The Importance of Species: Perspectives on Expendability and Triage 9781400866779

Citation preview

THE IMPORTANCE OF SPECIES

m

THE IMPORTANCE OF SPECIES

m Perspectives on E x p e n da b i l i t y a n d Tr i ag e

Edited by

Peter Kareiva and Simon A. Levin

PRINCETON UNIVERSITY PRESS

Princeton and Oxford

Copyright 䉷 2003 by Princeton University Press Published by Princeton University Press, 41 William Street, Princeton, New Jersey 08540 In the United Kingdom: Princeton University Press, 3 Market Place, Woodstock, Oxfordshire OX20 1SY All Rights Reserved Library of Congress Cataloging-in-Publication Data The importance of species : perspectives on expendability and triage / edited by Peter Kareiva and Simon A. Levin. p. cm. Papers presented at a symposium held in honor of Robert Treat Paine, upon the occasion of his retirement from the University of Washington. Includes bibliographical references. ISBN 0-691-09004-1 (alk. paper) — ISBN 0-691-09005-X (pbk. : alk. paper) 1. Conservation biology—Congresses. 2. Species diversity— Congresses. 3. Endangered species—Congresses. 4. Biological diversity conservation—Congresses. I. Kareiva, Peter M., 1951– II. Levin, Simon A. QH75 .I4 2003 333.95⬘16—dc21

2002025137

British Library Cataloging-in-Publication Data is available This book has been composed in Palatino Printed on acid-free paper. ⬁ www.pupress.princeton.edu Printed in the United States of America 10

9

8

7

6

5

4

3

2

1

Contents

m

Contributors ix Preface xiii Foreword xv

Pa r t I USING EXPERIMENTAL REMOVALS OF SPECIES TO REVEAL THE CONSEQUENCES OF BIODIVERSITY DEPLETION P. Kareiva and S. A. Levin 1 1. Native Thistles: Expendable or Integral to Ecosystem Resistance to Invasion? S. M. Louda and T. A. Rand 5

vi

contents

2. The Overriding Importance of Environmental Context in Determining the Outcome of Species-Deletion Experiments B. A. Menge 16 3. Species Importance and Context: Spatial and Temporal Variation in Species Interactions C.D.G. Harley 44 4. Effects of Removing a Vertebrate versus an Invertebrate Predator on a Food Web, and What Is Their Relative Importance? T. W. Schoener and D. A. Spiller 69 5. Understanding the Effects of Reduced Biodiversity: A Comparison of Two Approaches J. T. Wootton and A. L. Downing 85 Pa r t I I THE ANTHROPOGENIC PERSPECTIVE P. Kareiva and S. A. Levin 105 6. Models of Ecosystem Reliability and Their Implications for the Question of Expendability S. Naeem 109 7. Predicting the Effects of Species Loss on Community Stability D. Doak and M. Marvier 140 8. One Fish, Two Fish, Old Fish, New Fish: Which Invasions Matter? J. L. Ruesink 161 9. Ecological Gambling: Expendable Extinctions Versus Acceptable Invasions M. J. Wonham 179 10. Rarity and Functional Importance in a Phytoplankton Community D. E. Schindler, G. C. Chang, S. Lubetkin, S.E.B. Abella, and W. T. Edmondson 206

contents

vii

11. Community and Ecosystem Impacts of Single-Species Extinctions D. Simberloff 221 Pa r t I I I LINKAGES AND EXTERNALITIES P. Kareiva and S. A. Levin 235 12. Social Conflict, Biological Ignorance, and Trying to Agree Which Species Are Expendable E. G. Leigh Jr. 239 13. Which Mutualists Are Most Essential? Buffering of Plant Reproduction against the Extinction of Pollinators W. F. Morris 260 14. The Expendability of Species: A Test Case Based on the Caterpillars on Goldenrods R. B. Root 281 15. An Evolutionary Perspective on the Importance of Species: Why Ecologists Care about Evolution S. R. Palumbi 292 16. Recovering Species of Conservation Concern—Are Populations Expendable? M. Ruckelshaus, P. McElhany, and M. J. Ford 305 17. Virus Specificity in Disease Systems: Are Species Redundant? A. G. Power and A. S. Flecker 330 Conclusion P. Kareiva and S. A. Levin 347 References 353 Index 415

Contributors

m

Sally E. B. Abella, Department of Zoology, University of Washington, Seattle, WA 98195-1800. Current Address: King County Water and Land Resources, 201 S. Jackson St., Suite 600, Seattle, WA 98104 Gary C. Chang, Department of Zoology, University of Washington, Seattle, WA 98195-1800. Current Address: Department of Plant, Soil, and Entomological Sciences, University of Idaho, Moscow, ID 83844 Dan Doak, Department of Biology, University of California, Santa Cruz, CA 95064 Amy L. Downing, Department of Ecology and Evolution, University of Chicago, 1101 E. 57th Street, Chicago, IL 60637-1573. Current Address: Department of Zoology, Ohio Wesleyan University, Delaware, OH 43015 W. T. Edmondson (deceased), Department of Zoology, University of Washington, Seattle, WA 98195-1800 Alexander S. Flecker, Department of Ecology and Evolutionary Biology, Cornell University, Corson Hall, Ithaca, NY 14853

x

contributors

Michael J. Ford, National Marine Fisheries Service, Northwest Fisheries Science Center, 2725 Montlake Boulevard East, Seattle, WA 98112-2097 Christopher D. G. Harley, Department of Zoology, University of Washington, Seattle, WA 98195-1800. Current Address: Hopkins Marine Station, Oceanview Boulevard, Pacific Grove, CA 93950 Egbert Giles Leigh, Jr., Smithsonian Tropical Research Institute, Apartado 2072, Balboa, Panama Svaˇta M. Louda, School of Biological Sciences, University of Nebraska, Lincoln, NE 68588-0118 Susan Lubetkin, Quantitative Ecology and Resource Management, University of Washington, 407 Bagley Hall, Seattle, WA 98195 Michelle Marvier, Department of Biology, Environmental Studies Institute, Santa Clara University, Santa Clara, CA 95053 Paul McElhany, National Marine Fisheries Service, Northwest Fisheries Science Center, 2725 Montlake Boulevard East, Seattle, WA 98112-2097 Bruce A. Menge, Department of Zoology, Oregon State University, Cordley Hall 3029, Corvallis, OR 97331-2914 William F. Morris, Department of Biology, Duke University, Durham, NC 27708-0338 Shahid Naeem, Department of Zoology, University of Washington, 24 Kincaid Hall, Seattle, WA 98195-1800 Stephen R. Palumbi, Department of Biological Sciences, Hopkins Marine Station, Stanford University, Oceanview Blvd., Pacific Grove, CA 93940 Alison G. Power, Department of Ecology and Evolutionary Biology, Cornell University, Corson Hall, Ithaca, NY 14853 Tatyana A. Rand, School of Biological Sciences, University of Nebraska, Lincoln, NE 68588-0118 Richard B. Root, Department of Ecology and Evolutionary Biology, Cornell University, Corson Hall, Ithaca, NY 14853 Mary Ruckelshaus, National Marine Fisheries Service, Northwest Fisheries Science Center, 2725 Montlake Boulevard East, Seattle, WA 98112-2097

contributors

xi

Jennifer L. Ruesink, Department of Zoology, University of Washington, Seattle, WA 98195-1800 Daniel E. Schindler, Department of Zoology, University of Washington, Seattle, WA 98195-1800 Thomas W. Schoener, Section of Evolution and Ecology, University of California, 1 Shields Avenue, Davis, CA 95616-8755 Daniel Simberloff, Department of Ecology and Evolutionary Biology, University of Tennessee, Knoxville, TN 37996 David A. Spiller, Section of Evolution and Ecology, University of California, 1 Shields Avenue, Davis, CA 95616-8755 Marjorie J. Wonham, University of Washington, Department of Zoology, Box 351800, Seattle, WA 98915-1800. Current Address: University of Alberta, Centre for Mathematical Biology, CAB 632, Edmonton, AB, Canada T6G 2G1 J. Timothy Wootton, Department of Ecology and Evolution, University of Chicago, 1101 East 57th Street, Chicago, IL 60637-1573

Preface

m

A born naturalist, Robert Treat Paine has firmly established himself as one of the leading ecologists of this—or any—century. His ability to carry out unparalleled and innovative field experiments, informed by solid natural history and deep insights about the structure and functioning of ecological communities, has made his work among the most influential in the development of ecology over the past thirty years. He selected intertidal rocky shores as the ideal system for testing seminal questions about community dynamics, and his long-term commitment to elucidating ecological relationships in the intertidal has provided one of the benchmark ecological systems. In particular, he has illustrated the power and necessity of experimental approaches to community organization and has combined this approach with powerful theory to elucidate food web complexity, trophic dynamics, patch dynamics, and community energetics. R. T. Paine has been not only a great scientist but a model mentor. His students have been his primary concern, and he has inspired them (and others) through example and guided them to brilliant careers. Community ecology in general, and intertidal ecology in particular, have been deeply influenced by him. Robert Paine has been often and justly honored—as one of the few ecologists in the National Academy of Sciences and the American Academy of Arts and Sci-

xiv

preface

ences; by selection as the third Tansley Lecturer of the British Ecological Society; and through his receipt of the MacArthur Award of the Ecological Society of America, the Wright Award of the American Society of Naturalists, and the Ecology Institute Prize. His influence, always great, continues to grow.

Foreword

m

Not everyone is enthusiastic about protecting biodiversity. The practice of conserving species and biodiversity requires major societal sacrifices, and in many cases cannot be accomplished without substantial economic costs. Given these costs, it is not surprising that members of the public often ask conservationists whether every species must be protected. Of course, many of the hard decisions surrounding conservation have nothing to do with science—they represent a choice among values and are of a political, ethical, and philosophical nature. Still, there is a need for scientists to respond to questions about the consequences of losing particular species or segments of biodiversity. In this book, leading ecologists, evolutionary biologists, and conservation biologists explore questions about the “value” (or conversely the “expendability”) of species. The book provides an in-depth scientific exploration of what kinds of evidence and research can inform society about the choices it must make regarding species extinctions. Contributions were initially presented as talks at a symposium held to honor Robert Treat Paine, on the occasion of his retirement as a Professor of Zoology at the University of Washington. Much of Bob’s research focus, and the focus of those inspired by him, has been on species interactions. Consequently, whereas several other recent books

xvi

foreword

address the value of biodiversity from the perspective of “ecosystem services” (Daily 1997, Baskin 1997), this monograph emphasizes how the value of species is revealed through the particulars and generalities of species interactions. Although “biodiversity” is the rallying call of many conservation groups, the protection of biodiversity often comes down to individual species and their interactions. When the public at large questions conservation efforts, the challenge commonly involves queries about why we should care about a particular species such as the spotted owl or the snail darter. Although ecology and evolutionary biology are not in the business of offering reasons for caring, science can make clear what is likely to happen if different species are eliminated. Indeed, one of the dominant activities of ecologists, following the lead of Bob Paine, has been research that involves the experimental removal of species to assess the response of the surrounding community to the loss of particular organisms. Challenging biologists to assess the “expendability” of species pushes them to the limits of their science. Discussions about species loss in the popular press often resort to metaphors. According to the “rivet metaphor,” for example, losing species is like popping rivets out of an airplane: at first losing a few rivets appears inconsequential, but eventually, if too many rivets are popped out, the plane is not one on which anyone would like to be a passenger (Ehrlich and Ehrlich 1981). Such metaphors provide powerful imagery and have played an important role in sensitizing people to the problem of species extinction. This monograph takes the next step, moving beyond metaphors to struggle substantively with the difficult question of what it can mean to lose a species from a community, however unimportant that species might appear on casual (or even careful) observation. Bob Paine has used a mix of experiments, theory, and natural history to illuminate ecology, and the contributors to this book use a similar blend to examine the importance of species.

THE IMPORTANCE OF SPECIES

m

Pa r t I

m USING EXPERIMENTAL REMOVALS OF SPECIES TO REVEAL THE CONSEQUENCES OF BIODIVERSITY DEPLETION

For decades, Bob Paine has exhorted ecologists to conduct field experiments in which species are removed and the communitywide responses to those removals noted. The lessons learned from these manipulations of natural communities have been impressive and have reshaped our understanding of ecological systems. Given the central role of experiments in community ecology, it is surprising that the large body of results from species-removal experiments has been so neglected in the debate surrounding the importance of biodiversity. After all, there is no more straightforward way of examining the importance of a species than to remove it. Thus, experimental investigations of species interactions should speak directly to the question of species expendability. In the following five contributions,

2

pa r t i

we find compelling evidence that the loss of even a single species can have severe ramifications for ecosystem structure and functioning. But we also find clear limitations of the experimental approach as an indicator of species importance. Louda and Rand begin by describing the interactions between native thistles and the insects that feed on those thistles. They make a persuasive case that insects can play a major role in limiting the reproductive success of thistles. Pointing out that the insects feeding on native thistles also feed on noxious invasive thistles, Louda and Rand argue that native thistles, by harboring populations of herbivores, provide resistance to invasion by exotic thistles. The argument is logical and is based on careful natural history and a detailed understanding of insect-plant interactions. Missing, however, is the definitive experiment that uses species removal to test whether weedy thistles have an easier time invading areas where native thistles have been removed. Louda and Rand provide an excellent example of why we have come to value species-removal experiments as a research tool; without such experiments, even the best quantitative and natural history studies leave us with uncertainty about the role a species plays. Although we suspect that Louda and Rand are correct in their assessment regarding the importance of native thistles to community resistance, we cannot be convinced entirely until the critical removal experiment is performed. The four remaining contributions in this section report the results of actual removal experiments. The chapters by Menge and Harley both deliver a message of “context dependency”: that a single species can appear inconsequential in one experiment but of central importance in a different removal experiment. Harley makes this point by modifying a measure of interaction strength advocated by Paine (1992) and calculating this new metric of interaction strength for more than 150 experimentally perturbed pairs of species. It is a clich´e to say that organisms, populations, and communities vary according to time and place. Harley goes beyond this clich´e to quantify what this variation is likely to mean for a species’ importance as determined by measurements of interaction strength in one community or context. Specifically, Harley finds that the measure of interaction strength for a species pair in one context explains, on average, only 37% of the variation in the interaction strength observed in a different year or place. Menge remarks on the same phenomenon but goes on to suggest that the variation in interaction strengths observed in marine systems may well be predicted on the basis of large-scale variation in the physical environment. Menge predicts, for example, that the removal of preda-

experimental manipulations

3

tors will be especially important in highly productive upwelling systems, which tend to be characterized by strong top-down interactions. Schoener and Spiller synthesize the results of several years of removal experiments involving a dominant lizard species and a dominant spider species in the Bahama Islands. The high point of their analysis is its anticipation that based on trophic position, removing lizards should have a much greater impact than the removal of spiders. Secondly, whereas much ado is made about “ecosystem function” when discussing the value of biodiversity, Schoener and Spiller add considerable meat to the concept. In particular, they show that the impact of lizard removal on increased herbivore density carries through to primary production, noting that leaf damage increases by threefold in areas where lizard densities have been reduced. Finally, Wootton and Downing argue that the popular approach to assessing typical responses to reductions in biodiversity misses the clearest lesson of the last twenty years of species removal experiments: that the effect of a species-removal is highly idiosyncratic. As an alternative, they suggest a hybrid approach in which targeted species removals are combined with general diversity manipulations. Targeted removals are championed as invaluable, because they provide an experimental manipulation that can lead to predictions about the consequences of particular extinctions. The chapters by Wootton and Downing and by Schoener and Spiller illuminate a path for future research that is likely to be especially valuable for those interested in the conservation of biodiversity. These authors redefine popular questions about biodiversity in a way that is amenable to ecological theory. Schoener and Spiller point out that “expendability” is largely a question for values but that conservation is often motivated by an urge to “keep things the same.” By understanding community dynamics, we can predict the role of species in “keeping things the same”—exactly as Schoener and Spiller were able to understand why lizard removal altered ecosystem function but spider removal did not. Wootton and Downing argue that the “average effect” of biodiversity is not as interesting as the effect of particular losses of biodiversity in particular settings. Thus, instead of asking whether every species is irreplaceable or whether every reduction of biodiversity represents environmental degradation, they seek a predictive theory that allows one to understand why the losses of some species have large impacts while the losses of others have negligible ecological impacts. Whereas many ecologists lose their way when discussing “expendability,” Schoener, Spiller, Wootton, and Downing derive clarity from their focus on species interactions and the consequences of targeted species removals.

4

pa r t i

At its most basic level, research involving species removal is significant simply because it allows ecologists to accumulate data regarding species interactions. Now that we have accumulated a large suite of results from targeted removal experiments, contributors to this book are able to synthesize the consequences of removal experiments in a manner that sheds light on the likely consequences of species extinctions.

Chapter 1

m Native Thistles: Expendable or Integral to Ecosystem Resistance to Invasion? Svaˇta M. Louda and Tatyana A. Rand

One way of addressing the question of whether some species are expendable is to ask what role, if any, a minor species, even one that seems obnoxious, plays in the functioning of its community. Thistles (Cirsium spp.) are prickly plants native to North America that are numerically minor and are often considered unattractive or undesirable. So, thistles might be considered expendable. Yet, can we assume that such minor, seemingly undesirable species can be eliminated without disrupting important interdependencies or losing key ecological services? Our long-term studies of thistle-insect interactions are beginning to provide evidence that even such species may play important, unexpected roles in ecological dynamics and in economic welfare. These studies suggest that determining the cost of losing a species requires criteria other than relative abundance and general attractiveness (see Root this volume). In this chapter, we summarize the natural history of the native Cirsium species, which we study in the upper Great Plains. Then, we

6

chapter 1

briefly review experimental evidence that the native insects that feed on native thistles can restrict their abundance and weediness. Finally, we present new observational data suggesting that these native insects, moving over from a native species such as tall thistle (Cirsium altissimum (L.) Spreng.), are likely involved in limiting the invasion of bull thistle, Cirsium vulgare (Savi) Ten., in the tallgrass prairie region of Nebraska. Bull thistle is a Eurasian species that is highly invasive elsewhere (Austin et al. 1985; Randall 1991; Julien and Griffiths 1998; Olckers and Hill 1999) and often becomes expensive to control in agronomic regions around the world. Such invasions of nonindigenous species can present major economic and ecological threats to ecosystem structure and functioning (Mooney and Drake 1986; Drake et al. 1989; Simberloff et al. 1997). We hypothesize that our study represents a case of a numerically minor species that, acting as a reservoir of native insects, provides an ecologically and economically valuable ecosystem service: resistance to invasion by an alien weed. The case also suggests that more research on the potential biotic resistance provided by natural enemies may help us better understand those factors that influence whether an alien species becomes invasive after naturalization. Based on our studies, we argue that there are practical as well as aesthetic and ethical reasons for working to maintain minor, even seemingly obnoxious, species and their interactions. In particular, this case suggests that we are not yet in a position to predict the cost associated with the decline and loss of a specific species, since its ecological function and economic value may not be obvious.

Natural History Background The thistle genus Cirsium (L.), indigenous in Eurasia, North America, and North and East Africa, contains about 250 species (Bremer 1994). The North American contingent of this genus is represented by at least 96 indigenous, endemic taxa (USDA, NRCS 1999). We have extensive quantitative data on four native species that are characteristic of the prairie grasslands of the upper Great Plains. All typically occur singly or in small stands, and none are considered major weeds (McCarty et al. 1967; Louda et al. 1990; Stubbendieck et al. 1994). The data presented here are for the native tall thistle (Cirsium altissimum (L.) Spreng.), a late-flowering monocarpic species in the tallgrass region of Nebraska (McCarty et al. 1967), and the naturalized Eurasian thistle species bull (spear) thistle (Cirsium vulgare (Savi) Tenore), also a lateflowering monocarpic species that occurs as small stands in disturbed roadside or overgrazed grassland in Nebraska. Although several alien

ecosystem service of thistles

7

thistles are listed as noxious weeds in Nebraska, bull thistle is not. Our studies of tall and bull thistles were conducted in Lancaster County, east of the City of Lincoln. As fugitive species, the performance and density of thistles are related to the availability of seed, the level and spacing of disturbance, and the vigor of grass competition (Hamrick 1983; Hamrick and Lee 1987; Louda et al. 1990, 1992; Popay and Medd 1990; Louda and McEachern 1995; Louda and Potvin 1995; Bevill and Louda 1999). Seed availability usually limits local thistle seedling density in open grasslands (see de Jong and Klinkhamer 1986; Louda and McEachern 1995; Louda and Potvin 1995). Thistles have a suite of adapted insects (Zwolfer ¨ 1965, 1988; Lamp and McCarty 1979, 1981, 1982a, b; Redfern 1983; Zwolfer ¨ and Romstock-V¨ ¨ olkl 1991). The most common insects specializing on Cirsium species in Nebraska are picture-winged flies (Tephritidae), weevils (Curculionidae), moths (Pyralidae, Pterophoridae), butterflies (Nymphalidae, Hesperiidae), lacebugs (Tingidae), aphids (Aphididae), and sucking bugs (Hemiptera: Cicadellidae, Membracidae, Miridae, Pentatomidae). Several studies demonstrate that these native insects significantly affect key components of individual plant fitness (Louda et al. 1990, 1992; Louda and Potvin 1995; Guretzky and Louda 1997; Stanforth et al. 1997; Bevill 1998; Jackson 1998; Bevill et al. 1999). There is also evidence that thistle-feeding insects often adopt similar hosts in alternate or novel environments. For example, several thistle-feeding insects have been imported from Eurasia and released in the United States by the U.S. Department of Agriculture as biological control agents for exotic thistles (Julien and Griffiths 1998). At least two of the weevils being distributed for the control of alien thistles have also adopted native species as hosts; these include Rhinocyllus conicus Fr¨ol. (Louda et al. 1997; Louda and Arnett 2000) and Larinus planus (F.) (Louda and O’Brien 2002). In sum, since insects can reduce plant performance of thistles and since host range expansion in thistle-feeding insects occurs, the potential clearly exists for native insects to adopt potentially invasive thistles as hosts. If this is the case, these insects likely play a role in limiting the reproduction and spread of nonindigenous thistles that are closely related, or ecologically similar, to native species.

Native Insect Herbivores Limit Densities of Indigenous Thistles Multiple studies of the role of coevolved insects in the population dynamics of native thistles in prairie grasslands have demonstrated

8

chapter 1

clearly that native insects significantly decrease the survival, growth, reproduction, lifetime fitness, and density of native thistles under field conditions. These experiments document significant reductions by insect herbivores of (1) juvenile survivorship and growth (Guretzky and Louda 1997; Stanforth et al. 1997; Bevill et al. 1999), especially in the context of competition with grasses (Louda et al. 1990, 1992); (2) subsequent flowering effort of surviving plants (Bevill 1998; Bevill et al. 1999); (3) successful seed maturation (Louda et al. 1990, 1992; Louda and McEachern 1995; Louda and Potvin 1995; Jackson 1998; Louda 1999a, b; Maron et al. unpub. data); (4) lifetime fitness (Louda and Potvin 1995); and (5) seedling, juvenile, and adult densities (Louda and Potvin 1995). Populations of the monocarpic thistles have been shown to be seed-limited in undamaged prairie grassland (Louda and Potvin 1995). Platte thistle density, for example, increased 100–600% when floral insect herbivores were reduced, especially in disturbances but also in ungrazed prairie (Louda et al. 1990, 1992; Louda and Potvin 1995). Individually, these studies show that coevolved thistle-feeding insects significantly reduce key parameters of individual plant performance. Collectively, the studies suggest that the chronic pressure exerted by a diverse, dependent assemblage of adapted natural enemies often limits population density and patch regeneration of native thistles in grasslands under indigenous conditions.

Native Thistle Harbors Insects that Attack an Exotic Thistle Ecosystem resistance to invasion has been listed as an important property of intact ecosystems (e.g., Daily 1997). One mechanism that produces such resistance to exotic plant invasion is competition from native plants (see Drake et al. 1989; McKnight 1993; Mack 1996). Herbivory by native insects could be another mechanism that reduces the invasiveness of some nonindigenous plant species (Louda 1999b). In this case, insects from a native thistle contribute significant resistance to the potential for invasive population growth and spread by bull (spear) thistle, Cirsium vulgare (Savi) Ten., in eastern Nebraska. Bull thistle, a Eurasian species, is an aggressive weed in many areas of the world. Its invasive potential is clear from published descriptions, studies, and control efforts in grazed grasslands of Australia, New Zealand, and South Africa (e.g., Austin et al. 1985; Julien and Griffiths 1998; Olckers and Hill 1999) and in natural areas of California (Randall 1991). However, although bull thistle has been present in the tallgrass prairie region of eastern Nebraska for at least 35 years,

ecosystem service of thistles

9

its numbers remain relatively low (C. P. Andersen and S. M. Louda, unpub. data). Contemporary agricultural practices control weeds, including bull thistle, within crop fields, but numbers have also remained low in roadsides and perennial pastures despite disturbance and grazing. Bull thistle is not common enough to be classified as a major weed throughout eastern Nebraska. Native insects from the indigenous tall thistle have included the potentially invasive Eurasian thistle in their diet. When native insect herbivores fed on developing buds and flowering heads, the seed production of bull thistle in Nebraska was severely reduced. In Lancaster County, for example, insects destroyed 88% of the potential seed production by bull thistle plants sampled in 1997, 79% in 1998, and 71% in 1999, significantly reducing the reproductive success of bull thistle (fig. 1.1). At least four insects were often found feeding on or in the reproductive shoots and developing flower heads of bull thistle. These insects were Platyptilia carduidactyla (Riley) (Pterophoridae), Baris subsimilis Casey (Curculionidae), Papaipema mitela Guen. (Noctuidae), and Paracantha culta Wiedeman (Tephritidae). All these species typically feed on or in the reproductive shoots and developing flower heads of the native, tall thistle (Louda, pers. obs.). Feeding by these insects severely reduced the number of flower heads that matured and the number of viable seeds that were produced by the native tall thistle from 1994 to 1995 (Jackson 1998) and from 1997 to 1999 (fig. 1.2), significantly reducing reproductive success. The levels of use of bull thistle were also high, though not as high as those observed on the native tall thistle (see fig. 1.1 vs. fig. 1.2; Jackson 1998; Louda, unpub. data). Thus, the adapted, dependent insects of native thistles are exerting tremendous “pest resistance pressure” on this exotic, potentially invasive weed under its newly adopted conditions in eastern Nebraska. Our previous experiments, in addition, have shown that herbivory by native insects on tall thistle further limits the survival and growth of young plants (Guretzky and Louda 1997) and the flowering success of older individuals (Jackson 1998). Subsequent competition with grasses for nutrients and moisture, known to restrict the success of established thistles (Austin et al. 1985; Hamrick 1983; Hamrick and Lee 1987; Popay and Medd 1990), likely reduces plant density further by limiting the survival and growth of the seedlings that do establish. If the consequences of insect feeding on bull thistle are similar to those for tall thistle—and this is currently being tested (L. M. Young, unpub. data)—then population growth and the development of high densities of bull thistle are opposed by the pressure exerted on this exotic thistle by native insects, insects—that are adapted to and main-

10

chapter 1

*

*

*

Figure 1.1 Average number (X, SE) of potential seeds as florets initiated; florets and seeds destroyed by feeding of native insects; and viable seeds released per plant by the Eurasian bull (spear) thistle, Cirsium vulgare (Savi) Ten. in Lancaster County, Nebraska (N ⳱ 10, 5, and 10, in 1997, 1998, and 1999, respectively). Native insects, transferring from the co-occurring, late-flowering native thistle (Cirsium altissimum, tall thistle), reduced seed production by this exotic, potentially invasive thistle by 88%, 79%, and 71% in 1997, 1998, and 1999, respectively. * ⳱ p ⬍ 0.05 in orthongonal contrasts.

tained by a thistle native to this region. Disruption of these interactions, through a loss of the native thistle and its reservoir of native insects, would be expected to increase the probability of a full-blown invasion by bull thistle, as has occurred elsewhere. Thus, loss of the native thistle and its insects could create a more noxious, economic weed out of a currently innocuous exotic plant.

Discussion The reasons for preserving species are scientific, functional, and practical. First, studies of thistle dynamics and interactions have contributed basic ecological insights into how biological interactions can structure, limit, and influence the numerical abundance and distribution of native plants in grasslands (e.g., Louda et al. 1990, 1992; Louda

ecosystem service of thistles

*

11

*

Figure 1.2 Average number (X, SE) of potential seeds as florets initiated; florets and seeds destroyed by feeding of native insects; and viable seeds released per plant by the native thistle, Cirsium altissimum (L.) Spreng. (tall thistle) in Lancaster County, Nebraska (N ⳱ 5 and 10, in 1998 and 1999, respectively). The coevolved, inflorescence-feeding insects reduced seed production by this native thistle species 99.3% and 94.8% in 1998 and 1999, respectively. * ⳱ p ⬍ 0.05 in orthongonal contrasts.

and Potvin 1995; Guretzky and Louda 1997; Jackson 1998; Bevill et al. 1999). Furthermore, parallel studies have added to an understanding of the role of interactions in plant rarity (Louda and McEachern 1995; Stanforth et al. 1997; Bevill and Louda 1999; Bevill et al. 1999). Second, thistles contribute to the support of a broad array of animal species. In the Great Plains prairies, for example, at least 35 other species use native North American thistles (Louda, unpub. data). These species range from microscopic plant parasites (that harbor potentially useful secondary compounds) to macroscopic animals, including charismatic ones (e.g., as the American Goldfinch) and their predators (e.g., raptors) (Louda et al. 1998). Would such species decline if the native thistle were eliminated? No good answer exists yet. Third, our data suggest that a native thistle can support a set of herbivorous insects that contribute to a major ecosystem service: the limitation of a potential weed. Native insects limit the seed production and density of populations of native thistles in native and disturbed grasslands (Louda and Potvin 1995; Guretzky and Louda 1997;

12

chapter 1

Bevill et al. 1999). In addition, these insects are contributing to the suppression of the potentially invasive, exotic bull thistle, Cirsium vulgare. What important role, then—if any—can minor, seemingly obnoxious species play? In this case, the data suggest that a prickly, inconspicuous native plant, tall thistle, supports insect herbivores that dramatically reduce the seed production of an incipient invasive species, likely constraining the density and the rate of spread of a potentially serious economic and environmental weed. Thus, we hypothesize that elimination of the native tall thistle would likely have at least one negative ecosystem response with economic implications. Its reduction and loss would be expected to cause a reduction of temporally synchronized adapted insects, decreasing resistance to invasion by the exotic species. Bull thistle would likely become a much greater problem—reaching a status similar to that in other rangeland regions— without indigenous thistles to harbor adapted, thistle-feeding insects. Following Bob Paine’s example, several key experiments are now underway to test the hypothesis of significant ecosystem resistance provided by tall thistle. These experiments will quantify bull thistle response to (1) increased seed input, to address the question of whether population density is proportional to seed availability; (2) the exclusion of native insect herbivores, to address the question of whether potential seed production is limited by resources or factors other than intensive herbivory; and (3) the removal of neighboring native thistles, to address the question of whether proximity of native thistles influences the native insect use of bull thistle directly, as suggested by observational data (C. P. Andersen and S. M. Louda, unpub. data). We propose that our studies, along with the knowledge that communities are often structured by trophic interactions (Elton 1927; Pimm 1982; Polis and Winemiller 1996), argue that even a naturally sparse, potentially noxious, native species like a thistle cannot necessarily be assumed to be “expendable.” Interestingly, the potential functional significance of tall thistle and its dependent insect herbivore guild became evident only in the context of an external challenge: the establishment and potential invasion by bull thistle. Thus, this case illustrates the difficulty of identifying and quantifying an ecosystem service until it is needed. Similarly, Root (this volume) has pointed out that the role a species plays may be altered, or may only become apparent, as a result of changing conditions, such as the colonization of an ecosystem by an exotic species. The novel interactions that may arise as a consequence of such changes make it difficult to predict the role that a species may play in the future. It is also worth noting that even thistles are not interchangeable in

ecosystem service of thistles

13

their ecosystem functions. In this case, saving an early flowering thistle (e.g., Platte thistle) and its earlier-feeding insects would not be equivalent to saving the late-flowering thistle (tall thistle) whose insects are providing some resistance to invasion by bull thistle. Unlike biotic resistance provided by plant competitors, in which there is likely to be a substantial amount of functional redundancy among plant species, the function provided by specific plants that serve as reservoirs of coevolved natural enemies is likely to be sensitive to the elimination of a specific plant species. Few studies have quantified the ecosystem resistance to invasion that is provided by indigenous insect herbivores in other systems. In fact, detection could be impossible in many cases, for example if resistance is sufficient to prevent initial colonization and naturalization. Mack (1996) presents floristic data suggesting that naturalization is enhanced for alien species that lack native relatives. He argues that this pattern may be due to the decreased likelihood of host extensions, or shifts, by native natural enemies that have coevolved with native plants onto exotics that are more distantly related to the native flora. Our data are consistent with this suggestion. Clearly, though, the ecosystem service provided by native plants when naturalization is prevented will be exceptionally difficult to detect and document. Some indirect evidence suggests, however, that feeding by native insects on non-native plant species may be more common and potentially more important than generally thought. For example, host range expansion of native insects onto non-native plant species has been documented for a diverse group of plants, including other exotic thistles such as milk thistle (Silybum marianum Gaertn.) and Italian thistle (Carduus pycnocephalus L.) in southern California (Goeden 1971, 1974), a variety of other nonindigenous herbs (Wheeler 1974; Chew 1977; Berenbaum 1981; Thomas et al. 1987; Evans et al. 1994); and even introduced trees (Fraser and Lawton 1994). Such crossover effects are also common in agricultural systems (e.g., Strong 1974; Strong et al. 1977; Tabashnik 1983) where native host plants serve as “reservoirs” of crop pests (e.g., Herzog and Funderburk 1986). Furthermore, native insects have been manipulated successfully for biological control of exotic weeds (Sheldon and Creed 1995; Newman et al. 1998). A prominent theory used to explain the invasiveness of exotic plants within novel habitats is that they have been released from their adapted natural enemies, such as insect herbivores, which are assumed to keep them in check within their native habitats (e.g., Andres and Goeden 1971; Crawley 1987; Blossey and Notzold ¨ 1995; Mack 1996). The indirect evidence showing that native insect herbivores often expand their host range onto introduced plants leads to

14

chapter 1

the hypothesis that such herbivores could also play a role in limiting exotic plant success in the new environment. This result could keep exotics below invasive levels or even prevent naturalization all together, especially when related native plants support an herbivore guild within the habitat. Thus, the theory, the indirect evidence, and our data suggest that more investigations are merited. To clarify the role of adapted, indigenous herbivorous insects in providing resistance to plant invasion. More generally, an elucidation of important indirect linkages between plants that are mediated by mobile insects, herbivores, or pollinators will be critical to our ability to predict the community-level implications of the extirpation of individual plant species. Few individuals are known to admire thistles (although the Scots are a notable exception). Most express a disregard for thistles, native or not, without knowing much about them or their ecological interactions and ecosystem functions. The impact on native North American thistles was not a major factor in the decision to release the biocontrol weevil Rhinocyllus conicus in 1969, in spite of evidence that it would use Cirsium species (Boldt 1997; Gassmann and Louda 2001). Nor is the potential impact of Larinus planus on thistle populations, exotic or native, a factor influencing its redistribution within North America. Distribution continues, in spite of evidence that each of these weevils accepts Cirsium species into its diet and that each can have a major nontarget effect on native species (Louda et al. 1997, 1998; Louda 2000; Louda and Arnett 2000; Louda and O’Brien 2002). Perhaps a disregard for noncharismatic species, especially prickly ones, is understandable. However, neither a lack of charisma nor relative rarity provide an adequate scientific basis for deciding whether a species has a significant ecological function or indirect economic value.

