feed on plant sap (phloem). Based on body composition alone, the optimal food for an aphid would be another aphid, but in fact they perform best on a nutrient ratio that is far higher in carbohydrate and lower in the concentration of amino acids than aphid tissues (Abisgold et al. 1994). Part of the extra carbohydrate is

132  |  Chapter eight

needed by aphids for energy, but that is not the full explanation. Aphids perform best on sap because they possess a suite of adaptations, including a modified gut and a special association with symbiotic bacteria, that help them to void excess sugar and improve the balance of amino acids in phloem sap (Douglas 2003). These adaptations efficiently translate plant sap to aphid tissues, but cause problems if the aphid eats a diet that is much lower in carbohydrate and richer in amino acids than sap (see fig. 6.5). If we are to understand the flux of nutrients within food webs, and ultimately explain the functioning of ecosystems, we must consider both the nutritional requirements for growth and respiration of each species (and stage within the life cycle—see chapter 6), and how ingested nutrients are apportioned to yield the body compositions that may in turn become food for other animals. Despite the substantial variation in nutritional targets and regulatory responses between and within species (chapter 6), two high-level generalizations can be made about the nutritional biology of different trophic levels. First, generally speaking, an average herbivore has a greater range of possible food compositions available to it than does the average predator. An extreme example is insects such as bees, for which nectar and pollen, between them, encompass the entire protein-carbohydrate nutrient plane. The fact that herbivores have a wide range of dietary options has encouraged the evolution of a degree of nutritional specialization that is not commonly seen between species at higher trophic levels, with an associated greater spread of intake targets. For example, intake targets for herbivorous insects span 10% to 80% of total energy as protein—from aphids, which specialize on sugar-rich plant sap, to caterpillars that eat high-protein seeds and young leaves (see figs. 6.5 and 6.6). Predators by contrast typically require around 50% of their energy as protein in an optimal diet (figs. 8.2 and 9.1A). The second high-level generalization is that, within the range of food compositions available, an average predator will experience a higher frequency of protein-rich items than will an average herbivore. As a result, predators—especially higher-level carnivores—are adapted to use protein both as a primary source of nitrogen and as energy (Eisert 2011). With these two generalizations in mind, let us next consider how trophic interactions can be considered within geometric plots comprising two nutrient dimensions—protein and nonprotein energy (carbohydrate and fat) (fig. 8.3; Raubenheimer et al. 2009). For simplicity, we will assume that all ingested nutrient is absorbed and used by the animal to enhance its fitness, without wastage. This amounts to the situation where the intake target is the same as the nutrient target, because as we explained in chapter 2 the intake target exceeds the nutrient target in each nutrient dimension by the amount of nutrient that is eaten but not used

How Does Nutrition Structure Ecosystems?  |  133

for some beneficial function. We also presume that the targets are fixed, although as we have seen in chapter 6, they vary throughout an animal’s lifetime and across species within a trophic level. For the purposes of our model we can think of these fixed targets as averages representing animals at a given trophic level. Our final simplification is that the growth target (that portion of ingested nutrients that is optimally allocated to growth—see chapter 2) is, on average, the same for herbivores and carnivores. Here we have ignored the variation that may occur in the protein and fat content of animals at different life stages and under different environmental circumstances (discussed in the previous section). Having dealt with the simplifying assumptions (none of which affect the main conclusions to be drawn), let us return to the structure of food webs and consider the first trophic interaction between plants and herbivores (fig. 8.3A). To ground the model in real data, the herbivore in our example is based on a locust (Raubenheimer and Simpson 1993). It has a nutrient target (NT) that is slightly carbohydrate-biased. To get to the growth target (GT), which represents its optimal growth and body composition (for a locust 70% protein, 20% fat, and 10% carbohydrate), the herbivore must use some of its ingested protein (the line segment labeled D(u)p for “dissociated, utilized protein”) and a greater proportion of ingested carbohydrate and fat (labeled D(u)cl for “dissociated, utilized carbohydrate and lipid”) as fuel. The total energy requirements of the animal, E, on an optimal diet are therefore D(u)p + D(u)cl. The optimal fuel blend for the animal is D(u)cl parts carbohydrate and fat to D(u)p parts protein, but total energy requirements could in theory be met by ingesting any blend of protein and nonprotein energy that sums to E. The possible blends all fall along the dashed line AB. This assumes that protein, fat, and carbohydrate are all equally usable by the herbivore as energy sources, which need not be the case—for example, some herbivore species are less able than others to convert protein to energy (see chapter 6). The value of E is approximately 10 times the sum of energy at the GT; hence, there is a 10% efficiency of transfer of energy from one trophic level to the next, the rest being lost as heat. This is known as the trophic pyramid (Odum 1971). If all herbivores achieved an optimal diet (met their NT), the range of body compositions available to their predators would be restricted to the range of compositions represented by the growth targets across different herbivore prey species and life stages (see chapter 6). In our example of a single herbivore with a fixed growth target, all prey would be of the same composition. However, because of ecological circumstances, herbivores will not always achieve an optimal diet. The gray zone in figure 8.3A shows the range of food compositions that might be available to the herbivore, both within and between host plants. The distribution of those

134  |  Chapter eight

Available carbohydrate and/or lipid (kJ)

A. Herbivore (trophic level 2)

NT rail (near mean of food compositions)

A

Range of food compositions available, reflecting variation within and between plants. NT

D(u)p

E

GT rail D(u)cl

B

GT Available protein (kJ)

Available carbohydrate and/or lipid (kJ)

B. Primary carnivore (trophic level 3) Range of food compositions available has narrowed due to 50% effective homeostatic regulation of growth in herbivore. NT rail

A

D(u)p

E

GT rail

NT D(u)cl B

GT Available protein (kJ)

Available carbohydrate and/or lipid (kJ)

C. Secondary carnivore (trophic level 4) A Range of food compositions available has narrowed due to 50% effective homeostatic regulation of growth in primary carnivores. E

D(u)p

GT rail = NT rail

NT D(u)cl

GT Available protein (kJ)

B

How Does Nutrition Structure Ecosystems?  |  135

food compositions within the range is not shown, but may well be biased toward lower-protein items given the composition of many plant tissues. Note in our example in figure 8.3A that the position of the NT bisects the distribution of plant food compositions. We have done this to capture the idea that animals might be expected to evolve postingestive adaptations that place (and in the limit center) their intake requirements within the range of available food compositions in their nutritional environments (see chapter 6). We know from our experiments on locusts (Raubenheimer and Simpson 1993, 2004; see fig. 2.5) that if individual locusts were confined to a single diet composition spanning the range indicated in figure 8.3A, the range of resulting body compositions would be about half that of the range of diets, owing to mechanisms of postingestive regulation (chapter 3). This narrowed range of body compositions provides the nutritional landscape for primary carnivores, which brings us to figure 8.3B. A key point in figure 8.3B is that the narrowing in range of herbivore body compositions relative to plants (the gray region) is not symmetrical. Animal bodies, irrespective of their trophic level, contain higher concentrations of protein than the average plant (Eisert 2011), notwithstanding that some plants and plant parts are high in protein. Therefore, the diet of an average primary carnivore is both narrower in range and shifted toward a higher ratio of protein to nonprotein energy than that of an average herbivore. There are two important consequences of this shift in diet composition across trophic levels 2 (herbivores) to 3 (primary carnivores). First, because they are adapted to feed on animal tissues, the nutrient target (NT) of an average primary carnivore will be more protein-biased than that of an average herbivore (although some herbivores too have high protein intake targets; see chapter 6). Second, a greater proportion of the total energy requirements, E, of an average primary carnivore must come from protein than from carbohydrate and fat. This is indicated by the lengthening of line segment D(u)p relative to D(u)cl.

Fig. 8.3. An illustration of how regulation of macronutrient intake and postingestive regulation of body composition will successively affect different trophic levels as a food chain is ascended. See text for detailed explanation of logic, symbols, and conventions. (From Raubenheimer et al. 2009, courtesy of John Wiley and Sons.)

136  |  Chapter eight

Like herbivores, primary carnivores possess postingestive mechanisms to help adjust for dietary imbalances. These regulatory mechanisms will presumably narrow the range of body compositions among primary carnivores available as prey to secondary carnivores (gray region in fig. 8.3C), just as similar mechanisms narrowed the range of herbivore body compositions available to primary carnivores relative to plants. And once again the narrowing, such as it is, will be biased toward protein because of the constraints of animal tissues having high protein content. Consequently, the majority of an average secondary carnivore’s energy needs must be met from protein, and D(u)p is now much longer than D(u)cl. An extreme example of this is seen in so-called hypercarnivores such as cats, which rely heavily on ingested protein to supply glucose for brain function (Eisert 2011) and are ill equipped to deal with dietary carbohydrate (Hewson-Hughes et al. 2011; chapter 9). In summary, we can see from figure 8.3A–C that as trophic levels are ascended, regulatory feeding and growth responses will progressively narrow the range of food compositions and, more important, shift the mean composition toward a higher proportion of protein (fig. 8.4A, which is based on the gray regions in fig. 8.3A–C). This predicted pattern is supported by data on invertebrates showing that the percentage of body nitrogen increases as trophic levels are ascended, whereas the ratio of carbon to nitrogen (C:N) in food versus in the body of consumers (C:N food / C:N consumer) narrows progressively (Denno and Fagan 2003; fig. 8.4B). Denno and Fagan (Denno and Fagan 2003; Fagan and Denno 2004) concluded from these trends in body elemental composition that carnivorous arthropods are nitrogen (protein)–limited. This, they suggested, explains the tendency for many predators to demonstrate trophic omnivory—to feed not only on herbivores but also on other carnivores (which, as we have seen, tend to have higher nitrogen concentration). However, taking into account the respiratory needs of carnivores, we would argue that the reverse is true: consumers become progressively nonprotein energy–limited as trophic levels are ascended. This will be especially, but not exclusively, the case when body fat stores need to be increased to meet the demands of periods of starvation, hibernation, reproduction, or migration (Raubenheimer et al. 2007). We suspect that it is this energy shortage that motivates feeding back down the food chain (Raubenheimer et al. 2009) and perhaps explains why predators prioritize fat intake over that of protein (see previous section), and why populations of predatory marine birds and sea lions declined in the Gulf of Alaska after the loss of high-fat species of fish (Rosen and Trites 2000; Romano et al. 2006; see above). Recently, Wilder and Eubanks (2010) have reiterated and elaborated on the logic of this argument.