Conclusion Our data suggest that thistles—minor species with a mixed or even unfavorable public image—may provide a vital ecosystem service: resistance to invasion by a putative economic and environmental weed. The conflict between our results and the general attitude toward thistles illustrates the need to develop scientific criteria for determining whether a species might be ecologically redundant and therefore potentially dispensable. Even given the motivation to develop such criteria, however, the question remains: could the interdependencies and numerical consequences of eliminating an indigenous thistle be anticipated and predicted in the context of evaluating its expendability?

ecosystem service of thistles

15

The task would be difficult. In our study of tall thistle, intensive studies of thistles and their dependent species were required to document the demographic interconnections and to discover the potential economic value of preserving the species. Furthermore, the ecosystem service provided would not have been detected without the challenge from a potentially invasive species. Thus, it seems clear that we are not in a position to define all species-specific ecological or practical roles. Rather than expendability, the critical issue seems to be this: how can we use and manage natural systems in a way that will minimize the probability that component, potentially important species will be lost?

Acknowledgments We thank the many students, colleagues, and family members who have shared in and contributed to these studies of thistles. Bob Paine and his students, Bruce Menge and Paul Dayton, introduced Svata Louda to the challenge of understanding biological interactions in the field; and subsequent academic advisors, including Boyd Collier, Joe Connell, Tom Ebert, Bob Luck, Bill Murdoch, Jim Rodman, and Paul Zedler, continued that process. They all have her appreciation, but they also may bear some responsibility for the outcome! Our studies of thistle-insect interactions could not have been done without the funding provided by the Research Council of the University of Nebraska (1984–1997), the National Science Foundation (DEB92-21065, DEB96-15299), and the Nature Conservancy’s Rodney Johnson and Katharine Ordway Stewardship Endowments to S.M.L. and the David H. Smith Postdoctoral Fellowship to T.A.R.

Chapter 2

m The Overriding Importance of Environmental Context in Determining the Outcome of Species-Deletion Experiments Bruce A. Menge

Classic ecological experiments such as those of Paine (1966, 1974) leave little doubt that species loss can have profound consequences on a community. At the same time, there can be striking variation in these consequences. Thus, to understand the consequences of a loss of species from a community, one must learn the factors that are responsible for the variation observed in studies of removals, losses, introductions, or invasions of species. An ultimate goal in such efforts is prediction: can we forecast what will happen when a species is deleted from a community or ecosystem? It is still far from clear whether a meaningful prediction can be made of population or community consequences of species loss. The most that we can accomplish may be to understand the changes that have occurred and, from

environmental context

17

this understanding, make only general predictions of the outcomes of activities that are likely to lead to species loss. Predicting the outcomes of species loss depends on discovering the rules that regulate species interactions and their aftereffects, and how these vary with environmental conditions (Belyea and Lancaster 1999). My goal in this contribution is to probe a small part of this issue. My primary focus is on environmental context, both abiotic and biotic, and the extent to which it dictates the aftermath of species loss in natural communities. Because a vast literature has developed on the consequences of species loss, here I restrict myself largely to examples that involve consumers in marine hard-bottom communities. Much of my treatment summarizes and synthesizes examples taken from my own research activities during a 30Ⳮ-year career that was launched with the guidance and insights of my dissertation advisor and mentor, Bob Paine. It is with the deepest respect that I dedicate this paper to Bob on the occasion of his retirement from the Department of Zoology at the University of Washington. The debt I owe him is large. My career-long focus on the factors that structure and regulate ecological communities was given its initial push—plus numerous prods along the way—by Bob.

Concepts of Species Impact Species are far from equivalent in their impacts on communities. This fact was revealed clearly by Paine’s sea star manipulations in Washington, New Zealand, and Chile (Paine 1966, 1971, 1974; Paine et al. 1985). The keystone species concept that was fostered by this work (Paine 1969a ) has become an integral feature of ecological theory and practical application, particularly in some schemes of ecosystem management (Mills et al. 1993). Despite recent controversy (Mills et al. 1993; Paine 1995; Power et al. 1996b), the concept is robust and broadly relevant (Menge and Freidenburg 2001). As was clarified recently, a keystone species is one “whose impact on its community or ecosystem is large, and disproportionately large relative to its abundance” (Power et al. 1996b). An important implication of this definition is that keystone species have relatively high per capita interaction strengths—a quality that, when translated to the community level, is termed “community importance.” Other concepts of species impacts include “dominant species,” “strong interactors,” and “weak interactors.” Dominant species are those that have large effects on community or ecosystem structure but whose impact is a function of their high abundance or biomass. Such species would therefore have rela-

18

chapter 2

tively low per capita or per biomass interaction strengths, depending on the metric. By the definition followed here, strong interactors include keystone species, dominants, and all other species that have a large impact on populations as well as communities or ecosystems. Weak interactors are those that have little influence on any level of organization. This does not mean, however, that weak interactors are always unimportant in community dynamics. Under certain conditions, some weak interactors may increase their impacts on species dramatically, temporarily becoming strong interactors (Berlow 1999; Navarrete and Menge 1996). In other cases, many individual weak interactors together can jointly have strong impacts. An example is “diffuse predation,” in which a guild or group of predators can collectively exert “strong” predation even though each by itself has a small effect (Hixon 1991; Menge et al. 1986a, 1994; Robles and Robb 1993). Spurred in part by recent critiques of the keystone species concept, and by Paine’s experimental analysis of interaction “strength” in a guild of molluscan herbivores (Paine 1992), ecologists have begun to quantify the magnitudes of species effects on per capita and per population bases (Berlow et al. 1999; Fagan and Hurd 1994; Navarrete and Menge 1996; Raffaelli and Hall 1996; Ruesink 1998; Wootton 1997; Harley, this volume). Although still in its infancy, this work has begun to quantify community patterns inferred long ago by Paine (e.g., Paine 1980): communities consist of a few strong interactors and many weak interactors. What remains unclear, however, are the ecological and evolutionary bases for differences in ecological impact among species. For example, what are the ecological mechanisms and evolutionary conditions that produce keystone species?

Environmental Context: A Primary Determinant of Species Impact Much evidence suggests that variation among species in their impact, and thus in the consequences of their loss to the community, is determined by environmental context (e.g., Harley, this volume). By environmental context, I mean both abiotic and biotic environments. Abiotic context, for example, can include regimes of stress, disturbance, or productivity. Biotic context includes characteristics of the assemblage in which a species carries out its activities, such as diversity, species composition, traits of specific interactors, or the rate of replenishment (e.g., recruitment, growth) of resources. To simplify the discussion, I discuss each of these categories of environmental context

environmental context

19

separately, although they are of course usually tightly interlinked in determining the outcome of species impacts.

Physical Context As John Sutherland and I have argued, a variable regime of environmental stress can lead to dramatic differences in the outcome of species deletions (Menge and Sutherland 1987; see also Dunson and Travis 1991; Grime 1977). As on most rocky shores, for example, sites in New England occur along striking gradients in wave forces (horizontally along the shore) and exposure to air (vertically on the shore). The varying physical regime along these gradients leads to large differences in the impact of the predatory whelk Nucella lapillus (Menge 1976, 1978a, b; Menge and Sutherland 1976). Environmental Conditions: Hydrodynamic Forces and Thermal Stress New England Shores. New England rocky intertidal regions harbor communities that have relatively few species, each of relatively high abundance (Dudgeon et al. 1999; Lubchenco and Menge 1978; Menge 1976). Zonation is sharply defined and varies with wave exposure. In wave-exposed areas, there are high-barnacle (Semibalanus balanoides), mid-mussel (Mytilus edulis), and low-mussel-kelp zones (M. edulis overlaid by a canopy of Alaria esculenta during the summer). On shores with intermediate exposure to waves, zones from high to low are dominated by barnacles, fucoids (Fucus evanescens [formerly F. distichus] and F. vesiculosus) overlying a mosaic of mussels and free space, and red algal turfs (Chondrus crispus, Mastocarpus stellatus). At more sheltered sites, zones are dominated by barnacles, the fucoid Ascophyllum nodosum overlying largely free space, and Chondrus crispus turfs. In the middle and high zones at intermediate to wave-exposed sites, the only predator is the whelk Nucella lapillus. In more sheltered areas, Nucella is joined by small individuals of the green crab (Carcinus maenas), and in low zones in intermediate and sheltered areas by sea stars (Asterias forbesi, A. vulgaris) and brachyuran crabs (Carcinus maenas, Cancer irroratus, C. borealis). Using cage exclosure experiments in the high and middle zones of high and intermediate wave exposure, I found that predation by whelks varied strikingly (Menge 1976, 1991; fig. 2.1). In high zones, whelks had no effect on barnacles. Similarly, in mid-exposed zones, whelks had no effect on barnacles or on mussels. In sharp contrast, whelks had strong

20

chapter 2

Figure 2.1 Summary of studies of community dynamics along a waveexposure gradient in New England (after Menge 1976). Patterns of community structure are shown with the left set of histograms in each panel. Results are summarized from experiments in high (A, B) and mid (C, D) zones of different combinations of predation and competition among dominant space occupiers (presented as average abundance at the end of the season; generally autumn). In the high zone, neither predation nor interspecific competition affected community structure, regardless of exposure. In the mid zone, interspecific competition was the major determinant of community structure at exposed sites, whereas predation was the major determinant of community structure at protected sites. Error bars in this and the other figures are standard errors of the mean.

environmental context

21

effects on both prey species in the intermediate mid-zones and, with algal whiplash, were the major determinants of low mussel and barnacle cover in such areas. Field observations led me to surmise that physical constraints were the basis of subhabitat differences in the effects of whelks. To test this hypothesis, experiments quantified predation in relation to wave exposure, height on the shore, and other factors such as canopy cover (Menge 1978a, b). These experiments showed that whelk foraging rates were reduced greatly under conditions of high wave action and at higher shore levels. Foraging was decreased further when the canopy was removed, and whelks suffered high mortality under high-shore, canopy-free conditions. Although physical conditions could be evaluated only crudely in those days, it seemed clear that whelk foraging activity was inhibited by turbulent conditions at wave-exposed sites and by exposure to heat and desiccation stress higher on the shore. Later studies have confirmed the susceptibility of whelks to hydrodynamic and thermal stress (e.g., Bertness et al. 1999; Dahlhoff et al. 2001; Denny 1988; Garrity 1984; Leonard et al. 1998). Thus, the role of Nucella lapillus in controlling the abundance and zonation of barnacles and mussels and, indirectly, of algae in mid-zones, was contingent on environmental conditions. Predation was weak where thermal or desiccation stress and hydrodynamic forces were great, whereas it was strong where these factors were more moderate. Oregon Coast. Studies of the keystone species Pisaster ochraceus in Oregon have offered further insights into how species impact varies with abiotic context (Menge et al. 1994, 1996). As demonstrated by the studies of Paine (1966, 1974) and Dayton (1971), P. ochraceus can have overwhelmingly powerful effects on prey populations and community structure. Observations along a wave-exposure gradient on the Oregon coast, however, suggest that the impact of this predator on its prey varied under some conditions. In particular, in some subhabitats (e.g., wave-sheltered areas), sea stars were so scarce that it seemed unlikely that they played a major role in determining the lower limit to the mussel bed and low-zone community structure. In other subhabitats, sea stars were abundant, but sand burial appeared to be an alternative—or the only strong structuring agent in the low intertidal zone. Both situations were located in relatively wave-sheltered conditions. Because the outcome of standard exclusion experiments can be contingent on the rates of prey recruitment (see section on oceanographic context), prey-transplant experiments were established as an alternative and supplementary approach for studying sea star preda-

22

chapter 2

tion. Mussels (M. californianus, the preferred prey of Pisaster) were transplanted from mid-zone beds to low intertidal regions (where these bivalves are normally almost absent) to plots with and without Pisaster (Ⳮsea stars and ⳮsea stars, respectively). Experiments were conducted at wave-exposed and wave-protected areas at each of two sites. Predation was quantified by determining the rate of loss of mussels in Ⳮsea star and ⳮsea star plots. To quantify variation in per capita interaction strength, the rate of mussel loss was standardized to a per–sea star basis (Menge et al. 1996). At Boiler Bay, an area with little sand transport in sheltered areas, per capita predation rates were relatively low in wave-exposed plots and relatively high in waveprotected plots. These per capita rates were related inversely to the abundance and predation impact of sea stars, which were both greater at wave-exposed areas. This result suggests that the foraging activity of individual Pisaster was less at wave-exposed than at waveprotected sites, just as was found for whelks in New England and on Tatoosh Island in Washington State (Quinn 1979). Similar experiments at the other site, Strawberry Hill, also showed that per capita predation rates were lower at wave-exposed sites, suggesting that feeding may be reduced by hydrodynamic forces. These experiments at Strawberry Hill also revealed that another abiotic process—sand transport and the resulting periodic burial of waveprotected low intertidal areas—could override the influence of Pisaster in determining low-intertidal community structure. In such areas, mussel mortality in transplant experiments occurred at similar rates in plots with and without sea stars. Quantification of mussel survival in relation to sand burial suggested that, although mussels could apparently tolerate periods of a few weeks of burial, sand smothering was the dominant cause of death of the transplanted mussels. More generally, field observations from more than a decade of research at this site indicate that sand burial is the primary agent that structures low-intertidal communities over large areas of relatively wave-sheltered rocky shore. The usual absence of mussels and the barnacle Balanus glandula are determined largely by burial events, whereas sandtolerant algae (e.g., Gymnogongrus linearis, Laminaria sinclairii), surfgrass (Phyllospadix scouleri, P. torreyi), and barnacles (Chthamalus dalli) persist as community dominants. Persistent populations of Pisaster occur in the area and often aggregate around and feed on clumps of M. californianus that have been dislodged from wave-exposed reefs to seaward. Other observations suggest that these “times of plenty” can have a severe cost, however. Sea stars in these areas are often found clustered on the tops of buried rock outcrops, lying moribund or dead on the substratum (with evidence of damage from tumbling about

environmental context

23

over sand and rocks) or attempting to crawl on the sand away from recently buried outcrops. Because Pisaster is adapted for living on hard surfaces (its tube feet have suckers that are useless on sand), wave-sheltered areas at Strawberry Hill appear to be a suboptimal habitat. Thus, although much evidence suggests that Pisaster’s role as a keystone species is probably general in a biogeographic sense, its community impact is clearly variable (see also Harley, this volume) and is contingent on both hydrodynamic forces and substratum stability. Curiously, the strongest effect of this sea star occurs in those habitats in which individual foraging appears to be most inhibited by wave turbulence. The Oregon studies suggest that the large impact of Pisaster at wave-exposed areas is a function of the higher density of sea stars. Possible reasons for this higher density are considered later (see “Biotic Context”). Other Examples. Physical constraints are widely known to influence the impact of marine species; examples range from whelks on temperate and tropical rocky shores (Quinn 1979; Garrity 1984; Garrity and Levings 1981) to echinoderm predators (Lawrence 1990; Witman and Grange 1998). Similarly, the idea that the physical environment limits the likely impact of species removal is also thought to apply widely in terrestrial communities from biogeographic to local scales (Collinge and Louda 1988; Gillette 1962; Louda 1982; Louda and Collinge 1992; Louda et al. 1987; White 1978). Oceanographic Context: Productivity and Recruitment Recent studies suggest that, on relatively large spatial scales, nearshore oceanographic conditions may underpin significant differences in community impacts of individual species (Bustamante and Branch 1996; Bustamante, Branch, and Eekhout 1995; Bustamente, Branch, Eekhout, et al. 1995; Leonard et al. 1998; Menge 1992, 2000; Menge et al. 1997a, b, 1999). Variation in nearshore oceanography (nutrient concentrations, upwelling intensity, currents and mixing, phytoplankton blooms) can generate alongshore differences in phytoplankton productivity, rates of larval transport and delivery, and macrophyte production. Through bottom-up effects (e.g., nutrient and plant production impacts on food webs that “flow up” from the bottom of the web), such factors can alter communities both directly and indirectly through the modification of top-down effects. This latter effect is of particular interest, since it suggests a potential condition that might underlie variation in consumer-prey interaction strength and thus, perhaps, help explain the evolution of keystone species. In the follow-

24

chapter 2

ing pages, I present three examples that illustrate the importance of oceanographic context. I explore how oceanographic influences can govern variation in phytoplankton productivity, a bottom-up effect, and in turn how this may underlie the importance of top-down effects that are associated with predator species. Example 1: Productivity and Prey Recruitment Effects in Oregon. Along the Oregon coast, phytoplankton concentration can vary dramatically on scales of 10s to 100s of km. Sampling initiated in 1993 has shown that phytoplankton concentrations (as estimated by chlorophyll-a concentration, hereafter termed chl-a) at Boiler Bay are consistently lower than those at Strawberry Hill (Menge et al. 1997a, b). This difference is particularly dramatic after summer upwelling events. Within 2–3 days of relaxation of an upwelling, chl-a reached concentrations greater than 30 ␮g/l at Strawberry Hill, while concentrations changed only slightly at Boiler Bay. Associated with this difference were consistent differences in the growth rates of sessile filter feeders, including mussels and barnacles (Menge 1992; Menge et al. 1994; Sanford and Menge 2001). From 1990 to 2001, growth of the mussel M. californianus in annual transplant experiments has always been greater at Strawberry Hill. Similar measurements show comparable differences for the barnacles Balanus glandula and Chthamalus dalli (Menge 1992; Sanford and Menge 2001), and field observations suggest that the mussel M. trossulus also grows faster at Strawberry Hill. What are the causes of these differences in growth rates of sessile filter feeders? Our initial hypothesis was that differences in growth rate were driven by between-site differences in phytoplankton concentration (Menge et al. 1997a). That is, mussels and barnacles grew faster because a likely major food source, phytoplankton and phytoplankton-derived detritus, frequently reached higher concentrations at Strawberry Hill. This explanation was not complete, however, because higher growth rates at Strawberry Hill appeared to be sustained during fall and winter, when phytoplankton blooms do not occur and detritus concentrations do not differ. A recent study confirmed these impressions (Sanford and Menge 2001). Quantification of growth rates of individual uncrowded barnacles showed that both B. glandula and C. dalli did indeed grow faster in response to phytoplankton blooms. The analysis also showed, however, that increased growth rates continued well beyond the period of high phytoplankton concentration. Further investigation suggested that faster barnacle growth was associated with two additional factors. First, since barnacles are known to consume invertebrate larvae (e.g., Navarrete and

environmental context

25

Wieters 2000), it seemed possible that sustained growth after the cessation of blooms was due to the capture of zooplankton that were being transported past by currents. A comparison of short-term growth rates with recruitment data for barnacles during the same period was consistent with this notion. Moreover, recruitment of both barnacles and mussels at these sites has consistently been greatest during autumn months (September to December; B. Menge, unpubl. data). Thus, if zooplankton are in fact an important food source for filter feeders, the apparently high growth rates of filter feeders sustained through fall and winter may be a direct consequence of higher rates of onshore transport of zooplankton. Second, a spurt in growth in late summer and early fall may also be influenced by a more favorable thermal regime. Water temperatures during summer often fall below 10⬚C due to upwelling events, which largely cease in September. As a consequence, temperatures tend to stabilize at around 12–14⬚C, gradually falling to about 10⬚C in winter (Sanford 1999a; B. Menge unpub. data). In contrast, air temperatures experienced by barnacles in the intertidal region tend to decline as summer gives way to fall because of seasonal changes and because lower low tides shift from daylight (morning) to nighttime in October. Because more moderate temperatures foster greater efficiency in the conversion of food to growth and reproduction, increased growth in autumn may also reflect a more favorable thermal regime. What are the potential factors underlying the differential onshore transport of larvae and particulate food for filter feeders? Elsewhere (Menge et al. 1997a), evidence suggested that differing nearshore oceanographic conditions might lead to the dilution of plankton offshore from some coastal sites (e.g., Boiler Bay) and the concentration of plankton offshore from others (e.g., Strawberry Hill). Satellite imagery and coastal surveys of phytoplankton concentration suggested that the variable width of the continental shelf was associated with an uneven pattern of current flow nearshore that might tend to transport plankton away from the coast at Boiler Bay and concentrate plankton in a gyre adjacent to Strawberry Hill. Recent quantification of surface currents using high-frequency radar (which quantifies surface currents with resolution of about 2 km) is consistent with this hypothesis. Currents during upwelling tend to be strongly offshore toward the southwest off Boiler Bay and weak and variable off Strawberry Hill (M. Kosro, unpub. data). Assuming that the surface currents reflect patterns of the movement of larvae, such physical patterns would lead to low residence times for propagules and detrital particulates in waters off Boiler Bay but high residence times for particulates in waters off Strawberry Hill. Further research is necessary to sort out the

26

chapter 2

relative influences of phytoplankton and particulate concentration, larval condition, other meroplankton (larvae of benthic species) food sources, and thermal regime as determinants of the higher growth of filter feeders in rocky intertidal regions. In summary, the evidence that oceanographic processes drive between-site differences in prey productivity—a bottom-up process—is strong. Do such differences influence top-down processes as well? The mussel transplant experiments used to test predation rates (summarized earlier) suggest that they do. In three separate experiments, two using M. californianus and one using M. trossulus (Menge 2000; Menge et al. 1994; Navarrete and Menge 1996), predation (per population) rates were always greater at Strawberry Hill, and were generally greater at wave-exposed areas than at wave-protected areas. Longerterm (approximately 3-yr) removal of Pisaster led to changes similar to those seen by Paine (1966, 1974): the lower edge of the mussel bed migrated downward, presumably due to the lack of sea stars “browsing” on mussels that were pushed downward by lateral pressure from growth, and to the recruitment of mussels within the bed. In a separate experiment that tested the impact of predation, predators (primarily Pisaster) were excluded using standard caging techniques in the low zone in wave-exposed and wave-protected areas at each site (fig. 2.2). All experiments began with surfaces free of sessile biota (scraped but not sterilized) in July 1989 and ran until August 1990. Results showed that in the presence of predators, the barnacle C. dalli settled and grew to persist as the dominant space occupant at both sites, ranging in final cover from about 30% to about 70%. In the absence of predators, results differed between sites. At Boiler Bay, C. dalli was generally the most abundant sessile organism, although both B. glandula and M. trossulus were present in low abundance. In contrast, at Strawberry Hill, mussels (M. trossulus) recruited rapidly to the cages, overgrowing and displacing the barnacles. Mussels also became abundant over the winter in Ⳮpredator treatments at the wave-exposed site at Strawberry Hill but were eaten by July of the following summer (see fig. 2.2G, H). The apparent between-site difference in predator impact in this experiment (see fig. 2.2) is most simply interpreted as a difference in the rate at which mussels colonized cages. Thus, the results of the mussel transplant experiments appear to provide a clearer picture than do cage experiments of the actual predation regimes at each site, even though the results in both transplant and cage experiments superficially led to similar interpretations (i.e., predation is more intense at Strawberry Hill). A final feature of community dynamics at these sites is that population densities of Pisaster ochraceus are higher at Strawberry Hill than

environmental context

27

at Boiler Bay and are also much more abundant at wave-exposed sites than at wave-protected sites (Menge et al. 1994). As also observed on Vancouver Island (Robles et al. 1995), Pisaster respond quickly to changes in prey concentration, forming large feeding aggregations that often consisted of hundreds of individuals. These aggregations are common in the low zone at Strawberry Hill (Sanford 1999a, b; B. Menge pers. obs.). Mussel patches developing over winter are usually surrounded by sea stars by spring and eliminated by mid-summer. Thus, dense concentrations of prey may affect the abundance of sea stars at least on short-term and local scales, through a numerical response. It seems reasonable to suggest that longer-term, larger-scale variation in sea star abundance is also a function of consistent between-site differences in prey populations, such as those documented at Boiler Bay and Strawberry Hill. Another factor that could affect sea star abundance is recruitment. Although quantitative estimates suggest that the recruitment of sea stars may be somewhat higher overall at Strawberry Hill, differences were not large, and monthly peaks of recruitment were sometimes higher at Strawberry Hill and sometimes higher at Boiler Bay. Further, size structure (g of wet mass) samples indicated that juveniles (ⱕ 50 g wet mass) were actually proportionately more abundant at Boiler Bay (Menge et al. 1994). I conclude that between-site recruitment differences are unlikely to explain the differences in sea star abundance or effect. Collectively, these results suggest that two environmental characteristics—recruitment and secondary production of prey—drive between-site differences in predation regime. In other words, more intense predation is associated with higher rates of prey production. In this case at least, variation in bottom-up processes can therefore underlie differences in top-down effects. Thus, the community impact of the keystone predator Pisaster ochraceus seems at least partly dependent on the oceanographic context. Example 2: Prey Productivity and Recruitment Effects in New Zealand. In general, the rocky intertidal region of the South Island of New Zealand has typical patterns of community structure (Knox 1953; Menge et al. 1999). High, middle, and low zones are dominated by barnacles, mussels, and macrophytes, respectively. Patterns on the east coast contrast sharply with those on the west coast, however. At sites on Banks Peninsula on the east coast, mussels (Mytilus galloprovincialis, Perna canaliculus) dominate the middle and low zones, often ranging down to the upper edge of the infralittoral zone, abutting a zone of the massive brown macrophytes Durvillea willana and D. antarctica.

environmental context

29

Predators (e.g., the whelk Thais orbita and sea stars Coscinasterias calamaria and Stichaster australis), though present, are scarce. In contrast, while mussels dominate the middle zone at sites on the central west coast, the low zone consists of free space (bare rock and algal crusts) on the upper portion and red algal turf (mostly mixed Champia sp. and Gigartina spp.) on the lower portion. Predators (mostly S. australis but also T. orbita at some sites) are abundant. The northern half of the west coast lies in a region of intermittent upwelling. On colliding with the west coast, the southeastwardflowing Tasman current separates into the northeastward-flowing West-land current and the southwestward-flowing Southland current (Menge et al. 1999; Neale and Nelson 1998). These conditions, and summer orographic effects, cause intermittent upwelling along the northwest coast. The remainder of the west coast as well as the south and east coasts lie in a downwelling region of the Southland current, which wraps around the island (Vincent et al. 1991). Studies that quantify key community processes and environmental conditions (predation rate, grazing effects, recruitment, mussel growth, air and sea temperature, wave forces) were conducted at sites on the east and west coasts to test the prediction that both bottom-up and top-down forces would be greater in the west coast upwelling ecosystem. Briefly, results were consistent with these predictions. Predation was more intense on the west coast than on the east coast. In particular, the sea star Stichaster australis had a powerful influence on community structure; sea stars removed most transplanted mussels within 2 to 3 months. Further, in the absence of this sea star, mussel abundance increased steadily in cover (mostly by recruitment and growth), occupying 20–30% of the previously bare low zone in only 6 months. With Paine’s (1971) studies on the west coast of the North Island of New Zealand, these results suggest that this sea star is a keystone species. Mussel and barnacle recruitment rates were also in general orders of magnitude greater, and mussel growth, as indexed by RNA:DNA

Figure 2.2 Consequences of exclusion of Pisaster ochraceus and whelks (Nucella ostrina, N. canaliculata) on sessile prey abundance (% cover, estimated from photographs using random-dot methods) at wave-exposed and wave-protected areas at Boiler Bay and Strawberry Hill from 1989– 1990. Treatments were marked plot (MP; Ⳮpredators), roof (Rf; Ⳮpredators), and cage (Cg; ⳮpredators). Both predation impact and the rate of colonization by mussels were greater at Strawberry Hill than at Boiler Bay.

30

chapter 2

ratios (Dahlhoff and Menge 1996; Menge et al. 1999), appeared greater on the west coast as well. Temperature patterns of the seawater exhibited clear upwelling-characteristic trends on the west coast (periodic sharp drops of 3–4⬚C, with rapid recovery after days to a week or more) and downwelling-characteristic trends on the east coast (mostly steady trends with only slight temperature fluctuations). Limited nutrient samples suggested that upwelling generated elevated nutrient levels on the west coast while east coast nutrient concentrations were consistently low. Dynamometer data suggested that wave forces were similar at both sites (Menge et al. 1999). Although nearshore oceanographic conditions remain only sketchily known in New Zealand, these results, occurring on a biogeographic scale in contrasting oceanic regimes (upwelling vs. downwelling) are consistent with the Oregon studies, which were conducted on smaller spatial scales within an upwelling regime. Stronger bottomup effects were associated with stronger top-down processes. Thus, oceanographic context may be critically important in determining patterns of rocky intertidal community structure and dynamics. As in the northeast Pacific, keystone predation in these New Zealand sites was associated with an upwelling ecosystem. Example 3: Nutrient and Kelp Inputs in South Africa. A final example of strong oceanographic effects on rocky intertidal community structure offers interesting contrasts as well as similarities to the results in Oregon and New Zealand. Around the South African coast, upwelling is strong on the west, less on the south, and weak on the east (Bustamante, Branch, Eekhout, et al. 1995b). Nutrient concentrations and estimates of benthic algal productivity are correlated positively with these oceanographic changes, as are grazer biomass and filterfeeder biomass. Field experiments suggest that grazer biomass is strongly related to the rate of delivery of macrophyte detritus from nearshore kelp stands (Bustamante, Branch, Eekhout, et al. 1995a), and thus, by extension, that the correlation between benthic algal productivity and grazer biomass is driven by nearshore oceanographic conditions. Grazing impacts by the limpets on resident macrophytes were high where productivity was highest and lower where productivity was low. Unlike most other coasts in upwelling regions, however, the South African rocky intertidal ecosystem evidently does not harbor a dominant keystone predator, such as the sea stars or whelks discussed earlier (R. Paine pers. comm.; G. Branch pers. comm.). Under some circumstances, oystercatchers (Haematopus moquini) can decimate limpet abundance, leading to increases in algal cover (Bosman et al.

environmental context

31

1986; Bosman and Hockey 1986; Branch et al. 1987). These dynamics tend to be localized in regions that have high inputs of guano and where oystercatchers are not regarded as keystone predators. Thus, in the Benguela upwelling ecosystem, no keystone predator has evolved that exhibits a community function parallel to those seen in the northeast Pacific, the west coast of New Zealand, or the central Chilean coast (Duran and Castilla 1989; see later, under Potential for Compensation). Whether this exception offers further insight into the evolution of keystone species awaits further research.

Biotic Context As noted earlier, the biotic context of a species includes community diversity, particulars of species composition, and characteristics of key interactors. Another factor, the rate of resource replenishment (e.g., recruitment and growth), was considered from a large-scale oceanographic perspective. Aspects of recruitment effects can be considered an important component of biotic context as well, as I discuss in this section. As in other sections, I restrict myself largely to studies of consumer-resource interactions and concentrate on examples taken from my own research activities. Potential for Compensation Community diversity can dictate the response to the loss of a species by regulating the remaining species’ role in compensating for the lost interactor’s effect. Such responses can be key determinants of the persistence and adjustment stability of a community (e.g., Sutherland 1974). For example, in some New England middle and high zones, the absence of predators other than Nucella probably means that wholesale changes in community structure would occur if this whelk were to disappear from the ecosystem. Experiments reported earlier suggest that under this scenario, mussels would become the dominant occupants of midzone space in areas of intermediate wave exposure and perhaps even in sheltered rocky habitats (but see Bertness et al. 1999). Judging from the results of small-scale caging experiments, this change would probably lead to a greatly reduced abundance of fucoids because mussels, when dense, can trap algal fronds with their byssal threads and smother them (Menge 1976). The widespread loss of such an important community component is not far-fetched. Studies in Great Britain in the 1980s revealed that TBT (tri-butyl tin), a chemical used as an antifouling agent, led to sterilization of female whelks, ultimately leading to sharply reduced

32

chapter 2

population sizes (Bryan et al. 1986). The effect of TBT was observed on rocky shores outside of boat harbors and was geographically widespread. I am unaware of studies of the community impact of this loss, but the similarities between New England and Great Britain communities suggest that large increases in mussel cover on rocky shores are likely. This prediction could be tested readily. In Chile, harvesting by humans has drastically reduced the abundance of a large whelk, Concholepas concholepas, with dramatic consequences to the community (Castilla and Duran 1985; Duran and Castilla 1989). At this site, Mussels (Perumytilus purpuratus), normally limited to a narrow band high on the shore, spread downward and reduced the abundance of a variety of algal and herbivore species. In this case, alternative predator species (sea stars Stichaster striatus, Heliaster helianthus) co-occurred within the system. Separate experiments suggested that these sea stars could maintain significantly lower mussel densities than occurred in their absence, suggesting that there exists some potential for compensation (Paine et al. 1985). Sea star effects were not of the same magnitude as those of the whelk (compare Paine et al. 1985 with Duran and Castilla 1989), however, suggesting that in this system the sea stars were relatively weak interactors. Further study would be necessary to test this inference. Although much evidence suggests that keystone species occur in many communities and in all habitat types (Power et al. 1996b), few studies have explicitly tested the keystone species hypothesis (Menge et al. 1994; Menge and Freidenburg 2001). If the deletion of a single species results in a large change in the community, the inference that the species is a keystone species is robust, because the change occurred despite the presence of other consumers in the community (Paine 1966, 1971, 1974). However, the deletion of a putative keystone species with no resulting change does not mean that a keystone species is not present, since the experimenter might have chosen the wrong species. To test keystone predation rigorously, the experimental design should include treatments that remove all individual potential keystone species as well as separate treatments that involve the removal of all consumers. As in the rocky intertidal system studied by Paine on the outer Washington coast, communities on the Oregon coast include a diverse guild of predators, including Pisaster ochraceus and several species of whelks (Nucella ostrina, N. canaliculata, and Searlesia dira). The abundance of these predators can vary strikingly among sites (see earlier for data on Pisaster). Paine’s Pisaster removal experiments clearly implied that, in terms of community impact, whelk predation at his sites was weak. Other studies in the same and in other systems, however,

environmental context

33

suggest that whelks can be strong interactors with large community effects (Connell 1970; Dayton 1971; Luckens 1975; Menge 1976; Duran and Castilla 1989; Fairweather and Underwood 1991). These studies and extraordinarily high densities of whelks at Strawberry Hill on the Oregon coast suggest that under some conditions, whelks might be strong interactors in the low zone (Navarrete and Menge 1996). To test this possibility in the context of keystone predation by Pisaster, we quantified the survival of transplanted M. trossulus in all combinations of Pisaster and Nucella species’ presence and absence. Experiments were established at wave-exposed and wave-protected areas at sites of naturally low and high predator density (Boiler Bay and Strawberry Hill, respectively). Results indicated that, as expected, Pisaster had an overwhelmingly powerful impact on mussel survival whether or not whelks were present. In the absence of Pisaster, however, whelk impacts were surprisingly strong, although still less than that of the sea star. These effects were largely a consequence of high population densities of whelks; the per capita, interaction strength of sea stars was 2–3 orders of magnitude greater than that of whelks. Thus, although Pisaster was unquestionably a far stronger interactor, the normally weakly interacting whelks still have the potential to compensate partially for the loss of sea stars with a increased impact on prey populations. Effects of Species Composition Another example from New England suggests that the resilience of other portions of the rocky intertidal zone could be much greater than that suggested for the middle zones. In the low zone at sites of intermediate and low wave exposure, several other predator species join whelks as predators (see earlier, under Environmental Conditions, and Menge 1983). Caging experiments that were used to test the total impact of predation on this community (Lubchenco and Menge 1978) showed that mussels (M. edulis) eliminated red algal turfs, the normal dominant species at intermediate and sheltered areas. But which predators were responsible for this effect? By combining the feeding rates of each species observed in field experiments with rates of mussel recruitment and predator densities, I could estimate the relative contribution of each species to the predation regime (Menge 1983). These calculations suggested that although one or two species tended to be the dominant predator at each site, the actual dominant varied among sites (fig. 2.3). The crab Carcinus maenas was evidently the most important predator at Canoe Beach Cove, for example, while sea stars (Asterias vulgaris and A. forbesi) were most important at Chamberlain, and whelks dominated at Grindstone Neck. The clear implication is

34

chapter 2

environmental context

35

that in the low zone, predation is generally strong but that the species responsible for this impact varies in space. Further, the community is likely to be resilient to the loss of one or two consumers from this guild. Hence, in a diverse predator guild, compensatory tradeoffs among species may buffer the consequences of species losses in the low zone. A final example comes from the rocky shores of the Pacific coast of Panama, where community diversity is very high. The results from this site further exemplify the buffering effects of a diverse guild of consumers. Compared to the temperate rocky intertidal communities, the rocky shores along the tropical east Pacific coast are nearly barren of sessile organisms. Barnacles, mussels, and oysters are sparse, and macrophytes are very small (generally ⬍2 cm in thallus length) and limited mostly to thin turfs in the low zone (Lubchenco et al. 1984; Menge and Lubchenco 1981). Mobile invertebrates are relatively abundant but are largely hidden away in crevices and holes. The exceptions are grazing gastropods (neritids), which cluster in the high zone, and limpet species (Siphonaria spp., Fissurella spp.), which home to scars on open rock surfaces of the mid zone. In this system, the types of consumers (crabs, fish, whelks, sea stars) and the number of species per type were diverse (Lubchenco et al. 1984; Menge et al. 1986b). Patterns of space occupancy and zonation in this community appeared relatively constant. During 7 years of investigation, the only observed changes in community structure were brief and slight increases in the cover of ephemeral green algae during the early part of the dry season (December–January) (Lubchenco et al. 1984). The high diversity of consumers in this community sharply con-

Figure 2.3 Consequences of the exclusion of predators at Chamberlain, ME; Little Brewster Cove, MA; Grindstone Neck, ME; and Canoe Beach Cove, MA; with estimates of per-species effects. Total predation effects were estimated as the final difference in percent cover of mussels in Ⳮpredator (marked plots, roofs) and ⳮpredator (cages) treatments (see Lubchenco and Menge 1978). The effects of whelks (Nucella lapillus), sea stars (Asterias vulgaris at Grindstone Neck; A. vulgaris and A. forbesi at the other sites), rock crabs (Cancer borealis), and green crabs (Carcinus maenas) were determined using estimates of feeding rates on mussels (Mytilus edulis) in field cage experiments, field predator and prey densities, and mussel recruitment rates quantified using collectors (10-cm ⳯ 10-cm pieces of shag rug glued to marine plywood; see Menge 1983). Predation was generally strong at all sites, but different predators or sets of predators varied in importance among sites. Potential compensation for the loss of a predator species should be high in such a community.

36

chapter 2

strained the design of experiments that tested the effects of predation. Because the removal of individual species would have generated an impossibly large number of treatments, not to mention experimental units, we focused on the exclusion of species in four types of consumers. These were slow-moving grazers (mostly limpets and chitons), slow-moving predators (mostly whelks), fast-moving large fishes (mostly omnivorous consumers of sessile biota), and fastmoving small, benthic consumers (crabs and fishes such as blennies). Using an orthogonal design would have yielded 16 treatments. Despite this simplified design, however, it was impossible to delete the small fishes and crabs while allowing large fishes to enter the plots, so the final design was limited to a (still daunting) 12 treatments in each of three zones: high, middle, and low (Menge and Lubchenco 1981; Menge et al. 1986a). These experiments suggested that in this community, predation was strong but was a cumulative effect of several consumers. The cover of sessile biota increased in the absence of consumers, but the largest increase, by far, was observed only in the total absence of consumers (Menge et al. 1986a). Further, the exposure of dense prey concentrations to consumers revealed that predation was intense; within hours to weeks, even well-defended prey were decimated (Menge et al. 1986b). In Panama, the predation regime was evidently not characterized by a single, or even two or three, strongly interacting species, but by a minimum of four and most likely more consumers. Although by manipulating the consumer groups, we were removing many species and not just one in each treatment, each group was numerically dominated by two to three species (Lubchenco et al. 1984). This predation regime, now termed “diffuse” predation (Hixon 1991; Robles and Robb 1993; Menge et al. 1994), seems to consist of a group of consumers whose effects are additive rather than nonlinear, as would be expected if a keystone predator were present. By itself, each group had a small contribution to total predation (fig. 2.4), and this effect was not detected readily in the treatments involving the removal of only a single group (Menge et al. 1986a). Only by averaging across treatments (i.e., removals of one, two, and three species) differing by the presence or absence of a particular group was the effect in figure 2.4 detectable. Although strictly speaking, these experiments do not reveal the role of individual species, our results suggest that the highly persistent structure of this community is dependent on a taxonomically and functionally diverse array of consumers. Judging by their abundance, size, and foraging patterns, dominant consumers in this assemblage

environmental context

37

Figure 2.4 Effect of consumers, collectively and by each of four single groups (fishes, F; crabs, C; whelks, W; molluscan grazers, G) on total prey abundance in the rocky intertidal region at Taboguilla Island, Panama. Prey included sessile invertebrates (mostly barnacles and bivalves), colonial invertebrates (hydrozoans), and algae. The estimate of the effect of consumers was based on the % cover of prey in the absence and presence of all consumers (total) and each group separately. Singlegroup effects were estimated by determining the difference between treatments that differed in whether the particular group was included in the treatment. For example, fish effects were estimated by treatments ⳭFⳭCⳭWⳭG vs. ⳮFⳭCⳭWⳭG, ⳭFⳭCⳮWⳭG vs. ⳮFⳭCⳮWⳭG, etc. Error bars are 1 SE. Predation was strong but diffuse in this community, which should be highly resilient to the loss of predator species. Data from Menge et al. (1986a), after Menge and Freidenburg (2001).

included the porcupinefish (Diodon hystrix), a wrasse (Bodianus diplotaenia), a parrotfish (Scarus perrico), a damselfish (Stegastes sp.), a predaceous crab (Ozius verreauxii), three whelks (Acanthina brevidentata, Purpura pansa, Thais melones) three limpets (Fissurella virescens, F. longifissa, Siphonaria gigas), a chiton (Chiton stokesi), and an omnivorous crab (Grapsus grapsus). The great diversity of trophic apparatus included in this heterogeneous group is undoubtedly an important factor in holding a diverse group of prey populations at consistently low abundance. The results of this experiment also suggest that the community was

38

chapter 2

highly resilient to species losses. Presumably due to compensatory changes—such as speedy aggregation to concentrations of prey, shifts in prey preference, and population increases—consumer species in this assemblage were able to suppress fluctuations in prey rather quickly. Effect of Species Characteristics A crucial but still unresolved issue in community ecology is determining what factors underlie the interaction strength between consumers and prey. In particular, understanding the ecological and evolutionary conditions that generate keystone species is an important but unsolved problem (Menge et al. 1994; Menge and Freidenburg 2001). Although a variety of species traits and ecological conditions have been proposed as possible explanations, and some of these are undoubtedly important, the evidence remains equivocal. In rocky intertidal communities at least, four factors stand out as potentially critical characteristics: (1) the presence of a strong, competitively dominant space occupier, (2) a consumer that preys disproportionately on the competitive dominant, (3) relatively large sizes of these critical species, and (4) high prey production. As noted elsewhere (Menge et al. 1994; Menge and Freidenburg 2001), the first two traits seem to be associated with strong predation systems in general, not just keystone-dominated systems. There are also apparent exceptions to the second two traits. If whelks can be considered keystone predators in some systems (e.g., Menge 1976, Duran and Castilla 1989), they are not always large relative to competitively dominant prey species (e.g., on New England rocky shores). Some evidence suggests that systems that have keystones are located in upwelling ecosystems with high rates of nearshore productivity, prey recruitment, and prey growth (northeast Pacific, west coast of New Zealand, central coast of Chile). However, as noted earlier, the highly productive Benguela Current ecosystem on the west coast of South Africa is a glaring exception (R. Paine, G. Branch pers. comm.). Here, evidently, there is no strongly interacting intertidal predator. Further study may reveal the special circumstances that have generated such different community dynamics, but for the moment it seems clear that an understanding of some crucial part of the puzzle that leads to the evolution of keystones is still lacking.

Discussion Understanding, and ultimately predicting the likely consequences of species loss seems critically dependent on understanding how the

environmental context

39

abiotic and biotic environmental contexts shape species interactions. In this sense, I argue that environmental context is the overridingly important determinant of a species’ impact. To what extent can we predict the consequences of species deletion based on present knowledge? The consequences of abiotic stress seem clear, at least for invertebrate predators on rocky shores. Under conditions of high thermal stress, desiccation stress, or high hydrodynamic forces, the community impact of predators is likely to be diminished. With respect to abiotic stress, however, at least two issues remain to be resolved. First, at the level of herbivore-plant interactions, heightened stress can increase rather than decrease consumer impacts (Menge and Olson 1990; Louda and Collinge 1992; Olson 1992; B. Menge and A. Olson unpub. data). The basis for this is not yet clear, but it depends in part on size differences among plants and herbivores as well as the greater susceptibility of certain plants to thermal and desiccation stress relative to their consumers. Although these trends have been apparent for some time, progress in understanding them has lagged and further research would be rewarding. Second, the quantification of stress effects on living organisms has, until recently, been stymied by the lack of appropriate techniques for detecting stress. Mass mortality is readily observable but helps little in quantifying the impact of abiotic stress in the context of species interactions. Promising new tools developed by molecular physiologists, however, may spur progress on the issue of detecting and quantifying sublethal stress, how this varies under varying abiotic (including oceanographic) conditions, and its ecological consequences. Methods of determining levels of heat-shock proteins (HSPs) and RNA:DNA ratios, for example, provide relatively quick and inexpensive ways of quantifying the impacts of thermal stress and short-term growth responses to food availability, respectively (Hofmann and Somero 1995, 1996a, b; Dahlhoff and Menge 1996; Roberts et al. 1997). Combining these methods of quantification with field experiments that test species interactions under conditions of different levels of stress (e.g., Dahlhoff et al. 2001; J. Burnaford unpub. data) will likely fuel dramatic advances in the understanding of the roles of variation in stress and food levels. The consequences of varying oceanographic conditions may also be predictable. The studies of rocky intertidal communities in upwelling ecosystems summarized earlier suggest that communities typified by strong top-down species interactions occur where bottom-up processes—specifically, rates of larval delivery and phytoplankton productivity—are high. In Oregon and along the west coast of New Zealand’s South Island, high recruitment and growth dynamics of prey

40

chapter 2

populations seemed fueled by oceanographic conditions that favor the retention of nearshore plankton and the delivery of plankton concentrations to intertidal areas on a frequent basis. High rates of such bottom-up effects seem to underlie strong top-down effects and, with some as-yet unknown condition or process, may underlie keystone species–dominated ecosystems.

Further Observations: New Zealand Comparisons between the west and east coasts of the South Island of New Zealand, and between different sections of the California current ecosystem offer further insight. As summarized earlier, at sites in New Zealand, bottom-up processes were low on the East Coast, an apparent downwelling-dominated ecosystem, and high on the West Coast, an area of intermittent upwelling (Menge et al. 1999). The hypothesized consequence of this difference was strong top-down effects on the west coast and weak top-down effects on the east coast. Preliminary observations during ongoing studies examining the impact of nearshore oceanography suggest that, as predicted, onshore community structure and dynamics vary strikingly as a function of oceanic conditions (B. Menge et al. unpub. data). Communities occurring in downwelling conditions have few carnivores and sparse to near-absent populations of filter-feeding invertebrates. Observations at a total of six sites around the South Island indicate that zonation patterns at some sites are difficult to discern, both because zoneforming invertebrates (mussels and barnacles) are sparse to absent, and because the sharp zone boundaries imposed in other ecosystems by competition or predation are diffuse. Recruitment rates appear very low as well. Only grazers do well on the east coast; high densities of grazers (limpets, chitons, coiled gastropods) occur at all sites. Although dense stands of Durvillea spp. and red algal turf occur in low zones, macrophytes are sparse to absent in the middle and high zones from wave-exposed New Zealand rocky intertidal communities. These communities (near Kaikoura, Christchurch, and Dunedin on northeast, central, and southeast parts of the east coast) contrast strongly with two sites in the northwest upwelling region of the South Island. At these west coast sites, communities are vibrant: filter feeders are dense, grow fast, recruit at high densities; carnivores are abundant and large; and predation is strong (B. Menge et al. unpub. data). A site on the southwest coast, Jackson Head, near Haast, appears intermediate between the communities in the upwelling zone to the north and the east coast sites. Here the two mid- and low-zone mussels (Perna, M. galloprovincialis) are absent, but a small high-zone mus-

environmental context

41

sel Xenostrobus pulex is abundant. Barnacles (Chamaesipho columna, Epopella plicata) dominate space in the high and middle zones. Sea stars and whelks (T. orbita) are common but less abundant than at the northern upwelling-influenced sites. Preliminary evidence suggests that the recruitment of mussels and barnacles is intermediate between the high rates that occur in the upwelling region and the low rates seen on the east coast. These observations and our earlier studies are consistent with the notion that oceanographic conditions can predict patterns of rocky intertidal community structure and dynamics. As expected, mussel and barnacle recruitment rates evidently decline with distance from the upwelling region on the west coast of the South Island. A reduced abundance of adult filter feeders at the southwest, southeast, and northeast sites is consistent with the hypothesis that larval transport (for these species, at least) is low in downwelling ecosystems. The central east coast site, Godley Head on Banks Peninsula, with high barnacle and mussel covers, is an apparent exception to this trend. However, the northward-flowing Southland current is in the lee of the Banks Peninsula, which may generate larvae-concentrating eddies and relatively high subsequent recruitment rates. These possibilities are currently under investigation.

Looking to the Future In asking whether the effects of species loss could be predicted, I have described in some detail the results of much of my own research, in which the context of species interactions exerts an overriding influence on the expected impact of a species loss. A danger in my approach, however, is that the reader may get lost in the detail and miss my attempt to outline a framework for predicting consequences of species extinction. My point is that predicting responses to species loss depends on the context (hence the detail) but that certain principles guide how context should be defined. Key features of context are environmental influences (e.g., productivity and physical stress) and, to some extent at least, the responses to species loss along gradients of these influences is predictable. In attempting to make predictions about species loss, it is clear that a large-scale, long-term perspective that goes well beyond particular species interactions is needed. Using such a perspective will be hard work, and predictions will need to be explored in terrestrial and freshwater as well as marine habitats. To succeed, such endeavors will require new funding, new research partnerships, and the adoption of creative new approaches to the integrative study of ecosystem

42

chapter 2

dynamics. Then will also require a stronger sense of cooperation among scientists that have the expertise required to investigate problems on large scales that necessarily cross disciplinary boundaries. Ecologists will need to incorporate new technologies and methodologies into their research ventures. An understanding of the mechanisms that underlie the phenomena that occur on larger or longer scales will require the use of techniques and technologies that are both under development and in use in fields as diverse as landscape ecology, physical and chemical oceanography, remote sensing, physiology, molecular biology, and genetics. Although it is impossible to forecast the level of precision that might be achieved in predicting the consequences of species loss, I am optimistic that dramatic strides in predictive capacity will be made over the next decade. As witnessed by his scientific legacy, which is at least partly reflected in the articles in this volume, Bob Paine has had a profound impact on ecology. Many of the consequences of species loss, for example, were first revealed by his experiments. These ecological insights attracted attention, and his energy and charisma attracted many students and colleagues eager to participate in the excitement and ferment of his laboratory. Inevitably, many of these students, postdocs, collaborators, and colleagues have also contributed substantially to the ideas and details of ecological systems. Above all, we were encouraged by Bob to be naturalists—to let nature guide us in our probing and prodding and hopefully provoke us to ask questions that would reveal the innermost secrets of the workings of natural systems. For many, the thrill of such activities came in large part from having solved a puzzle—from satisfying one’s curiosity and learning how nature works. These days, of course, nature is reeling, suffering from the tsunami of humanity and its many negative effects, including massive species loss. Many of us hope that the knowledge fostered by Bob and his associates will contribute to efforts to reduce these insults to the Earth. Although humanity has, to date, lacked the political will to take serious action to stem the onslaught, I remain cautiously optimistic that grassroots efforts and our growing scientific awareness will force a global adoption of strong and appropriate measures, and that we can at least slow the growing tide of species loss.

Acknowledgments I thank Simon Levin and Peter Kareiva for the opportunity to contribute to this symposium volume and for their comments, and those of

environmental context

43

three reviewers, on the manuscript. By taking a chance on a na¨ıve Minnesotan with no understanding of how science was done, Bob Paine provided me with a priceless opportunity and a major redirection of my life. I hope that, in a small way at least, my contributions to scientific progress have justified his decision and his ongoing support throughout my career. Much of the work and insights I summarized in this paper was also influenced by collaborations with students and colleagues. In particular, I thank my mate and research partner Jane Lubchenco for her help, support, and tough critiques throughout my career. Colleagues and former students deserving particular mention for their important insights into community dynamics include Gary Allison, Eric Berlow, Elizabeth Dahlhoff, Steve Gaines, Brian Grantham, Patti Halpin, Mark Hixon, Gretchen Hofmann, Sergio Navarrete, Karina Nielsen, Annette Olson, Eric Sanford, George Somero, Ted Strub, and Pat Wheeler. I thank all those in the “Lubmengo” group for their never-ending efforts to keep me humble and young in mind, if not body. This chapter is contribution number 19 from PISCO, the Partnership for Interdisciplinary Studies of Coastal Oceans, a Long-Term Ecological Consortium funded by a grant from the David and Lucile Packard Foundation. My research has been supported by grants from the National Science Foundation, the Andrew Mellon Foundation, a John Simon Guggenheim Fellowship, an endowed professorship from the Wayne and Gladys Valley Foundation, and the David and Lucile Packard Foundation.

Chapter 3

m Species Importance and Context: Spatial and Temporal Variation in Species Interactions Christopher D. G. Harley

The science of ecology must deal with two emerging realities: (1) the human presence on Earth is having an effect on virtually all species, and (2) these effects do not always work in humanity’s best interest. Many academic disciplines—forestry, fisheries biology, and conservation biology, to name a few—focus on developing strategies for managing and sustaining nonhuman species, and for minimizing the harm that we indirectly inflict upon ourselves through resource extraction, food production, and economic development. The common thread uniting these fields is the idea that the benefits of human endeavor (redwood decks, grilled salmon, and sport utility vehicles) must be weighed against the costs (deforestation, fisheries collapse, and atmospheric pollution). The notion that one species might be more or less “expendable” than another is one manifestation of this cost-benefit paradigm.

s p a t i a l a n d t e m p o r a l va r i a t i o n

45

The global extinction of a species is irreversible; a component of the biosphere is lost forever. However, there are situations in which a human-mediated extinction could be acceptable or even desirable (e.g., smallpox), as long as the cost of losing the species is outweighed by the benefits. Therein lies the rub. Determining the costs and benefits of an extinction—or, conversely, the costs and benefits of preventing an extinction—is a complex process. Policymakers typically seek to account for the ecological, economic, cultural, and philosophical components of “cost” and “benefit.” Unfortunately, because so many of these components involve subjective valuation, the question of species expendability on the whole is not a scientific one. Nevertheless, scientists can and should play an important role in informing the debate. Specifically, ecologists can provide valuable information by determining the ecological importance of a species in a community or ecosystem using empirical methods (e.g., Power et al. 1996b; Hurlbert 1997). To relate ecological importance to expendability, we need to define what is meant by “expendable.” For the purposes of this chapter, I define an ecologically expendable species as one whose loss would have a minimal effect on the structure or dynamics of the community or ecosystem as a whole. This definition requires a way of measuring the effect of a species. One common approach involves the experimental removal of the species of interest. The effect of that species can then be determined by comparing removal and control areas in terms of one or more community variables (e.g., densities of the remaining species) or ecosystem variables (e.g., productivity, nutrient cycling, and species richness). This technique can provide estimates of interaction strength (sensu Paine 1992), community importance (sensu Power et al. 1996b), and species importance (sensu Hurlbert 1997). In essence, then, the expendability of a species is inversely proportional to its importance (i.e., the strength with which it interacts with other species and with ecosystem processes). However, the strength of an interaction between a species and a given response variable is likely to vary spatially and temporally, depending on the abiotic and biotic context (for reviews, see Price et al. 1980; Werner and Gilliam 1984; Price et al. 1986; Thompson 1988; Dunson and Travis 1991; Michalakis et al. 1992; Bronstein 1994; Power et al. 1996b; Travis 1996; Lafferty and Kuris 1999; Wellnitz and Poff 2001; Menge this volume). The challenge for ecologists trying to evaluate species importance is to determine the range of variation, and causes of variability, in empirical assessments of importance. In this chapter, I explore the context-dependent nature of species interactions and the variability of species importance through space

46

chapter 3

and time. I begin by defining a metric for estimating interaction strength derived from the formulation by Paine (1992). I use this metric as a basis for several aggregate indices of species importance. I apply these indices to a set of case studies from a rocky intertidal system to highlight differences in interaction strength and species importance among sites and years. I then take a broader look at the literature to gauge the generality and extent of variability in interaction strengths and species importance. I use these results to assess the feasibility of achieving an ecologically defined concept of expendability.

Species Interactions in High Intertidal Communities The typical wave-protected, high-shore rocky intertidal community in the northeast Pacific consists of mobile consumers, sessile invertebrates, and algae. The mussel Mytilus trossulus is the dominant competitor for space, which is the limiting resource (Carroll and Highsmith 1996). The barnacle Balanus glandula is an intermediate competitor, and the smaller barnacle Chthamalus dalli and various crustose algae (predominantly Hildenbrandia sp. and the tetrasporophytic phase of Mastocarpus papillatus) are weak competitors (Dayton 1971; Farrell 1991). Erect algae (Fucus gardneri, Mastocarpus papillatus, and Endocladia muricata) grow preferentially on and among sessile invertebrates and are facilitated by the presence of Balanus (Farrell 1991). These sessile plants and animals provide biogenic habitat for a suite of smaller organisms (Harley 2001; Knox 2001) and serve as a prey base for several consumers. The dominant predator in the system is the sea star Pisaster ochraceus, which preys on Mytilus and Balanus, but rarely eats the smaller Chthamalus (Menge 1972; Menge this volume, pers. obs.). These community interactions are represented schematically in figure 3.1. Notice that Pisaster can influence this assemblage dramatically through direct predation on Mytilus and Balanus, as well as through indirect pathways involving competition or facilitation among the sessile species. My specific interests lie in the temporal and spatial variability of these interactions.

Variation in Time Beginning in early June 1998, I conducted a series of Pisaster exclusion experiments on Saddlebag Island, near Anacortes, Washington. Saddlebag Island marks the outer boundary of the Padilla Bay National Estuarine Research Reserve. Padilla Bay is an orphaned estuary with

s p a t i a l a n d t e m p o r a l va r i a t i o n

47

Figure 3.1 High intertidal interaction web for protected shores in the northeast Pacific. Filled arrows represent negative interactions, and open arrows represent positive interactions. The indirect effects of Pisaster ochraceus mediated by intermediate species are not diagrammed but can be inferred by the chains of direct effects. See text for details.

little freshwater input and minimal wave action. Saddlebag itself features mildly to steeply sloping rock benches that drop off into a deepwater shipping channel along the western flank of the island. The intertidal community is fairly typical of protected sites in the San Juan Islands. Pisaster is the most conspicuous predator, although larvae of the dipteran Oedoparena glauca also inflict substantial mortality on barnacle populations (Harley and Lopez 2002). Predatory whelks (Nucella spp.) are extremely uncommon, and crabs are rarely seen foraging in the intertidal. A more detailed site description is provided elsewhere (see Harley 2001). Two Pisaster-removal experiments (one initiated in 1998, the other in 1999) were conducted on the same rock benches to compare the effects of Pisaster between years. The removal experiments consisted of cages, from which Pisaster was excluded, and open areas, to which Pisaster had access, in a paired design. Cages were constructed from nylon-coated wire mesh with a mesh size of 2.5 cm. Vexar (mesh size ⳱ 6 mm) was attached to the side walls of the cages to deny access to smaller Pisaster. The overall cage dimensions were approximately 25 cm ⳯ 25 cm ⳯ 5 cm. Cage artifacts were minimal: longterm biological trends in cage-control treatments were similar to uncaged treatments on Saddlebag, and caged and open areas were similar at additional sites where Pisaster did not forage in the high intertidal (Harley 2001). For brevity, I present only data from open areas and full cages here.

48

chapter 3

Figure 3.2 Results of Pisaster ochraceus exclusion on Saddlebag Island in 1998 (N ⳱ 10). Although Pisaster directly consumes only Mytilus trossulus and Balanus glandula, it can have strong indirect effects on other taxa. One-way ANOVAs, blocked by treatment pair, were calculated for each sampling date using arcsine square root transformed data (*p ⬍ 0.05, **p ⬍ 0.01, ***p ⬍ 0.001). Error bars indicate standard error. The loss of significance for some taxa at the last sampling date may have resulted from the wintertime reduction of Pisaster foraging. Nevertheless, p ⬍ 0.10 for all comparisons at this time.

At the beginning of the 1998 removal experiment, unusually heavy spring recruitment had produced an ambient cover of Balanus glandula of about 55% (fig. 3.2). Over the next few months, Balanus cover increased in areas from which Pisaster had been excluded, until the dominant barnacles began to be replaced by Mytilus trossulus. The increase in space occupancy of the dominant invertebrates was accompanied by decreases in the competitively inferior Chthamalus dalli and crustose algae. In the open treatments, the spring Balanus set was quickly consumed by Pisaster, and Mytilus never became abundant. As a result of this suppression of the dominant competitors for space, the percentage cover of the competitively inferior Chthamalus and crustose algae increased in the open areas. The result of Pisaster exclusion was a shift from a community dominated by Chthamalus and bare space to one dominated by Balanus and, increasingly, Mytilus (see fig. 3.2).

s p a t i a l a n d t e m p o r a l va r i a t i o n

49

Figure 3.3 Results of Pisaster ochraceus exclusion on Saddlebag Island in 1999 (N ⳱ 7). All else as per figure 3.2. Note the slow (relative to 1998) rate of treatment differentiation for Balanus glandula and the lack of strong indirect effects on Chthamalus dalli and algae.

In the experiment initiated in 1999 (fig. 3.3), the recruitment of both Balanus and Mytilus was substantially lower than in 1998. (Compare the initial values for each in fig. 3.2 and 3.3). In addition to, or as a result of, these recruitment differences, Pisaster appeared to be a less effective predator in 1999. Pisaster’s impact on Balanus was less distinct in 1999, and treatment differences took longer to develop. Chthamalus cover, which was initially high, responded only slightly to treatment differences. Mussels, erect algae, and crustose algae were similar in open and caged areas throughout the experiment (see fig. 3.3). The overall effect of Pisaster exclusion was less striking in 1999 than in 1998. To better quantify the interannual variability in Pisaster’s effects, I calculated Pisaster’s population-level interaction strength (spop, see box 3.1). Because the removal experiments consisted of paired cages and control areas, I was able to calculate spop for each treatment pair. This allowed me to derive means and variances for the strengths of interaction between Pisaster and the sessile community members. Pisaster’s mean interaction strengths were generally weaker in 1999, although the difference was significant only for Chthamalus (fig. 3.4a).

50

chapter 3

BOX 3.1 Quantification of Interaction Strength Interaction strength can be defined as the ␣ij in a community matrix model (e.g., Laska and Wootton 1998). This narrow usage of interaction strength is insufficient for describing the net effect of a species, because it neglects important indirect effects such as trophic cascades and exploitation competition. From the standpoint of total species impact, a more appropriate definition is that developed by Paine (1992), where the sum of the direct and indirect effects of a species are accounted for. Although a number of techniques have been used to calculate interaction strengths (see Laska and Wootton 1998, Berlow et al. 1999 for reviews), I have chosen to base my formulation on Berlow et al.’s (1999) reinterpretation of Paine’s (1992) original metric, which is expressed in the following ratio: NⳮD DY

.

(1)

In this equation, the per capita interaction strength of a focal species on a target species is a function of the number of individuals of the focal species (Y), and the abundance, biomass, or percent cover of the target species in the natural community (N) and in the community from which the focal species has been deleted (D). (Note that the “target species” could just as easily be an ecosystem variable such as productivity or species richness.) One drawback to this formulation is that a strong negative interactor, where a single individual completely removes the target species, has an interaction strength of ⳮ1, while a strong positive interactor, without which the target species cannot exist, has an interaction strength of Ⳮ ⬁. This makes direct comparisons between positive and negative interactors difficult. In order to compare positive and negative interactions on the same scale, I modified Paine’s original formula such that positive interactions have magnitudes ranging from 0 to Ⳮ1.0. This is easily done by redefining per capita interaction strength (sp.c.) as: sp.c. ⳱

NⳮD max(N,D)Y

(2)

where term max(N,D) refers to the larger of the two values, N and D. Thus, a negative interaction is measured relative to the community from which the focal species has been deleted, whereas a positive inter-

s p a t i a l a n d t e m p o r a l va r i a t i o n

51

(Box 3.1 continued) action is measured relative to the “natural” community with the focal species present. Using this formulation, sp.c. can vary from ⳮ1 to Ⳮ1. When there is no net effect of a species, sp.c. ⳱ 0. While per capita interaction strength is a useful quantity, it does not necessarily reflect the importance of a species. For example, a very abundant species with a low sp.c. may be a stronger interactor at the population level than a very rare species with a high sp.c.. Population interaction strength (spop) can be calculated by multiplying sp.c. by the abundance of the focal species (Y), resulting in the following: spop ⳱

NⳮD max(N,D)

.

(3)

The impact of a species is best described by the interaction strength of the entire population of that species (spop). Nevertheless, sp.c. can provide insight into whether changes in spop are due to changes in per capita interaction strength, changes in focal species abundance, or both. ■

Pisaster’s population-level impacts will obviously scale with Pisaster abundance. However, because Pisaster’s predation rate also varies with the size of the Pisaster present (Menge and Menge 1974), the total Pisaster biomass per unit shoreline is likely to be a more accurate reflection of Pisaster’s population-level impact. The observed interannual differences in spop could therefore arise from differences in total Pisaster biomass, differences in Pisaster’s effects per unit biomass, or both. I counted and measured sea stars along fixed transects on Saddlebag Island in both years and then used these data to calculate total Pisaster biomass per unit shoreline (table 3.1). There were no differences in Pisaster abundance or biomass per unit shoreline between years. Thus, when biomass is accounted for using a biomass-corrected calculation of interaction strength (i.e., by substituting Pisaster biomass per unit shoreline for Pisaster abundance in the calculation of per capita interaction strength, sp.c.; see box 3.1), interannual differences in Pisaster’s interaction strength per unit biomass were qualitatively the same as the differences in population-level interaction strength (see fig. 3.4b). Therefore, it seems likely that the betweenyear differences in spop were due to changes in interaction strength per unit Pisaster biomass, as opposed to changes in Pisaster biomass itself.

52

chapter 3

Figure 3.4 Interannual comparison of (A) Pisaster’s population level interaction strength (spop) (B) interaction strength corrected for Pisaster biomass per meter of shoreline. Note the qualitative and quantitative similarities between population interaction strength and per-unit biomass interaction strength, which suggest that interannual differences are due to changes in interaction strength per unit Pisaster biomass. The data (mean Ⳳ s.e.) were calculated from individual treatment pairs (oneway ANOVA, *p ⬍ 0.05, **p ⬍ 0.01, ***p ⬍ 0.001).

s p a t i a l a n d t e m p o r a l va r i a t i o n

53

TABLE 3.1 Spatial and temporal variation in Pisaster abundance, size, individual mass, and total biomass per meter of shoreline SI98 and SI99 represent Saddlebag Island in 1998 and 1999, respectively. CC99 represents Colin’s Cove in 1999. Note that there are no differences between years (Saddlebag only) but substantial differences between sites. Pisaster Characteristics

Site and Date

p (ANOVA)

SI98

SI99

CC99

SI98 v. SI99

SI99 v. CC99

Abundance (噛/m) mean s.e. (n)

2.95 0.33 (3)

2.63 0.31 (4)

0.71 0.11 (4)*

0.52

0.0011

Ray Length (cm) mean s.e. (n)

9.67 0.38 (43)

9.25 0.67 (18)

12.1 0.27 (21)

0.57

0.0002

Mass (g)** mean s.e. (n)

296 24 (43)

276 45 (18)

478 28 (21)

0.68

0.0004

Biomass per Unit Shoreline (g/m) mean s.e. (n)

875 98 (3)

725 86 (4)

339 50 (4)*

0.30

0.0083

*CC abundance data include two sampling dates from 1999 and two from 1998. There were no significant differences between years, so the data were pooled. **Wet mass was calculated using the length-to-mass regression provided by Paine (1976).

The data obtained from the field experiments were used to calculate Pisaster’s importance in the 2 years in three separate ways (box 3.2). The first importance metric (Ia.d.) is based on absolute differences in sessile species cover during the 2 years. These differences were larger for all species in 1998 (compare fig. 3.2 and 3.3), and Ia.d. was significantly higher in 1998 than in 1999 (fig. 3.5). The second two importance metrics (Is and IE) are both based on population interaction strength (see box 3.2). Because spop was, on average, higher for all species in 1998 than in 1999 (see fig. 3.4a), Is and IE were significantly higher in 1998 than in 1999 (see fig. 3.5). Thus, Pisaster was ecologically more important in 1998, regardless of the metric used to estimate importance.

54

chapter 3

BOX 3.2 Aggregate Measures of Species Importance Assigning a single number to the importance (I) of a species in a community is a daunting task. I have taken three separate approaches. The first follows the simple index used by Hurlbert (1997):

IHurlbert ⳱

V

兺ⱍ N

i

ⳮ Di ⱍ

(4)

i⳱1

where V is the number of response variables (species in this case). As was the case for equations 1–3 (see Box 3.1), N and D represent the abundance, biomass, or percent cover of the target species in areas with and without the focal species, respectively. Hurlbert’s metric emphasizes each species according to its abundance. The extinction of a common target species is more important than the extinction of a rare one, because the result is a larger absolute change in abundance. I will use a similar index (Ia.d.) that is based on absolute differences between communities with and without the focal species: V

兺ⱍN

Ia.d.

i

ⳮ Di ⱍ

i⳱1

. V

max(

(5)

V

兺N ,兺D ) i

i⳱1

i

i⳱1

The term in the denominator represents the larger potential abundance, biomass, or cover of the species pool in the two communities, N and D. This normalization allows for direct comparison between studies that measure percent cover (0 ⬍ max(兺N,兺D) ⱕ 100) and those that measure abundance or biomass (0 ⬍ max(兺N,兺D) ⬍ ⬁). Ia.d. ranges from zero (no change) to two (complete turnover of species). The second index is a simple average of the individual population interaction strength values: V

兺 ⱍ(s

Is ⳱

pop)iⱍ

i⳱1

. V

(6)

s p a t i a l a n d t e m p o r a l va r i a t i o n

55

(Box 3.2 continued) As with Ia.d., V represents the number of target species interacting with the focal species. Is emphasizes the relative changes in the abundances of the measured species. Each target species is treated equally; the extinction of a very rare species is equivalent to the extinction of a common one. Is ranges from zero to one. It could be argued that, in a community with two species (A and B), a change from 10A, 10B to 5A, 5B is not as great a change as 10A, 10B to 0A, 10B. Neither Ia.d. nor Is can distinguish between these two alternatives. Therefore, I present a third, Euclidean index of importance:

IE ⳱

√

V

兺 (s

2

pop)i

i⳱1

(7) V

which disproportionately weights large changes in interaction strength. IE ranges from zero to one. ■

There are several possible explanations for these interannual differences. The slight decrease in Pisaster abundance and overall Pisaster biomass from 1998 to 1999 may have contributed, but interaction strength was generally stronger in 1998—and significantly so for Chthamalus, even after correcting for Pisaster biomass. Interannual differences in Pisaster’s importance are, therefore, likely to be due to changes in the effects per unit biomass. These differences may have arisen as a result of the large differences in sessile species recruitment between years, particularly for the two competitively dominant species (Mytilus and Balanus). Theoretical studies have predicted, and empirical studies have found, that higher recruitment is associated with stronger interspecific interactions on rocky shores (Connolly and Roughgarden 1999; Menge this volume). Additionally, water temperatures in Padilla Bay during the summer of 1999 were, on average, about 1⬚C cooler than during the summer of 1998 (D. Bulthuis, pers. comm.). Sanford (1999b) has demonstrated that cooler water temperatures over shorter time scales are associated with reductions in Pisaster feeding rates. Interannual temperature differences in Padilla Bay may, therefore, have contributed to the observed decrease in Pisaster’s effects per unit biomass from 1998 to 1999.