Frequency (unscaled)

How Does Nutrition Structure Ecosystems?  |  137

Predators

Herbivores

0

Plants

1 2 3 Ratio of carbohydrate plus fat to protein in body

4

800

Frequency

600 Plant : herbivore 400 200 0 0.1

Predator : predator

Herbivore : predator

1 Ratio of C:N in resource to C:N in consumer

10

Fig. 8.4. The overall effect of homeostatic feeding and growth responses across trophic levels (fig. 8.3) is that the range of body compositions available to consumers will narrow and the mean dietary composition will shift toward a higher proportion of protein and lower proportion of nonprotein energy. Panel (A) illustrates this shift, whereas (B) plots supporting data based on comparisons of elemental carbon and nitrogen ratios between food resources and consumers across trophic levels. ([A] from Raubenheimer et al. 2009; [B] from Denno and Fagan 2003, courtesy of the Ecological Society of America.)

Robert May noted in 1999 that why food chains are so short and consistent in length (four to six levels typically) is one of the big unanswered questions in ecology. The question remains unanswered. The textbook explanation is that metabolic activities ensure that with each successive trophic transaction energy becomes progressively depleted, until no further trophic levels are sustainable (the trophic pyramid effect). If this were the case, however, it would be expected that trophic pyramids with a large base (abundant primary productivity) would be higher (a greater number of trophic levels), but, in general, this is not the case (May 1999).

138  |  Chapter eight

Perhaps the downward pressure within food chains occasioned by increasing nonprotein energy limitation provides an explanation.

8.5 The Nutritional Niche We have seen that it may not be possible to predict the growth rate of a population or species from the nutritional responses of individuals, because of direct and indirect interactions with other organisms within food webs. To use the terminology of niche theory, the fundamental nutritional niche of a population may not be realized. The Geometric Framework shares basic features with Hutchinson’s niche concept. Hutchinson (1957) conceived of the niche as a region within multidimensional environmental space in which populations can persist. The environmental space includes axes for physical (“abiotic”) factors such as temperature, light, humidity, and pH, and other axes for food resources. The “fundamental niche” in environmental space is that set of combinations of environmental variables within which a population can persist (i.e., where pgr is greater than zero) in the absence of complicating interactions with other organisms. The “realized niche” is the reality once such interactions are present (Kearney 2006). The Geometric Framework concerns the “resources” component of a more inclusive Hutchinsonian environmental space. To broaden the power of GF models to predict the performance, population dynamics, and distribution of species, it will be necessary to incorporate abiotic influences as extra dimensions and to transpose this niche space onto real environments in space and time (Kearney et al. 2010). Behmer and Joern (2008) were the first to use GF models to explore species interactions in the context of the nutritional niche. They studied seven species of generalist grasshoppers (all species in the genus Melanoplus) that coexist in the Arapaho Prairie in Nebraska. Using artificial diets, they found that there were significant differences between all but one pair of species in the self-selected ratio of protein to carbohydrate. Maximal performance coincided with these differences in regulated intake. These results suggested that competition for nutrients might have driven partitioning among these species within nutritional space. Behmer and Joern’s (2008) study raises an important issue: what is the relationship between nutritional dimensions in environmental space and the foods that contain those nutrients? Throughout this book we have represented foods as nutritional rails, which is entirely appropriate when the focus is on the biology of the consumer. But in considering ecological and evolutionary relationships among consumers and their foods, the

How Does Nutrition Structure Ecosystems?  |  139

situation becomes more complex because, in addition to their nutritional composition, prey organisms can contribute other important properties to the interactions. Take, for example, the situation where the same food organism represents more than one food rail, as is very often true of plants—leaves, flowers, and seeds all have very different nutrient compositions. In the “nutritional niche” concept of Behmer and Joern, an animal that ate leaves would not be in competition with another herbivore species that ate the seeds of that plant, because their respective intake targets would differ. In reality, however, these species might well be in competition, because depletion of leaves would damage the plant’s ability to produce seeds and, conversely, without seeds future generations of leaves would be jeopardized. Here the fate of the two foods and their constituent nutrients are inextricably linked by the biology of the plant, and to understand the competitive interaction between the herbivores this biology needs to be taken into account. It is also easy to imagine the converse situation, where animals with similar intake targets do not compete for nutrients. For example, a grasshopper that specialized on grasses would not be in competition with another that specialized on broad-leaved plants, even if the nutrient requirements (intake targets) of the two grasshoppers were identical. This is not to say that we disagree with Behmer and Joern that the grasshoppers in their study are able to coexist because they have different intake targets. Rather, we wish to emphasize that ecological interactions should not, by default, be reduced to nutrient compositions, until it has been established that crucial higher-level factors that cannot be expressed directly in nutritional terms have not been omitted from the model. In the above example, the mutual reliance of seeds on leaves of the food plant is an organism-level factor that needs to be taken into account to understand the nutritional dynamics of the interaction, but there might well be other ecological factors that need to be included in the model too (more on this in the following sections). To generalize this point, we have argued that effective models in nutritional ecology must be not only “nutritionally explicit” (able to express a situation in terms of nutrition, as we have done using GF throughout this book) but also “organismally explicit” and “ecologically explicit” (able to incorporate organism-level and ecological factors in the model) (Raubenheimer et al. 2009). Crucially, models that are nutritionally, organismally, and ecologically explicit should be able not only to incorporate nutritional, organismal, and ecological factors but also to address questions that lie at the level of nutrition, the organism, and ecology. Examples of nutritionally explicit and organismally explicit analyses abound in previous chapters of the book. For example, in chapter 4 we

140  |  Chapter eight

address the question of how nutrition affects the life cycle (longevity) of flies, whereas in chapter 6 we show how aspects of the biology of animals (such as the possession of symbionts, and diet breadth) affect the patterns of nutrient regulation. In the sections that follow we show some ways in which GF can be made more ecologically explicit by incorporating abiotic and biotic aspects of the ecological environment, enhancing the capacity of this approach to integrate nutrition into ecosystem-level analyses.

Incorporating Abiotic Factors In addition to food resources, Hutchinson’s niche model includes dimensions for abiotic factors such as temperature, light, humidity, and pH. A research field known as “Biophysical Ecology” has developed to represent the “climatic” niches of organisms in “climate space” (Porter and Gates 1969). These models define the combinations of environmental variables that are suitable for survival and reproduction for a species, and have been applied at the scale of landscapes to predict the distribution, temperature, energy, and water relations of organisms (Kearney and Porter 2009). Abiotic factors can shape nutritional interactions in two major ways. The first is through influences on the physiology and behavior of organisms. For example, changes in body temperature alter metabolic and water loss rates, amounts and ratios of nutrients required, and tolerance to toxins and diseases. Hence, a decrease in ambient temperature results in rats increasing their carbohydrate or fat consumption but not changing their protein intake (Aubert et al. 1995), and changes in temperature affect the tolerance of white-throated wood rats for plant secondary metabolites (Dearing et al. 2008). Miller and colleagues (2009) showed that locusts developed faster but converted protein and carbohydrate less efficiently to growth if kept at 38°C rather than at 32°C. If given a choice of temperatures under conditions of food abundance, locusts selected 38°C, therefore prioritizing development rate over utilization efficiency. In a subsequent experiment, Coggan and colleagues (2011) allowed recently fed locusts to select a temperature within a thermal gradient, and after 50 minutes (the average between-meal interval) they either fed the locusts again or left them without food. Locusts initially selected a temperature of 38°C and returned to that temperature after having been fed again. However, if they were not fed at 50 minutes, they soon moved to a cooler region of the gradient, 32°C; they were now prioritizing efficient extraction of nutrients from the previous meal over rapid growth in response to the absence of food. The second way in which abiotic factors influence nutritional interactions is through the distribution of environmental conditions in space and

How Does Nutrition Structure Ecosystems?  |  141

time. Environmental gradients, concentrations, and ratios of chemical elements in soil or water, microclimatic conditions, and other habitat features will shape patterns of food availability over multiple spatial scales and timescales and will interact with an animal’s size, mobility, and regulatory behavior (Pincebourde and Casas 2006; Kearney and Porter 2009).

Dynamic Energy Budget Theory as a Computational Engine Kearney and colleagues (2010) proposed that a platform known as Dynamic Energy Budget theory could be used as a “computational engine” for integrating GF and Biophysical Ecology, allowing population responses to be modeled dynamically within environmental space (fig. 8.5). We will not cover this complex topic in much detail here, but a brief description and some examples are warranted. Dynamic Energy Budget (DEB) theory was developed by Kooijman and collaborators (Kooijman 2009; Sousa et al. 2008). In brief, DEB was designed to model how organisms acquire and allocate energy and matter. It assumes that energy and matter are assimilated in proportion to surface area, and are directed first to a “reserve pool” and then allocated to structural components. Reserves are continually used and replenished, and structure is retained and requires energy for its maintenance in direct proportion to its volume. The process of transforming ingested substrates into reserves (assimilation) is defined by the “synthesizing unit,” which is a construct based on classical enzyme reactions involving more than one potentially limiting substrate (Kooijman 2009; Poggiale et al. 2010). Development, growth, and reproduction are calculated dynamically according to an allocation rule, which defines how energy and matter flow from the reserves to growth, maintenance, maturation, and reproduction. The scaling of metabolic rate with body mass results from the relative amounts of reserve and structure, and from other costs such as growth and endothermy (Kooijman 2009). A special type of DEB model is needed to integrate DEB theory with GF (fig. 8.6), one that includes a separate reserve for each nutritional component (Kearney et al. 2010). An example of how a two-currency DEB model could be used as a computational engine for implementing GF designs to predict population responses is shown in plate 7. Kearney and colleagues (2010) applied a DEB model developed by Kuijper and coworkers (2004) to predict egg production in a copepod as a function of ingested carbohydrate and protein (plate 7). The rate of egg production increased with protein and carbohydrate consumed, and production could occur in the absence of carbohydrate but required protein. The response surface for lifetime reproductive output shifted when longevity was linked to the ratio of protein to carbohydrate in the diet, using values

142  |  Chapter eight Biophysical Model Target (optimal) range

Survivable range Available range humidity

Environmental gradients (GIS data)

Selected environments through time

r Su ce

fa

d win ed pe s air t emp erat ure

ter wa Available food 1

Fo rag i an ng lo d t ca im tio es ns

Target (optimal) range

carbohydrate Food rail

Physiological rates

Survivable range

Available food 2

pro tein

body temperatures

Selected foods through time

Water balance

Dynamic Energy/Mass Budget κ

Food ingested

e um ol ,V

ea ar

Geometric Framework

Reserve 1–κ

Growth

Structure

Somatic maintenance Maturity maintenance

Metabolic rate

Reproductive maintenance

Fig. 8.5. A representation of the niche derived by integrating models from Biophysical Ecology, the Geometric Framework, and Dynamic Energy Budgets. The Biophysical Ecology component is a climate space diagram, whereas GF contributes a nutrient space diagram. Only a subset of the many possible niche dimensions is shown in each case. Organisms can survive within defined regions of niche space and regulate both behaviorally and physiologically toward target states within this survivable region. Regulatory behaviors respond to the environment (and in turn influence it), yielding nutritional and biophysical outcomes that act as driving variables for the DEB model, which computes rates of growth, development, reproduction, and aging. (From Kearney et al. 2010.) Fig. 8.6. A schematic illustrating the components of a two-reserve (protein and carbohydrate) Dynamic Energy Budget model. Synthesizing units (SU) control assimilation and allocation of ingested nutrients to growth, maintenance, and reproduction. (From Kearney et al. 2010.)

Metabolic work Gonads

How Does Nutrition Structure Ecosystems?  |  143 Food ingested

pA κP

lumped assimilated material protein fraction of assimilate

pC κG pS pG pM pR κE

pa mobilization rate fraction to growth/maintenance somatic maintenance growth allocation Assimilation SUs maturity maintenance repro allocation fraction returned to reserves pA*κP pA*(1–κP)

Feces

Carbohydrate reserve Return flux

Protein reserve pC (1–κM)

κ

pS

pC κ

pG pG

(1–κM)

pS

Maintenance SU Structure

Somatic maintenance

pM

Somatic maintenance

pR

pR

Return flux κR

Return flux κR

Growth SU Maintenance SU

pM

Maintenance SU Reproduction Maintenance SU SU

Maturity maintenance Rejection

Maturation/ reproduction

Maturity maintenance Rejection

144  |  Chapter eight

extrapolated from other invertebrates (see chapter 4). When hypothetical costs for storing nutrient excesses were imposed (see fig. 2.2), the surface developed a peak. Hence, DEB could generate at the level of populations (in the absence of other biological interactions) the types of responses seen in GF studies on individuals.

8.6 Agent-Based Modeling of Nutritional Interactions: From Individuals to Ecosystems An appropriately defined DEB model offers a means of implementing GF models and calculating the fundamental niche of an organism, but the problem of dealing with interactions among organisms—of calculating the realized niche—remains. Since interactions among individual organisms are the most fundamental units from which ecosystems are built, might it be possible to use agent-based models as introduced in chapter 7 to model trophic interactions? Agent-based models (ABMs) have been used to explore patterns of resource exploitation in foraging animals (Oom et al. 2004; Grimm and Railsback 2005; Nonaka and Holme 2007), but they have yet to take account of multiple nutrient dimensions; in other words, they are not yet nutritionally explicit (Raubenheimer et al. 2009; Simpson et al. 2010). However, if ABMs were to incorporate aspects of GF models, they would offer the promise of dynamic, spatially explicit models of nutritional interactions, which could take account of an individual organism’s simultaneous membership in a group, population, community, and ecosystem. From such “heterarchical” models (Raubenheimer et al. 2009) may well emerge step-shifts in organizational complexity, such as have been observed already at the level of groups and populations (chapter 7). What would such an agent-based modeling platform look like? It would need to involve resource-seeking agents interacting locally and evolving within dynamic environments. The first step would be to define a generic structure for representing individual agents, which could be parameterized to represent any specific type of organism. The next requirement would be to define the rules whereby agents interact with one another and with aspects of their environment. Next, it would be desirable if agents were able to adapt and evolve in response to their environment. Finally, the agents would need to be set free to interact within simulated environments that include key physical features at multiple spatial scales, and which respond dynamically to the activities of agents operating within them (Simpson et al. 2010). The risk of any modeling enterprise is that models can become too com-

How Does Nutrition Structure Ecosystems?  |  145

plex and case-specific. Agents would need to be encapsulated by as few basic elements as possible in order to capture reality and yet be sufficiently generic that they can be tailored to specific cases by adjusting parameter values (plate 8). Agents would need to exhibit a set of key actions—feed, move, defend themselves, metabolize, grow, excrete, develop, reproduce, senesce. Agents would also need to have states and conditions that affect their actions: nutritional state, levels of toxins ingested, developmental stage, age, sex, size, and level of experience (memory). Each agent would have a nutritional composition and some ability to prevent access to those nutrients by other agents (escape behavior, production of a toxin or immune response), which would determine whether, and by how much, it is eaten by a predatory agent. Having eaten, the consumer (and consumed) would change state. This change in state would then affect behavior according to predefined mathematical functions. Assimilation and allocation of ingested nutrients could be implemented in DEB models as described above. Evolutionary change could occur by allowing targets, rules of compromise, responses, and linking functions to be mutable. As evidenced from previous chapters, much of the information required for such a model already exists, and other data could readily be collected (Simpson et al. 2010). Such models will also benefit from recent functional modeling of optimal foraging decisions, which have been derived within a GF context that takes account of environmental unpredictability (Houston et al. 2011). The aim will be to use such a platform to explore the evolution of nutritional phenotypes, to model populations, and to help explain the structure and dynamics of species assemblages within changing, spatially explicit environments. If successful, the outcome will be a synthesis of nutritional ecology that is dynamic as well as nutritionally, organismally, ecologically, and spatially explicit.

8.7 Conclusions We have seen that the ecological consequences of nutrition are not restricted to the effects of diet on individual organisms but include as well the direct and indirect interactions occurring among individuals within populations and between species. Understanding the complex network of interactions that produce food webs and structure ecosystem dynamics requires that we understand the participants’ differing nutritional requirements, priorities, and regulatory capacities. GF analyses have shown that these features differ between species and across trophic levels. Until recently it had generally been as-

146  |  Chapter eight

sumed that predators do not regulate their intake of macronutrients, but we now know that they do—with important implications for the functioning of food webs. Nutritional space is one part of the fundamental niche of an organism, and there is a need to integrate nutrition with the biophysical ecology of organisms. Evolutionary processes also need to be taken into account, and agent-based models offer promise toward development of a new understanding of the evolutionary ecology of nutrition. In the next two chapters we apply some of these findings and ideas within a practical context. First we consider artificially selected systems: food-animal production and companion animals, and the conservation of endangered species (chapter 9). Then, and finally, we turn the spotlight onto ourselves and consider the human obesity crisis (chapter 10).

500 450 20

400

25

350 Carbohydrate eaten (mg)

20

25

30

300 35

25

250 200

20

150

15

100 50 20

0

0

50

25

30

200 100 150 Protein eaten (mg)

35

250

300

Plate 1. Results of an experiment in which caterpillars of Spodoptera littoralis were fed chemically defined diets over the final larval stadium. Thirty-five treatment groups were each provided with a single food, varying in the ratio and concentration of protein and carbohydrate. A performance landscape (growth rate × survivorship) has been fitted onto the resulting array of nutrient intakes. The landscape has its maximal elevation in the deep red and its lowest height in dark blue. When offered the opportunity to select their diet composition in one of five complementary food pairings, caterpillars regulated to an intake target that sat at the summit of the performance landscape (bull’s-eye). (Data from Simpson et al. 2004.)

1. Training (48 h) Protein food

Carbohydrate food

2. Deprivation (4 h) Protein- or carbohydratedeficient food 3. Testing

?? 4. Results Preference score Prefer green Prefer yellow

10 8

Trained

Naïve

6 4 2

Training Deficient nutrient eaten in: Yellow Green

0 −2 −4 −6 −8 −10

Carbohydrate

Protein

Protein & carbohydrate Deficient nutrient

Plate 2. An experiment showing macronutrient-specific learning in locusts. Insects were first allowed access to both protein-rich and carbohydrate-rich foods for 48 hours. Each food was positioned at the end of a compartment projecting from a central chamber and associated with either the color yellow or green. Then locusts were rendered either proteinor carbohydrate-deficient for 4 hours, in the absence of color cues, before being tested to see whether they responded differently to the color cues depending on their nutritional state. When locusts were deficient in protein, they were attracted toward the color previously paired with protein-rich food; whereas when they were carbohydrate-deprived, they were attracted by the color previously paired with carbohydrate-rich food. Naïve insects showed a weak preference for the color yellow. (After Raubenheimer and Tucker 1997.)