56

chapter 3

Figure 3.5 Interannual comparison of the absolute difference index (Ia.d.), the interaction strength index (Is), and the Euclidean index (IE) of species importance. Means and standard errors were calculated from withintreatment pair comparisons (one-way ANOVA, *p ⬍ 0.05, **p ⬍ 0.01, ***p ⬍ 0.001). Although the three metrics produce qualitatively and quantitatively similar estimates of importance, they are not necessarily correlated strongly with one another (correlation coefficients, with importance index comparisons in brackets, for 1998: r[Ia.d., Is] ⳱ 0.556; r[Ia.d., IE] ⳱ 0.685; r[Is, IE] ⳱ 0.983; for 1999: r[Ia.d., Is] ⳱ 0.673; r[Ia.d., IE] ⳱ 0.530; r[Is, IE] ⳱ 0.971).

Variation in Space When the 1999 series of exclusions was initiated on Saddlebag Island, an additional set of exclusions was established at Colin’s Cove, San Juan Island. This site is approximately 34 km to the west of Saddlebag, and both sites experience similar environmental conditions and harbor a similar suite of species (Harley 2001). However, Colin’s Cove has lower recruitment levels for sessile invertebrates (Harley, unpub. data). Furthermore, Pisaster ochraceus are less common at Colin’s Cove, and Pisaster biomass per meter of shoreline is less than half that of Saddlebag Island (see table 3.1). Although both Mytilus trossulus and Balanus glandula appear in the

s p a t i a l a n d t e m p o r a l va r i a t i o n

57

Figure 3.6 Results of Pisaster ochraceus exclusion at Colin’s Cove in 1999 (N ⳱ 8). All sessile taxa were rare relative to those in Saddlebag Island, and there were no significant differences between treatments at any time (␣ ⳱ 0.05). Note that the scale of the y-axis differs between this figure and figures 3.2 and 3.3.

diet of Pisaster on San Juan Island (Mauzey 1966; Menge and Menge 1974, pers. obs.), experimental exclusion of Pisaster at Colin’s Cove had minimal effects on the sessile community. By the end of the 10month experimental period, there were no significant differences between exclusion plots and controls (fig. 3.6). When compared to Saddlebag in the same year, Pisaster’s effects in Colin’s Cove are slightly weaker for most taxa and significantly weaker for Balanus (fig. 3.7a). When differences in Pisaster biomass are accounted for, there are no significant differences in interaction strength between the two sites (fig. 3.7b). In contrast to the interannual comparison on Saddlebag, it is probable that the observed differences in Pisaster effects between Saddlebag and Colin’s Cove are a result of differences in overall Pisaster biomass, rather than differences in interaction strength per unit biomass. I calculated the importance of Pisaster for both sites in 1999. Only the importance index based on absolute differences in percentage cover (Ia.d.) differed significantly between sites (fig. 3.8). This difference is likely due to the substantial, and significant, differences in

58

chapter 3

Figure 3.7 Between-site comparison of (A) interaction strength of the entire Pisaster population (spop) and (B) interaction strength corrected for Pisaster biomass. Pisaster as a species has a larger impact on Balanus glandula on Saddlebag Island, although this difference disappears when interaction strength is corrected for between-site differences in total Pisaster biomass. The data (mean Ⳳ s.e.) were calculated from individual treatment pairs (one-way ANOVA, *p ⬍ 0.05, **p ⬍ 0.01, ***p ⬍ 0.001).

s p a t i a l a n d t e m p o r a l va r i a t i o n

59

Figure 3.8 Intersite comparison of the absolute difference index (Ia.d.), the interaction strength index (Is), and the Euclidean index (IE) of species importance. Pisaster’s importance varied between sites when estimated with Ia.d. but not when estimated with the other two importance indices. At Colin’s Cove, as with the 2 years on Saddlebag Island, the three indices are correlated, but not perfectly so (correlation coefficients for Colin’s Cove, with importance index comparisons in brackets: r[Ia.d., Is] ⳱ 0.773; r[Ia.d., IE] ⳱ 0.729; r[Is, IE] ⳱ 0.974). All else as per figure 3.7.

Pisaster abundance and total biomass between the two sites (see table 3.1), although intersite differences in prey recruitment and abiotic factors may have contributed as well. The similarity of Pisaster’s importance at the two sites in terms of Is and IE, which are both based on interaction strength, is due at least in part to low sessile species cover at Colin’s Cove. Low species cover increases the frequency of zero values in the percentage cover data, which in turn increases the odds of generating spop values of Ⳮ1 and ⳮ1 by chance. This computational artifact amplifies the variance of interaction strengths among plots. However, because this potential inflation occurs in both the positive and negative directions, the calculated interaction strength should not suffer a directional bias. The calculation of Is and IE, on the other hand, relies on the use of absolute or squared values of interaction strength. Thus, Is and IE (but not Ia.d.) suffer from artificial infla-

60

chapter 3

tion when rare species are included in the analyses. Clearly, this is an important consideration for future determinations of species importance.

Generality of Context Dependence in Interaction Strength and Species Importance The predatory sea star Pisaster ochraceus has had well-documented effects on sessile invertebrate communities—it is the original keystone species (Paine 1969a). Its impact, while often very strong, is known to vary spatially (Paine 1980; Menge, this volume; Menge et al. 1994) and temporally (Sanford 1999b). My results from Saddlebag Island and Colin’s Cove also demonstrate spatial and temporal variability in Pisaster’s interaction strength, as well as in Pisaster’s ecological importance. Pisaster’s maximum and minimum population level interaction strengths, as calculated for Saddlebag Island in 1998 and Colin’s Cove in 1999, respectively, varied by factors of 1.5 and 2.9 for the algal functional groups, and by factors of 3.7 to 5.9 for the sessile invertebrate species. The maximum and minimum calculated differences for importance indices varied by approximately a factor of 2 for Is and IE, and by more than a factor of 5 for Ia.d.. To assess the generality of context-dependent species importance, I culled species-removal experiments from 3 years of Ecology and Oecologia (1995–1997) and 5 years of Ecological Monographs (1995–1999). Acceptable studies were those that removed a single species in multiple contexts. Species-removal experiments were considered to occur in separate contexts if they were performed in separate locations or at different times, or if the experiments involved different combinations of additional interacting species. I excluded studies that removed multispecies units (e.g., “predators” or “molluscs”) and studies that precluded full community responses (e.g., prey-tethering experiments). These criteria yielded 21 studies (see appendix), representing a total of 154 interaction pairs (129 consumer-resource interactions, 15 competitive interactions, and 10 facilitative interactions). The use of ecosystem response variables was rare, so I confined the analysis to species-species interactions. Because the natural abundances of the species removed were not reported consistently, I was able to calculate population-level interaction strength (spop) but not per capita interaction strength (sp.c.). There is considerable variability in spop among contexts (fig. 3.9). It is clear that many, if not most, of the measured interaction strengths are context dependent. In some cases, species that are weak interactors in one situation are strong interactors in another (points lying

s p a t i a l a n d t e m p o r a l va r i a t i o n

61

Figure 3.9 Scatter plot of interaction strengths in multiple contexts, based on data from the literature. Each point represents one interaction pair (i.e., one focal and one target species) in each of two contexts. The first context described in a publication was arbitrarily designated Context 1, and the second was designated Context 2. Open squares and triangles represent the interannual comparison (1998 vs. 1999 on Saddlebag Island) and the intersite comparison (Saddlebag Island vs. Colin’s Cove in 1999) described in this chapter, respectively. The dashed line represents no context dependency; it represents a ⱍ:ⱍ relationship (i.e., no difference in spop from one context to the other.

near the axes). In other cases, species that have a strong negative effect in one context have a strong positive effect in another (points lying near the upper-left and lower-right corners). Overall, the measured spop in one context explains little more than one-third of the variance in the other context (r2 ⳱ 0.37). In the absence of per capita data, context dependency due to per capita changes and due to changes in focal species abundance cannot be teased apart. Of these 21 studies, 11 (comprising 18 separate experiments) measured the responses of 3 or more species per context per experiment. I use these studies to investigate the context dependence of species importance, as estimated earlier for Pisaster. All three metrics (Ia.d., Is,

62

chapter 3

Figure 3.10 Context-dependency scatter plots for the absolute difference index (Ia.d.), the interaction strength index (Is), and the Euclidean index (IE) of species importance, based on data from the literature. Symbols as in figure 3.9. Although the different indices show slightly different patterns of context dependency, correlations between indices are high (r[Ia.d., Is] ⳱ 0.833; r[Ia.d., IE] ⳱ 0.858; r[Is, IE] ⳱ 0.986; Pisaster data are included). The choice of an index depends on what the investigator considers to be the most accurate reflection of importance, given the available data.

and IE) showed considerable variability from context to context (fig. 3.10). For comparative purposes, I recalculated the index values for my Pisaster work based on the experimental means (rather than individually for each experimental pair, as was done in the previous section). Pisaster’s importance in 1999 is relatively similar between Saddlebag Island and Colin’s Cove (see fig. 3.10, open triangles). Pisaster’s importance between years on Saddlebag is strikingly different

s p a t i a l a n d t e m p o r a l va r i a t i o n

63

(see fig. 3.10, open squares), although no more so than in examples from the literature. Although species importance does vary substantially from one context to another, this context dependency does not appear to be as large as the context dependence of interaction strength (see fig. 3.9). This muting of variability presumably arises because the importance indices are aggregate measures that tend to smooth out some of the variability seen in the interactions between individual species pairs. Furthermore, species with context-independent importance values may still play radically different roles in different situations. As long as the interactions between the focal species and the other community members average out, importance does not change even when the identities of the strongly or weakly affected species vary dramatically from one context to another.

Utility of the Expendability Concept Virtually everything that humans do affects some species somewhere. In many cases, our actions result in extinctions. Given that we can prevent some, but not all, of these human-mediated extinctions, it is critical to understand how important any given species is so that we can prioritize our conservation efforts. As demonstrated in this chapter, it is possible to calculate the importance of a species. However, the quantification of species importance may overlook the ecologically significant, context-specific roles of a species. The measured importance of a species can change through space and time with variation in the abundance of that species (Fauth 1999; also, the intersite comparison in this chapter), and with variation in its per capita effects on other species or processes (Navarette and Menge 1996; Wootton 1997; Ruesink 1998; the interannual comparison in this chapter). Both the abundance and the per capita effects of a species are known to vary with biotic and abiotic contexts. At the population level, the age structure, size structure, and genotypic characteristics of interacting populations influence the outcome of interspecific interactions (Lynch 1978; Hairston 1980; Werner and Gilliam 1984; Thompson 1988; Cushman and Whitham 1989). At the community level, the impact of a focal species obviously depends on the identity of the species with which it interacts directly. Furthermore, the presence of additional species can modify the magnitude and even the sign of any particular interspecific interaction (Park 1948; Smith 1968; Cushman and Whitham 1989; Michalakis et al. 1992; Wootton 1993; Bronstein 1994; Nowlin and Drenner 2000). These di-

64

chapter 3

rect and indirect interactions may be nonlinear functions of the density of the focal or target species (Knowlton 1992; Ruesink 1998). All the biological interactions listed previously may be modified further by environmental variables. Species abundances and per capita interaction strengths may depend on temperature (Park 1954; Aizen and Patterson 1995; Sanford 1999b), pH (Warner et al. 1993), salinity (Dunson and Travis 1991), light (Dudt and Shure 1994), precipitation (Polis et al. 1997; Linthicum et al. 1999), humidity (Park 1954), desiccation (Trowbridge 1998), nutrient availability (Pace and Funke 1991; Set¨al¨a et al. 1997), substrate characteristics (Power 1992), soil characteristics (Tansley 1917), elevation (Preszler and Boecklen 1996), depth (Palumbi 1985), wave exposure (Menge 1978b, this volume; Navarette and Menge 1996), and fire regime (Collins et al. 1998). This long list of potentially important—and potentially interacting—influences on species importance makes it exceedingly difficult to separate those species that are generally or occasionally important from those that are never important and thus “ecologically expendable.” Assuming there is sufficient funding and motivation, improved estimates of importance (using the indices presented here or other metrics) could be made by repeating experiments in multiple locations and at different times. But how many places and times are sufficient? If a species interacts strongly but only under temporally or spatially rare conditions, those rare occasions may nevertheless influence long-term community dynamics. For example, gypsy moths (Lymantria dispar) reach outbreak proportions only once per decade or so, but these outbreaks have lasting impacts on forest dynamics (Davidson et al. 1999). Other outbreaking species (e.g., crown-of-thorns starfish, the pathogen associated with die-off of the sea urchin Diadema antillarum in the Caribbean) have similar long-term effects on communities and ecosystems (Lessios 1988; Seymour and Bradbury 1999). Another challenge is to incorporate those species that provide important energetic links across habitats (e.g., Bustamante et al. 1995; Willson et al. 1998; Nakano et al. 1999). In many cases, species that do not occur in the system of interest may nonetheless be important to that system. The oceanic production of plankton, for example, indirectly influences inland riparian habitat via salmon migration (Willson et al. 1998; Hare et al. 1999), and high-elevation forests may influence aquatic communities far downstream via control over the hydrological regime (Wootton et al. 1996; Poff et al. 1997). In light of the spatial and temporal complexities involved in species interactions, any measurements of expendability must be made at appropriate spatial and temporal scales. Given the challenges and complexities just outlined, the term ecologically expendable cannot be applied with confidence in the absence

s p a t i a l a n d t e m p o r a l va r i a t i o n

65

of substantial empirical data. However, the converse is not true; the following are general starting points for the a priori identification of ecologically important, nonexpendable species in the absence of experimental evidence: 1. Species that attain or maintain high abundance are likely to be important, because even the weakest per capita interactors can exert influence through sheer numbers. 2. Species with very high consumption rates may be disproportionately important, as they are likely to have strong per capita effects on their prey. Even if these species are typically rare, their importance may be manifested in occasional outbreaks. 3. Species that have many consumers (key-industry species, Elton 1927) are likely to be important both directly, by supporting a diverse group of consumers, and indirectly, through the impacts these elevated predator populations may have on alternative prey species (e.g., apparent competition; Holt 1977). 4. Species that modify the environment or provide habitat (e.g., ecosystem engineers; Jones et al. 1994) are likely to be important due to their potential to influence the abundances of and interactions among other species. 5. Any species that is the sole representative of a particular functional group will likely be important. Bluegill sunfishes, for example, are strong interactors when they are the only piscine predators in the system, but they are weak interactors when other predatory fish are also present (Nowlin and Drenner 2000). 6. Species whose natural history is tightly linked to any of the species in categories 1 through 5 may also be important. This ecological trait is the hallmark of a keystone species (Paine 1969; Power et al. 1996b), although in terms of species importance it applies equally to keystone and nonkeystone species. Species in this group include mutualists, pollinators, seed dispersers, endosymbionts, parasites and hosts, disease organisms and vectors, specialist predators, and preferred prey items.

Conclusions The desire for a metric of expendability is understandable; such a metric would allow planners and resource managers to make betterinformed decisions regarding a variety of human actions. However, assigning an “expendability quotient” to species should not be undertaken lightly. The cost of error can be high, since extinctions and their

66

chapter 3

corresponding losses of genetic material are irreversible. Because expendability can vary substantially through space and time, estimating expendability is risky if done poorly but expensive if done well. Nevertheless, the world will continue to turn, policy decisions will continue to be made, and some species will of necessity be prioritized higher than others. Given this reality, ecologists must provide data to allow policy makers to make informed decisions. Even without substantial empirical data, a solid foundation in natural history will make the selection of potentially inexpendable species enormously easier. Adaptive management is a particularly promising approach to incorporating science into management decisions, and vice versa. Increasing our understanding of how and why species interactions vary from one context to the next will be invaluable in making management decisions in the inevitable absence of perfect knowledge.

Acknowledgments I express my sincere gratitude to Bob Paine, who has fostered my interest in marine ecology, tolerated my attempts to prove him wrong, and provided me with invaluable support, encouragement, food, insight, and more food throughout my graduate career (and beyond). The ideas presented in this manuscript benefited from the input of E. Crozier, M. Graham, P. Kareiva, R. T. Paine, J. L. Ruesink, M. D. Scheuerell, D. E. Schindler, M. J. Wonham, and four anonymous reviewers. Fieldwork on Saddlebag Island would have been unenjoyable, if not impossible, without the assistance of D. A. Bulthuis, M. “Cap’n” Olson, and S. R. Riggs. Funding was provided by an Achievement Rewards for College Scientists Foundation fellowship, a National Science Foundation doctoral fellowship, and a Padilla Bay National Estuarine Research Reserve research assistantship. STPB.

Appendix to Chapter 3 Studies used to assess the context specificity of interaction strength. Those marked with an asterisk were used in the calculation of Ia.d., Is, and IE.

Balˇciunas, ¯ D., and S. P. Lawler, 1995. Effects of basal resources, predation, and alternative prey in microcosm food chains. Ecology 76: 1327–1336.

s p a t i a l a n d t e m p o r a l va r i a t i o n

67

*Berlow, E. L. 1997. From canalization to contingency: historical effects in a successional rocky intertidal community. Ecol. Monogr. 67: 435–460. Diehl, S., and P. Eklov. 1995. Effects of piscivore-mediated habitat use on resources, diet, and growth of perch. Ecology 76: 1712–1726. *Downs, B. J., P. S. Lake, E.S.G. Schreiber, and A. Flaister. 1998. Habitat structure and regulation of local species diversity in a stony, upland stream. Ecol. Monogr. 68: 237–257. *Estes, J. A., and D. O. Duggins, 1995. Sea otters and kelp forests in Alaska: generality and variation in a community ecological paradigm. Ecol. Monogr. 65: 75–100. Ferguson, K. I., and P. Stiling, 1996. Non-additive effects of multiple natural enemies on aphid populations. Oecologia 108: 375–379. Green, P. T., D. J. O’Dowd, and P. S. Lake. 1997. Control of seedling recruitment by land crabs in rain forest on a remote oceanic island. Ecology 78: 2474–2486. *Guti´errez, J. R., P. L. Meserve, S. Herrera, L. C. Contreras, and F. M. Jaksic. 1997. Effects of small mammals and vertebrate predators on vegetation in the Chilean semiarid zone. Oecologia 109: 398–406. Halpern, C. B., J. A. Antos, M. A. Geyer, and A. A. Olson. 1997. Species replacement during early secondary succession: the abrupt decline of a winter annual. Ecology 78: 621–631. Hill, W. R., M. G. Ryon, and E. M. Schilling. 1995. Light limitation in a stream ecosystem: responses by primary producers and consumers. Ecology 76: 1297–1309. Kupferberg, S. J. 1997. Bullfrog (Rana catesbeiana) invasion of a California river: the role of larval competition. Ecology 78: 1736–1751. McAlister, S. 1995. Species interactions and substrate specificity among loginhabiting bryophyte species. Ecology 76: 2184–2195. Menge, B. A., B. A. Daley, J. Lubchenco, E. Sanford, E. Dahlhoff, P. M. Halpin, G. Hudson, and J. L. Burnaford. 1999. Top-down and bottom-up regulation of New Zealand rocky intertidal communities. Ecol. Monogr. 69: 297–330. *Meserve, P. L., J. R. Guti´errez, J. A. Yunger, L. C. Contreras, and F. M. Jaksic. 1996. Role of biotic interactions in a small mammal assemblage in semiarid Chile. Ecology 77: 133–148. O’Connor, T. G. 1995. Acacia karroo invasion of grassland: environmental and biotic effects influencing seedling emergence and establishment. Oecologia 103: 214–223. *Pennings, S. C., and R. M. Callaway. 1996. Impact of a parasitic plant on the structure and dynamics of salt marsh vegetation. Ecology 77: 1410– 1419. *Proulx, M., F. R. Pick, A. Mazumder, P. B. Hamilton, and D.R.S. Lean. 1996. Effects of nutrients and planktivorous fish on the phytoplankton of shallow and deep aquatic systems. Ecology 77: 1556–1572. *Stadler, B., and T. Muller. ¨ 1996. Aphid honeydew and its effect on the phyllosphere microflora of Picea abies (L.) Karst. Oecologia 108: 771– 776.

68

chapter 3

*Vanni, M. J., and C. D. Layne. 1997. Nutrient recycling and herbivory as mechanisms in the “top-down” effect of fish on algae in lakes. Ecology 78: 21–40. *Vanni, M. J., C. D. Layne, and S. E. Arnott. 1997. “Top-down” trophic interactions in lakes: effects of fish on nutrient dynamics. Ecology 78: 1–20. *Weltzin, J. F., S. Archer, and R. K. Heitschmidt. 1997. Small-mammal regulation of vegetation structure in a temperate savanna. Ecology 78: 751– 763.

Chapter 4

m Effects of Removing a Vertebrate versus an Invertebrate Predator on a Food Web, and What Is Their Relative Importance? Thomas W. Schoener and David A. Spiller

Ecologists routinely simulate species loss. At first, these simulations were carried out mainly in theory, but over the last couple of decades, they have increasingly been carried out in practice via socalled “removal experiments.” It is tempting to think that such experiments bear on the question of species importance, since they involve comparisons among communities with and without the removed species. The spatial scale is of course contracted: species are typically eliminated from very small spatial units (Kareiva and Andersen 1988). For procedural reasons as well as the need to capture the underlying spatial structure, the bias toward a small spatial scale may be important (Tilman and Kareiva 1997; Peterson and Parker 1998; Inouye 1999; Schoener and Spiller 1999a). The temporal scale of removal experiments may also be a problem. The indirect effects of species removal, for example, may take a long time to be realized fully (Yodzis 1989;

70

chapter 4

Schoener 1993); although this may not be as much of an issue as once thought, both because of increasingly longer experiments and, in some systems, the surprising rapidity of indirect effects (Menge 1997). Although the absence of an effect in a species-removal experiment is no guarantee of a lack of value, the presence of an effect is some evidence for concern about that species’ loss. In fact, if one uses economic impact in an evaluation (see later), then a species showing few or no community-wide effects should ceteris paribus be less important than one having such effects.

An Experiment that Evaluates the Relative Impact of Species Removal Fortuitously, and perhaps fortunately, approximately when the topic of this symposium was crystallized, we had just completed an experiment in which the most abundant lizard and web-spider species were removed from the same site. The design was a 2 ⳯ 2 factorial—the presence or absence of the lizard species crossed with the presence or absence of the spider species—yielding the following four treatments (fig. 4.1): (1) control (both unmanipulated), (2) lizards removed, spiders unmanipulated, (3) spiders removed, lizards unmanipulated, and (4) both removed. The purpose was to evaluate predator-mediated interactions among prey (Spiller and Schoener 2001), but the data collected for that purpose, as well as other data collected incidentally, can be applied toward the goals of the present symposium. Details of the experimental procedures are given elsewhere (Spiller and Schoener 2001). Briefly, the site was located on Staniel Cay, Bahamas, in an area of shrubby vegetation consisting mostly of Coccoloba uvifera (sea grape) 0.5 to 1.5 m high. Large enclosures surrounded each treatment arena and had barriers impeding lizard movement in or out. Throughout the experiment, spiders as well as errant lizards were removed manually where prescribed, and removal efficiency was virtually identical for the two species. Response variables were monitored at 2- to 3-month intervals throughout the 30-month duration of the experiment. The spider variables measured were the number (and properties) of Eustala cazieri, the second most-abundant spider, and numbers of the rarer spider species (methods in Spiller and Schoener 1998). As in previous studies, the abundance of aerial arthropods in two size classes was determined with sticky traps. Foliage damage, however, was measured differently than in previous experiments. Specifically, each year in May, we tagged the apical ends of 15 branches in each plot. The distribution of tagged branches in

e f f e c t s o f p r e d a t o r r e m o va l

71

Figure 4.1 Experimental design for simultaneous removal of the most common lizard (Anolis sagrei) and spider (Metepeira datona).

each plot consisted of 5 uniformly spaced sites with 3 branches per site. In the next three sessions, we tagged all new leaves that emerged on each branch. We measured (to the nearest mm) the maximum diameter and the diameter perpendicular to the maximum diameter of each leaf and each damaged area on the leaf (as in Spiller and Schoener 1990b, 1994, 1997); total leaf area and damaged areas were estimated by assuming that they were elliptical. This present method of measuring all new leaves is yielding amounts of damage 4 times higher than our previous method (which quantified damage using only fully expanded leaves). Apparently, most damage occurs before full expansion (as in Coley 1983; Lowman and Box 1983). The impact of removal was determined for each of the five variables listed in table 4.1, using comparisons of time averages from mid-July 1995, to mid-December 1997 (the experiment began in midMay 1995). The four treatments allow two sensible ways of measuring impact according to which treatments or combinations of treatments are being compared. First, the control treatment, in which no species is manipulated, can be compared with the treatment in which only one of the species (e.g., the lizard) is removed. This approach uses two of the four treatments. A potentially more accurate, although less

72

chapter 4

TABLE 4.1 Effect magnitudes of the removal of A. sagrei (lizard) and M. datona (spider) Web Spiders

Method 1 A. sagrei Ratio Sign M. datona Ratio Sign

Aerial Arthropods

Leaf Damage

E. cazieri

Rare species

Small

Large

1.80 ⳮ

3.74 ⳮ

1.18 Ⳮ

1.31 Ⳮ

2.98 ⳮ

1.44 Ⳮ

1.95 Ⳮ

1.11 ⳮ

1.02 Ⳮ

1.15 ⳮ

2.18 ⳮc

4.68 ⳮc

1.14 Ⳮ

1.26 Ⳮ

2.42 ⳮc

1.09 Ⳮ

1.24 Ⳮ

1.15 ⳮc

1.01 ⳮ

1.18 Ⳮ

a

Method 2b Ratio Sign M. datona Ratio Sign

a 1 vs. 2 for A. sagrei or 1 vs. 3 for M. datona; numbers refer to means of treatments as labeled in fig. 4.1. b 1 Ⳮ 3 vs. 2 Ⳮ 4 for A. sagrei or 1 Ⳮ 2 vs 3 Ⳮ 4 for M. datona; numbers as in footnote a. c Raw P ⬍ 0.05 in full ANOVA without correcting for multiple comparisons (see text). Note that statistics are not done for method 1, only method 2.

representative, method uses all four treatments: to determine the effect of lizard removal, for example, one compares both treatments without lizards to both treatments with lizards (see fig. 4.1). Because the number of replicates per treatment (in this case, three) is the same for all four treatments, one can compare the average of treatments 1 and 3 with the average of treatments 2 and 4 (the spider comparison would use 1 and 2 vs. 3 and 4). The second method would be questionable if there were a significant statistical interaction between the factors (i.e., if the difference between the control and the removal of spiders were significantly more [or less] in the presence than in the absence of lizards). However, because no response variable showed a significant interaction in its respective ANOVA, the second method appears valid. For the second method only, we obtained P values for each removed species from the ANOVA on the entire factorial design. Table 4.1 indicates which of these values is less than 0.05 by itself, i.e., without correcting for multiple comparisons. A multiple-comparison analysis is given in Spiller and Schoener 2001. To compare effect magnitudes with other approaches, we use the

e f f e c t s o f p r e d a t o r r e m o va l

73

ratio of removals to nonremovals—or vice versa, whichever is larger (see table 4.1). We have been reporting this measure (e.g., Schoener and Spiller 1987) instead of the absolute difference in logs, to which it is monotonically related, because it is easier to intuit. Were we interested in statistically analyzing effect magnitudes, we would use the log of the ratios (⳱ the difference in logs) as in Spiller and Schoener (1995). The latter meshes well with our ANOVA analysis of the factorial table, since values for the first four response variables in table 4.1 are log transformed (and the fifth is arcsine–square root transformed) before statistical analysis.

Experimental Results The removal of A. sagrei caused about a doubling of the commonest unmanipulated web-spider species, E. cazieri, and about a four-fold increase in the combined abundances of rarer spider species (fig. 4.2, table 4.l; raw Ps for method 2 ⬍ 0.05). In contrast, the effect of M. datona removal on the same spider variables was much smaller, with raw Ps exceeding 0.05 (Spiller and Schoener 2001). Neither the removal of A. sagrei nor M. datona much affected the number of aerial arthropods in sticky traps (fig. 4.3). Whereas A. sagrei removal increased total leaf damage somewhat less than three-fold, with raw P less than 0.05 (fig. 4.4), the M. datona effect on leaf damage was much smaller and not significant. Importantly, the overall compilations of results were generally similar, regardless of which method was used to construct “effect ratios” (see table 4.1): most or all ratios were greater for A. sagrei than for M. datona. Finally, another way of evaluating the relative effects of removal is to count the number of times a raw P value of 0.05 or better is obtained. Removal of A. sagrei gave three such values, whereas the removal of M. datona gave only one. (The apparent directional changes given in table 4.1 should be viewed with these statistical evaluations in mind.)

Significance, Interpretation, and Implications For the purposes of this symposium, perhaps the best measure of effect magnitude is the effect of removal of the species as a whole. It is this effect that would be of issue for species loss. By our version of this measure, the most common lizard species has a much greater effect than the most common spider species. If, however, we were interested in the per capita effects of removal, the difference reported here between lizards and spiders would be much exacerbated, since

74

chapter 4

Figure 4.2 Mean number of web spiders (left, Eustala cazieri; right, rarer species) in the four treatments during the removal experiment. Bars represent 1 standard deviation (SD) of the plot means within each treatment.

Figure 4.3 Mean number of arthropods (per enclosure) caught in sticky traps (left, arthropods between 1 and 4 mm; right, arthropods ⬎ 4mm) during the removal experiment. Bars as in fig. 4.2.

e f f e c t s o f p r e d a t o r r e m o va l

75

Figure 4.4 Mean leaf damage (% total area damaged/% total leaf area; this measure of damage is Ds in Schoener 1988 and is further justified therein) during the removal experiment. Bars as in fig. 4.2.

the removed spider species occurs at about 10 times the density of the removed lizard species. If we were ecosystem ecologists interested in the effects per gram of predator, the relative effects would shift back again to somewhere near the original comparison. We now ask three questions about our result. First, is it representative (i.e., likely to characterize removal experiments with other lizards or spiders that may be performed at different times or with different procedures)? Second, why do lizards have so much more of an effect than spiders? Third, what do the relative effects of removing lizards versus spiders imply about their expendabilities, if anything? Are the Present Results Representative of Relative Effect Ratios in This System? Because no other experiment (or comparative study) was designed to show the effect of removing the most abundant species of spider versus lizard simultaneously, we cannot obtain a direct answer to this question. However, two oblique approaches allow a partial evaluation. First, previous data on lizard removal per se, some at the same site, allow us to ascertain whether lizard effect ratios in the present experiment were extreme. Second, a previous experiment at the same site removed all web spider species versus all diurnal lizard species in a crossed design (Spiller and Schoener 1994). In the case of lizards, this is practically the same as removing only A. sagrei, since that spe-

76

chapter 4

cies comprises 93% of all lizard individuals observed. In the case of spiders, the most abundant species comprised only 73% of all spider individuals combined, which means that the removal of all web-spider species might show a substantially greater effect than the removal of the most common species. However, if the effect of removing all spider species is small, then the effect of removing the most common species will also be small (an inference that can be made because there is no evidence, experimental or comparative, for any negative interactions among the web spider species populations in this system). Lizard removal was studied in two previous experiments at the present site (Staniel Cay) using the same kind of enclosures. The first experiment (Spiller and Schoener 1988), lasting 3.5 years, did not distinguish between spider species in measuring the overall lizard effect on numbers; the effect ratio was 3.1. (These and most other ratios given below are summarized in table 3 of Schoener and Spiller 1999a.) In the second experiment, lasting 4.5 years, the ratio for E. cazieri was 2.6, and ratios for the rarer species ranged from 3.7 to 53.0 (Spiller and Schoener 1998). Compared with the present ratio of 1.8 (or 2.2; table 4.1) on E. cazieri and 3.7 (or 4.7) on rarer species, the lizard effect on spiders was, if anything, weaker in the present experiment than typically observed. In turn, lizard effects on all web spiders combined at the large island of Staniel Cay tend to be weaker than those on the medium to large islands (5.7) or on the small to medium islands (5.0– 7.0). Of these latter two island data sets, the first is comparative, but the second is experimental and illustrates the devastating effect lizards can have on spiders on the smallest islands studied. In that experiment (Schoener and Spiller 1996), A. sagrei was introduced onto islands without lizards, while similarly sized islands with or without lizards served as “bracketing” controls. Within 2 years of introduction, lizard-introduction islands (islands on which lizards were introduced) became identical in spider density and species richness to those naturally having lizards (fig. 4.5). The proportion of species becoming permanently extinct (i.e., not reimmigrating during the 7 years of the experiment) was 12.6 times higher for lizard-introduction islands than for islands without lizards (fig. 4.6). In this experiment and in a lizard-removal experiment at the Staniel site (Spiller and Schoener 1998), spider species that became extinct were the rarer ones. The effect of lizards on aerial arthropods was not significant in the present experiment, and effect ratios were substantially lower than those of lizards on spiders, ranging from 1.1 to 1.3. The comparable ratios from previous experiments at this site were also generally small; only the negative effect on large arthropods in the first experiment (Spiller and Schoener 1990a) had a ratio exceeding 1.3 (it was 1.7, the only

e f f e c t s o f p r e d a t o r r e m o va l

Figure 4.5 Numbers of species (above) and densities of individuals (below) of diurnal web spiders through time on islands in the introduction experiment (Schoener and Spiller 1996); leftmost values are pre-experimental. Symbols give mean (n ⳱ 4 islands in each case); bars represent Ⳳ 1 SD of the plot means within each treatment at the time of census.