A. Female Drosophila Carbohydrate eaten (mg per day)

Life span 0.2

0.1

0.1

0

0

0.1

0.2

B. Female Queensland fruit fly Carbohydrate eaten (mg per day)

3.5

Life span

2.1

2.1

1.4

1.4

0.7

0.7 0

0.7

1.4

2.1

2.8

3.5

50

Life span

0

40

30

30

20

20

10

10 0

10

20

30

40

50

D. Male field cricket 50

Life span

0

40

30

30

20

20

10

10 0

40 10 20 30 50 Protein eaten (mg per day)

0

0.2

Lifetime eggs

0

0.7

1.4

2.1

2.8

3.5

Lifetime eggs

0

50

40

0

0.1

50

40

0

0

3.5 2.8

C. Female field cricket Carbohydrate eaten (mg per day)

0

2.8

0

Carbohydrate eaten (mg per day)

Lifetime eggs

0.2

10

20

30

40

50

Lifetime singing

0

40 10 20 30 50 Protein eaten (mg per day)

Plate 3. Response surfaces showing how the intake of protein and carbohydrate affects life span and reproductive output in adults of three insect species. Insects were given ad libitum access to one of 28 (Drosophila and the Queensland fruit fly) or 24 (field cricket) diets varying in the ratio and total concentration of protein to carbohydrate (P:C). Fitted surfaces for longevity and reproductive output are plotted onto nutrient intake arrays. The measure for reproductive output is lifetime egg production for females, and for male crickets, the amount of time spent singing to attract females across the lifetime. Surfaces rise in elevation from dark blue to dark red. Unbroken red lines show the dietary P:C that maximized the response variable; dashed lines indicate isocaloric intakes. Data are replotted from Lee et al. 2008a (Drosophila), Maklakov et al. 2008 (field crickets), and Fanson et al. 2009 (Queensland fruit fly).

A. Lysozyme activity

B. Phenoloxidase activity 250 Carbohydrate eaten (mg)

Carbohydrate eaten (mg)

250 200 150 100 50 0

0

50 100 150 200 Protein eaten(mg)

250

200 150 100 50 0

0

50 100 150 200 Protein eaten(mg)

250

Plate 4. Response surfaces showing the effects of protein and carbohydrate eaten by Spodoptera littoralis caterpillars on two components of the innate immune system, lysozyme activity (associated with fighting microbial pathogens) and phenoloxidase activity (part of the encapsulation response to invading organisms). Note how the two immune traits have very different response surfaces in relation to macronutrient intake. (From Cotter et al. 2010.)

A. No choice

Carbohydrate in diet (g per L)

100

Slime mold

Food 1 cm

Density of growth (mg cm–2)

80 60 40 20 0

0

B. Pair-wise choice

80 40 120 Protein in diet (g per L)

160

7 6 5 Growth (mg)

6:1 Food 1:2

4 PC2:1 vs PC1:6 PC3:1 vs PC1:9 PC4:1 vs PC1:8 PC6:1 vs PC1:2 PC8:1 vs PC1:4 PC9:1 vs PC1:3

3 2 1 0

C. Multiple choice

1:9

9:1

0

0.6 0.2 0.4 0.8 1.0 Proportion of carbohydrate in diet

6:1 4:1

1:6 1:4

3:1

2:1

1:3 1:2

1:1

Plate 5. Nutritional wisdom in the acellular slime mold Physarum polycephalum. (A) When provided with one of 35 foods varying in protein to carbohydrate ratio (P:C) and concentration, slime molds grew best on a 2:1 P:C diet. The colored surface plots the density of growth (mass / area grown) for slime mold plasmodia across diet composition space. The red line indicates the optimal ratio for growth (2:1 P:C, which represents a carbohydrate proportion of 0.33). (B) When offered one of six pairwise choices of nutritionally complementary food blocks, the slime molds grew to distribute their mass in precisely the proportion required to achieve an optimal diet. Note how the colored points sit at the peak of the growth curve at a dietary carbohydrate proportion of 0.33. (C) When provided with a clock face of dietary choices varying in P:C, slime molds migrated to the optimal food block. The red line encompasses the average proportion of growth found on the various foods. (From Dussutour et al. 2010, copyright 2010 National Academy of Sciences, U.S.A.)

Lipid eaten (mg/mg beetle)

0.8 Egg content

0.6

0.32

0.4

0.34

0

0.3

0.26

0.2

0.18 0.16

0

0.2

0.4

0.2

0.6

0.22

0.28

0.24

0.8 1.0 1.2 1.4 Protein eaten (mg/mg beetle)

1.6

1.8

2.0

2.2

Plate 6. A response surface for egg content (eggs contained in the abdomen at the end of the experiment) derived from over 500 mated female ground beetles, Anchomenus dorsalis (formerly Agonum dorsale), that were confined to one of 25 diets differing in protein and fat content. The bivariate mean (± standard error) lipid and protein intake point of the beetles that were allowed to compose their diet from two nutritionally complementary foods is plotted on the landscape in white. Nutrient intakes of beetles that were provided with ad libitum access to only one of the 10 foods are plotted on the landscape in black (means ± standard error). In this case the standard errors are diagonal, because intakes were free to vary only along the respective nutritional rails. (After Jensen et al. 2012.)

0.20

0.20

0.15

0.15

0.10

Lifetime egg production (μg) (no costs to nutrient excesses)

8 7 6 5 4 3

0.10

0

C.

80 70

0.05

0

0.05 0.10 0.15 Protein eaten (μg)

2

0.05

0.20

0

1 0

0.05 0.10 0.15 Protein eaten (μg)

0.20

Lifetime egg production (μg) (costs to nutrient excesses)

0.20

Carbohydrate eaten (μg)

B.

Egg production (pg d–1)

60 50 40 30 20 10

Carbohydrate eaten (μg)

A.

0.15

0.10 0.7.6 0 .5 0 .4 0 .3 0 2 0. .1 0

0.05

0

0

0.05 0.10 0.15 Protein eaten (μg)

0.20

Plate 7. Calculation of the nutritional niche (dark colored areas in each panel) for the copepod Acartia tonsa, based on a Dynamic Energy Budget model with separate reserves for protein and nonprotein assimilates and represented using the Geometric Framework. Egg production rate (A) and lifetime egg production (egg production rate × longevity) are plotted, both without (B) and with (C) costs for storing excess nutrients. Longevity is assumed to be affected by dietary protein to carbohydrate balance in the same way as reported for other arthropods (see chapter 4). (From Kearney et al. 2010.)

A.

Plate 8. Simplified (A) and expanded (B) flowcharts setting out a general model to represent foraging agents, for use in individualbased simulations. Solid arrows indicate influences, and dashed lines are mathematical linking functions, which are based on the difference between the current and target states of the agent. Within the central boxes, states and conditions of the agent are shown in boldface type, and behavioral and physiological responses (actions) are in normal type. P, C, and PSM indicate protein, nonprotein energy (carbohydrate + fat), and plant secondary metabolites, respectively, as found in resources (r) or required (target, t). (From Simpson et al. 2010, copyright 2010, with permission from Elsevier.)

Environment Individual Target state

Current state

Mapping functions Behavior Fitness Evolution Social interactions

B. Evolutionary, population & community outcomes

Natural enemies Global risk or attack by other agents

Target states Intake, growth, storage, metabolism (Pt, Ct, PSMt) Mapping functions ( )

Knowledge state

Abiotic influences

Fitness

Nutritional environment (including other agents)

Metabolize and excrete Body composition

Reproduce Immune response

Grow and develop

Nutritional state

Evoke defensive response

Consume resource

Sensory responsiveness Move

Encounter resource of composition Pr, Cr, PSMr Social interactions (attraction, repulsion, alignment)

nine

Applied Nutrition

In chapter 1 we commented on the pervasiveness—or wide-ranging significance—of nutritional biology, and in subsequent chapters have provided examples where nutritional geometry has been used to unravel a range of questions involving diverse organisms. Varied as they are, a factor that is common to these examples is that they all concern fundamental or basic questions, with no intended applied significance. They might, of course, ultimately contribute to some useful application, but the key point about these studies is that they were designed to help us to understand issues in biology, rather than specifically to solve practical problems. Clearly, however, the pervasiveness of nutrition extends beyond fundamental biology, to a wide range of problems that demand scientifically designed management solutions. There are many examples, but among the most important is the need to feed production animals (agriculture), to feed companion animals, to conserve endangered species, and to manage the impacts of the rapidly changing human diet on health. In this chapter we demonstrate how the Geometric Framework has been applied to design nutritional regimes for agricultural and companion animals, and to manage the nutritional environment of endangered species. In the next chapter we focus the methods of nutritional geometry on the modern human diet.

9.1 Domestication As we have seen in chapter 6, the intake targets of animals and rules of compromise they adopt when constrained from reaching their intake target are sculpted over evolutionary time to produce favorable outcomes within the particular ecological circumstances of the animal. In the process of domestication, humans have altered both the characteristics of the animal (through artificial selection) and the environment, thereby disrupting the ancestrally evolved match between the nutritional biology of the animal and its nutritional environment.

148  |  Chapter nine

An important challenge of managing domesticated animals is to ensure that they are provided with foods that enable them to meet their nutritional needs. But how do you decide what the optimal requirements of a domesticated animal are? The usual procedure for wild animals involves two steps (chapter 2). The first is to provide the animal with suitable combinations of nutritionally complementary foods, thereby enabling it to move freely in nutrient space and demonstrate the composition of the preferred diet. In the second step, the performance of the animals on the preferred diet is compared with the performance of other groups of experimental animals that are confined to diets that differ in the balance of the nutrients of interest, and response surface methodology is used to relate performance to diet composition. The same procedures can be useful in deciding on the best diet to feed domesticated animals. The case of domesticated animals is, however, complicated in some respects compared with the examples we have presented to this point. One complicating factor is that nutritional research into domesticated animals usually aims to provide applied recommendations, and the consequences of getting it wrong can be more serious than is the case for studies that aim to understand how the natural diet of animals evolved. On the other hand, as we will see at the end of this chapter, there are also examples where misunderstanding the evolved dietary priorities of animals can have catastrophic applied consequences—for example, in the conservation of endangered species. A second complication concerns the question of which factors we wish the diet to optimize. To this point in the book we have assumed that the ultimate factor that should be taken into account in deciding on the composition of the optimal diet is the evolutionary fitness of the animal. To a rough approximation, this means that the diet composition that supports maximal lifetime reproduction is the optimal diet (e.g., chapter 4). But for domesticated animals things can be more complex, because the concept of “evolutionary fitness” in the normal sense no longer applies. There is, for example, no diet that will maximize the reproductive success of a neutered cat, and yet we still need to aim at providing our pets with suitable foods. An obvious alternative is to aim at diets that maximize the health and welfare of the cat. As comforting as this might seem, however, the situation can be more complex. In addition to the interests of the cat, owners would want to ensure that the diet they provide also caters to other needs that are not directly concerned with improving the animal’s health and welfare; indeed, these other concerns might even adversely affect the health of the animal. For example, compared with herbivores and omnivores, many predators are adapted to “running on empty” (Huey et al. 2001; Arrington et al. 2002)—taking large meals relatively infrequently. But, from the viewpoint of the owner, managing the diet of a pet cat in

Applied Nutrition  |  149

this way would not be ideal. It could encourage the cat into “hunting mode” causing it to roam and, to the distaste of most pet owners, increase the frequency with which garden birds perish. Cat owners usually, therefore, encourage their cats to feed more frequently than is the case for wild cats, and perhaps as a result domestic cats are less active, tend to eat more, and are at risk of becoming obese and suffering associated diseases such as type II diabetes (Martin et al. 2010). There are many other potential conflicts between the health and welfare of the domestic animal and the interests of the owner that need to be resolved in deciding on the optimal diet and feeding regime. Economics plays a role (many cat owners are not in the position, or in the mood, to shoulder the costs of fresh steak or salmon on a daily basis), as do logistics (it is far easier to open a can of cat food that to ensure a continuous supply of fresh meat) and ethical concerns (few would be prepared to maintain a colony of live mice to feed their cat, even if this would optimize its health). Such conflicts can be even more pronounced when considering production animals—animals that are raised with the sole purpose of entering the human food chain. This is especially the case in a commercial setting, where an overriding concern is the time that it takes for the animal to reach marketable size, and diets that reduce this time are preferable to diets that extend it. Clearly, eating a diet that hastens its delivery to the slaughterhouse would not normally be considered a strategy that optimizes the health and welfare of a pig, and yet from the farmer’s point of view such a diet might well be optimal. Likewise, force-feeding geese until they develop fatty liver disease favors the interest of the foie gras producer and gourmand more than that of the goose. On the other hand, when considering commercial interests, there is an overriding optimization criterion, which approaches the dominance that “fitness” enjoys in evolutionary studies: profit. Unlike fitness, however, the drive for profit is often (to an extent) moderated by higher purposes such as animal welfare and reducing impacts on the environment. In some respects, therefore, it is a very different challenge to design optimal diets for domesticated animals than to understand the evolution of diet optimization in undomesticated species. We should not, however, overstate the differences, because on closer inspection it can be seen that these two tasks share much in common. In the same way that conflicts between different optimization criteria must be resolved in designing diets for domesticated animals, the evolution of natural diets is also constrained by a web of trade-offs. In chapter 4, for example, we showed how fruit flies can select a diet that maximizes life span or reproduction but not both, while in chapter 7 we saw how foraging ants have to select a diet that optimizes not their own health but that of the brood that they

150  |  Chapter nine

feed. We have also shown how geometric analysis can be used to model these processes, and now provide an example illustrating how a similar approach can be used to help design diets for domesticated animals.

Feeding Our Companions: Pet Nutrition A recent study by Hewson-Hughes and colleagues (2011) used geometric analysis to test whether domestic cats balance their macronutrient intake, if so to what target composition, and how they respond when confined to a diet that differs from the target composition (i.e., what is the rule of compromise?). The study also tested the extent to which experience of imbalanced diets helps cats to combine nutritionally complementary foods into a balanced diet. This complex set of experiments represents the most thorough study of macronutrient regulation done on a carnivore to date. One complicating factor involved the dimensionality of the study. For most carnivores, the relevant macronutrient space is two-dimensional: protein and fat are the most important macronutrients, whereas carbohydrate is rare in natural foods and plays a lesser role in nutritional regulation (Eisert 2011; chapter 8). In considering carnivorous pets, however, the situation is different because carbohydrate is a significant component of the foods that manufacturers produce for these animals to eat. The study performed by Hewson-Hughes and colleagues therefore involved three-dimensional nutrient spaces, including carbohydrate as well as protein and fat. A further complicating factor is that commercial cat foods come in two very different forms—as moist, tinned foods and as dry pellets. To gain a more comprehensive overview of the relationships between domestic cats and the foods we feed them, the study therefore involved both wet and dry foods. Including both food types in the study had the added advantage of enabling us to test the strength of macronutrient regulation as a factor in the dietary priorities of the cats. This is because these foods are very different in many respects—they differ in water content, presentation (pellets vs. gel), consistency (hard vs. soft), and nutrient composition—and should we observe similar patterns of macronutrient regulation on the two food types, this would suggest that macronutrients play an overriding role in the nutritional priorities of cats. Further complicating the study was the decision not only to test the ability of cats to regulate to a macronutrient target but also to test the extent to which learning plays a role in this process. To achieve this, all cats were tested in three phases. In the “naïve self-selection” phase, they were given a combination of three nutritionally complementary foods, which enabled them to move freely within the nutrient space. At this stage

Applied Nutrition  |  151

they had had no previous experience with these foods, and were therefore considered “naïve” to the challenge that faced them. In the second phase, termed the “monadic phase,” each cat was provided with only one of the three foods at a time, but the foods were alternated daily such that over a 3-day period each cat had been provided with all three foods but on separate days. This provided the cats with an opportunity to learn the consequences of eating for a prolonged period (in this case a full day) foods that contained an excess of protein, fat, or carbohydrate without being able to mix these to provide a balanced diet. In the third phase of the experiment, the cats were once again provided with all three foods—that is, this phase was identical to the first phase except now the cats had had an opportunity to experience the effects of feeding for prolonged periods on imbalanced diets. This was therefore called the “experienced self-selection” phase, and it enabled us to test the extent to which experience helps cats to blend nutritionally complementary foods to compose a balance diet. This study produced several interesting results. First, when provided with food combinations that enabled them to do so, the cats clearly did regulate to an intake target, at which approximately 52% of their energy intake was derived from protein, 36% from fat, and 12% from carbohydrate. Similar results, where fat and protein are the major and carbohydrate the minor energetic components of the selected diet, have been demonstrated for predatory fish (Sánchez-Vázquez et al. 1999; Rubio et al. 2003; Ruohonen et al. 2007) (see chapter 8). Second, when confined to diets that prevented them from achieving the target intake, the cats showed a complex pattern of compromises. On low-protein diets they overate both fat and carbohydrate to approach their target protein intake, but carbohydrate represented more of an impediment to protein gain than did fat. Where protein was low relative to carbohydrate, there was an absolute level of carbohydrate intake that the cats would not exceed, even if this resulted in a large shortfall in protein, whereas the limit on fat intake was more flexible. This suggests that cats are better adapted to deal with variation in the protein to fat balance of their diet than the protein to carbohydrate balance, which probably reflects the low levels of carbohydrate in their predomesticated ancestral diets (see also Eisert 2011). Third, learning played a role in enabling the cats to reach their macronutrient target by mixing their intake from nutritionally complementary foods. Most important, it enabled them to better avoid excessive intakes of carbohydrate, largely through increasing the protein to carbohydrate balance in the selected diet compared with the diet they had mixed in the naïve selection phase. This, again, is consistent with a lack of evolutionary experience of carbohydrate regulation, suggesting that

152  |  Chapter nine

cats have a poorly developed innate ability to avoid carbohydrate excess and rely on individual experience to do so (see chapter 3). While studies have yet to be performed to directly test the consequences of macronutrient imbalance for domestic cats, the experiments of Hewson-Hughes and colleagues (2011) have provided some important information that needs to be taken into account in the decision of what to feed our cats. First and foremost, cats in the laboratory clearly do have a preferred balance of macronutrients in their diet, where carbohydrate is low relative to protein and fat. This has since been confirmed for the diets of free-roaming feral cats (Plantinga et al. 2011). In many cases, the balance of macronutrients in commercially available cat foods differs from the regulated target balance, containing higher carbohydrate levels than was preferred by the cats, this being especially the case for dry, pelleted diets. If given the opportunity, cats are capable of composing the preferred diet by mixing their intake from nutritionally complementary foods, but in reality they are seldom provided by their owners the luxury of several complementary foods. They might, however, seek such foods elsewhere, possibly finding them in the form of birds and other small animals that comprise the urban ecology. When unsuccessful, cats fed commercial foods will be held in a chronic state of nutritional imbalance. This raises welfare concerns, and will also in the long term likely affect cat metabolism and health (Martin et al. 2010).