77

78

chapter 4

Figure 4.6 Mean % extinction (proportion of species initially present that became extinct and never recolonized) of spider species in the introduction experiment (Schoener and Spiller 1996). Bars represent 1 SD of the plot means within each treatment.

statistically significant effect), and others were 1.0 to 1.1. On very small islands, lizards significantly increased small arthropods (Schoener and Spiller 1999b), but ratios were again small (1.2–1.4); ratios for large arthropods were slightly more (1.5–1.7) but varied in direction and were never significant. The lizard-effect ratio on total sea-grape leaf damage in the present experiment was 3.0 (or 2.4; see table 4.1). This value is rather similar to others obtained in this system experimentally (2.0, 3.3) or comparatively (3.2), despite major differences

e f f e c t s o f p r e d a t o r r e m o va l

79

in methodology and the absolute amount of damage obtained (see earlier). In conclusion, the effect ratios for lizard removal in the present experiment are either similar to or less than the effect ratios of lizards measured previously. We can also examine effect ratios for spider removal in previous studies, although only for the removal of all web spider species (Spiller and Schoener 1994). Effect ratios for spider removal were 1.1 (method 1) or 1.2 (method 2) on small arthropods, and 1.3 (both methods) on large arthropods. While the effect on small arthropods was nearly identical to that of the present experiment, the effect on large arthropods was somewhat larger (see table 4.1). Indeed the effect of spider removal was statistically significant for either size class, whereas in the present experiment only the spider effect on small arthropods was significant (but only using the raw P). The effect on total leaf damage was nonexistent in the previous experiment (values for spider removal alone were virtually identical to those of the control); while it was somewhat larger in the present experiment, it was still not significant. After the first experiment at the Staniel site (Spiller and Schoener 1990b), we hypothesized that one type of leaf damage (galls of the cecidiomyiid midge Ctenodactylomia watsoni) was reduced by spiders and hence indirectly increased by lizards; it was significantly greater in controls than in lizard removals. However, the second experiment, designed specifically to test for this effect, was foiled by extremely low abundances of the midge. During the present experiment, neither M. datona nor A. sagrei significantly affected galls. In summary, the effect of total web-spider removal, where it could be compared, was the same or only slightly greater in a previous experiment than the effect of M. datona removal in the present experiment. The overall conclusion is that, if anything, the present effect as summarized in table 4.1 is a conservative one for testing the claim that the effect ratios of the most common spider are substantially smaller than those of the most common lizard. Why Do Spiders Have Weaker Effects than Lizards? As we have already reviewed, the answer to this question does not lie in the relative abundances of the two groups: the most common web spider averages 10 times the number of individuals of the most common lizard. Of course, lizards are much larger per individual, and although consumption might be expected to rise with some fractional power of body weight (note that both species are ectothermic), again we are dealing with a difference of several orders of magnitude. Both lizards and spiders are generalized predators. Both eat a range of prey sizes, although both prefer the larger sizes of those available

80

chapter 4

(Spiller and Schoener 1990b). On average, A. sagrei eats larger prey than M. datona. Because of their less restricted prey-capture style, greater mobility, and broader range of habitat, as well as their larger size lizards can perhaps efficiently capture more kinds of prey than can spiders. Certainly, a major example of a kind of prey more affected by lizards than web spiders are the herbivorous Lepidoptera. Lizards, but generally not web spiders, are able to capture these insects in their larval stage; while the opposite might be thought true for the adults, butterflies and moths are notorious in being able to “slip through” spider webs, probably largely as a result of their detachable scales (Eisner et al. 1964). In the present experiment, hole damage, caused in large part by lepidopteran caterpillars, was by far the most abundant type of damage. Thus, the lizard effect on total herbivore damage would be expected to be substantially greater than that of the spiders. Of the possible response variables, the one spiders might be expected to affect the most is dipterans. In fact, that group constituted 59% of the prey caught in the webs of M. datona (Spiller and Schoener 1990a). Dipterans do constitute the majority (59–68%) of small aerial arthropods caught in sticky traps, and indeed the spider effect is strongest for small, aerial arthropods. Although a few of these dipterans should be gall midges, the majority may be detritivores, and we have suggested (Spiller and Schoener 1996) that spiders might be important for that functional group. Nonetheless, their effect ratio even on small arthropods is small compared with most lizard effect ratios. Although special explanations are thus possible for certain response variables, the consistently low effect of spiders, even where one expects it to be strongest, seems to require a more general explanation. We have proposed that, in general, spider populations are held well below their maximum size by predators, physical factors, or a combination of these. In the present experiment, even when lizards were removed, spiders did not compete (Spiller and Schoener 2001), and indeed this is consistent with most other studies of web spiders (Wise 1993, but see Spiller 1984a b). “Diffuse” predation (i.e., predation by organisms other than lizards, such as birds and wasps; Hixon 1991; Menge et al. 1994) may hold spider populations down even if lizards are removed; we are presently considering excluding birds as a partial test. Although we are conducting no comparable experiment for physical disturbances, long-term observations suggest that they are likely to play a major role in controlling spider densities. For example, in October 1996, the second year of the present experiment, a major hurricane passed over Great Exuma, located about 100 km south of the Staniel study site (Spiller et al. 1998). The hurricane devastated

e f f e c t s o f p r e d a t o r r e m o va l

81

spider populations in the former area, and although the effects at Staniel were not as dramatic, the abundance of M. datona at the following census (December 1996) dropped to one of the lowest levels recorded during the experiment. Similarly, in the first experiment at the Staniel site, we recorded a precipitous drop in spider abundance following a tropical storm (Spiller and Schoener 1988). Cold fronts during the winter may have similar effects at Staniel and other sites. Both the effect of predators and the effect of physical factors on lizards versus spiders are reflective of the general proposition that small organisms are more immediately vulnerable than large organisms to these mortality sources (Schoener 1974, 1983; Connell 1975). Indeed, immediately after the aforementioned hurricane (Spiller et al. 1998), spider populations were substantially more devastated than lizard populations in the region of moderate impact: 41% of spider populations became extinct compared with no lizard extinctions, and 79% of spider individuals (mean per island) were lost compared with 34% of lizard individuals. (In the region of high impact, all populations were exterminated.) On average, vertebrates often appear to be closer to their carrying capacities than invertebrates (Schoener 1986). A corollary of this observation is that we might expect the experimental removal of vertebrates, at least in terrestrial systems, to have more impact on average than the removal of invertebrates. Of course, not all terrestrial vertebrate removals will be notable, as experiments by Wise and Chen (1999) and Wiens et al. (1991) have shown, but many have had quite strong effects (e.g., Holmes et al. 1979; Gradwohl and Greenberg 1982; Pacala and Roughgarden 1984; Joern 1986; Parmenter and MacMahon 1988; Atlegrim 1989; Fowler et al. 1991; Bock et al. 1992; Belovsky and Slade 1993; Marquis and Whelan 1994; Dial and Roughgarden 1995). Indeed, the evolutionary appearance of vertebrate predators on land may have been a very devastating event for many terrestrial invertebrate taxa (see fig. 4.6). On the other hand, studies of spiders elsewhere, particularly in agroecosystems, have sometimes found strong effects of spiders (e.g., papers in the 1999 Symposium on Spiders in Agroecosystems). In summary, the larger lizards, which occupy a higher trophic position than spiders, eat more kinds of food (including spiders) and appear to have weaker controlling agents (in the form of predators and physical factors) than do the smaller spiders. One is tempted to speculate that trophic position and effect size may be positively related in general. Of course, trophic position is related to the kinds of effects that can occur (e.g., top down vs. bottom up, direct vs. indirect, within vs. between trophic level), and much argument exists concerning them

82

chapter 4

(e.g., Hairston et al. 1960; Menge and Sutherland 1976; Fretwell 1977; Paine 1980; Oksanen et al. 1981; Schoener 1989; Wootton 1994b; Abrams et al. 1996; Polis et al. 1996; Power et al. 1996a; Strong et al. 1996). Another caveat is that the topmost predators are often rare. Yet this is perhaps artificially true as often as not, thanks to the double-whammy of direct persecution and indirect (food-chain) poisoning from humans. This brings us to our third issue, one that is an inevitable consequence of human activities in general: the necessity of dealing with species expendability. What Are the Implications of These Results for the Importance of Species? So far in this presentation, we have asked the rather well-defined question: which of two species has a stronger effect on other elements of its community as measured by removal experiments? One of the two species considered, a vertebrate, was found to have a substantially greater effect than the other species, an invertebrate. Indeed, we were able to confirm statistically that the latter, the most common web spider, only affected one trophic element—small aerial arthropods—and the latter changed only by 15% on spider removal. Does this mean that, in some sense, the spider is more “expendable?” Expendability is a concept that involves values, and to evaluate values, we have to go well beyond basic ecological research into economics, aesthetics, and perhaps even ethics. It is reasonable however, to assert that the particular ecological community that exists naturally in an area, with its component species and attendant relative abundances, constitutes the most valuable state. Given this assumption, the larger a deviation from that state a species removal causes, the less expendable is that species. Whether and how to include the removed species themselves in a measure of overall deviation, and how to deal with removals of species combinations, are two of a number of technical issues that might have to be considered with this approach. If we ignore such complications, then the lizard A. sagrei is clearly less expendable than the spider M. datona, so at least we have determined relative expendability. This proposition, which we can call “the assertion of lesser effect,” has many flaws, of course, but it does have the great advantage of being tractable both theoretically and empirically. Moreover, what many people mean by “conservation” is “keeping things the same,” or more specifically keeping some aspect of the natural world in the same condition as it was (e.g., when they were growing up). From another perspective, the natural community present in an area consists of species that are individually and collectively (in a proportion

e f f e c t s o f p r e d a t o r r e m o va l

83

that is debated) adapted to the physical conditions of its site. By multiple measures of “success”—biomass, productivity, and species diversity of the total system—the natural state is often considered to rank near the top, certainly when compared with randomly selected species sets that might be introduced in its place. This ecological optimum may translate into an economic or aesthetic optimum. Highly productive species, if they happen to be harvestable as food, are valuable in a sense that most would define as economic. High species diversity, and maybe even high biomass, are also valuable to some in an aesthetic sense, and this may translate into an economic currency. Finally, one might simply adopt an “ethos” based on the principle that the most natural state is the most desirable state, never mind other types of values. Here are some of the worst flaws with this argument. Species occurring at a site in minimal abundance or productivity may be more valuable in some traditional economic sense, say as food, than that community’s most abundant species. If we knew these economic values, we could use them to weight the effects on response variables in evaluating species removals—and we would have to consider directions of change as well, since species removals both increase and decrease response variables. In agroecosystems related to our particular study system, for example, certain producer species would be weighted highly, and the lizard’s much greater reduction of foliage damage than the spider’s would contribute even more to the former being less expendable. Chances are, however, that in such an exercise, no species would be found to have much, if any, traditional economic value. This perception—low absolute values of everything—has in the past led humankind, with its terrible ingenuity, to replace entire natural communities with near monocultures of introduced species, a procedure that seems to optimize certain short-term economic goals but is anathema for those who value diversity and natural landscapes. Another problem with measuring a species’ expendability by lack of “deviations from natural conditions” on its removal is that we are not quite sure what “natural” means; if we mean minimally disturbed by humans, this would need to be ascertained. Do we mean “modern” humans, or do “earlier” humans count? In addition to the alterations in abundances and diversity of resident species from human activity, how should we take into account species introduced as an intentional or unintentional result of human activity? Species naturally introduce themselves to new locales all the time, so perhaps we should only devalue introductions from afar. Even if the community is entirely natural, one might—by at least an aesthetic criterion—not consider it optimal. For example, a subset

84

chapter 4

of the community might reside in a state of low species diversity, perhaps because of an especially voracious and effective predator. For a possible case in point, consider the data in fig. 4.6, showing how the introduction of the lizard A. sagrei wiped out most web-spider species occurring on certain small islands of the Bahamas. These islands are extremely natural by any criterion of human disturbance: the Arawaks were ruthlessly deported hundreds of years ago, and few people set foot on them today. Anolis sagrei did not occur naturally on the very small islands to which we introduced it, but it does occur naturally on most other islands in the vicinity. In fact, it may occasionally have occurred on the former islands—in which case our introduction may be considered a restoration by some. If one has an inordinate fondness for spiders, then the natural state of affairs on the vast majority of islands is less desirable than the state without lizards. Of course, one can argue that we have demonstrated only local extinction whereas it is global extinction that is really of concern. But how many invertebrate species were exterminated globally by the appearance of these very abundant, generalized predatory lizards? Were anoles eventually to colonize other parts of the world lacking any diurnal arboreal lizards (e.g., the Australian tropics) and (it seems inevitably) wipe out many spider species, would this be desirable, or does the latter judgment depend on the “naturalness” of the invasion? In conclusion, our assertion of lesser effect may cut both ways economically and aesthetically. Some may react by abandoning it altogether, whereas others may find it a good starting place from which we can modify slightly or moderately. As scientists, however, experimentally determining the impact of species removals is something we can do objectively, and we are doing this increasingly well. The resulting information should contribute substantially to the evaluation of species importance.

Acknowledgments We thank the National Science Foundation for supporting this research and P. Kareiva for comments on the manuscript.

Chapter 5

m Understanding the Effects of Reduced Biodiversity: A Comparison of Two Approaches J. Timothy Wootton and Amy L. Downing

This volume poses the question of when, if ever, a species is expendable. Even if we believe that species are never expendable, many species are at risk of extinction, and we need to prioritize efforts to conserve them. Before we can even begin to address these issues, however, we must be capable of determining the consequences of species extinction, with the ultimate goal of predicting an extinction’s effects on the ecosystems on which we depend. Meeting this challenge also requires developing an understanding of the effects of extinction in natural ecosystems. Therefore, it is worth considering what goals we should have for studying the loss of biodiversity, and what different approaches we should pursue to attain these goals. The most obvious reason for humans to worry about species extinction is the possibility that we could lose organisms of direct value to humanity. If the effects of extinction were limited only to the loss of

86

chapter 5

a particular species, setting priorities would be relatively straightforward; conservation measures would be focused on species that benefit humans directly and in proportion to the benefits they provide. Of course, it is worth noting that we already have managed to drive species to extinction that offered clear benefits, such as passenger pigeons (Ectopistes migratorius) and great auks (Penguinus impennis)— both sources of food. But the more important point is that the loss of a species is likely to impact other parts of the ecosystem, thereby raising the possibility of complex indirect effects. The idea that humans and other species might be intimately interconnected through webs of interactions among species and the physico-chemical environment appears repeatedly throughout the history of ecological thought (e.g., Darwin 1859; Camerano 1880; Forbes 1887; Shelford 1913; Elton 1927) and stems in part from the idea of a “balance of nature.” Even if nature is not delicately balanced (Pimm 1991), the potential remains for the effects of species loss to ramify through ecosystems. Until recently, however, these ideas were largely speculative. Paine’s classic 1966 study, in which he experimentally removed the starfish Pisaster ochraceus and observed the subsequent response of the rocky intertidal community, changed this situation. Two points of the study were particularly important. First, the experiment demonstrated rigorously what had previously been speculated: that reducing biodiversity could cause cascading effects on other members of the community, potentially even causing a radical collapse of the system. Second, this study demonstrated a clear way for ecologists to identify and begin to understand the effects of reduced biodiversity by creating an experimental, local extinction in a natural ecosystem. This work helped spawn a plethora of studies in which local species extinctions have been imposed experimentally in a range of ecosystems (Connell 1983; Schoener 1983; Sih et al. 1985; Goldberg and Barton 1992; Wootton 1994b; Menge 1995; Abrams et al. 1996; Leibold et al. 1997 contain partial reviews of these experiments). In many cases, these “targeted species reductions” have verified the findings of Paine (1966), demonstrating that species extinction can have direct and indirect effects on other parts of ecosystems. Additionally, studies taking advantage of the recovery of endangered and threatened species have demonstrated that species on the verge of complete extinction can have profound effects on their ecosystems, including sea otters (Enhydra lutris; Estes and Palmisano 1974; Estes et al. 1978; Duggins et al. 1989) and peregrine falcons (Falco peregrinus; Paine et al. 1990). In the past few years, an alternative approach to “targeted species

effects of reduced biodiversity

87

removals” has emerged, which we refer to as the “biodiversity manipulation” (e.g., Ewel et al. 1991; Naeem et al. 1994; Tilman and Downing 1994; Tilman et al. 1996, 1997; Hooper and Vitousek 1997, 1998; McGrady-Steed et al. 1997; Naeem and Li 1997; Hector et al. 1999; Petchey et al. 1999; Naeem et al. 2000). Previous to this body of research, the current views of biodiversity’s effects on ecosystem processes consisted of verbal hypotheses (Lawton 1994), observations and natural experiments (Jones and Lawton 1995; Schulze and Mooney 1993), and the “targeted species removals” studies. Despite growing concern over the loss of large fractions of biodiversity (i.e., more than one or a few species as in the targeted species studies), there had been no rigorous experimental tests of the diversity—functioning hypotheses and proposed relationships. The first “biodiversity manipulation” experiment (Naeem et al. 1994), although heavily critiqued (Grime 1997; Huston 1997; Hodgson et al. 1998), has been provocative and has spurred much discussion and further experimental work on the role of species in ecosystems (Lawton et al. 1998). Curiously, the biodiversity manipulation studies have developed independently of the large body of information generated by studies that involve the targeted removals of species. This circumstance raises several questions: what role should classic species-removal experiments play in biodiversity research? What do we hope to learn from the biodiversity studies? What could we gain by synthesizing the two bodies of work? and, What approaches should we take in the future to further our understanding of the consequences of species loss? Thinking about these questions also requires that we pause and consider what we actually want to learn from studies of species loss.

A Comparison of the Approaches What are the characteristics and distinguishing features of the two approaches? In our view, the two approaches can be categorized by how they are carried out and, to a lesser extent, by the response variables they measure (table 5.1). The targeted species-removal approach repeatedly removes one or more selected species or taxa, generally from an intact, natural ecosystem or in semi-natural mesocosms with an assemblage of background species typically observed in natural ecosystems. In general, the approach has concentrated on metrics of population, biomass, or trait changes of different species in the system, although this is certainly not a requirement of the design. The biodiversity-manipulation approach differs from the targeted species approach by varying biodiversity through constructing communities

88

chapter 5

TABLE 5.1 Comparison of targeted species reduction and biodiversity manipulation for exploring the effects of extinction Approach

Targeted Species Reduction

Biodiversity Manipulation

Experimental Design

Remove or reintroduce one or more species from a natural or seminatural community

Randomly assemble communities of different diversities in lab, mesocosms, or field

Typical Response Variables

Species abundance, biomass, traits (size structure, behavior, etc.)

Changes in nutrient cycling, productivity, aggregate standing biomass, total respiration, decomposition

of different species diversities, as opposed to removing selected species. Typically these constructed communities are assembled randomly from a pool of species that presumably live together in nature. Because of the nature of these biodiversity manipulations, this work is usually carried out in laboratory microcosms or mesocosms in the field, although there are exceptions. The response variables on which the biodiversity manipulations have focused have usually been changes in ecosystem-level variables such as aggregate standing crop, productivity, respiration, decomposition, or nutrient cycling rates. An initial issue to consider is whether the response variables characteristic of biodiversity manipulations (i.e., “ecosystem function” variables) are more relevant for understanding the consequences of species extinctions than those of traditional species manipulations. We argue that they are not, and that both approaches explore responses that matter for ecosystems (see also Mooney et al. 1995). First, it is worth remembering Tansley’s (1935) original definition of the ecosystem as “including not only the organism-complex, but also the whole complex of physical factors.” This definition clearly includes both the biological community and the physico-chemical environment. From this perspective, the biological response variables measured in traditional targeted species-reduction studies are variables representing “ecosystem function” as much as are aggregated chemically oriented variables such as standing biomass, rates of primary productivity, and nutrient cycling. Second, it is reasonable to ask what sorts of variables should be of most interest to humans. Humans generally interact with the environment on a species-specific basis (see also Mooney et al. 1995, p. 297).

effects of reduced biodiversity

89

Therefore, changes in particular species or “community-level” responses are often more immediately relevant to humans than changes in “ecosystem-level” responses such as the biogeochemical cycles, gross biological production, etc., the focus of many of the biodiversity manipulations studies. For example, we care whether biomass in the ocean is housed in salmon versus algae, or whether terrestrial biomass is contained in fir trees versus shrubs, or in lettuce versus slugs. On the other hand, “aggregate or ecosystem-level” responses such as nutrient recycling rates and CO2 production may have broad and long-lasting implications for many species. For example, scenarios of rising CO2 levels and global warming predict major shifts in community composition (e.g., Hobbie and Chapin 1998; Daly et al. 2000; Rogers and Randolph 2000), which could have large implications for humans if the species are commercially important (e.g., timber production). We suspect that the immediate implications for humans to changes in aggregate-level responses are generally less tractable in the short term, whereas in the long term these changes may prove to have large effects and are therefore worthy of additional study. Third, it is appropriate to consider the sensitivity of indicators for the effects of reduced biodiversity. In investigations in which both community-level and ecosystem-level responses are monitored, such as in the experimental studies lakes (Schindler et al. 1987; Carpenter et al. 1993), species-specific variables generally appear to be more sensitive to environmental impacts than are aggregated ecosystem-level variables. The stability-complexity debate also highlights the need to distinguish community variables, which respond strongly to perturbations, and aggregate variables, which respond less dramatically (May 1974; Tilman 1996; Tilman et al. 1998; Doak et al. 1998). Consequently, focusing solely on ecosystem-level variables may inappropriately minimize the effects of reduced biodiversity, especially if the species is a central focus of human concerns. In summary, community and ecosystem variables are both important aspects of ecosystem functioning but may have different shortand long-term implications for humans and ecosystems. Although we advocate a continued focus on species response variables, we also believe that studies exploring responses of aggregate ecosystem variables deserve further attention. In fact, changes in ecosystem-level variables will probably be associated with large changes in community-level variables, so finding changes in ecosystem-level variables is of particular interest. This argument suggests an alternative definition of Paine’s (1969a) keystone species: a species that produces a detectable change in ecosystem-level variables.

90

chapter 5

What Are the Goals of Biodiversity Studies? It is worth beginning by asking if the goals of the biodiversity manipulations differ from the goals of targeted species-removal experiments. Both approaches share a general interest in knowing whether species extinction affects other parts of an ecosystem. In targeted speciesremoval studies, the investigator is interested in what happens when a particular species is removed, with an eye toward understanding the effects of its extinction and perhaps even predicting the consequences of extinction. In contrast, biodiversity manipulations are interested in slightly different questions (Lawton 1994): First, can species affect the nature and function of an ecosystem? Second, what is the direction and shape of the relationship between species number and ecosystem functioning? Third, do different species have redundant effects, effects of similar magnitude, or idiosyncratic (variable) effects? Fourth, what is the average effect of one or more extinctions on an ecosystem? An interesting issue is whether targeted species removals can address similar questions to those addressed by the biodiversity manipulations. Targeted species reductions have long ago and repeatedly answered the first question; losing species can affect a variety of ecosystem variables. The second question regarding the direction and shape of the relationship between species and ecosystem functioning has not been addressed by the targeted approach, primarily because of the nature of the response variables measured in the targeted studies. Community-level responses depend both on the nature of species interactions within a community and on how the imposed manipulation alters those interactions. Community properties such as mussel biomass, forb biomass, bullfrog biomass, or Euglena biomass are all specific to a particular system. One may make predictions regarding the response of these species to a particular manipulation, but the direction of the predicted response will differ depending on the composition of rest of the food web and on the manipulation itself. There is no reason to expect a general response of a particular species to an arbitrary extinction. In contrast, a goal of the biodiversity approach is to determine if general, predictable relationships of the same direction or shape exist between biodiversity and ecosystem processes. For example, a central question driving the biodiversity manipulation approach has been to determine if productivity always increases with diversity or decreases with extinctions. To answer this question, aggregate variables can be measured in different communities (grasslands, aquatic microcosms, ponds, and terrestrial food webs) to esti-

effects of reduced biodiversity

91

mate productivity in some manner. This provides a “common currency” to compare among systems and with which to examine the relationship between biodiversity and ecosystem function. What about the effects of species loss and question of whether species are redundant, equivalent or idiosyncratic? It is important to first consider what these characterizations imply. Redundancy implies that many species are able to perform functions similarly: the loss of redundant species will have a negligible effect on function until all redundant species have been lost (Lawton and Brown 1993; Naeem 1998). Redundancy might apply generally to all species in an ecosystem; or it might arise in subsets of species that act synergistically, thereby compensating for the loss of other species, such as when species occupy one of several functional groups. We define the former as redundant and the latter as redundantly synergistic. The key feature of redundant or synergistic species for our discussion is that single species removals produce no effect if species are redundant. Species equivalence implies that all species have some effect on ecosystem functioning but that the species identity is unimportant (i.e., species are interchangeable). If species were truly equivalent, we would expect a linear relationship between diversity and ecosystem function. Naeem et al. (1995) have suggested, however, that a nonlinear pattern might arise, implying that the effect of a species might vary with diversity but not with species identity. Idiosyncratic patterns imply that individual species have variable effects (positive, negative, or none) on ecosystem processes, and would include the presence of keystone species at one extreme. Catalytic synergisms among species also might occur in this scenario if all species must be together to have an effect. For example, removing either a legume or its Rhizobium symbiont would be expected to have an immediate impact on nitrogen dynamics by reducing nitrogen fixation, whereas subsequently removing the other member of the pair might have relatively minor effects. The biodiversity manipulation approach has explored the redundancy, equivalence or idiosyncratic effects of species by looking for the pattern of change in ecosystem function over a wide gradient of species diversities (Naeem et al. 1995). Using biodiversity manipulations, if species are redundant, the prediction is that a gradient of species removals will produce little change until very few species are left, resulting in a strongly concave-downward response curve (fig. 5.1a). If species have equivalent effects on an ecosystem, the prediction is that there will be essentially a constantly changing response curve (see fig. 5.1b). This response curve might be linear or nonlinear (Naeem et al. 1995). Finally, if species are idiosyncratic or catalytic in their effects, one would predict a response curve that exhibits a

92

chapter 5

Figure 5.1 Summary of expected patterns of species contributions to the effects of reduced biodiversity across an experimental biodiversity gradient (curves following Naeem et al. 1995) or in response to a set of experimental single-species deletions (shown as Xs). Error bars represent within-treatment variation in the single-species removal experiments and are shown for the control (all species present, where S ⳱ number of species). A. All species are redundant. B. All species make equivalent contributions to the ecosystem parameter of interest (curve could be linear or nonlinear). C. Species have idiosyncratic effects on the ecosystem (positive, negative, or no effect). D. species have synergistic effects, such as compensation within functional groups. In D and E, different curves represent different orders of species removal, with ⳮA and ⳮB representing the removal of two particular synergistic species. A sixth situation in which the biota has no effect on ecosystem processes (the null pattern of Naeem et al. 1995) is not shown.

highly sinuous shape varying with the order of species removed (see fig. 5.1c). Removing species with strong effects would cause rapid changes in the relationship between diversity and ecosystem function, whereas removing weak species would produce relatively little change (i.e., flat sections). If multiple orders of species removals were imposed, the combination of such curves would produce an overall pat-

effects of reduced biodiversity

93

tern with a wide envelope of variation in the central section of the curve. When species are removed in different orders, the effect of particular species removals would be the same for idosyncratic species but different for catalytic species (see fig. 5.1d). Redundantly synergistic species would also exhibit sinuous and variable response curves but initially might not exhibit effects of reduced biodiversity if the remaining species could compensate for the functional role of the deleted species (see fig. 5.1e). Like catalytic synergists, the effects of removing particular species of redundant synergists depends on the order in which they are removed (see fig. 5.1e). Although this approach seems straightforward, some problems have recently been identified in the interpretation of these types of experiments (Huston 1997, Allison 1999; Huston et al. 2000). In particular, designs with purely random combinations of species may be unable to use a line of best fit to distinguish between species equivalence, redundancy within functional groups, or idiosyncratic responses. For example, if species are idiosyncratic, species with both large and small effects are removed at each level of diversity. If the response curve is constructed from the average response at each level of diversity, this procedure would result in a response curve similar to the equivalent species curve, but perhaps with large error bars. The size of the error bars around the general curve could provide some differentiation between the two but would raise the question of how large the error must be to claim an idiosyncratic effect. To accurately distinguish between the idiosyncratic and equivalent response, single species removals and groups of species removals would have to be replicated, which would allow patterns of variation among different treatments to be compared with the expected variation within a treatment. To date, this has not been done (Petchey 2000).

Can Targeted Species Reductions Answer Key Questions that Underlie Biodiversity Manipulations? The targeted species-removal approach can provide answers to questions that are typically posed in biodiversity manipulations when multiple single-species manipulations have been carried out in the same system, or when several different species are manipulated in the same experiment. In this case, we would expect no detectable change in our response variable to different species removals if species are redundant either generally (see fig. 5.1a) or within functional groups (see fig. 5.1d). If all species are generally redundant with respect to an ecosystem function (see fig. 5.1a), for example, if terrestrial plants in

94

chapter 5

Figure 5.2 Results of manipulations of amphibian diversity on frog production (carbon flux from aquatic to terrestrial ecosystems) in experimental ponds by Wilbur and Fauth (1990). Curve represents interpolated line of best fit.

grassland communities all contribute to primary productivity, many species may be lost before primary productivity would be altered. If, instead, species are redundant within functional groups (e.g., trophic levels or feeding guilds) (see fig. 5.1d), changes in ecosystem function might occur in discrete jumps every time a species extinction results in the loss of an entire functional group. If species are equivalent (see fig. 5.1b), each species extinction will have similar effects on ecosystem function and we would expect a linear relationship. Finally, extinctions of idiosyncratic species (see fig. 5.1c) would create large variability in the direction and magnitude of ecosystem responses. For example, the extinction of a predator species may result in a net decrease in primary productivity by indirectly increasing grazing pressure, whereas the extinction of an efficient grazer species may enhance productivity. In many instances, multiple species deletions have already been conducted with the targeted species-removal approach, although they are usually designed and interpreted to answer slightly different questions. Results from targeted studies show clearly that species have idiosyncratic effects on ecosystem components. For example, Wilbur

effects of reduced biodiversity

95

and Fauth (1990) reduced diversity in different ways within experimental ponds in North Carolina. When their results are graphed in a manner similar to those of biodiversity manipulations (in which we could term the response variable, frog biomass, carbon export from aquatic to terrestrial systems, in ecosystem parlance) the loss of species clearly has quite different effects depending on the species being manipulated (fig. 5.2). The pattern observed when multiple species were removed in the Wilbur and Fauth study (1990) further illustrates that answers about the redundancy, equivalence, or idiosyncratic nature of species may still be murky even when there is a diversity gradient (see fig. 5.2). Here, the line of best fit is similar to that expected under the equivalence hypothesis, but the actual experimental points deviate wildly from the curve, clearly lending further support to the idiosyncratic view. Experimental studies in the rocky intertidal zone of the Bay of Panama (Menge et al. 1986a) have also clearly demonstrated an effect on aggregate properties of the ecosystem, but they also show that characterizing the responses of a system as redundant, idiosyncratic, or equivalent may be difficult. Plots of plant cover in response to species reductions (roughly related to primary productivity of the ecosystem) suggest that species have relatively equivalent effects on the ecosystem (fig. 5.3a), whereas the response of total primary and secondary cover to species reductions (related to standing biomass or carbon storage of the ecosystem) suggests a redundant or idiosyncratic response of this variable (see fig. 5.3b). Because there were significant differences among treatments in this study, an idiosyncratic view would be favored, at least among functional groups of consumers. Additionally, the response of animal cover to species reductions is to increase (see fig. 5.3c), whereas the possibility of increases in aggregate ecosystem-level responses with reduced diversity is rarely considered in most discussions of the effects of biodiversity on ecosystems. The type of system and the response variable being explored may have some relationship to whether increases in ecosystem-level responses are expected with increasing diversity. In studies focused only on productivity rates and the diversity of basal species in the trophic web, one might predict increases in productivity with diversity, whereas if the diversity of species at multiple trophic levels is considered, the relationships may be much less predictable because consumers remove producer biomass. Clearly, targeted species reductions can answer many questions about the effects of reduced biodiversity. To date, results synthesized from such experiments provide some of the clearest evidence for the role of biodiversity:

96

chapter 5

effects of reduced biodiversity

97

1. Reducing biodiversity affects other parts of ecosystems, often substantially. Following Paine (1966), numerous studies have found impacts of experimental extinction on other variables (reviewed in Connell 1983; Schoener 1983; Sih et al. 1985; Goldberg and Barton 1992; Wootton 1994b; Menge 1995; Abrams et al. 1996; and Leibold et al. 1997), as described previously. 2. Extinctions of different species have variable (idiosyncratic) effects on ecosystems; although some species appear to be redundant, others have very strong effects. For example, Paine (1992) found a wide array of impacts of grazer species on kelps in an intertidal community, with many species having relatively slight effects but several species having very strong effects. 3. Extinctions affect ecosystem resilience. Resilience refers to the speed at which a system returns to steady state after a disturbance. While not usually couched in these terms, the study of succession is the study of resilience (although succession does not necessarily imply that a system returns to a stable equilibrium point). After the paper by Connell and Slatyer (1977) laid out an experimental program for investigating succession (i.e., the rate and trajectory of a system following a disturbance), targeted species removals have provided insight into the effects of species loss on resilience. For example, experimental studies in marine nearshore areas demonstrate that removing mobile consumer species can change the rate, and even the outcome, of the succession of sessile species after a disturbance (e.g., Sousa 1979; Lubchenco 1983; Wootton 1993b). In these studies, the exclusion of consumer species changed the rate of recovery of disturbed sites relative to similarly disturbed control sites with consumers. 4. Extinction affects classical “ecosystem-response” variables. Although many targeted species removals look for responses in the populations of other species, some studies have also investigated

Figure 5.3 Results of manipulations of species diversity on rocky intertidal shores in the Bay of Panama (data from Menge et al. 1986). All species removed were consumers. Aggregate response variables for effects of species deletions: (A) Plant cover (an index of primary productivity); (B) total cover of organisms (an index of standing biomass); (C) animal cover. Lines of best fit from polynomial regressions either with (solid curve) or without (dashed curve) the influence of the logical requirement that each index will be zero when all species are eliminated. Cover includes the sum of primary and secondary space, so can potentially be ⬎ 100%.

98

chapter 5

the effects on patterns of productivity, standing crop, or nutrient and energy flow. For example, the classic study by Likens et al. (1970) on the Hubbard Brook watershed is essentially a treeremoval experiment. Similarly, Vanni and Findlay (1990), Vanni and Layne (1997) and Schindler et al. (1993, 1997) have shown that removing fish species can strongly affect nutrient and carbon cycles in lakes, and Sterner (1986) has found a similar result for zooplankton. Duggins et al. (1989) showed that reversing the extinction of sea otters profoundly changed the carbon cycling of nearshore ecosystems because of the increased contribution of kelp detritus. 5. Extinction of multiple species can have synergistic effects. A possible limitation of using targeted species-removal experiments to address the general effects of reduced biodiversity is that synergisms might exist among species. Therefore, removing multiple species might cause different effects than the additive effects of removing several individual species. A large body of work in the targeted species-removal tradition that manipulates more than one species simultaneously addresses this question directly under the rubric of higher order interactions and interaction modifications. These studies have demonstrated that synergisms among species occur, and have uncovered the mechanisms giving rise to these synergisms (e.g., Wilbur 1972, 1987; Neill 1974; Morin et al. 1988; Wilbur and Fauth 1990; Werner 1992; Wootton 1992, 1993a; Wissinger and McGrady 1993; Morin 1995; Schmitz 1998; see reviews in Wootton 1994a, b). 6. Different species play important roles under different conditions (reviewed in Dunson and Travis 1991). Consequently, a diverse assemblage is likely to be buffered from a temporally fluctuating environment. For example, the importance of the starfish Pisaster ochraceus in the rocky intertidal zone of the Pacific coast varies dramatically with changes in physical conditions, ranging from being a keystone species on which a number of other species depend to being “just another starfish” (Paine 1966; Menge et al. 1994; Harley this volume). Similarly, the importance of mussels, kelp, barnacles and salt marsh plants to shore communities varies with tidal height (e.g., Connell 1961, 1970; Paine 1966; Paine and Levin 1981; Bertness and Leonard 1997), phytoplankton dominance depends on physical and chemical conditions (Tilman 1977; Tilman et al. 1981), and terrestrial systems are dominated by different plants, depending on soil conditions (e.g., Tansley 1917).

effects of reduced biodiversity

99

Strengths and Shortcomings of the Approaches In addition to providing the insights just mentioned, targeted species reductions have several unique benefits. In particular, these studies have been capable of identifying mechanistically why particular results were obtained. Ultimately, we will need to know why reducing biodiversity alters other components of the ecosystem if we are to begin to develop a predictive framework. Targeted species reductions direct investigators toward identifying the specific ecosystem components that have the most important effects on the system, and therefore lead toward a predictive framework for anticipating the effects of specific extinction events. Predicting the consequences of particular extinctions may not be the aim of biodiversity research, however; alternatively, we may be more interested in predicting the “average” effects of randomly selected species, or in estimating the potential variability in ecosystem response given a random set of extinctions. If reducing biodiversity indicates either redundant or similar effects among species, then concentrating on an average response is reasonable and would make prediction much easier. However, if species effects are idiosyncratic, it becomes more important to understand and predict the variation in response to the removal of particular species. Importantly, biodiversity studies will be able to provide estimates of the magnitude of the variation due to composition relative to the average effects of species loss. These estimates may be critical because although species extinctions do not occur randomly in communities (Wardle 1999), they may be difficult to predict. A drawback of biodiversity manipulations, however, is that they will be less able to provide clear mechanisms for the variation in responses. Because empirical results repeatedly indicate the idiosyncratic effects of species removals, a focus on speciesspecific effects as opposed to “average effects” will ultimately be necessary if predicting the response of ecosystems to species extinction is a goal of biodiversity research. The preceding discussion does not argue for one approach versus the other, but instead that both approaches are ultimately needed to understand and predict the consequences of species extinctions from ecosystems. Biodiversity manipulations have generated public interest as well as a deeper thinking on the consequences of species loss. These studies (e.g., Tilman 1996; McGrady-Steed et al. 1997; Naeem and Li 1997; Doak et al. 1998; Naeem 1998; Tilman et al. 1998) have focused the attention of ecologists on how biodiversity affects the in-

100

chapter 5

herent variability of aggregate ecosystem variables and how stable these variables are in the face of environmental impacts. In addition, the results of these studies have pointed out a need to critically reevaluate the causal connections that underlie many important patterns related to productivity and biodiversity. Observational studies have generally assumed that productivity determines biodiversity patterns (e.g., Connell and Orias 1964; MacArthur 1972; Huston 1979; Rosenzweig and Abramsky 1993; Waide et al. 1999), but biodiversity manipulations have shown the opposite (e.g., Naeem et al. 1994; Tilman et al. 1996; Hector et al. 1999; Downing 2001). These sorts of insights have not been forthcoming from targeted species removals because of a lack of attention to aggregate properties, and to large gradients of species removals. Another issue to consider is whether the results of biodiversity manipulations that randomly assemble communities will differ from those of targeted species reductions that disassemble natural communities. Theoretical studies have suggested that complex communities such as those found in nature are a nonrandom subset of all possible community configurations (Pimm 1979, 1982), and that patterns of community assembly may strongly determine the final community configuration (Drake 1989, 1991; Lockwood et al. 1997). The properties that allow long-term persistence of complex communities might, therefore, produce communities that have different characteristics following disassembly than would randomly assembled communities. Although preliminary evidence from microcosms supports an important effect of assembly pattern on community structure (Drake 1991) and food web diagrams appear nonrandom (Cohen et al. 1990; de Ruiter et al. 1995), experimental comparisons of random assembly versus disassembly of intact communities would be enlightening.