Feeding Our Food: Diet Optimization for Animal Production Unlike the cat study of Hewson-Hughes and coworkers (2011), an experiment that did directly measure the consequences of nutritional imbalance, in this case for food production animals, was performed by Ruohonen and colleagues (2007) on a salmonid fish species reared in aquaculture, the European whitefish (Coregonus lavaretus). The aims were to use the Geometric Framework to define the optimal macronutrient composition in the diet of the whitefish, to measure the behavioral responses of fish to diets that depart from this composition, and to optimize multiple performance criteria, taking account of economic, environmental, and animal welfare concerns. The protocol for the analysis involved four main stages. The first step was to plot the pairwise relationships between the three macronutrients separately (fat vs. protein, carbohydrate vs. protein, and fat vs. carbohydrate). This showed that, as long as carbohydrate levels were below a damaging threshold (like cats, whitefish suffered when carbohydrate was above a relatively low level), the fish regarded lipid and carbohydrate as interchangeable sources of energy. Nutrients that the animal regards as interchangeable can for practical purposes be regarded as a single re-

Applied Nutrition  |  153

source, enabling us to collapse two axes into one (nonprotein energy) and plot it against protein. The second step was to construct nutrient intake and growth arrays, according to the usual procedures in the GF (fig. 9.1A). Relative to an estimate for the intake target (derived from selfselection studies in other salmonids), whitefish overate nonprotein energy to gain limiting protein when restricted to diets low in protein. They did this to a greater extent than they overconsumed protein to gain limiting nonprotein energy on diets containing a high percentage of protein, possibly reflecting the toxic limits to voiding excess protein as ammonia via the gills. Protein growth was more tightly regulated than lipid growth, as indicated by the vertical elongation of the growth array (fig. 9.1A). Thus, fattiness of the flesh was a result of overeating carbohydrate and lipid to gain limiting protein on diets containing a low percentage of protein, while diets high in protein were associated with low levels of body fat as a result of fish having a limited capacity to overeat protein. As discussed below, this is reminiscent of macronutrient regulation in wild spider monkeys, rodents and also humans (chapter 10), but different from other (terrestrial) predators, which were more willing to overeat protein to gain limiting fat (chapter 8). In the third step we superimposed measures of performance and other response variables onto the intake array. Rather than plot these as a sequence of response surfaces, we have presented them as a series of separate two-dimensional plots (fig. 9.1B), in which the x axis represents the diets (labeled using letters in fig. 9.1A) and the y axes show a range of consequences. The wet weight growth of fish was consistent across the intake array, but this growth was composed of differing amounts of protein and fat, with flesh protein content rising as the dietary percentage of protein increased. Feed efficiency and energy retention efficiency (the ratio of growth to feed consumption in g wet weight and kJ, respectively) showed little change across the intake array, but nitrogen waste (the difference between nitrogen eaten and nitrogen retained as growth) rose with the percentage of protein in the food. Commonly used welfare indicators, such as plasma glucose, plasma cortisol, liver glycogen, and liver somatic index (weight of liver in relation to weight of fish) fell as the percentage of dietary protein rose. (The interpretation of such “welfare measures” can be problematic—a matter we will leave aside for now.) The final step was to choose a set of performance responses to be considered in diet optimization, and to normalize and scale these relative to one another. This involved making a priori judgments about which variables were relevant, what each was “worth,” and, for each variable, whether high or low values were desirable. Fish and fish feed have a market price, and there is a price premium for high-quality flesh. Environmental and welfare costs are harder to measure but can be targets for

154  |  Chapter nine

taxes and licensing restrictions, which do have a measurable cost. We did not conduct a full economic analysis but for illustration chose four scenarios, in which production costs, flesh quality, environmental impact, and animal welfare were prioritized. These are shown in figure 9.1C. It can be seen that the decision functions in the figure have different shapes, with maximum values associated with different diets. Were the fish to have chosen their own diet (gray vertical zones in fig. 9.1C), they would have performed well under all four criteria. They almost maximized the production and welfare goals, while scoring less well on the quality and environmental goals. Fish on diets similar in composition to their intake target ate the minimum amount of protein needed to maximize protein

Non-protein energy intake or lipid growth (kJ per fish)

A. 1800 k c

1600 j

1400

j

1200 1000 j j

800

j

600 400

j

k m kk ml

n ff mf e hl n no c l o e ee r hg igpr i qi p i rrqq p

m

Intake array

m

f fff

m

e l

l

e

e

n

n

g

p

h

ooh

o

g

pp qi rqq

ii i i r

Growth array

200 0

j

j

kk

0

200

400 600 800 1000 1200 1400 1600 1800 Protein intake or growth (kJ per fish)

Fig. 9.1. (A) Nutrient intakes and growth of European whitefish fed one of 15 experimental foods varying in macronutrient composition, represented in a nutrient space consisting of protein and nonprotein energy. Foods are labeled with letters that correspond to formulations given in table 1 of Ruohonen et al. (2007). Duplicate letters represent intake and growth for replicate tanks of fish. A curve (intake array) was fitted through the array of intake points, estimated using a cubic spline regression with 95% confidence limits (dotted lines). The crosses indicate intake and growth for diets that were estimated from data on similar species to be similar in protein to nonprotein energy ratio to that which would be self-selected by the whitefish (i.e., the intake and growth targets).

Applied Nutrition  |  155 B. a. Wet wt. growth (g per fish)

b. Protein growth (kJ per fish) 700

180

600

140

500 400

100 j 1200

k

m

f e

n o

g

p

i

j

f e

n o

g

p

i

g

p

i

n o

g

p

i

f e

n o

g

p

i

f e

n o

g

p

i

3.0 2.0

400

1.0 j

k

m

f e

n o

g

p

i

j

e. Feed efficiency

k

m

f e

n o

f. Nitrogen waste (g per fish)

1.5

6

1.3

4

1.1

2 j

k

m

f e

n o

g

p

i

j

g. Energy efficiency (%)

k

m

f e

h. Plasma glucose 0.76

65

0.74

55

0.72

45

m

d. Body protein:lipid ratio

c. Lipid growth (kJ per fish)

800

75

k

0.70 j

k

m

f e

n o

g

p

i

j

i. Plasma cortisol

k

m

j. Liver glycogen 10

40

8

20

6

0

4 j

k

m

f e

n o

g

p

i

g

p

i

j

k

m

k. Liver-somatic index (%) 40 20 0

j

k

m

f e

n o

(Fig. 9.1. continues on next page)

156  |  Chapter nine C. Weighting index

Production cost valued most highly 0.8 0.4 0 j

k

m

f

e

n

o

g

p

i

e

n

o

g

p

i

e

n

o

g

p

i

m f e n o g Diet (protein increasing to the right)

p

i

Weighting index

Flesh quality valued most highly 0.60 0.45 0.30

j

k

m

f

Weighting index

Environment valued most highly 0.8 0.6 0.4 0.2 j

k

m

f

Weighting index

Fish welfare valued most highly 0.8 0.6 0.4 0.2 j

k

Fig. 9.1. (continued) (B) A selection of response variables superimposed onto the intake array shown in (A) and laid out as flat strips (“performance ribbons”). The x axes of these ribbons are essentially the curved intake arrays from 9.1(A) straightened into a line, with the tick marks and letters along its length indicating the intersection of individual diets with the array. Dots denote the original observations, and the line is the best-fitting function together with its 95% confidence limits (dotted lines) estimated using cubic spline regression. (C) Decision functions for four diet optimization goals with different emphasis. The functions plotted on the y axes (labeled “weighting index”) are the sums of predicted wet weight growth and feed efficiency (production cost), body protein to lipid ratio (flesh quality), nitrogen waste (environment), and blood cortisol (fish welfare) from. 9.1B. These responses were first normalized and then weighted with respect to each other according to one of four optimization goals (see text). (From Ruohonen et al. 2007.)

Applied Nutrition  |  157

growth. Since we made the assumption that flesh quality increases monotonically with lowered fat content, which is probably not the case, it is likely that the fish would have been better on the flesh quality criterion than figure 9.1C suggests. This study illustrates an important advantage of using the Geometric Framework in applied nutrition: that it places not just foods, but the interaction of the animal with those foods, at the center of diet optimization decisions. Taking account of the pre- and postingestive regulatory responses of the animal simplifies the problem of identifying optimal diet formulations, while multicriterion optimization is a matter of deciding how to select and weight normalized response variables, then summing these to arrive at the decision function.

9.2 Wildlife Conservation As we have seen above, an important goal of applied nutrition is to provide a nutritional environment—foods and feeding regimes—that meets the needs of domesticated animals. A related challenge arises in conservation ecology when the natural habitat of wild animals is diminished or altered, and decisions need to be made about providing a nutritional environment that will sustain the species. In some cases this will involve setting priorities for conserving remnants of the natural habitat, or identifying suitable habitats for establishing translocated populations. In other, usually more extreme cases, the natural diet is not available in sufficient quantities and therefore needs to be supplemented or replaced by alternative foods. We now provide examples where the Geometric Framework has been applied in both scenarios: understanding the natural nutritional ecology of wildlife, and supplementing the diet of critically endangered species. These scenarios share in common that the central objective of both is to understand nutritional properties of the natural foods, and how the regulatory systems of the animal interact with these.

Conserving Habitat At one level, it might be argued that there is no need to perform nutrientlevel analyses of the foraging habitats of wildlife, and even less so analyses of the nutritional regulatory responses of the target species. In this line of thinking animals simply eat foods, they are pretty good at knowing what they want and how much to eat, and as long as management decisions ensure sufficient quantities of the relevant foods are conserved, then from the nutritional point of view the job of the conservationist is done. This logic might well work in situations where the evolved link between the distribution of foods in the environment and the animal’s nutritional

158  |  Chapter nine

regulatory system remains intact. However, where this is not the case— for example, where the habitat in which the regulatory systems evolved is diminished or altered—then there might be a mismatch between the animal’s regulatory systems and the altered environment in which these systems need to operate (as we suggest in chapter 10 is the case for humans in modern nutritional environments). To understand, and avoid, or manage this mismatch, we need to think in terms of foods as well as their nutrient and other (chapter 5) components. There are several reasons why nutrient-level analyses are different from food-level analyses and, as we said in relation to ecological competition and niche divergence (chapter 8), both are important. First, a given category of foods can vary tremendously in its composition. For example, as discussed in section 8.3, the body chemical profile of a species of prey animals can vary widely, even in the same developmental stage, depending on such factors as the levels of fat storage; plants, likewise, vary in composition depending, among other things, on available light, water, and nutrients (e.g., Chapman et al. 2003). Therefore, a given food category, for example, the fruits of a particular species of plant, can be represented by more than one nutritional rail, and to understand the consequences for a consumer of introducing, conserving, or eliminating that food from the environment, we need to look beyond foods to their chemical constituents. Second, different foods can have similar nutrient content and might thus be substitutable in the diet of an animal. Again, this is important information for efforts to conserve, restore, or re-create the habitat of a species. Third, nutrient-level analysis can help us to explain variations in the performance of a species in different habitats. For example, if the fecundity and/or longevity of a species differed in different habitats, then we might test the prediction that the habitats differ in the relative availability of protein and nonprotein energy (chapter 4). Finally, as shown in several of the previous chapters, to understand and predict the consequences when an animal is faced with nutritional heterogeneity in the environment, we need to measure not only the distribution of nutrients in the environment but also how the regulatory systems of the animal respond to that distribution. Particularly useful information would be the position of the intake target, the ability to regulate to the intake target in different circumstances, and the likely feeding response when the animal is unable to reach the target (i.e., the rule of compromise). A recent example where nutritional geometry was used to understand the habitat requirements of a priority conservation species was Felton and colleagues’ (2009a, b) study of Peruvian spider monkeys (Ateles chamek) in their natural forested habitat in Bolivia. Felton first spent 5 months habituating the monkeys. When they were sufficiently comfort-

Applied Nutrition  |  159

able with her presence, she followed individual monkeys throughout their feeding day (from dawn to dusk), recording everything they ate during the 12-hour observation period and estimating the amount of each food that was eaten. Representative samples of each food item were collected and later taken back to the laboratory, and their nutritional compositions analyzed. From these data, 38 daily intakes of protein, fat, and carbohydrate were estimated and analyzed using the Geometric Framework. Several interesting results emerged. First, although primarily frugivorous, the monkeys ate a range of foods that differed widely in their macronutrient composition. When available, they specifically targeted ripe figs (Ficus spp.), but when these were not available in sufficient quantities, they mixed their intake from foods with high and low protein to nonprotein energy balances, thereby moving in a zigzag pattern through nutrient space (as illustrated schematically in fig. 2.1B). Interestingly, whether the monkeys were eating figs or composing their diet from diverse foods, the macronutrient balance of their daily diet did not change. This suggests that figs are a nutritionally balanced food for the monkeys, and when they are not available the monkeys compose a balanced diet through mixing nutritionally complementary foods. Another pattern of feeding was observed during the late wet season, when ripe figs were scarce but other fruits, particularly high in sugars and fats, were abundant. In this period, the monkeys ate very similar amounts of protein as when eating figs or diverse complementary foods (this most likely being the target level for protein), but in so doing ate considerably more carbohydrates and fat than when eating figs or mixing complementary foods. The resulting intake array was thus a vertical line, with little variation on the protein axis compared with the axis for nonprotein energy (fig. 9.2). This is a more extreme form of the rule of compromise that has been observed in experimental studies of humans, which, as we will show in chapter 10, might be an important contributory factor to the epidemic of obesity that has swept many countries in recent decades. The study by Felton and coworkers (2009a, b) has several implications for understanding the habitat needs of Peruvian spider monkeys. First, the results provide an estimate of the position of the macronutrient intake target of the monkeys, and suggest that figs (Ficus spp.) are nutritionally balanced with respect to that target. This underscores the importance of regulating the commercial harvesting of Ficus timber in the habitat of spider monkeys. Second, the study suggests that the monkeys are not reliant on balanced foods such as figs but can compose a balanced diet through mixing nutritionally complementary foods. This shows that the monkeys are to some extent flexible in their foraging behavior, and also provides an indication of which combinations of foods are important. Third, when exposed to foods with a low protein to nonprotein

Carbohydrate + lipid eaten (MJ/day)

160  |  Chapter nine

6 5 4 3 2 1 0

0

1 2 3 Protein eaten (MJ/day)

4

Fig. 9.2. Geometric plot of protein and nonprotein energy intakes during 38 day-long observations of individual Peruvian spider monkeys. Each point summarizes the intake for one “monkey day.” The low variance around the vertical line, compared with the much higher variance along the y axis, suggests that protein intake was strongly prioritized by the monkeys, despite variation in foods available in the cafeteria of their forest home. (From Felton et al. 2009a.)

energy balance, spider monkeys will defend their protein intake but overeat nonprotein energy. This helps us to predict how these monkeys would respond to distortions in the macronutrient composition of their foraging environment. Specifically, it suggests that prolonged exposure to foods with a low protein to nonprotein energy balance is unlikely to result in protein deficit, but rather a surplus of nonprotein energy and, conversely, that these monkeys are unlikely to overeat protein to meet their requirements for nonprotein energy. It remains to be determined, however, whether being constrained to suffer macronutrient imbalance would affect the monkeys and if so, how. It is easy to imagine a situation in which energy shortage, as might be expected to result from high-protein diets, could be a limiting factor to wild primates. But might a sustained energy surplus, as predicted on lowprotein diets, also adversely affect spider monkeys? Indirect evidence raises a cautionary note. Like their human cousins, nonhuman primates are prone to obesity under certain conditions, which include exposure to readily available foods high in nonprotein energy (Hansen 2001). While obesity is most common in captive primates (e.g., in zoos), it has also been observed in free-ranging populations of baboons (Altmann et al. 1993) and macaque monkeys on the island of Cayo Santiago off the coast of Puerto Rico (Schwartz et al. 1993). In both cases obesity was associated not with natural foods but with access to garbage dumps (baboons)

Applied Nutrition  |  161

or with the management practice of supplementing the natural diet with primate chow (macaques). At the very least, all direct and indirect evidence warrants careful consideration of the foods that should be conserved in the habitat of spider monkeys. It also provides information regarding the required macronutrient composition of any supplementary or captive diets for this species, should its habitat become too degraded for unmanaged survival. Can the prioritization of protein observed in spider monkeys and humans (chapter 10) be generalized to other primates? Work by Rothman and colleagues (2008, 2011) suggests that the answer is no. Detailed observations were performed of feeding in a troop of wild mountain gorillas (Gorilla beringi) in Bwindi Impenetrable National Park, Uganda (Rothman et al. 2008), recording food choice and amounts eaten over a full annual cycle. By analyzing samples of each food for its nutritional composition, Rothman and coworkers were able to perform both a food-level and a nutrient-level analysis. An interesting aspect of Rothman’s results was the seasonal variation in the diet. The main foods of gorillas were leaves and fruits, but in 8 months of the year intake was dominated by leaves with only approximately 10% of the diet coming from fruits. However, in two separate 2-month periods (February to March and June to July), when the fruits of some tree species were abundant, fruit comprised on average 40%–50% of the diet. Geometric analysis of the data (Rothman et al. 2011) showed some interesting results. First, the seasonal variation in available foods provided a natural experiment for assessing the responses of the gorillas to variation in the macronutrient composition of their diet. The leaves they ate contained a much higher ratio of protein to nonprotein energy than the fruits, and consequently for two 2-month periods of the year, when both fruits and leaves were abundantly available, these gorillas had access to a wide area within a protein / (fat + carbohydrate) nutrient space. Could this be regarded as a self-selection experiment for establishing the position of the intake target (as in fig. 2.1B)? The geometric analysis of Rothman and collaborators suggests that this is, indeed, the case (fig. 9.3). In the months when significant amounts of both fruit and leaves were eaten, the gorillas selected a diet with approximately 19% of energy derived from protein. There are strong indications that this represents the intake target for mountain gorillas. First, the gorillas selected this diet in a situation where they had access to a wide region of nutrient space, and this was true of silverbacks (adult males), adult females, and juveniles alike. Second, the selected macronutrient balance is similar to the balance of 17% protein selected by a larger-brained species of great ape on which manipulative experiments have been performed, Homo sapiens (chapter 10), and to the balance recommended by the American Heart

mi rilla

Hu ma

20

10

Le

0

Go

n in t ak

e

30

Fru it p er i od

Carbohydrate + lipid eaten (MJ/day)

lk

162  |  Chapter nine

0

2

iod er

p af

4 6 8 10 12 Available protein eaten (MJ/day)

14

Fig. 9.3. Geometric plot of mean daily intakes (± standard deviation) by gorillas during predominately fruit- (triangles) and leaf-eating months (circles). Black-filled symbols represent adult males, white symbols adult females, and gray symbols juveniles. Nutritional rails indicate the balance of available protein (AP) to nonprotein energy (NPE) in fruit and leaf periods, the balance of AP to NPE in midlactation gorilla milk, and the recommended balance of AP to NPE for humans to maintain healthy diets. The gray area spans the upper and lower recommended limits of AP to NPE for humans. (From Rothman et al. 2011.)