Hybrid Approaches Might some of the shortcomings of the two approaches be minimized by alternative strategies? Several hybrid approaches appear promising, depending on the goals of the investigator. Downing (2001; Downing and Leibold 2002) has conducted one such study in experimental pond communities with realistic multitrophic structures. A major goal of the pond study was to determine the effects of species diversity and composition on ecosystem processes. To manipulate species diversity independently from composition, replicates within a given level of diversity consisted of random assemblages of species. The initial pool of manipulated taxa contained 21 species divided evenly

effects of reduced biodiversity

101

TABLE 5.2 Functional groups and species manipulated in the pond biodiversityecosystem function studya Functional Group Species

Common Name

Macrophytes Utricularia vulgaris Myriophyllum verticullatum Ceratophyllum demersum Valisneria american Elodea occidentale Potamageton zosteriformi Potamageton crispu

Bladderwort Millfoil Coonstail Eel-grass Water-weed Eel-grass pondweed Curly muckweed

Mesograzers Rana clamitans Rana catesbeiana Gammurus americana Hyallela azteca Helisoma trivolis Physa gyrina Trichocorixa sp.

Green frog Bullfrog Amphipod Amphipods Ram’s horn snail Pond snails Corixid beetle

Invertebrate Predators Belostoma flumireum Ambrysus sp. Notonecta undulata Notonecta sp. Acilius sp. Gyrinus sp. Neoplea striola

Water bug Water bug Notonectid Notonectid Ditiscid beetle Whirligig beetle Pleid beetle

a

Study described in Downing 2001; Downing and Leibold 2002.

among each of three functional groups: macrophytes, mesograzers, and invertebrate predators (table 5.2). Species diversity was manipulated while keeping the number of functional groups constant. For each of 3 diversity levels (3, 9, and 15 species), 7 unique combinations of species were created, and each species combination was replicated. Ecosystem response variables measured in this study included productivity, respiration, decomposition, and standing crops. The hybrid design employed in this pond experiment differs from the traditional biodiversity manipulation studies by introducing replication of each of the species combinations that were chosen randomly to represent each diversity level. The additional replication of species

102

chapter 5

composition within each diversity level allows statistical isolation of the effects of species composition beyond the effects of overall biodiversity levels. This experimental design provides a method for detecting idiosyncratic patterns of changes in diversity, and an assessment of the roles of particular species in the system. Furthermore, examining numerous species in various combinations at once provides a more efficient method to screen for the strongly-interacting species than provided by traditional targeted species removals. The results of this study (Downing 2001; Downing and Leibold 2002) identify the effects of species diversity on ecosystem responses, but they also reveal a consistent and strong effect of species composition beyond its effects on species diversity. There is no way to completely disentangle changes in diversity from changes in composition, because every change in species diversity is necessarily associated with a change in species composition; however, random species selection can minimize the chances that diversity patterns are driven by single species (Tilman 1997b). By drawing attention to the compositional effects, an analysis of this type of experiment leads naturally to an analysis of the effects of individual species on ecosystem responses. Some species combinations are associated by chance, so there is some risk of erroneously identifying a species as being important. Because species combinations are assigned randomly, however, the degree of colinearity in species presence should generally be quite weak, thus minimizing the risk of misidentifying strong interactors. Although this type of analysis is possible in other biodiversity designs, it has not yet been performed, perhaps in part because the effects caused by species composition have not been separated from the effects caused by species diversity in biodiversity manipulations. The results of using this hybrid approach (Downing 2001) to detect individual species contributions to ecosystem function were both interesting and unexpected. The ecosystem-response variables of the ponds were well summarized by two principal components (PC) axes: one was dominated by concordant variation in primary productivity, plant standing crop, and respiration; whereas the other differentiated plant standing crop from plant productivity. The problem of detecting the contributions of single species in this multivariate ecosystem function space can be addressed by comparing replicates with and without each species. A species that has an effect will be associated with a difference along either PC axis. One prediction of this study was that manipulations in the plant community would have the strongest effects on variation in plant productivity. This turns out not to be the case. Of the 21 species manipulated, 24% (5 species), showed clear

effects of reduced biodiversity

103

Figure 5.4 Flow diagram of a hybrid approach to determine the general nature of species contributions to ecosystem processes. By conducting sequential experiments as necessary, the approach can potentially reduce the number of treatments required to assess the effects of reduced biodiversity.

associations with ecosystem function. Of the important species, 60% were predators (Belostoma flumireum, Acilius sp. Ambrysus sp.), 40% were herbivores (Rana clamitans, Helisoma trivolis), and 0% were plant species. A second interesting result was that predators had more variable effects than would be predicted if they were redundant with respect to ecosystem functioning. The results of this experiment suggested that the observed variation in ecosystem function was driven by top-down effects and that more detailed mechanistic studies should focus on food web interactions to understand the effects of reduced biodiversity in this system. A second hybrid approach (fig. 5.4) could be used to efficiently and effectively explore the general nature of the roles of species in ecosystems (see Huston 1997 and Allison 1999 for reviews of limitations of the current approaches). Initial studies would take the targeted species-removal approach, involving replicated deletions of a number of single species. These experiments would be analyzed in a standard manner for differences among particular treatments and controls. If treatment effects were found, the results would then be analyzed to determine whether the variance in differences among treatment means was significantly larger than the variance within treatments. If differ-

104

chapter 5

ences existed, this would indicate idiosyncratic effects of species (see fig. 5.1c). Otherwise, species equivalence (either linear or nonlinear; see Naeem et al. 1995) would be indicated (see fig. 5.1b). If the singlespecies manipulations did not produce treatment effects, a second set of replicated experiments would be required in which different combinations of the manipulated species were deleted. If these experiments yielded significant differences from controls, the results would suggest redundant synergisms among species (see fig. 5.1e), including the compensatory interactions that underlie redundancies within functional groups (e.g., Hooper and Vitousek 1997; Tilman et al. 1997). If these experiments produced no differences from controls, a final experiment deleting all species could be performed to differentiate between the species redundancy hypothesis (see fig. 5.1a) and a hypothesis that organisms play no role in the ecosystem process being studied (the “null” pattern of Naeem et al. 1995). Although finding a “null” pattern would probably be very surprising to ecologists, it might be possible for some ecosystem processes such as nutrient cycles, which have a strong geochemical component (Schlesinger 1997).

Conclusions Experimental reductions of biodiversity provide one of the most effective methods of exploring the consequences of species extinction in complex natural ecosystems. Current approaches that manipulate aggregate levels of biodiversity have produced interesting patterns and insights into the effects of extinction. Perhaps less appreciated, a substantial body of work involving targeted species reductions exists that has increased our understanding of the effects of extinction and presents an untapped opportunity for synthetic studies. In addition, for ecological science to contribute to the development and prioritization of conservation plans for endangered species, it is necessary to move beyond documenting the effects of changes in biodiversity; we must begin to understand and predict the consequences of particular extinctions. Targeted species reductions can contribute greatly in this endeavor and should therefore remain an important approach concerning extinctions and expendability. Hybrid approaches linking targeted species reductions and biodiversity manipulations may prove particularly profitable in advancing our understanding of the effects of reduced biodiversity.

Pa r t I I

m THE ANTHROPOGENIC PERSPECTIVE

Most contemporary discussions of biodiversity and species expendability begin and end with humans—begin, because we humans are the dominant agents changing the number and identity of species in most communities; and end, because we humans expend considerable energy attempting to affix value to biodiversity in terms of the services it provides to us. The following six chapters follow in this tradition, dwelling on the services of species and biodiversity, and on the dramatic impacts of humans on species assemblages. Two theoretical papers take very different modeling approaches to examine the value of biodiversity. Naeem argues for focusing on ecosystem function rather than the value of particular species when exploring the consequences of species losses. He presents “ecosystem reliability” and “biological insurance” as ecosystem services that could not be measured by standard removal experiments or community models, but which may depend critically on high biodiversity. Whereas traditional models in community ecology highlight popula-

106

pa r t I I

tion dynamics and life history attributes, Naeem embraces reliability theory, which focuses on the typology of system design and the consequences of component failure for system failure. Doak and Marvier also use models, but in a style almost the opposite of Naeem’s approach. They use more mechanistic classical community models in an attempt to identify traits that predispose a species to being an important contributor to community stability. Their findings reinforce the message that it is difficult to predict a priori which species are important for the overall stability of communities. Perhaps most notably, the abundance of a species does not equate to importance at the community level. The contrast between chapters by Naeem and by Doak and Marvier, however, is not as sharp as it might first appear. Both contributions make a case for developing a general mathematical theory of “biodiversity value”; the fact that they adopt different formalisms is a modest distinction given the contrast of their methods with the experimental, observational, and anecdotal approaches that prevail in most discussions of species expendability. The question of expendability naturally turns our attention to extinction. But if removal experiments and extinction can inform us about the role of species, Wonham and Ruesink argue that so, too, can invasions by exotic species. Ruesink’s analysis of invasions by fish reinforces the message of “context dependency” delivered earlier in this book by Menge and Harley. Although fewer than one quarter of the 1408 cases of successful fish invasions yielded any reported effects, many species produced a dramatic effect in one place but not another. Wonham takes a different approach, and explores in detail two marine invasions: Atlantic cord grass and Asian eel grass in the Pacific Northwest. These case studies are especially interesting because policymakers have essentially decided that Atlantic cord grass is harmful to ecosystem functioning (and hence requires active control) but that the Asian eel grass is acceptable (and hence is not controlled). However, when the many direct and indirect food web connections of these species are examined, the basis for this decision is suspect. An interesting distinction between invasions and extinctions arises at the policy level in the United States; invasions are categorized as either harmful or harmless, whereas all extinction is viewed as undesirable. Wonham points out that the uncomfortable labeling of certain invasive species as harmless suggests that when it comes to extinction, the most risk-averse approach may indeed be to simply assume that any species loss is undesirable. Schindler and colleagues tap into a remarkable long-term data set collected by the late W. T. Edmondson in Lake Washington. Edmondson pioneered long-term ecological research, sampling zooplankton

anthropogenic perspective

107

and phytoplankton in Lake Washington for almost 40 years in a consistent and frequent manner (on average every other week) and with unparalleled taxonomic accuracy. This quantitative time series saw Lake Washington through the initial stages of eutrophication, followed by clean-up (thanks largely to Edmondson’s research), and several shifts in community dominants. These data are a gold mine for theoretical community ecologists. Schindler et al. find that of the 176 phytoplankton species found in Lake Washington, most are frequently absent or very rare. For example, half of all the species ever observed there are found in fewer than 1 in 5 years. These data reflect a remarkable turnover of species, with the majority of species dropping in and out of the system on an irregular basis. Few ecosystems have been sampled so thoroughly or consistently, so it is impossible to know whether their findings are general. However, Schindler and colleagues suggest that high turnover, with most species appearing erratically and disappearing, may in fact be the standard for communities but goes largely unnoticed thanks to our focus on dominant and conspicuous species. They conclude that it is futile to try to estimate the expendability of species in a system experiencing such turnover; instead, they offer the more fruitful question of what processes maintain the overall species pool and the persistence of productivity in the face of constant change. Finally, Simberloff marshals all the data he can find documenting the effects of species extinction. He focuses on the loss of species for which we might expect to find major impacts, such as the widespread disappearance of community dominants (e.g., the American chestnut) or hugely abundant consumers (e.g., the passenger pigeon). Remarkably, he finds that the measured impacts of major extinctions are not well substantiated. This does not mean that we can dismiss species loss as a negligible concern. Rather, it means that the science of ecology has not been astute enough to document the effects from species losses, either due to an absence of quantitative “before and after” data, or to a superficial understanding of a species’ natural history. In cases that document major effects, those effects often arise through indirect and surprising pathways. The six chapters in this section illustrate the breadth of approaches currently used to ascertain ecological measures of species importance. The methods run the gamut from mathematical models to an examination of historical events surrounding species invasions and species extinctions, to a search for pattern in long-term data sets. Collectively, these chapters illustrate that it will not be easy to predict how humans affect and are affected by biodiversity loss.

Chapter 6

m Models of Ecosystem Reliability and Their Implications for the Question of Expendability Shahid Naeem

The realization that a single species in a community may be the equivalent of a keystone in an arch has profound implications for the other—non-keystone—species in a community. Paine’s seminal contribution (1966) produced a compelling picture of nature in which the loss of a single species could lead to the local extinction of other species while its presence could foster the local coexistence of many species. The unintended and somewhat dangerous corollary of this powerful idea, however, is that many species are of lesser importance. In the same way the keystone of an arch is often protruded and ornamented, a keystone species gains special status, relegating the nonkeystone species to seeming obscurity. Building on the same arch analogy, the voussoir, the entrado, the extrado, the springer, the abutment, the pier, and the impost stones are regarded as pedestrian. In architecture of course, each of these functional groups of stones is

110

chapter 6

significant, and each stone within each functional group is important in contributing overall structure and function, irrespective of how each stone may be adorned. In communities, the same is true. Species occupy functional groups, and each species within each functional group contributes to community and ecosystem properties. The business of ranking species by their impact on a community is a fairly established, necessary, but dicey area of contemporary ecology. Species are either competitive dominants or subordinates (Hardin 1960), abundant or rare (Preston 1962; May 1975; Sugihara 1980), keystone or non-keystone (Paine 1966; Mills et al. 1993; Power et al. 1996b), weak or strong interactors (Hughes and Roughgarden 1998; Laska and Wootton 1998; McCann et al. 1998), have high or low communityimportance values (Power et al. 1996b), are ecosystem engineers or nonengineers (Jones et al. 1994), or are redundant (Walker 1992, 1995; Lawton and Brown 1993; Naeem 1998; Gitay et al. 1996). Irrespective of the more complex issues surrounding each of these ecological constructs, it is difficult not to consider a rare, weak, redundant, subordinate, non-keystone species of low community importance that is freeloading off the engineering activities of others, to be anything other than expendable. This is the dicey part of the business. Justified or not, the notion of expendability is contrary to our gut-level feelings about the sanctity of species. The minute species are ranked, the specter of species expendability rears its ugly head. Preservation realists, stewards of a sinking ark, might embrace these many and varied means for assessing species expendability as a rational means of tackling the unpleasant business of managing extinction in the face of inevitable human expansion. Preservation idealists, on the other hand, as stewards of the same ark, are chagrined that we could even imagine not defending the preservation of all the Earth’s species. Bring human welfare into the picture, as indeed we must, and the business of assessing expendability and executing species becomes even more complicated. This chapter focuses on a unique perspective of expendability that has arisen from the recent, slowly emerging synthesis of population and ecosystems ecology—one that addresses the role of the Earth’s biota in the functioning of ecosystems (Chapin et al. 1997; Chapin et al. 1998; Loreau et al. 2001). Rather than focusing on the relative impacts of species on community structure or stability, these studies emphasize the functions of species in biogeochemical or ecosystem processes and their role in ensuring the reliable functioning of ecosystems. This perspective accords species no special ranking, but classifies them according to their contribution to overall ecosystem functioning. It is an approach that has generated much controversy, in

models of ecosystem reliability

111

part because it deemphasizes the importance of species identity (Kaiser 2000; Naeem 2000; Wardle et al. 2000), searching more for generalities about biodiversity than individual species or groups of species. Note that function refers to activity, not purpose. An ecosystem functions if it exhibits patterns of energy and nutrient flow that are within the boundaries of typical activity. An ecosystem function is any process that reflects the collective metabolic activities of species within a community; examples include community respiration, biomass production, rates of decomposition, and measures of nutrient retention or material flux. Although less glamorous than studying the marvelous and complex dynamics of populations in communities (compare measuring the rate of litter rot with studying predator-prey dynamics), measures of ecosystem functioning represent important indices of whole-system activity. Neither the magnitude nor the stability of these measures is necessarily correlated with community dynamics (May 1974; McNaughton 1978, 1985, 1993), although one expects that in most cases they are correlated. This chapter examines how focusing on species contributions to ecosystem functioning—rather than on species impacts on community properties—provides an important, possibly overlooked dimension to the complex issues that surround species expendability. Although perhaps not measuring up to the Deep Ecologist’s ethical call for biocentric equality (Devall and Sessions 1985), this focus on ecosystem function rather than community impact levels the playing field within which the relative importance of species is judged. A species of decomposer bacteria that mineralizes organic nutrients, a plant that supplies that bacterial species with carbon while taking up nutrients mineralized from the bacteria, a species of ungulate that feeds on the plants, and a carnivore that hastens the entry of ungulate organic matter into the dead organic pool all contribute to system functioning. The impacts of predators (e.g., top-down effects) or the impacts of decomposers (e.g., bottom-up effects) on population dynamics, although obviously related, are less important. Rather, I argue that the functions of each of these species—the bacterium as an inorganicorganic matter transformer, the plant as a photoautotroph or supplier of carbon to the system, the herbivore and carnivore as regulators of organic matter cycling—elevate species to similar levels. One could rank species within functional groups in terms of their relative contributions to overall ecosystem functioning or quantify their impacts on the functioning of other species within or among functional groups, but by doing so one would inevitably return to a community approach. Indeed, most concepts of community dynamics have their analogies in ecosystem dynamics (De Angelis 1992; Loreau 1994, 1995;

models of ecosystem reliability

113

Grover and Loreau 1996). As I argue later, however, this could obfuscate a species’ value to ecosystem functioning and limit our ability to assess species expendability accurately. The different perspectives just reviewed are compared in figures 6.1 and 6.2. The community perspective is one in which the web of biotic interactions (e.g., competition, predation, parasitism, or facilitation) is of primary importance (fig. 6.1, left). When modeled, the impacts of a given species on the population dynamics of another become important factors (see fig. 6.2, top). For example, the loss of a species that dramatically changes relative abundance and community structure (fig. 6.2, top), is of great interest. In contrast, an ecosystem perspective places importance on the flow of organic and inorganic matter, coupled with energy flow among species (see fig. 6.1, right). Important factors, when modeled, are the number of functional groups, and the number of species per functional group (see fig. 6.2). How the loss of a species affects the functioning of the system is of interest rather than its impact on population sizes. Clearly, communities and ecosystems are inseparable in the sense that energy and material fluxes affect communities, just as communities affect fluxes. Traditionally, however, as most current ecological textbooks suggest, communities and ecosystems are treated as if they are independent, as illustrated in figure 6.1. The ecosystem-function perspective provides a different view of expendability than does a community perspective. From an ecosystem perspective, how a species affects biogeochemical or ecosystem processes becomes more important to the question of expendability

Figure 6.1 Community vs. ecosystem function perspective for a simple community. The left figure represents a rocky intertidal community with a sea star as the top predator feeding on molluscs and barnacles that in turn feed on algae and the plankton community (including microbes). Lateral arrows indicate intratrophic group interactions, such as competition for space or nutrients. The circle indicates a magnified view of the plankton community, which has its own web of interactions. The right figure shows the ecosystem function perspective of the same system, illustrating the biota (as part of the biosphere) moving elements and compounds among the atmosphere, hydrosphere, and lithosphere. Ecosystem or biogeochemical processes (e.g., immobilization of nutrients by microbes, mineralization by microbes and other heterotrophs, or N or C fluxes) are the principle measures in this system. Rates of decomposition, measured as the biomass of autotrophs (A), decomposers (D), consumers of autotrophs (PA), or consumers of decomposers (PD), represent ecosystem functions.

114

chapter 6

Figure 6.2 Fundamental differences between community and ecosystem function perspectives. The community perspective emphasizes populations (circles), population sizes (size of circle), and biotic interactions such as predation, competition, facilitation or parasitism (arrows). Community impacts of species on densities provide a means for measuring expendability, such as the effect of the loss of species B—note how the community structure (the number, type, and strengths of interactions) has changed. The ecosystem perspective emphasizes functional roles. Species are classified into serially dependent functional groups (e.g., A– C or B–D). Within functional groups, the assumption is that species may substitute for one another. Expendability concerns changes in the functioning of an ecosystem if a species is lost. In the example illustrated, the loss of species B from the system on the left does not affect functioning, whereas the loss of species D from the system on the right alters ecosystem functioning. As a consequence, we call the system on the left more reliable than the system on the right.

than how it affects the relative abundance of species in the community. Of particular importance is whether a species—even if occasionally rare, a weak interactor, or of no importance to community stability—may nevertheless contribute significantly to ecosystem functioning at some time in the future. There are two distinct ways in which seemingly insignificant species may become important. First, a species may become abundant or increase its activity as environmental conditions change. Although this may appear to be mere opportunism, it may represent “biological insurance” from an ecosystem perspective. By guaranteeing functioning during hard times when other

models of ecosystem reliability

115

species may become rare or temporarily cease functioning, the presence of such species can be important to long-term functioning (Yachi and Loreau 1999). For example, drought-tolerant plants that are rare during periods of normal rainfall may become abundant and play an important role during the ecosystem production during drought years (e.g., Tilman and Downing 1994) even if they have no community role in intervening decades of normal rainfall. Second, a species can substitute for others in the same functional group should any species become locally extinct (Walker 1992, 1995; Lawton et al. 1996; Naeem 1998). After an outbreak of chestnut blight (Simms 1996), other tree species can replace chestnuts. This is a sort of extreme form of biological insurance, much as life insurance (monetary compensation for the loss of an individual) is distinct from health insurance (monetary compensation for the temporary loss of an individual’s functioning). By borrowing from reliability engineering, in which systems with larger numbers of redundant parts are inherently more reliable, we can define ecosystem reliability as the probability that an ecosystem will function at a given level over a given unit of time and also show that seemingly expendable species may be invaluable for ensuring ecosystem reliability (Naeem and Li 1997; Naeem 1998). How ecosystem functioning, functional groups, species redundancy, biological insurance, and ecosystem reliability (collectively the “ecosystem perspective”) relate to expendability is the focus of this chapter. I employ elementary models to explore expendability from this perspective. Although my emphasis is on using ecosystem functioning and ecosystem reliability to address the question of expendability rather than using relative abundance and community dynamics, ultimately my goal is to encourage the use of both community and ecosystem perspectives rather than abandoning one for the other.

Community Dynamics versus Ecosystem Functioning Whereas the central principles of community ecology are population based and focus on the causes and consequences of variation in the distribution and abundance of species (a vision of ecology succinctly captured in the works of Andrewartha and Birch: Andrewartha 1961; Andrewartha and Birch 1984), the central principles of ecosystem ecology are based on the biota’s role in governing material and energy flows (Odum 1953, 1994). Although born synthetically, ecosystems and community ecology have worked separately in recent history (McIntosh 1985; Likens 1992; Grimm 1995). Global change triggered the modern resynthesis of community and systems ecology, one

116

chapter 6

in which biodiversity (largely the domain of community ecology) and biogeochemistry (largely the domain of systems ecology) are merging (Schulze and Mooney 1993; Vitousek and Hooper 1993; Jones and Lawton 1995), although there is a long road to a full synthesis. Space does not permit visiting the many empirical, theoretical, and controversial issues that surround the emerging ecosystem function perspective. To facilitate the treatment of expendability in this chapter, I provide a brief review, but I do not review some of the controversial issues (Garnier et al. 1997; Grime 1997; Huston 1997; Tilman et al. 1997; Wardle et al. 1997; Hector 1998; Loreau 1998; Allison 1999; Aarssen 1997); these are reviewed elsewhere (Naeem 2000; Wardle et al. 2000, Loreau et al. 2001). The central principle in ecosystem functioning is that living biomass shuffles elements (e.g., C, H, N, O, P, S, K, Fe, Si, Ca, N, Mg) and other biologically important elements and their compounds between three material pools (dead organic matter, living organic matter, and inorganic matter) and in so doing creates an environment that sustains life. From a global perspective, the biota comprise the biosphere that moves materials among the hydrosphere, lithosphere, and atmosphere (see fig. 6.1, right). Although not as captivating as a vision of our biosphere packed with predators, prey, parasites, and pollinators, the collective biogeochemical activities of our biota actually keep the planet from drifting into a horrifically inhospitable chemical equilibrium state somewhere between that of Mars and Venus (Lovelock 1979; Ernst 2000). Without ecosystem functioning, the Earth would have an anoxic atmosphere of 98% CO2, a mean annual temperature of 290⬚C, and an atmospheric pressure of 60 atm (Lovelock 1979). From a local perspective, the biota make up the pool of living organic matter and either mineralize nutrients (move nutrients in organic matter to inorganic nutrient pools), respire carbon into the inorganic pool, or die, shed, exude, secrete, or otherwise move material into the organic pool (fig. 6.3). Thus, soil fertility, water quality, production, and the innumerable ecosystem services that sustain us are derived from the activities of our biota (Baskin 1997; Daily 1997). One study estimated that ecosystem functioning provides around $36 trillion (1994) of goods and services per year (Costanza et al. 1997). While one might quibble over the exact value, it is safe to say that human welfare is intimately tied to the functioning of its ecosystems. In community ecology, the numbers of individuals or densities are important, although biomass can often be used as well. From an ecosystem perspective, however, biomass is the principle measure of interest. The pool of living organic matter (living biomass) can be divided into autotrophic, decomposer, and consumer biomass (see fig.

models of ecosystem reliability

117

Figure 6.3 Fundamental ecosystem structure. An ecosystem consists of two groups of matter (inorganic and organic), with one part of the organic matter representing living biomass and the other dead organic matter. Living organic matter is divided into functional groups that exchange materials among each other as well as between inorganic and dead organic matter pools. C represents carbon, which measures energy flows, whereas M represents nutrients and other elements used by living organisms. Both C and M are found in organic or inorganic molecules. Only decomposers have the capability of transforming organic matter to inorganic matter. (If supplied with sufficient energy, they can use inorganic matter as well, but this is left out for simplicity.) Ultimately, however, only autotrophs can build new organic material from inorganic matter. These abilities are shown as heavy arrows.

6.3). Note that autotrophy means to acquire carbon from inorganic sources, whereas heterotrophy means to acquire carbon from organic sources. The acquisition of energy is a separate issue. Photoautotrophs, for example, acquire energy from light, whereas chemoautotrophs acquire energy from chemical bonds. For our purposes, and for the models that follow, there are three main functional groups of interest: (1) autotrophs, the main providers of organic material most often produced by photosynthesis and nutrient uptake from inorganic materials (photoautotrophs); (2) decomposers, which possess extraordinary biochemical capabilities for transforming biogenic organic material to inorganic forms (organicinorganic matter transformers); and (3) consumers, which are divided into trophically defined functional groups (see fig. 6.3). Note that the first two groups describe an autotroph-decomposer codependency,

118

chapter 6

an antagonistic mutualism that represents probably the most important biotic interaction in solar energy-based ecosystems (Harte and Kinzig 1993; Naeem et al. 2000). Consumers encompass the bulk of described diversity and the focus of much of community ecology. These consumers affect primarily the rates of movement of autotrophic and decomposer living organic matter into the dead organic pool. From an ecosystem perspective the effect of consumption—be it herbivory, bacterivory, fungivory, or carnivory—is primarily on rates of turnover (Loreau 1995; Grover and Loreau 1996; Zheng et al. 1997). Consumers themselves, of course, contribute to the dead organic matter pool as well, but they can also mineralize (via excretion) and return organic carbon to inorganic carbon as CO2 (via respiration). Consumers feed on one another, and this activity affects standing biomass, production, and cycling rates of autotrophic and decomposer biomass. The role of consumers (trophic groups) in governing the distribution of biomass among autotrophs and heterotrophs has stimulated community ecologists for decades (Lindeman 1942; Hairston et al. 1960; Oksanen et al. 1981; Hairston and Hairston 1993; Wootton and Power 1993; Del Giorgio and Gasol 1995; Naeem and Li 1998). It is important to recall that although consumers generally suppress standing levels of biomass (the classic “HSS” patterns inspired by Hairston, Smith, and Slobodkin [1960] or trophic cascades), their presence can either increase or decrease productivity. A system that produces plants as fast as they are eaten will be barren but have high productivity. A system that is flush with plant biomass may be lush and green but nevertheless have very low productivity. Thus, although a properly designed density-based model could allow us to predict material cycling (De Angelis 1992; Loreau 1994; Loreau 1995; Grover and Loreau 1996), densities of populations in an ecosystem by themselves do not necessarily inform us about ecosystem functioning. Similarly, relative abundance does not necessarily tell us anything about biotic interactions, so it is equally uninformative in the contexts of both ecosystems and communities.

Ecosystem Reliability and Species Expendability The issues concerning expendability and functioning concern mainly how local extinction (loss) or colonization (replacement) affect ecosystem processes. It is instructive to use an engineered system as an analog to an ecosystem. In an engineered system, the failure (loss) of parts affects functioning. In an ecosystem, local extinction or the loss of species is analogous to the failure of parts, whereas recolonization

models of ecosystem reliability

119

is analogous to the replacement of lost parts. Redundant species (species that can substitute for one another) are equivalent to having spare parts or parts that can substitute fully for other parts. Ecosystems are not engineered systems, so there is no given function. Rather, the investigator must declare what aspect of ecosystem functioning is of interest. This factor may be an ecosystem’s propensity to exhibit a particular level of autotrophic production over 10 years (e.g., carbon dioxide sequestration), the ability of a freshwater ecosystem to remain aerobic over several years (e.g., water quality), or the ability of a marine system to sustain certain levels of heterotrophic biomass production (e.g., fish production). For example, if a pond becomes anaerobic due to eutrophication, for all intents and purposes it is no longer functioning, even though it is still, in every sense of the word, an ecosystem. I use reliability models to explore these ideas. Key to these models are extinction and colonization probabilities, both of which are difficult to estimate in nature and often involve many factors. Reliability models, however, can be developed readily to contend with such complications, and I offer examples of how this can be done. Specifically, I examine how the order (sequence) of colonization and extinction events and the nonindependence of extinction events can be incorporated into these models. I also offer examples of how one might derive estimates for parameters in reliability models.

The Relationship between Reliability and Insurance The reliability of a system, as described earlier, is the probability of a system functioning over a given unit of time. By definition, a system’s reliability is equal to 1 initially (t ⳱ 0) and declines immediately, eventually reaching zero when all parts have failed. By analogy, ecosystem reliability is equal to 1 initially but declines over time due to local extinction. Reliability is related to biological insurance. The insurance value of biodiversity is a venerable concept in ecology (Folke et al. 1996) but is largely a theoretical construct rather than an empirically derived property of biodiversity (but see McNaughton 1985 and 1993 for a clear discussion of the concept from an empirical perspective, though he does not use the term). Simply put, biological insurance represents the guarantee provided when extra species compensate for the loss of others. A system that is more reliable is better insured against failure. There are two aspects of biological insurance (Yachi and Loreau 1999), only one of which relates directly to ecosystem reliability. One

120

chapter 6

aspect concerns the “buffering” of functioning provided by diversity in the face of environmental heterogeneity (Yachi and Loreau 1999). The other aspect concerns “performance enhancing” effects that occur at the global or regional level (Yachi and Loreau 1999). The buffering effect more closely reflects ecosystem reliability, whereas the performance effect is a separate issue. The performance-enhancing effect observed in Yachi and Loreau’s model concerns global functioning (functioning summed across all local ecosystems), not local functioning. The global functioning of replicates that draw species from a large regional species pool outperforms the global functioning of replicates that draw from a small species pool. This difference occurs because environmental heterogeneity in the model (i.e., a highly diverse regional pool) permits replicate communities that differ in composition each to experience optimal conditions over the course of a long period. When replicate systems have similar species composition (as would happen when the regional pool is species poor), then only during certain times over the course of a long period would conditions be optimal for some of the replicates. This pattern is distinct from Doak et al.’s (1998) statistical averaging, the kind of insurance gained by having a portfolio of mixed stocks rather than a portfolio of a few stocks (Doak et al. 1998; Tilman et al. 1998). A mixed portfolio insures “stability” (in this case, low variability of performance) and higher yields because the performance of the portfolio is an average of entities whose performances covary positively (Tilman et al. 1998; McCann 2000). Ecosystem reliability, however, does not necessarily guarantee higher performance, only longer running, consistent functioning.

Fundamental Ecosystem Reliability The simplest treatment of ecosystem reliability is for an isolated ecosystem in which there is no colonization from other communities. Over a short period, functioning is generally guaranteed, but over a longer period, the loss of species due to local extinction increases the likelihood that the system will no longer be able to function. Local ecosystem functioning is determined by the nature and composition of its community. Although the boundaries of an ecosystem’s functioning are set by local physical and chemical conditions, an ecosystem’s biota ultimately drives biogeochemical or ecosystem processes. The biota can be classified by similarities in contribution to overall system functioning (fig. 6.3 provides a simple example of such a division). All autotrophs fix inorganic carbon, all decomposers perform inorganic-organic matter transformations, all herbivores con-

models of ecosystem reliability

121

sume autotrophs and affect rates of carbon fixation, and all consumers can be classified into trophically defined functional groups that affect rates of turnover and cycling. Ecosystem functioning is the result of serially dependent functional groups. As described earlier, autotrophs and decomposers are mutually dependent, and groups of consumers are dependent on their prey. Remove any one of these groups, and functioning is likely to cease or change such that it no longer resembles the original ecosystem. Species within functional groups are considered redundant. That is, each species can substitute for the other. The ability for such complete substitution is rare, but reliability models can be modified readily to deal with such issues. A reliability block diagram is used to model serial dependency, species redundancy, functional groups, and species per functional group. Figure 6.4 provides an example of the sort of reliability model that might be used to depict the ecological systems illustrated by figures 6.1 and 6.3. The example in figure 6.4 considers six functional groups, each with six redundant species. In most ecosystems, however, there are likely to be many more complications such as crosslinkages created by mixotrophy or nutrient immobilization caused by the production of plant defense compounds. Again, these issues can be addressed, but the models become complex and would not serve well for the purposes of illustration. From the preceding overview of the relationships between species and ecosystem functioning, it follows that the fundamental aspects of ecosystem reliability concern the numbers of species, extinction probabilities, the numbers of functional groups, and the numbers of species within functional groups. For simplification, I assume that extinction probabilities are small and effectively identical for all species. I also assume that extinction probabilities are independent, although I relax this assumption later. I also assume that within a functional group, the extinction of one species is compensated by the compensatory growth of another species in that group. Such compensatory growth is observed widely in nature among many kinds of organisms (McNaughton 1983; Tonn 1985; Cowling et al. 1994; Fownes 1995; Hartnett and Wilson 1999; Brown et al. 2001). Note that I use ␦ to represent the probability of extinction over a given unit of time, and ␭ to represent the probability of colonization over this same unit. This choice of symbols follows the conventions used by others but differs from my usage elsewhere (Naeem 1998). If we consider extinction to be equivalent to failure, then the probability of the failure of a species over time is equivalent to its probability of extinction (␦) over time, or

Figure 6.4 Use of a reliability block diagram to depict functional groups and redundant species within an ecosystem. Species are boxes, with symbols representing trophic positions. Six species per functional group are shown. All six functional groups are shown as serially dependent, whereas within each functional group, six species reside in parallel configuration, symbolizing the ability of one species to substitute for another within a functional group.

models of ecosystem reliability

123

Rij (t) ⳱ eⳮ␦t, where R is the reliability, or the probability that the ith species in the jth functional group will exist to contribute to its functioning. Expanding to consider a system with F functional groups, the probability that the jth functional group will contain the species it needs to contribute to the system’s functioning is, Sj

Gj(t) ⳱ 1 ⳮ ⌸ [1 ⳮ Rij (t)]. i⳱1

This formula makes the functioning of a group of species dependent on the presence of only a single species within the group. Expanding to consider an ecosystem made of several functional groups yields F

H(t) ⳱

⌸ Gj(t),

j⳱1

where H represents the reliability of the system. With this model, we can now compare the reliability of an ecosystem with respect to the number of functional groups and the number of species per functional group (fig. 6.5). One hardly needs these mathematical formalisms to demonstrate that an increase in functional groups lower a system’s reliability whereas an increase in species per functional group increases reliability (fig. 6.5). There are two values, however, to performing this simple exercise. First, it is useful to lay out the basics before building more elaborate models. Indeed, textbooks on reliability engineering (Lewis 1987; Billinton and Allan 1983; Dhillon 1983; Ebeling 1997) often begin with similar elementary exercises. Second, the “obvious” for some is nevertheless difficult to grasp by others (Naeem 1998).