Association for humans to achieve optimal health (15% protein). Third, the macronutrient balance selected in fruit-eating periods is similar to the balance in a food that has evolved specifically to support growth and development in gorillas: midlactation gorilla milk (approximately 16% protein, Whittier et al. 2010). By contrast, in the periods when leaves dominated the diet, protein comprised 30.5% of dietary energy. This is close to the maximum of 35% recommended for humans to prevent chronic illness, and comparable to high-protein weight-loss diets such as the Atkins diet (around 30%). Overall, this evidence suggests that the mountain gorillas in Bwindi are able to reach their intake target (approximately 19% of energy from protein) during 4 months of the year when edible fruits are abundantly available, and in the remaining 8 months they are restricted to a diet of leaves with surplus protein concentration (30.5% protein). This provided a natural equivalent of a standard geometric experiment in which a range of foods is provided to assess the position of the intake target (fig. 2.1B), and then intakes on a restricted imbalanced diet are compared with the

Applied Nutrition  |  163

target intake to assess the rule of compromise (fig. 2.1D). Following this logic, the analysis showed that in the months when the gorillas fed predominately on leaves, their intake of nonprotein energy did not differ from the estimated target intake, but protein intake significantly exceeded the target value for this macronutrient. It therefore appears that the rule of compromise for the Bwindi mountain gorillas is to defend the intake of nonprotein energy, even if this involves ingesting a surplus of protein. This contrasts with humans and Peruvian spider monkeys, which defend their protein intake and in so doing over- or undereat nonprotein energy. Several interesting issues arise from these findings. First, they call into question the widespread assumption that the main nutritional limitation on terrestrial herbivores is shortage of protein compared with protein and other nutrients—for gorillas, the opposite was the case. Second, they give rise to the question of whether there are fitness costs to this threatened species (approximately 800 individuals remain worldwide) of ingesting surplus protein for eight months of the year. Alternatively, a long history of coping with seasonal diets might have selected for the ability to excrete the surplus protein with minimal cost. The fascinating question remains of whether this capability, if it exists, is associated specifically with the amino acid profile of plant tissues, or whether it applies for proteins more generally. Third, the results show that it cannot be assumed that protein prioritization, and the associated vulnerability to surplus energy intake in low-protein environments, can be generalized across primates. This underscores the need in conservation to take into account on a case-by-case basis the details of nutritional environments and the ways that the regulatory systems of animals respond to them.

Supplementary Feeds In extreme cases where suitable habitat is no longer available to conserve endangered species, it is necessary to supplement the natural diet with provisioned foods. In some respects, the challenges in doing this are similar to the challenges discussed above that are encountered in devising feeding regimes for domesticated animals. An example of where the Geometric Framework has been applied to designing supplementary feeds for a critically endangered animal concerns one of the rarest birds in the world, the New Zealand kakapo (Strigops habriptila). Once widespread throughout New Zealand, this vulnerable flightless parrot was decimated by the colonization of New Zealand by humans and associated introduced mammals including rats, domestic cats, and mustelids (ferrets, stoats, and weasels) (Powlesland et al. 2006). Between 1980 and 1997 the New Zealand Department of Con-

164  |  Chapter nine

servation transferred all known surviving kakapo to predator-free islands. This almost certainly saved the species from extinction, but narrowly so: in 1995 the world population numbered a mere 51. Protection from predation has allowed the population to increase over the ensuing 15 years, to the point where at the time of writing kakapo number 131. A remaining challenge is to increase the rate of reproduction among kakapo. On average, these parrots breed only once every 2 to 5 years (Elliott et al. 2001), a statistic that is not comforting for those concerned about the survival of this species. A reliable predictor of whether in a given year the birds will breed is the abundance of fruits of podocarp trees such as rimu (Dacrydium cupressinum), which are the principal food used by females for rearing the young. This strongly suggests that the decision by female kakapo of whether or not to breed is linked to nutrition. Consequently, in 1989 the New Zealand Department of Conservation introduced a supplementary feeding program in the hope that this would help to trigger breeding in the interim years when rimu fruits were not abundant. However, the effects of the supplementary provisioning on the rate of reproduction, if any, have been subtle (Elliott et al. 2001). There are, furthermore, indications that the feeding program had the unwanted side effect of skewing the sex ratio toward males (Clout et al. 2002), and in the case of one kakapo it has been linked to the development of obesity (Powlesland and Lloyd 1994). With the aim of better understanding the relationship between reproduction and the abundance of rimu fruits, we have used the Geometric Framework to compare the composition of rimu fruits and “muesli,” a supplementary feed that was historically used by the Department of Conservation (Raubenheimer and Simpson 2006). Our analysis revealed some potentially important discrepancies in the composition of rimu and this supplementary feed. First, the protein to lipid ratio was higher in rimu fruits (1.6) than in the supplementary feed (0.44). Consequently, kakapo feeding largely on supplementary feed would need to underingest protein and/or overingest lipid compared with the nutritional composition of a diet of rimu berries. Might this help to explain the case of obesity that has been documented in a supplementary-fed kakapo? Second, the ratio of protein (and total macronutrients) to calcium was much higher in the supplementary feed than in rimu (fig. 9.4). Our analysis showed that if a kakapo ate 100 g of the supplementary feed it would ingest only 38% of the amount of calcium and a massive 580% of the amount of macronutrients compared with 100g of rimu fruits. If it ate an amount of supplementary feed that gave it the same amounts of macronutrients it would get from 100g of rimu, it would ingest only 6.3% of the level of calcium when feeding on supplementary food compared with what it would on rimu. Alternatively, to have the same calcium intake on the two foods a kakapo

Applied Nutrition  |  165 180

120 100 80

Ca 1 ––––––––––

60

P+L +++++++

Protein + lipid eaten (g)

Supplement

140

40

Ca 3 ––––––––––––––––

20 0

P+L ++++++++++++++++++++++++

2

160

0

0.5

Green rimu fruit

1.0 1.5 Calcium eaten (g)

2.0

Fig. 9.4. Geometric model of macronutrient-calcium interactions for kakapo parrots eating green rimu fruit or a supplementary food (muesli). The circle denotes the intake achieved by eating 100 g of green rimu fruit. The squares denote the nutrient intakes achieved by eating the same amount (i.e., 100 g) of muesli (1), by eating muesli to the same calcium intake (2), and by eating muesli to the same protein intake (3). (After Raubenheimer and Simpson 2006.)

would need to ingest an enormous 15 times the amount of macronutrients when feeding on the supplementary food compared with rimu. Given the strong influence that macronutrients exert on the regulation of food intake, it is most likely that kakapo would not overeat macronutrients to any large extent, and would therefore suffer a shortage of calcium when feeding on the supplementary food. Since calcium is critically important for breeding in birds because it is a key component of eggshells (Tordoff 2001), it is possible that the high macronutrient to calcium balance of the supplementary feeds contributes to their ineffectiveness in triggering more frequent breeding among kakapo. These insights are currently being developed for incorporation into supplementary feeding regimes for the remaining kakapo. We are, however, happy to report that a series of naturally induced breeding seasons has prevented us from trialing the modified recipes.

9.3 Conclusions In this chapter we have shown how the Geometric Framework can help in designing foods for domesticated animals and in providing information for the conservation of endangered species. In each case, the key to

166  |  Chapter nine

this approach is to depict the animal, the environment, and the interaction between animal and environment in terms of nutrients. In conservation ecology, these nutrient-level depictions should not be considered an alternative to analyses that focus on foods—both food- and nutrientlevel analyses are important. We now turn to a third application of nutritional geometry, namely, to understand and manage the nutritional biology of our own species.

ten

The Geometry of Human Nutrition

In previous chapters we have shown how the Geometric Framework can be applied to a range of issues spanning the functional biology of animals, as well as higher-level phenomena such as the collective behavior of groups and community ecology. In this chapter we apply the geometric approach to an analysis of a key aspect of human nutrition, the topical subject of human obesity. Because of its importance, the link between nutrition and obesity is an area that has attracted a tremendous amount of research, which has exposed a complex network of contributing factors. Our aim is not to provide a comprehensive review of the topic, but rather to illustrate how the logic of geometric analysis can illuminate the subject in new ways. We will also build on a point that we raised in chapters 2 and 9: namely, that natural selection equips animals to regulate to nutritional targets within a defined range of nutritional environments, but beyond this range the outcomes might be very different.

10.1 The Modern Human Nutritional Dilemma It is conservatively estimated that more than 1 billion people worldwide are overweight or obese. This number is increasing, notably among the young, and the associated disease burden is immense (Must et al. 1999; Björntorp 2001; Hill et al. 2003). Figure 10.1A plots the relative risk of dying prematurely as an adult against body mass index (BMI), which approximates to body fatness and is calculated as body mass in kilograms divided by the square of height in meters. Clinicians categorize adults as underweight if they have a BMI of less than 18.5, as overweight if they have BMI values between 25 and 30, and as obese if they exceed 30. The curve is U-shaped, with the risk of dying prematurely increasing at both low and high values of BMI, and the target zone for health and longevity lying in between. The relationship between body fat content and risk of premature death in humans is very similar to what we have observed in the locust (fig. 10.1B). This is the same species that we showed in chapter 2 defends a

168  |  Chapter Ten A. Humans

USA 65%

2.8

Women Men

1.8 1.6 1.4 1.2

Obese

2.2

Overweight

Target zone for health and longevity

2.4 Underweight

Relative risk of death

2.6

2.0

30%

1.0

.0

.9

≥4 0

.9 35

.0

–3 9

.9 32

.0

–3 4

.9

–3 1

.0 30

28

.0

–2 9

.9

.4 26

.5

–2 7

.9 25

.0

–2 6

.4 23

.5

–2 4

.9 22

.0

–2 3

.4

–2 1

20

.5

–2 0

18

.5