Including Colonization in Ecosystem Reliability I modified the model just presented to consider recolonization (Naeem 1998) under the assumption that over a sufficiently long time, all species would be present again. Rastetter et al. (1999) expanded this special case to a general model in which recolonizing species face extinction again—such that a system is never as reliable in the future as it is

124

chapter 6

Figure 6.5 Relationships among ecosystem reliability (H), the number of species (S) per functional group, and the number of functional groups (F) in an ecosystem, following the model for ecosystem reliability in a closed system. Reliability was evaluated after 50 time steps (t ⳱ 50), and the extinction rate (␦) was assumed to be 0.01.

in the present. They also considered the sequence of colonization to be an important factor. All species may not have to be present simultaneously for functioning to occur. Their formulation for the reliability of the ith species in the jth functional group was Rij(t) ⳱

(␦ij␭ij ⳮ ␦ij)e(␦ij␭ijⳮ␦ijⳮ␭ij)t ⳮ ␭ij

␦ij␭ij ⳮ ␦ij ⳮ ␭ij

.

In my initial concept paper (Naeem 1998), I treated functional groups as serially dependent but not necessarily sequentially dependent. Sequential dependency means that successful colonization is dependent on the order of colonization. For example, imagine an ecosystem in

models of ecosystem reliability

125

which pollinators require a yearround source of nectar but plants are divided into early- and late-season flowering groups. If late-season flowering plants went extinct early, the ecosystem would essentially no longer be reliable even though the early-season plants are still supporting active pollinators. Rastetter et al. (1999) expanded the original approach to treat sequential dependency. Under their model—with the key features being continuous probabilities of extinction and colonization and sequentially dependent functional groups—it is difficult to model the reliability of an ecosystem. Rastetter et al. (1999) limited their analysis to three functional groups, with j representing the order in which the functional groups are sequentially dependent. Letting Fj(0) represent the probability of the jth functional group being present during any infinitesimal unit of time, and letting Fj(t) represent the probability of the jth functional group being present over the time unit t, then Fj(t) ⳱ Fj(0) Ⳮ

兰0t Gj(␶j)d␶j,

where ␶ is the place in the sequence where the functional group must be present to ensure functioning. Even for just three functional groups, a singularly ugly model emerges, which I present in tabular form to explain its various terms (table 6.1). The principle finding is that colonization rates strongly affect ecosystem reliability. This result adds the dimension of landscape to our considerations of reliability because colonization rates are often a function of nearness to other ecosystems and the vagility of the organisms that move among them. As landscapes become more fragmented, colonization rates will decline, and reliability will decline even if extinction rates are constant.

Including Nonindependent Extinctions in Ecosystem Reliability The favorite child of community ecology is biotic interactions, but as just modeled, interactions are not included in ecosystem reliability. A biotic interaction in community ecology typically refers to the impacts of the dynamics of one species on the dynamics of another. Reliability models focus on the probabilities of extinction and colonization, thus biotic interactions are more appropriately modeled as the effect of one species on the probability of extinction or successful colonization of another. To model this requires relaxing the assumption that extinction and colonization probabilities are independent. This adjustment is readily made, but the model becomes more complex, especially if

126

chapter 6

TABLE 6.1 Reliability of an ecosystem with three functional groups in which each group is sequentially and serially dependent Data modeled by Rastetter et al. (1999). The variable t represents the time over which one wishes to know the likelihood that an ecosystem will function. The variable ␶ represents the order of the sequence during the interval in which species colonized. The temporal pattern is displayed as a time sequence with the initial point in time, t, at the left of the column and the numbers referring to the presence of functional groups initially (leftmost end of dashed line) and over the interval prior to t. Term H(t) ⳱ F1(0)F2(0)F3(0) Ⳮ

F1(0)F2(0)

兰

t

F1(0)F3(0)

兰

t

F2(0)F3(0)

兰

t

0

0

0

G3(␶3)d␶3 Ⳮ

G2(␶2)d␶2 Ⳮ

G1(␶1)d␶1 Ⳮ

F1(0)

兰 兰

tⳮ␶2

F2(0)

兰 兰

tⳮ␶1

F3(0)

兰 兰

tⳮ␶1

兰 兰 t

0

t

0

0

t

0

t

0

0

0

tⳮ␶1

兰

0

G2(␶2)G3(␶3)d␶3d␶2 Ⳮ

G1(␶1)G3(␶3)d␶3d␶1 Ⳮ

G1(␶1)G2(␶2)d␶2d␶1 Ⳮ

tⳮ␶ⳮ␶21

0

G1(␶1)G2(␶2)

G3(␶3)d␶3d␶2d␶1

Explanation Ecosystem reliability equals the probability that F1 to F3 are all initially present at time t, plus . . . the probability that only F1 and F2 are initially present, but F3 will colonize prior to t and after F1 and F2, plus . . . the probability that only F1 and F3 are initially present, but F2 will colonize prior to t at the appropriate interval, plus . . . the probability that only F2 and F3 are initially present, but F1 will colonize prior to t at the appropriate interval, plus . . . the probability that only F1 is initially present, but F2 and F3 will colonize at the appropriate intervals prior to t, plus . . . the probability that only F2 is initially present, but F1 and F3 will colonize at the appropriate intervals prior to t, plus . . . the probability that only F1 is initially present, but F2 and F3 will colonize in the appropriate intervals prior to t, plus . . . the probability that none of the species were present initially, but by time t they were all present in their appropriate sequence.

Temporal Pattern

123

----------t

12

--------3 t

13

--- 2 --- t

23

1 ------- t

1

-- 2 -- 3 t

2

1 ----- 3 t

3

1 -- 2 -- t

1 -- 2 -- 3 t

models of ecosystem reliability

127

Figure 6.6 Simple community in which two autotrophs (A, and A2) and two decomposers (D, and D2) coexist. In this system, the presence of a second species within a functional group changes the extinction probability of the other species. Reliability for this system is simulated in figure 6.7.

colonization is included. To simplify the treatment, I focus only on extinction probabilities, but clearly species impacts on colonization probabilities are equally important. As I demonstrate, if biotic interactions are included, ecosystem reliability may actually decline with increasing diversity. Thus, the statement that greater species richness within functional groups increases ecosystem reliability no longer holds over all conditions. To include biotic interactions, I use a simple model containing only two functional groups, with two species in each functional group (fig. 6.6). The two functional groups are photoautotrophs (A1 and A2) and chemoheterotrophic decomposers (D1 and D2), two mutually dependent groups. The species within these functional groups are assumed to be redundant or substitutable, with the ability to compensate for the absence of each other. Recalling that the focus is on the loss of species, the ecosystem can assume 16 possible configurations depending on which species are lost. By definition, however, there must be at least one species in each functional group for the system to function. Thus, although 16 states are possible (table 6.2), 7 are nonfunctional, thereby leaving only 9 functional states. These nonfunctional states consist of (1) no species, (2) one of two different decomposers, (3) one of two different autotrophs, (4) two autotrophs and no decomposers, or (5) two decomposers and no autotrophs. Estimating the reliability of the system requires estimating and summing the reliabilities for each functional state, which is a function of the probability that the full system will, by local extinction, enter the other states. That is, the reliability in this model is modeled as a Markov process. This approach follows the methods of estimating the reliability of systems in which components share loads (e.g., if one of two generators servicing a plant fails, the other takes on the load). In

128

chapter 6

TABLE 6.2 Sixteen possible states of the near-minimal ecosystem portrayed in fig 6.6 1 means that a species is present, 0 that it is absent. State

A1

A2

D1

D2

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

1 0 1 1 1 0 1 0 1 0 0 0 1 0 1 0

1 1 0 1 1 1 0 1 0 0 0 0 1 1 0 0

1 1 1 0 1 0 0 1 1 1 0 1 0 0 0 0

1 1 1 1 0 1 1 0 0 1 1 0 0 0 0 0

such systems, the surviving generator has a shorter life span because it takes on a larger load. Such load-sharing models (Lewis 1987) differ from ecological models because designed systems would not incorporate two components in which the failure of one would improve the survivorship of the other. Natural ecosystems, however, are not designed, so anything goes. In ecological systems, for example, the reduction within a functional group from two to one competitors would actually decrease the probability of extinction; this scenario corresponds to inverse load sharing—something, as mentioned earlier, one would not expect in an engineered system. Estimating the reliability of a system with load sharing requires solving the transition probabilities among states, then summing the reliabilities of the functioning states (states 1–9). That is, R(t) ⳱

兺

Pi(t),

i∈o

where i∈o indicates the states in which the system is operating (states 1–9 in this analysis, as listed in table 6.2). Determining the transition probabilities for this system is straight-

models of ecosystem reliability

129

forward. The first state is assumed to be the state in which all species are present and t ⳱ 0. Therefore, P1 (0) ⳱ 1 and Pi (0) ⳱ 0 where i ⬆ 1. We now need to determine Pi(t), or the state transition probabilities. If species extinction probabilities within functional groups are independent, then the ␦fd are independent (equivalent to no load sharing), where f represents the functional group and s is the species (in our simplest model, f ⳱ A or D, autotroph or decomposer, respectively, and s ⳱ 1 or 2): dP1(t)/dt ⳱ ⳮ␦A1 P1(t) ⳮ␦A2 P1(t) ⳮ␦D1 P1(t) ⳮ␦D2 P1(t).

(1)

That is, the net change in the probability of the system being in state 1 is the sum of the decreases in probabilities caused by the probabilities of any component failing. Thus, dP1(t)/dt ⬍ 0, or with every moment in time the probability that the system is in state 1 decreases. This procedure essentially models reliability as a continuous-time Markov process (Karlin and Taylor 1975). The change in the probability of the system being in a particular state is, dP2(t)/dt ⳱ ␦A1 P1(t) ⳮ ␦A2 P2(t) ⳮ ␦D1 P2(t) ⳮ ␦D2 P2(t)

(2)

dP3(t)/dt ⳱ ⳮ␦A1 P3(t) Ⳮ ␦A2 P1(t) ⳮ ␦D1 P3(t) ⳮ ␦D2 P3(t)

(3)

dP4(t)/dt ⳱ ⳮ␦A1 P4(t) ⳮ ␦A2 P4(t) Ⳮ ␦D1 P1(t) ⳮ ␦D2 P4(t)

(4)

dP5(t)/dt ⳱ ⳮ␦A1 P5(t) ⳮ ␦A2 P5(t) ⳮ ␦D1 P5(t) Ⳮ ␦D2 P1(t)

(5)

dP6(t)/dt ⳱ ␦D1 P2(t) Ⳮ ␦A1 P4(t) ⳮ ␦A2 P6(t) ⳮ ␦D2 P6(t)

(6)

dP7(t)/dt ⳱ ␦D1 P3(t) Ⳮ ␦A2 P4(t) ⳮ ␦A1 P7(t) ⳮ ␦D2 P7(t)

(7)

dP8(t)/dt ⳱ ␦D2 P2(t) Ⳮ ␦A1 P5(t) ⳮ ␦A2 P8(t) ⳮ ␦D1 P8(t)

(8)

dP9(t)/dt ⳱ ␦D2 P3(t) Ⳮ ␦A2 P5(t) ⳮ ␦A1 P9(t) ⳮ ␦D1 P9(t)

(9)

For example, the probability of being in state 2 over a small unit of time is the probability that the first autotrophic species goes extinct, but this probability is decreased by the possibility that the second autotrophic species and either decomposer species may go extinct. Inverse load sharing is achieved in this model by assuming that the extinction rates of all species are ␦1 when all species are present within a species’ functional group, and ␦2 when one species within the functional groups is missing. In the case of interspecific competition among species within a functional group, ␦2 ⬍ ␦1 because the loss of a competitor enhances persistence. For facilitation among spe-

130

chapter 6

cies within a functional group, ␦2 ⬎ ␦1 because the loss of a facilitator reduces persistence time. The easiest way to treat load sharing is to allow ␦2 ⳱ a␦1, where a is a constant relating the extinction rate of a species in the presence of another species within its trophic group (␦1) with the extinction rate of a species in the absence of the other species (␦2). If a ⬍ 1, then the species are competing, whereas a ⬎ 1 indicates facilitation between the two species. The models using load sharing are as follows: P1(t) ⳱ eⳮ4␦1t P2(t) ⳱ P3(t) ⳱ P4(t) ⳱ P5(t) ⳱

1 2ⳮa

[eⳮ(2Ⳮa)␦1t Ⳮ eⳮ4␦1t]

P6(t) ⳱ P7(t) ⳱ P8(t) ⳱ P9(t) ⳱

(1 Ⳮ a) 2(2 ⳮ a)

2

[eⳮ2a␦1t ⳮ 2eⳮ(2Ⳮa)␦1t Ⳮ eⳮ4␦1t].

Reliability of this system is therefore, 9

Rs(t) ⳱

兺

Pi ⳱ eⳮ4␦1t Ⳮ

i⳱1

Ⳮ

4 2 ⳮa

[eⳮ(2Ⳮa)␦1t ⳮ eⳮ4␦1t]

2(1 Ⳮa) (2 ⳮ a)

2

[eⳮ2a␦1t ⳮ2eⳮ(2Ⳮa)␦1t Ⳮ eⳮ4␦1t].

Figure 6.7 plots the reliability of this system under different assumptions about the degree of interaction among species, with a ⳱ 0 (no load sharing), with a ⳱ 0.1 (species within trophic groups are interference competitors), and with a ⳱ 10.0 (species within trophic groups are facilitators). This model shows that communities in which coexistence leads to enhanced probabilities of local extinction (e.g., by suppression of population sizes) enhance the reliability of the system because the loss of a species increases the probability of a species persisting. Alternatively, systems that contain many facilitators that enhance survivorship decline in reliability if species are lost. Biodiversity, at least in terms of species richness, is not necessarily associated positively with reliability if extinction probabilities are not independent. The idea that community persistence is tied to one’s dependency on other species is not new, but explicitly tying the consequences of such dependency to

models of ecosystem reliability

131

Figure 6.7 Four different scenarios for reliability with load sharing or interacting species in a simple ecosystem. Ecosystem reliability is estimated for a 2–functional group ecosystem (autotrophs and decomposers only) with only 2 species per functional group, as illustrated in figure 6.6. Species within functional groups “interact” by affecting each other’s probability of extinction. Each line plots reliability under four different scenarios (A–D). A (dashed line). Functional groups are serially dependent, and the loss of either species within the functional group leads to failure of the functional group (species cannot substitute for one another). B (solid, thick line). Species are linked in parallel redundancy (one species can compensate for the loss of the other). C (dash-dot line). Species are linked in parallel redundancy, but within functional groups they compete with one another (a ⬍ 1) and the loss of a single species releases the other from competition (lowers probability of extinction). D (solid, thin line). Species are linked in parallel redundancy, but species within functional groups facilitate each other (a ⬎ 1), and the loss of a single species increases the probability of extinction of the other. Note that any of these interactions can dramatically alter the reliability of a system. For example, a system without parallel redundancy (A) is the least reliable for a short period (t ⬍ 100), but a parallel redundant system with facilitators in each functional group can be the least reliable over long periods (t ⬎ 100).

132

chapter 6

the reliability of system functioning puts a different perspective on the issue. Of course, biotic interactions and the impact of one species on the extinction probabilities of another are not the same, but both provide additional linkages among species within an ecosystem, and both show that conclusions about systems are strongly affected by such linkages. Note that if I included the impact of species on probabilities of successful colonization as I did for the extinction probabilities just shown, the general conclusion—that diversity could decrease reliability—would remain true, but the actual outcome would be dependent on how species’ impacts on extinction and colonization balance one another. One might expect that typically they work together, but it is easy to envision cases in which they do not. For example, a competitor that increases the probability of the extinction of another species by reducing available resources or allelopathy is also likely to reduce the success rate of colonists by the same mechanisms. The opposite, however, could be true. For example, the competitor may act as a “nurse” plant by shading a seedling of its competitor and thereby increasing its competitor’s ability to colonize, even if it reduces the probability of survival of an established plant. Such complexities always exist in nature and make generalizations as difficult in community ecology as in ecosystem reliability.

Estimating Parameters and Applying Ecosystem Reliability Applying ecosystem reliability to real-world problem solving can, in some cases, be less difficult than applying principles of community ecology. The latter is difficult because application may require measuring things—carrying capacity, the intrinsic rate of population increase, resilience, resistance, niche axes, coefficients of biotic interactions, and interaction strengths—that are very difficult to measure, let alone define to anyone’s satisfaction. Reliability, at least on the surface, may be more straightforward. Local extinction and colonization rates are readily defined. Presence-absence data are easier to collect than relative densities. Functional group definitions can be tailored to specific functions of interest. Of course, until actually applied to a real-world problem, this claim of greater utility and ease in determining reliability over other community-level processes remains unproven. Like carrying capacity, intrinsic rates of population increase, and other community variables, extinction probabilities can be equally difficult to determine, but in the next section I suggest how one might begin.

models of ecosystem reliability

133

Estimating Extinctions and Colonizations If there are S species in the global pool (SG), regional species pools (SR) are likely to be partial sets of these species, and the ith community is likely to contain an even smaller set of species (Si). The presenceabsence matrix is the simplest representation of this information, in which species represent the columns, community replicates represent the rows, and the cells indicate presence or absence. I assume that differences in composition are due to two processes: extinction and colonization. The variables that govern colonization might be habitat size or nearness to other habitats, whereas local extinction might be caused by processes such as competitive exclusion or stochastic events. Both colonization and local extinction are probabilities. Over long periods, the probability that a species will be extinct in a community is ␦t, provided that there is no colonization. If there is colonization, then presence and absence is a combination of colonization and extinction probabilities. The model that reflects this combined probability is ␦1 ␦t ⳱ [1 ⳮ (1 ⳮ ␭1 ⳮ ␦1)t]. ␭1 Ⳮ ␦1 The model for colonization is ␭1 ␭t ⳱ [1 ⳮ (1 ⳮ ␭1 ⳮ ␦1)t]. ␭1 Ⳮ ␦1 Clark and Rosenzweig (1994) demonstrate how these models can be used to estimate extinction and colonization probabilities from presence-absence data. Estimating Extinction Probabilities Extinction probabilities are difficult to estimate in part because they are determined by many factors. For example, patch area, patch age, and densities of resident species within guilds can be strong determinants of local extinction. One can obtain estimates of extinction probabilities using patch, disturbance, climate, population census, and other information collected over suitable periods of time. For example, Crooks et al. (2001) used logistic regressions to determine what extrinsic and intrinsic factors in a series of habitat islands served as the best predictors for the extinction of birds over a 50-year period. One or more local extinctions occurred in 16 of the 30 habitats over a 50year interval, and island size, island age, and density indices of bird species served as predictors for extinction (Crooks et al. 2001). This

134

chapter 6

method could be applied to other systems if the appropriate data existed. Determining the Independence of Extinction Probabilities Given that biotic interactions affect reliability, it would be useful to understand how the presence of one species affects the probabilities of other species being present. One possible approach is the variance test for biotic interactions (Schluter 1984), which estimates average interactions (associations) within a community from presence-absence data. This estimate requires N samples, which can be considered samples from N replicate ecosystems for our purposes. Within each sample, there are ni occurrences of the ith species. There are M species in the regional or global pool, with Tj species in each ( jth) sample. Table 6.3 presents imaginary data of this kind to illustrate the meanings of the terms. This test requires estimating the variance in the total number of species, ST2, and estimating the variance of occurrence (presence-absence), ␴i2. The formulas are N

ST2 ⳱ (1/N) 兺 (Tj ⳮ t)2, j

where t is the observed mean number of species per sample, and

␴i2 ⳱ pi(1 ⳮ pi), where pi equals ni/N. The value of interest is V ⳱ ST2冒␴i2, where V is the index of species association in samples. Although it would be more valuable to have a matrix of speciesby-species effects on each other’s extinction probabilities, this variance test can be used to temper estimates of reliability. If V ⳱ 1, then species do not covary, and we might infer that reliability is unaffected by interactions. If V ⬍ 1, then species covary negatively, while V ⬎ 1 indicates that species covary positively, the former indicating that estimated reliability may be greater than expected, and the latter indicating that reliability is worse than expected. Interestingly, Schluter (1984) noted that positive associations seemed to be more reflective of the data he examined from real communities. The variance test has its limitations. For example, unless we are

models of ecosystem reliability

135

TABLE 6.3 Determining the independence of extinction probabilities using the Schluter’s variance test This table outlines how terms in the variance test are derived using an imaginary ecosystem consisting of 3 functional groups as an example. Five replicate ecosystems are presented as local species pools (left-most column). X indicates a species is present (1), whereas a blank cell indicates the species is absent (0). For the variance test of biotic interactions, M ⳱ 9 (species A–K), N ⳱ 5 samples or local ecosystems. Functional Group 1

Functional Group 2

Functional Group 3

Ecosystem

A

C

D

F

G

Local Species Pool 1 Local Species Pool 2 Local Species Pool 3 Local Species Pool 4 Local Species Pool 5 ni Regional Species Pool Global Species Pool

X

X

X X X

X X X

X

1 X X

X 4 X X

X 2 X X

B X

X X 3 X X

X X 3 X X

X X X 4 X X

X 3 X X

E

0 X

H

K

S or Tj

F

X

5 3 4 3 6

3 3 3 1 3

8 9

3 3

1 X X

certain that presence-absence data from replicate samples reflect all possible combinations of species in an ecosystem, we cannot be certain how much confidence we should have in V. Estimating true interactions among probabilities of species extinction would require experiments, since covariance in presence or absence is not necessarily related to interactions—much as covariation in density is not necessarily related to biotic interactions between species (Schluter 1984).

Implications and Limitations There are numerous implications for assessing species expendability if we consider the functional roles of species rather than their roles in determining community dynamics; however, there are also important limitations to keep in mind. Whereas nonlinear dynamics, predatorprey theory, age-structured demography, simple diffusion approximations of extinction processes, and many other theories that are the wonderful stuff of population and community ecology have found their way into practical applications, the perspective of ecosystem functioning has not been employed as extensively. Species cannot be managed successfully if their linkages to other species via material

136

chapter 6

and energy flow are overlooked. As shown in figure 6.1, the roles of species in greenhouse gas regulation, soil fertility and water quality, pH, nitrogen cycling, and other biogeochemical processes are clearly seen when placed within the context of ecosystem functioning, yet are potentially overlooked when viewed only from a community perspective. Bringing production into balance by managing plant communities, for example, may be short lived without recognition of the importance of understanding and managing autotrophic linkages to decomposers (Harte and Kinzig 1993; Berman-Frank and Dubinsky 1999; Wall and Moore 1999; Naeem et al. 2000), herbivores (McNaughton 1979; Knapp et al. 1999; Mulder et al. 1999), and mycorrhizal processes (Van der Heijden et al. 1998; Hartnett and Wilson 1999). There are limitations to this approach, however, that have to be addressed. For example, defining functional groups is critical. This process has a long tradition in ecology and has been applied to many groups of organisms (Raunkier 1934; Root 1967; Meyer 1993; Steneck and Dethier 1994; Gitay and Noble 1997; Smith et al. 1997; Bisevac and Majer 1999), but approaches have been quite varied. For example, if a functional group is defined too narrowly, every species becomes singular. If it is defined too broadly, every species becomes redundant. This problem is analogous to those of defining species in conservation biology (Meffe and Caroll 1994). For example, birds of paradise (family Paradisaeidae) would be described as 42 species by the more restrictive biological species concept, but as 90 species by the phylogenetic species concept (Meffe and Caroll 1994). Similar problems arise when conservation is based on species rather than populations (Hughes et al. 1997). The solution to such issues is to be pluralistic (Meffe and Caroll 1994) and to eschew population, community, or ecosystems chauvinism. The timescale over which reliability is calculated is equally problematic. Ecosystem reliability is best estimated for short timescales, on the order of decades or centuries. For example, the reliability of a grassland ecosystem in Minnesota on a scale of tens of thousands of years is zero, since no species composition will survive periodic glaciation. Likewise, on a scale of a year or two, the local extinction of even rare species is unlikely and, even if it did occur, would have little influence on mean ecosystem performance. This trend is particularly problematic when dealing with species that have different temporal or spatial scales, such as a model containing microbes, plants, insects, and ungulates. The problems of scale plague most areas of ecological research, and in this regard, ecosystem reliability is no different.

models of ecosystem reliability

137

The models I employ are also relatively simple (although I provided a few examples of how these simple models can be developed readily to tackle more complex problems). These models use presence/absence data, do not contain colonization, and do not allow for evolution. They also make no assumptions about ecosystem design. Presence/absence data, however, are not only widely used in community ecology, they are also appropriate for expendability data that examine the consequences of a species being absent from its community or ecosystem. The phenomena that are being addressed concern ecological timescales (a few generations), so it is appropriate to exclude natural selection and the evaluation of species. Although the models used could be modified to address these issues of colonization, evolution, design constraints, and much more, it would be outside the scope of this chapter. Such further development, however, would not alter the basic conclusions that, from the standpoint of long-term functioning, species are likely to be more indispensable than a shortterm or community study might reveal. As one explores ecosystem reliability, doubtless many further complications will arise without ready analogy to mechanical systems. In the models presented here, for example, I employ serial dependency among functional groups and parallel redundancy. There can be nested sets of serial dependencies in which species within functional groups are dependent on one another. Such systems would seem particularly prone to failure, requiring either high rates of colonization or low probabilities of extinction to function. There can also be circular serial dependencies, as in many nutrient cycling pathways, which may also compound the probabilities of failure. There can be interactions among functional groups, much as there are among species within functional groups. Reliability block diagrams of ecosystems may be every bit as complicated as community web diagrams, but conclusions about species expendability are likely to differ considerably between these different types of analyses of ecosystem complexity. My plea, again, is to employ both perspectives, not one or the other. Central to the notion of reliability is the concept of redundancy—a concept that seems to imply no need for unique or “singular” species. A singular species is not substitutable, thus its loss is not compensated for by compensatory growth. For example, chinook or sockeye salmon, neither of which is a keystone species, are so unique that their loss from the Snake River is not likely to be followed by compensatory growth of another species in the river. In such cases, no method that attempts to quantify the importance of species relative to

138

chapter 6

others, either in terms of reliability or keystone status, would succeed in capturing the importance of unique species. This is a serious limitation of ecosystem-based management, again highlighting the need for pluralism in conservation biology.

Synthesis and Conclusions Expendability concerns the ecological consequences of species loss. If there are no consequences of losing a species, the species is considered expendable. The danger in this logic is that there is no easy way to know fully what all the consequences of losing a species might be. Here, I have surveyed modeling approaches that directly address the ecosystem consequences rather than the community consequences of biodiversity loss. I have argued that there is no necessary correlation between the impact of species loss on community properties and processes and the impact of species loss on ecosystem processes and properties. For illustration, I have chosen ecosystem reliability that shows clearly that rare, substitutable, competitively inferior, redundant species that may have little or no impact on communities or ecosystems over short periods may nevertheless be integral to reliable ecosystem functioning over long periods. Ecosystem reliability is a different way of looking at species importance or species expendability. As presented here, it uses presence/ absence data, examines ecosystem functioning rather than dynamics, and is distinctly nonevolutionary in its perspective. One can modify the models, relax assumptions, and even create a set of models that smoothly grade from pure community to pure reliability. Extinction probabilities, for example, can be based on population dynamics, treating species with high variability in densities as more prone to extinction. The strengths of biotic interactions can be used to estimate species impacts on extinction or colonization probabilities of species necessary for reliability models. Food webs can be converted to reliability block diagrams. Phylogenetic distances might serve as guidelines for estimating redundancy. Probabilities of extinction and colonization might be allowed to evolve in one’s model, just as some models examine the evolution of biotic interactions. If we were to carry out these and other exercises, we would achieve synthesis—and the distinction between community and ecosystem perspectives would vanish. Our definitions of expendability would also be more robust than those based on less synthetic approaches. The models presented not only serve as heuristic constructs, but can also be developed for use in real-world assessments of ecosystem

models of ecosystem reliability

139

reliability. Many of the foci of community ecology are difficult to apply to the issues of expendability. For example, using interaction strength as a measure of expendability is difficult. Berlow (1999) argues that higher variability can be associated with weaker interactions, which means that at critical times, weak interactions may have strong effects. McCann et al. (1998) argue that many species with weak interactions could dampen oscillations and thus stabilize the community. This complication does not diminish our ability to employ community approaches for estimating species expendability. Rather, it enriches our understanding of the problem. Ecosystem functioning, in relationship to biodiversity, is a new discipline, and the complications that arise (the heart of the controversies) similarly enrich our understanding of expendability from an ecosystem perspective. The concept of the keystone species, and Paine’s devotion to exposing the innumerable ways in which species persist in nature, spurred the complementary way of thinking that is expressed in this chapter. In the same way that a keystone species is defined by the voussoirs it binds together, the voussoirs define the keystone species by their relative positions. It is the arch, not the stones, that matter. The functioning of an ecosystem reflects the community as much as the community governs ecosystem functioning. Ecosystem functioning, not the richness of species, affects the environment, but the richness of species guarantees reliable environmental regulation by our ecosystems.

Acknowledgments I thank P. Karieva, S. F. Tjossem, and three anonymous reviewers for critical evaluation of this manuscript and support from the National Science Foundation (DEB 9996114) for model development.

Chapter 7

m Predicting the Effects of Species Loss on Community Stability Dan Doak and Michelle Marvier

The question of how species richness and community organization may influence ecological stability has fostered a longstanding debate, recently revived through a spate of new field and microcosm studies (e.g., Tilman 1996; Naeem and Li 1997; McGradySteed et al. 1997) and modeling efforts (e.g., Doak et al. 1998; Hughes and Roughgarden 1998; Yachi and Loreau 1999). This new research has led to renewed claims regarding the importance of species diversity for various community and ecosystem functions as well as increasing efforts to develop simple principles to guide conservation policy and practice (e.g., Pimm 1991; Naeem 1998; Schwartz et al. 2000; Hector et al. 2001). Nonetheless, the age of—and attention to—the stability-diversity question does not imply its resolution or direct application to questions of environmental management (McCann 2000; Schwartz et al. 2000; Hector et al. 2001). A particular problem with much of the ongo-

species loss and community stability

141

ing work is the neglect of details about species-specific life histories and ecological interactions. While the nuances of individual species are key in determining their community importance, most theoretical work on diversity effects glosses over such details. Our goals in this chapter are twofold. First, we review recent work that has explored the relationship between species diversity and community stability, emphasizing modeling efforts (for reviews of more empirical work, see Loreau 2000; McCann 2000; and Schwartz et al. 2000). In addition to examining the consequences of species richness, we also ask whether it is possible to predict the consequences of losing particular species from a community. Although the problem of predicting single-species effects is far messier than that of arguing over more general stability-diversity patterns, it is also far more important for conservation. Our approach comprises a Markov community model that addresses three general questions: (1) What aspects of life history or interactions with other species determine the importance of particular species for the stability of communities? (2) What features predispose a species to become dominant following the extinction of the original dominant species? And, most importantly, (3) can we accurately predict the importance of particular species removals in the absence of detailed field observations or experiments? Throughout this chapter, we gauge species importance in terms of its effects on the stability of aggregate community properties (e.g., total community abundance) or on community composition (May 1973; Power et al. 1996b; Tilman 1996).

Recent Studies of Stability-Diversity Patterns The stage was set for the current crop of stability-diversity modeling by the historic mismatch between May’s modeling results (1973) and the empirical work of McNaughton and others (Mellinger and McNaughton 1975; McNaughton 1977, 1985; Frank and McNaughton 1991). The models predicted declining stability with increasing species richness, whereas empirical observation generally suggested the opposite. Although there are numerous, competing ways of classifing the various mechanisms that contribute to stability-diversity patterns (e.g., Tilman 1999; Loreau 2000), we discuss them in terms of four simple categories:

• •

Simple sampling effects Niche complementarity and compensatory competition

142

• •

chapter 7

Averaging effects (a.k.a. portfolio effects) Addition of weakly interacting species

The first mechanism that can enhance community stability with increasing species number is a simple sampling effect: with a large number of species present, there is a greater chance that one or more of the species will be stable and dominant. In other words, the stability of more diverse communities may simply be due to the increased chance of a stable species being present. Although this simple sampling effect has more often been mentioned as a factor causing diversity-productivity relationships (Loreau 2000), it can also explain stability-diversity correlations. The most common explanations of stability-diversity relationships are all somewhat more complicated versions of the sampling effect idea. These include niche complementarity and compensatory competition. The basic idea is that diverse groups of species are more stable because complementary species compensate for one another if one species suffers severe declines (e.g., Tilman 1996); with more species, there is an increased chance of complementary species being in the community. Naeem (1998 and this volume), using models from reliability engineering to describe ecosystem processes, has extended this basic idea to predict that more species-rich communities will be less affected by species losses. He argues that (1) as more functional groups are needed for proper ecosystem functioning, ecosystem reliability declines, simply because there are more ways for the system to fail, but (2) more species per functional group (higher species redundancy) increase the reliability of the ecosystem. Nijs and Impens (2000) use a careful, though abstract, probabilistic analysis to reach similar results. Essentially, redundancy within a functional group means that when one species is deleted from the community, another species with similar function will “step in” and fill the functional role of the deleted species. Whereas the above ideas rely on the interactions of competing species, stability in aggregate measures (such as total productivity) is also expected to increase with species richness due to an averaging of random fluctuations in the growth of each species without any special competitive mechanisms. Using simple probability theory and simulations, Doak et al. (1998) demonstrated that as more independently varying species are added together, the sum of their abundances becomes more stable. However, the strength of this averaging effect depends on several factors (Doak et al. 1998; Tilman et al. 1998; Schwartz et al. 2000). First is correlation among the species’ fluctuations, with more positive correlations leading to reduced averaging effects, and

species loss and community stability

143

negative correlations increasing the strength of these effects on stability. Second, the averaging effect is strongest when species are on average equal in abundance, and weakest when one species holds the lion’s share of community biomass. Finally, the strength of averaging depends on the mean-variance relationships for individual species in the community. In the Doak et al. (1998) models, total community biomass was held constant and the variances of the biomass of individual species were assumed to scale with the square of the species’ mean abundance (i.e., standard deviation scales linearly with mean abundance, and coefficients of variation are constant across species). In response to this study, Tilman et al. (1998) demonstrated that if per capita variability in numbers increases dramatically with decreasing mean abundance, the averaging effect can be eliminated or reversed. Regardless, Yachi and Loreau (1999) have demonstrated the same averaging effect of diversity under much less restrictive assumptions about the relationship between the mean and variance of species’ biomasses. In their model, increasing the total number of species increases both the mean community productivity and the stability of community productivity with the dynamics of community biomass driven largely by the set of most abundant species. In particular, a species can contribute substantially to community stability only if its range of abundance overlaps with that of the most dominant species. In other words, the loss of species that are consistently rare is predicted to have little effect on either the productivity or the stability of the community as a whole. This is akin to Doak et al.’s (1998) result that stability-diversity relationships are weakened by highly skewed distributions of mean abundance. The studies just cited rely largely on statistical descriptions of community fluctuations. Other studies, however, have more explicitly included the effects of competitive interactions into models of stochastic community dynamics. Hughes and Roughgarden (1998, 2000), Ives and his coworkers (1999), and Lehman and Tilman (2000) use several alternative stochastic, density-dependent Lotka-Volterra competition models to explore the stability of community biomass. Hughes and Roughgarden (1998) consider only two species at a time, and they assess the stability of these “communities” as a function of the strength of the competitive interaction (indicated by the magnitude of the competition coefficients) as well as the degree of asymmetry between the two species’ competitive effects. They find that the stability of the aggregate, or community, biomass is relatively independent of the strength of the species’ competitive effects, but that community biomass becomes less stable as the disparity between the two species’

144

chapter 7

competitive effects grows. Similar models of multispecies systems (Ives et al. 1999; Hughes and Roughgarden 2000) show more generally that positive stability-diversity relationships are usually predicted and appear to be robust over many alternative model structures (Lehman and Tilman 2000). Most surprisingly, the details of competitive interactions seem to have relatively little influence on the generation of stability-diversity relationships (Ives et al. 1999). Ives et al. (2000) show that multitrophic models also show similar stability-diversity relationships, considerably broadening the class of models that show these effects. Finally, increases in diversity can enhance stability if the additional species have only weak interactions with the other members of the community. This possibility was first hinted at by Robert May’s analyses of diversity-stability relationships (May 1971, 1973), in which a tight relationship between interaction strength and instability was documented. However, the exact mechanism was only recently worked out by McCann et al. (1998), using community matrix models much like May’s, except with predator switching and nonequilibrium dynamics. Starting with simple food webs, McCann et al. added species while simultaneously titrating the interaction strengths of the new species. They found that diversity begets stability as long as the enhanced diversity comes in the form of species with weak interaction strengths. This pattern occurs because the addition of weakly interacting species dampens oscillations between strongly interacting species and thereby increases the stability of the overall community. In sum, past modeling and empirical work indicates that the mechanisms driving stability-diversity correlations include averaging, competitive release, symmetry in competitive effects, and the addition of weak interactions to food webs. However, most of the diversity-stability models, like all models, suffer from a lack of realism of one form or another. First, all of the models to date assume a closed system and a constant suite of species. In reality, even intensively managed field studies cannot obtain precise control of species richness. The turnover of species is an integral feature of natural communities that is important to consider when linking model predictions to empirical data (Schwartz et al. 2000; Hector et al. 2001). Second, the use of stability-diversity models to make meaningful statements about the preservation and management of ecological communities is limited, since little or no attention has been given to what makes a particular species important to the community. Instead, models have sought to link total species number with community patterns, a relationship that may rarely be of importance

species loss and community stability

145

for practical management. Finally, even less attention has been given to whether readily available data would provide any insight regarding species importance. In other words, within the stability-diversity literature, there has been little progress toward recognizing the set of traits that might identify which species are the most dangerous to lose and hence most crucial to preserve. We investigate these last two questions using a simulation model of competing, space-occupying species.

Methods Basic Model Structure We developed a stochastic simulation model for a set of competing species that is a direct extension of a first-order Markov model of community replacement (succession) dynamics (Waggoner and Stephens 1970; Horn 1975). Markov community models best represent groups of sessile, competing species in which priority effects predominate (an individual must die before there is any chance of replacement by an individual of the same or a different species) and consumer-prey interactions are of reduced importance. This modeling approach has proven to be a highly robust way of quantifying community structure and dynamics for a variety of ecological communities, including termites (Usher 1979), coral reefs (Tanner et al. 1994, 1996), rocky intertidal systems (Wootton 2001), forests (Waggoner and Stephens 1970; Horn 1975; Runkle 1981; Barnes and Dibble 1988), and desert plants (McAuliffe 1988). These models are also capable of accurately predicting the effects of species removals for real communities (Wootton 2001). We chose to use this framework both because of its success in predicting real community patterns, and also because these models are able to incorporate various aspects of species’ life histories and interaction types (McAuliffe 1988; Wootton 2001) while still having the virtue of minimal structural complexity. We first present the deterministic form of the model, then describe stochastic simulations based on this framework. Our model is represented as an (S1)  (S1) matrix for S species (the “1” is because we also keep track of empty space). By repeatedly multiplying the community transition matrix, C, by a vector, Nt, of the amount of space occupied by each species (or empty) at time t, we can project changes in the abundance of each species, ni, as well as of the total S

community, CB 

兺 ni through time:

i1

146

n1 n2 ... nS E

chapter 7

A B t1

s1  a1,1(1  s1) a2,1(1  s1) ... aS,1(1  s1)  C1  s  a1,1  a2,1  ... 1  aS,1)(1  s1)

[

1

[

(

)]

a1,S(1  sS) a2,S(1  sS) ... sS  aS,S(1  sS) a  a2,S  ... sS  1,S  aS,S)(1  sS)

(

a1,2(1  s2) s2  a2,2(1  s2) ... aS,2(1  s2) a  a2,2  ... 1  s2  1,2  aS,2)(1  s2)

[

(

a1,E a2,E ... aS,E a  a2,E  1  1,E ...  aS,E

)] (

... ... ... ...

)] ...

n1 n2  ... nS E

)D A B

t

In the matrix, C, cij is the probability that an individual of species j is replaced by an individual of species i in a single time step. For transitions between individuals of the same species (i  j), cjj  sj  (1  sj)ajj, where sj is the survival probability for an individual of species j, and ajj is the probability that an area vacated by species j is immediately colonized by new individuals of the same species. For transitions between species (i ⬆ j), cij  (1  sj)aij, where aij is the probability that an area vacated by species j is immediately colonized by new individuals of species i. The final (S1) row of C includes transitions to unoccupied space. The total amount of space remains constant through time due to the S

constraints that

兺 aij ⱕ 1 and that all vacated space not immediately

i1

colonized becomes “empty.” The final (S1) column of C includes the probabilities that empty space is colonized by a species i, aiE, or remains unoccupied. This model structure implicitly assumes that annual sampling of the community occurs just after the major recruitment season for most members of the community—if sampling occurs at some other season, or if recruitment is highly asynchronous, the transition probabilities cannot be divided so cleanly between survival (sj) and conditional replacement probabilities (aij). Although simple, this model allows a substantial amount of ecological reality. For example, the species-specific survival terms allow one to incorporate variable life spans, and the probabilities of transition between each pair of species allow the implicit inclusion of factors such as shared habitat requirements and preferential colonization by some species of space already occupied by others (McAuliffe 1988). However, this model is still quite limited in the types of communities

species loss and community stability

147

that it can portray accurately. These are largely communities in which adult survival precludes the establishment of new individuals, and in which the abundance of recruits is not limited by the local abundance of adults.

Stochastic Simulations We used the basic model structure just described to conduct stochastic simulations, picking new values for each sj and aij in each year based on predefined means and variances for each parameter. The basic model represents a community using 4S  2S2 parameters: the mean and variance for the annual survival of each species and for the conditional probabilities that determine how vacated space is reallocated among all combinations of species and unoccupied space. To simulate the dynamics of a broad range of randomly constructed species and community types, we first had to specify the rules that govern the behavior of each community matrix. To create a community (limited to 10 or fewer species in this paper), we randomly picked means for all sj values from a uniform distribution between the limits of 兵0,1其, 兵0,0.5其, 兵0.5,1其, 兵0.25,0.75其, or 兵0.5其, and means for aij values from a uniform distribution such that all lay between 兵0,1其 and S

兺

aij ⱕ 1. We also randomly assigned variances to each parameter,

i1

selecting values from a uniform distribution bounded by 0.001 and either 10% or 50% of the maximum possible variance given a parameter’s mean (cf. Doak et al. 1994). In these models, annual variation in aij and sj values was simulated by beta-binomial variables with no correlation or autocorrelation. In all, we used 10 sets of parameter value restrictions (5 sets of bounds on sj and 2 bounds on maximum variation) and created a minimum of 20 different random communities for each combination of rules. For a given set of parameter values, we simulated the dynamics of the entire community for 110 years. We recorded no information for the first 10 years (transient dynamics for a deterministic version of the model generally settled within 10 years) and then recorded the abundance of each species and of the entire community for each of the remaining 100 years. The total amount of space in a simulation always remained constant at 100(S1) spaces. The model was initiated with space divided evenly among the number of species  1. For example, in a community with 10 species, each species initially occupied 100 spaces with the remaining 100 spaces empty.

148

chapter 7

Over the 100 years of data collection for each simulation, we recorded the mean and variability in the abundance of each species. To measure community stability (or, really, instability) we used the coefficient of variation in total numbers, or community biomass, over the 100year sampling period, CVCB. We used the model in two ways. First, to examine the overall relationship between species diversity and community stability, we created communities that varied in the total number of species (S  2 to 10) and recorded CVCB over 100 years. Here, we allowed annual survivorships with means of 兵0,1其 and with standard deviations up to 50% of the maximum. We simulated 40 replicate communities per level of species richness. Second, to explore the consequences of removing particular species, we compared the stability of communities before versus after an extinction event. We began by simulating each full community to establish baseline dynamics and then explored the consequences of removing each of the original species one at a time. For S  5, for example, we ran a simulation with all 5 species, and then 5 additional simulations, each with 1 species removed and the other 4 remaining. For communities with 5 species, we ran 40 replicates, whereas for communities with 10 species we ran 20 replicates. Following a species removal, we reallocated the share of vacated space that would have been colonized by the removed species, by assigning the removed species’ share of space to all remaining species proportional to their mean aij values. For example, if species 1 were removed, the new replacement values (a’ij) for species i and j  2 through 5 would be recalculated as: 5

aij  aij  aij * a1j/

兺 aij.

i2

To compare the effect of each species’ removal on community stability, we used the percentage change in CVCB between the full model and each of the models with one species missing: PCVCB  100

(

CVCB,removal  CVCB, full CVCB, full

)

,

where CVCB,removal is the coefficient of variation in community numbers after a species removal, and CVCB,full is the coefficient of variation for the full community.

species loss and community stability

149

Using this measure of community importance, we asked how easy is it to predict the effects of a species’ loss with limited information (e.g., previous abundance, longevity, or ability to colonize unoccupied space). In addition, for the subset of simulations in which the dominant species was removed, we used discriminant function analysis to identify attributes that predispose a species to become the new community dominant. We also ran analyses using a measure of community compositional change after a species removal. In particular, to quantify shifts in the abundances of the remaining species after a removal, we calculated the multidimensional Euclidean distance between abundances in the full community and those after a removal for all species but the one removed (Collins 2000; Collins et al. 2000): ED 

√兺(n

i, full

 ni,removal)2.

Results Before turning to patterns of stability (or instability), it is important to note that the distribution of abundances among species in our model communities was altered dramatically by changing the extent to which survival rates varied among species and through time. The abundance of species in the simulated communities ranged from highly skewed to very even (fig. 7.1). The range of mean survival rates was far more important in generating these differences than was the maximum temporal variability in sj and aij values (data not shown). As expected, the more variable survival was among species, the more highly skewed was the distribution of abundances among species. Many of the results regarding patterns of stability and model details likely follow from changes in species abundance patterns such as those depicted in figure 7.1. First, like almost all other diversity-stability models, we found the typical, asymptotically declining relationship between community instability (CVCB) and species number (fig. 7.2). In other words, according to our model, the removal of species usually results in more temporal variation in community biomass, with the destabilizing effects of extinction being more pronounced in species-poor communities. However, despite the overall decrease in mean CVCB with increasing species number across communities, removing a particular species from a particular community can generate a wide range of effects on the percentage change in community variability, PCVCB (fig. 7.3). For example, 43% of removals from 5-species communities and 31% from 10-species communities actually increased stability (negative values

150

chapter 7

Figure 7.1 Distributions of species abundance for two community rules. The ten different community rules resulted in species distributions ranging from (A) highly skewed (survival rates 0.0–1.0) to (B) quite even (all survival rates  0.5). Results are shown for communities in which S  5 and annual variability in species’ survival rates ranged up to 50% of their maximum.

of PCVCB). For most species, removal has little effect on community stability, but for a few species, removal greatly destabilizes the community (see fig. 7.3). These results are consistent across both 5- and 10-species communities; for convenience, we present further results only for 5-species communities.

species loss and community stability

151

Figure 7.2 Relationship between community stability and species richness ( of species, S). CVCB, the coefficient of variation in community biomass, is used as a measure of instability. Here, mean annual survivorships range from 0 to 1, with standard deviations up to 50% of maximum possible. Means  1 s.e. are plotted for 40 replicate simulations at each diversity level.

The primary determinant of a species’ effect on community stability, when removed, was its abundance (or dominance) prior to its removal (fig. 7.4). For the most part, the species with very high PCVCB were those that were highly abundant in the original community (see fig. 7.4). Roughly speaking, if a species constituted 50% or more of the original community’s biomass, its removal nearly always destabilized the community to some extent. To ask more rigorously how well the magnitude of these effects could be predicted, we ran a suite of general linear models (GLM), using PCVCB values as the response variable. When running these analyses, we always included the set of community rules as a categorical variable. As potential predictor variables, we focused on the attributes of species from intact communities (i.e., prior to any extinction) that might be readily available. For example, perhaps the easiest information to collect on a species is the mean and variability of its abundance. For our communities, mean abundance alone could predict about 31% of the variability in speciesremoval effects. (Including the coefficient of variation for abundance did not increase this model’s predictive power: table 7.1). Running GLMs with more detailed species information gave negligible gains in predictive power. For example, the mean and temporal variance in survival rates, the probabilities of taking over space from other species, and the probabilities of colonizing empty space together predict only 22% of the variation in PCVCB (see table 7.1). Finally, taking all this information together allowed prediction of about 44% of the variation in PCVCB. The message seems to be that abundance alone

152

chapter 7

Figure 7.3 Histograms of % change in instability (PCVCB) following individual species removals. A. Removals from all simulations of 5 species communities. B. Removals from all simulations of 10 species communities. Many removals produce little change in stability, whereas only a small number produce a large decrease in stability (large positive values).

species loss and community stability

153

Figure 7.4 Relationships between change in community stability (PCVCB) and the dominance of the removed species. All communities shown consist of 5 species. Dominance is calculated as the mean proportion of total community biomass prior to a species’ removal. For our communities, abundance and dominance are basically interchangeable, with correlation coefficients ranging from 0.990 to 0.999 for different community rules. Results are shown for communities constructed with 5 different ranges of species survival rates (all with standard deviations up to 50% of maximum possible) and for all 10 sets of simulations combined.

is a reasonably good predictor of the effect of a species’ extinction. On the other hand, that predictive power is so poor that one could never guarantee that rarer species, if they were to go extinct, would have minor effects on stability. Because it is natural to overlook species of lesser abundance, it is worth focusing on those species alone. We found that the removal of species with lower dominance values could either stabilize or de-

154

chapter 7

TABLE 7.1 Ability to predict the importance of species removals with prior information Different general linear models (GLM) were run on output from the five species community simulations. Either PCVCB or ED was used as the dependent variable in each GLM, while information on the characteristics of the species prior to removal were used as independent variables. In all GLMs, the rules used to formulate the community were included as a 10 level categorical variable. The percentage of variance explained (r2) is used to gauge explanatory power.

Model Mean Mean Mean Mean Mean aiE

abundance and CV of abundance and CV of survival, mean aij, and aiE abundance and mean survival and CV of abundance, survival, mean aij, and

Models Predicting PCVCB

Models Predicting ED

r2

r2

0.311 0.311 0.223 0.324

0.747 0.747 0.543 0.747

0.435

0.762

stabilize the community (see fig. 7.4). Clearly, it would be useful to be able to predict the effects of losing a rare species a priori. To address this issue, we selected the subset of species that contributed less than 30% of total community biomass (dominance ⬍ 0.3). We then used discriminant function analysis to examine how well we could classify species as stabilizing (defined as PCVCB ⬎ 40%) versus destabilizing (PCVCB ⬍ 30%), given different limitations on the availability of data. We found that several single variables provided equally poor predictions; abundance, survival, and colonization abilities (aij and aiE) each led to correct classification of 55% of cases. Of these, mean abundance is by far the easiest to estimate in the field. We also found that very little predictive power is gained from embellishing the model (table 7.2). For example, when we included six predictor attributes (mean and cv of survival, aij, and aiE), we correctly classified only 66% of the cases, in comparison with 55% of cases with but one attribute. Of course, stability is not the only thing one is interested in when examining the effects of species removals. An additional way of measuring change entails the Euclidean distance between communities before and after an extinction. In contrast to our findings on the effects of extinctions on stability, the effects of extinction in terms of altered community composition responded in a much more predict-

species loss and community stability

155

TABLE 7.2 Predicting the importance of rare species for community stability For the subset of species with dominance ⬍ 0.3, discriminant function analysis was used to classify species as strongly stabilizing (PCVCB ⬎ 40%) vs. strongly destabilizing (PCVCB ⬍ 30%). Rare species with intermediate effects on community stability were excluded from the analysis. Model Mean Mean Mean Mean Mean

abundance and CV of abundance and CV of survival, mean aij, and aiE abundance and mean survival and CV of abundance, survival, mean aij, and aiE

% Correctly Classified 54.7 49.5 67.4 56.8 66.3

able manner. In particular, we ran a second suite of GLMs, similar to those described above, to predict the change in community composition, measured as ED (see table 7.1). Mean abundance before removal was by far the most important factor predicting ED, and accounted for 75% of the variance in removal effects on community composition. Adding other information did little to increase predictive power, with the full model (that included means and variability in abundance and per capita rates) predicting only 76% of the variation in ED (see table 7.1). While ED and PCVCB showed a modest correlation (fig. 7.5; Pearson r  0.39), this was driven almost entirely by a very few, extremely abundant species that were important for both measures. The importance of abundance in predicting the community consequences of species removal seems to be due in part to its integration of the effects of survival rates and colonization ability. GLMs show that survival rates alone do a poor job of predicting a species’ abundance, whereas adding the mean probability of taking over space from other species and the probability of occupying empty space allows prediction of 62% of the variance in abundance (table 7.3). In particular, high abundance and the resulting strong effects of removing an abundant species depend on high survivorship (fig. 7.6A) and strong colonization abilities (fig. 7.6B). Finally, we asked whether we could predict which species would become dominant following the extinction of the original community dominant. To address this, we selected a subset of simulations in which the single most dominant species was removed, then used discriminant function analysis to classify the remaining species accord-

156

chapter 7

Figure 7.5 Relationship between the effects of species removals on community composition (ED) and on temporal variability in community biomass (PCVCB).

ing to whether they became the new community dominant. Mean abundance in the intact community was the single most important variable; 76% of species could be classified correctly on the basis of abundance alone. In fact, in 287 out of 400 cases, the species that was the second most abundant in the original community became the new dominant after the original dominant was removed. In contrast, knowledge of those variables that are more difficult to measure (e.g., survival, replacement rates and colonization abilities) together allowed for correct classification of only 70% of species.

TABLE 7.3 Individual rates explaining mean species abundance Different general linear models (GLM) were run on output from the 5 species community simulations. Mean abundance is the dependent variable in all analyses. In all GLMs, the rules used to formulate the community were included as a 10 level categorical variable. The percentage of variance explained (r2) is used to gauge explanatory power. Mean aij is the average amount of vacated spaced freed by other species that the removed species occupied, and aiE is the mean probability of colonized previously empty space. Model Mean Mean Mean Mean

survival and CV of survival of survival, mean aij, and aiE and CV of survival, mean aij, and aiE

r2 0.386 0.390 0.610 0.617

species loss and community stability

157

Figure 7.6 Relationship between mean abundance of species and (A) mean survival rate and (B) mean aij (the average amount of space vacated by other species that a species takes over).

Discussion Current rates of species extinction are alarming—both in and of themselves, and because it seems clear that at some point so many extinctions may accrue that natural ecosystems will become irreparably damaged. A growing body of theoretical work has focused on the question of how many extinctions are “too many”—or, put differently, how many species we can afford to lose before permanent damage is done. These efforts have focused on the relationship between

158

chapter 7

species richness and either ecosystem function or ecosystem stability (McCann 2000). Although these questions are of theoretical interest, their relevance to the real world is probably limited (Schwartz et al. 2000). In particular, it appears that in real communities each extinction does not result in an incremental increase (or even a nonlinear or “threshold”-type increase) in damage at the community and ecosystem levels. Rather, the effects of extinctions are highly species- and context-specific (Power et al. 1996b; Soul´e and Terborgh 1999). For example, some real-life extinctions have resulted in minimal consequences for communities and ecosystems (Simberloff this volume). On the other hand, it is now widely appreciated that certain keystone species exert effects on communities or ecosystems that are disproportionately large relative to the species’ abundance (Paine 1966, 1969a, 1974, 1988, 1992, Power et al. 1996b). Thus, in some cases, real communities are strongly altered by the loss of a single species, but in other cases not; and the magnitude of the effects appears to depend on the specific traits of the deleted species or on the nuances of how the deleted species interacted with others in the community, rather than on the changes in the number of species per se. Here, we have addressed how the community-level effects of single-species extinctions vary on the basis of some easily measured (and some not-so-easily measured) species traits. In our models, the removal of more abundant, or dominant, species exerts the strongest effects on community stability and composition. However, even for these simple, space-limited communities, moving much beyond this broad generalization is difficult. The power to predict the effects of extinction on community stability is depressingly low, given the data that are usually available. We find that even for a very simple set of model assumptions, and even knowing all the rules of the game, the importance of removing particular species is difficult to predict. This result parallels that of models for specific communities, in which the sensitivity of damping ratios or community composition to specific rates is not easily related to species abundances or other species traits (Tanner et al. 1994; Wootton 2001). However, we reiterate the caveat that our models have only considered competing species; models that include trophic diversity may suggest worse, or better, predictive power regarding species importance. More positively, some aspects of community dynamics were highly predictable for our models. For example, easily obtained information such as rank abundance prior to an extinction was very useful in predicting which species would “take over” after the extinction of a community dominant. Also encouraging was the finding that mean abundance is a better predictor of species importance than more diffi-

species loss and community stability

159

cult-to-collect per capita rates or variability in abundance. Thus, to the extent that prediction of removal effects can be gauged ahead of time, information that is most easily gathered may be as useful as much more labor-intensive information. As with any model, our results are driven by our assumptions. Although our results do not apply directly to any particular natural community, they do provide a clear set of hypotheses to test about the factors that direct the importance of species losses in competitively structured systems. Specifically, relative abundance alone is a reasonable surrogate measure for a wide variety of effects (e.g., changes in stability, shifts in community composition, and which species will take over for the lost taxa). However, there are notable exceptions to the overall abundance-importance correlation. When we turn to the real world, we find that although abundant species are a good place to look for large effects of extinction, several extremely abundant species have been lost with no dramatic effects recorded (Simberloff this volume). The challenge that this work suggests for conservation is how to prioritize species-management efforts when clear rankings of species by community importance are not easy to make a priori. We hope that our general modeling approach will encourage the development of models that can be linked more accurately to data from real communities. In particular, models that take per capita rates for different species as a starting point are clearer and more understandable than are less “mechanistic” models or more complex frameworks (Wootton 2001). In addition, there is a clear need to incorporate trophic interactions into stability-diversity models. Like most others pursuing community stability ideas, we have taken the easy road here, explicitly considering only competitors (but see McCann et al. 1998; Ives et al. 2000). Given the striking importance that R. T. Paine and his long line of students and collaborators have shown for trophic interactions in determining average community structure, better modeling of such effects for community stability is clearly needed. Hopefully, such models can be developed hand in hand with manipulative field experiments to carefully test and expand assertions of community stability patterns in real-world communities. Such testing is needed both for the science of ecology and for clarification on how best to pursue pressing conservation goals.

Acknowledgments We thank The Last Paine Lab for conceiving and organizing the event that prompted our contribution. Conversations with Bill Morris and

160

chapter 7

Greg Dwyer helped clarify many of the issues discussed, and comments by Peter Kareiva and five anonymous reviewers greatly improved the presentation of this work. Support for Doak was provided by NSF DEB-9806722. Bob Paine’s tireless enthusiasm and prodding provided the main impetus for our efforts to enter the confusing world of community ecology; we give him our sincere thanks.

Chapter 8

m One Fish, Two Fish, Old Fish, New Fish: Which Invasions Matter? Jennifer L. Ruesink

The impacts of introduced species, both economic and ecological, grow on an almost daily basis as new species invade and new calculations are made (Simberloff 1981; Moyle et al. 1986; Humphries et al. 1992; OTA 1993; Cox 1999; Pimentel et al. 2000). Although the impacts of invasions capture public attention, most ecological theory focuses on the establishment of invaders, not their effects (Crawley 1986; Rejm´anek and Richardson 1996; Williamson and Fitter 1996a, b; Veltman et al. 1996; Crawley et al. 1997; Tilman 1997a; Goodwin et al. 1999; Shea and Chesson 2002; a notable exception is Parker et al. 1999). Understanding the factors that enable invaders to have high impacts has obvious value for targeting the prevention and control of invasions. In addition, studies of invasions may reveal how ecological systems respond to changes in food web structure. In this regard, analyses of the impact of invasion examine the question of “are species important?” from a different perspective. Instead of describing when

162

chapter 8

species can be lost without effect, impacts of invasions define the effects of additional species. Because invasions represent dramatic natural experiments in which a “reference state” prior to the invasion can be compared to a “treatment” afterward, they fit the classic framework for identifying how strongly species interact (Paine 1966, 1994). According to current theory, the course of an invasion emerges from the interaction of species’ traits with traits of the new environment (Simberloff 1986; Lodge 1993a; Ruesink et al. 1995; Parker et al. 1999). Environmental traits may be abiotic (e.g., temperature, disturbance, nutrient supply; Orians 1986) or biotic (e.g., food availability, natural enemies; Elton 1958; Goeden and Louda 1976; Crawley 1987; Mack 1996b; Chapin et al. 1998). An early step in the course of an invasion is establishment, which involves a critical anthropogenic component because humans often control the opportunity for invasion. In taxa as disparate as insects, birds, and terrestrial plants, establishment rises with the number of times or number of individuals introduced (Beirne 1975; Duncan 1997; Green 1997; Goodwin et al. 1999; Lockwood 1999). Thus, there are roles for species’ traits, recipient location, and human involvement in invasion. The contribution of these factors in affecting establishment has been explored statistically for cases in which both successful and failed invasions are known (Rejmanek and Richardson 1996; Veltman et al. 1996; Duncan 1997; Green 1997; Sorci et al. 1998). Except for the role of opportunity, few generalities have emerged concerning the types of species likely to invade or places likely to be invaded. In a recent review, Levine and D’Antonio (1999) found that most traits of successful versus unsuccessful invaders were context specific. Relative to establishment, there have been few general reviews of the factors that influence the effects of invaders (unless “weediness” is considered an impact; see Crawley et al. 1997; Daehler 1998). It remains largely unknown, therefore, how species’ traits, the biotic and abiotic environment at the site of invasion, and human involvement in moving species across biogeographic boundaries are related to impact. This inequality between analyses of establishment and impact in part reflects the fact that less is known about what an invader does than about where it is. Ecological effects are less well documented than are identities, vectors, and numbers of invaders (Levine and D’Antonio 1999; Parker et al. 1999). The effects of introduced fishes have been collated in FishBase, an international database of finfish biology (Froese and Pauly 1998). This database records cases of fishes becoming established in countries where they are not native. For many of these cases, information is also available on the size and diet of the species, country of origin,

w h i c h i n va s i o n s m a t t e r ?

163

year of first introduction, reason for introduction, and native fish in the invaded country. Reports of impacts of introduced fishes in this database are incomplete and are often based on expert opinion rather than published data, in contrast to the case studies presented elsewhere in this section of the book (Wonham this volume). Nevertheless, this information makes it possible to ask which factors are related to the impacts of freshwater fishes at the scale of countries. Here I use predictor variables of species’ traits (size, diet), abiotic environment (difference in latitude), biotic environment (species richness and endemism), and human involvement (when and why a species was introduced) to identify the factors that may produce high-impact invasions. General analyses of this sort are important both to address the pragmatic question of how to reduce high-impact invasions and because they could suggest theoretical hypotheses pertinent to what makes a species ecologically important or not.

Methods The fish introductions examined here were compiled initially by Welcomme (1988) and more recently incorporated into FishBase as one of more than 60 tables cataloging finfish species and their biology (Froese and Pauly 1998). Information on fish introductions has also been entered from published literature and from surveys of fisheries managers worldwide. The table of introductions includes fields for year of introduction, source and recipient countries, reason for introduction, whether the species established, and whether ecological effects were evident after establishment. Only first introductions are listed, and aquarium species are not considered introduced unless established. In the 1998 version of FishBase, the table of introductions contained 2751 cases, involving 443 species moved to 191 countries or regions. This analysis focused on freshwater invasions because the bulk of reported introductions (363 of 443) are of freshwater species. Through queries in Microsoft Access, these cases can be linked to other tables in FishBase, including species biology and species lists by country. A portion of the database can be viewed at www.fishbase.org. The compilation is certainly not exhaustive. A U.S. database, for example, records 75 established foreign fishes (Fuller et al. 1999), whereas FishBase records only 71 and does not include transfers within the continental United States. Nevertheless, the records of introduced fishes are comparable in number to plant and bird introductions worldwide (8000 weeds, Daehler 1998; 396 birds, Lockwood 1999) and cer-

164

chapter 8

tainly make FishBase the best source for information on fishes at a global scale.

Defining “High-Impact” Species As reported in FishBase, the ecological effects of invaders are essentially binary: yes (probably yes) or no (probably no), with a large proportion unreported or unknown. The impacts include changes in food availability, habitat structure, nutrient dynamics, or top-down trophodynamics. The ecological effects of invasive fishes are undoubtedly underreported, particularly for older introductions; no effect can be reported if no observations are made. Fishes may also be well established before researchers examine a site, so no baseline is available for comparison. It may be difficult to distinguish the impacts of particular species in highly invaded systems. In contrast to these false negatives, a fish might truly have no detectable effect if it remains at low abundance where introduced or replaces the functions of other species (redundant). For most analyses, I assumed that ecological effects would have been reported if they were obvious, so “unknown” cases were included as having no ecological effect. However, when possible, I also performed parallel analyses with three dependent variable categories: effect, no effect, and no data. I defined high-impact species as those species regularly reported to have ecological effects across all countries where they established—a category that does not incorporate any information about the type or magnitude of the effect.

Examining Specific Factors For the analyses of patterns of impact, tables in FishBase were adjusted as follows. From the table of introductions (Intrcase), which included 2751 records, I extracted freshwater species and removed duplicate records, then selected cases of established fishes (1408 cases). The year of first introduction was known within 25 years for 1055 cases, and most of these were pinpointed more accurately. The country of introduction was reported for all cases, and the originating country for 865. As an index of climatic match between origin and introduction site, I used the absolute difference in latitude (not distinguishing north vs. south) of countries’ capitals in the Countref table. Because of the inaccuracy of capitals as indicators of introduction sites, I divided transfers into three categories: ⬍10⬚ latitude, 10–30⬚ latitude, and ⬎30⬚ latitude. The Countref table also provided informa-

w h i c h i n va s i o n s m a t t e r ?

165

tion on native fish faunas. For an index of native fish diversity, I used the residual between each country’s richness and a best-fit speciesarea curve for all countries (ln(Richness) ⳱ ⳮ0.13096 Ⳮ 0.356973 ln(Area in km2); n ⳱ 178; F1,176 ⳱ 169, P⬍⬍ 0.001) (e.g., MacArthur and Wilson 1967). To calculate endemism, I divided the number of endemic species in a country by its total number of native species (including endemics). At least 17 reasons were given for introducing species, which I combined into four. Two categories include species established as byproducts of travel or trade (Moyle 1999): (1) imported and accidentally released and (2) accidentally introduced. One category includes species introduced because the fish itself was desirable (e.g., for a fishery or angling). The final category includes species introduced for their effects (e.g., pest control or providing forage for species of higher trophic levels). The reasons for introduction were unknown for 15% of established fishes. Maximum size (Species table) was available for all 254 established freshwater species, although units differed slightly (e.g., total length vs. standard length). Feeding information was available for only 110 species of established invaders, but I extrapolated feeding for an additional 45 when several congeners shared a particular set of resources (herbivore, carnivore, omnivore).

Data Analysis I first examined whether particular fish species tend to have similar effects (importance) wherever they establish. In other words, do particular species cause ecological change either more or less often than the average across all established cases? To do this, for each species I asked whether the proportion of cases with impacts differed from expected based on the overall reports of ecological effects. Species that fell above the 95% confidence limits for the binomial expectation were considered high-impact species. This confidence band shrank as species established in more places. I also examined whether higher taxonomic clades differentially contained high-impact species by comparing ecological effects among the 14 families with at least 4 established species (ANOVA with family as a factor and species as samples, weighted by the number of established cases per species). Subsequent analyses explored whether the impacts were related to species’ traits (size, diet), the recipient location (latitude difference, endemism, richness), or introduction characteristics (year, reason). The common method for exploring relationships among various factors and invasion involves multiple regression (Veltman et al. 1996; Wolf et al. 1996). Logistic regression is used when the outcome is

w h i c h i n va s i o n s m a t t e r ?

167

binary (established vs. not) as opposed to continuous (proportion established or length of time persisting). All of these above variables were available for 690 cases of established freshwater fishes, and I used these cases in multiple logistic regression to determine which factors were related to invasions with noted ecological effects. In addition, I examined separately the relationship between impact and each factor, using all cases with data for that factor. This procedure was a “check” of the multiple regression analysis to determine if similar patterns arise when different cases are examined. In other statistical analyses of invasion, samples have been transformed to account for phylogenetic nonindependence. The problem occurs because species are not independent samples but instead differentially share evolutionary history and therefore an entire suite of characteristics (Veltman et al. 1996; Wolf et al. 1996). In general, I did not correct for phylogenetic nonindependence because most of the variables in the analysis were not linked to species but to other factors in the invasion process (recipient environment, human involvement), and these cannot be represented by a common phylogenetic tree. In cases where phylogenetic nonindependence might be particularly relevant (e.g., body size), I investigated whether patterns were strongly influenced by relatedness. Phylogenetic relationships are well established at the order level for teleost fishes (Helfman et al. 1997). In addition, species-level trees have been published for a few taxa (e.g., Salmoniformes). From Helfman et al. (1997) and Nelson (1994), I developed the tree presented in figure 8.1, which shows that, where information was lacking, divergences were assumed to be polytomies. The timing of divergence events is not well known, so I performed numerous phylogenetically independent contrasts using a range of divergence times: orders (100–200 million years ago), families (70 million years ago), genera (20–50 million years ago), and species (1–10 million years ago). I used equations developed by Felsenstein (1985) to transform raw data into phylogenetically independent contrasts. Raw data were fish size (independent) and proportion of cases with ecological effects for each fish species, weighted by the number of establishments (depen-

Figure 8.1 Phylogenetic relationships among fishes. Data were used to build phylogenetically independent contrasts for regression of invasion impact on fish size. In parentheses are the number of species and the number of cases within each taxon (only the number of cases for salmon species).

168

chapter 8

dent). This dependent variable was continuous and available for all 254 established freshwater fish species.

Results and Discussion The 1408 cases of established freshwater fishes in FishBase included 254 species and 161 countries or regions. On a case-by-case basis, 313 (22.2%) had ecological effects whereas 69 (4.9%) had no impact. The remainder had unknown or unreported effects. On a species-byspecies basis, about one third had ecological effects in at least one location. This rate corresponds well with previous reports at more restricted spatial scales. For example, 30% of 111 fishes that are established invaders in the United States cause harm, although less than 4% of these invasive species cause severe harm (e.g., economic pests, disrupt ecological systems, threaten indigenous species; OTA 1993). In California, as many as 50 to 95% of introduced fishes have ecological effects (Herbold and Moyle 1986). In general, fishes appear to exceed the “tens rule” (Williamson and Fitter 1996a, b), since substantially more than 1 in 10 established species reportedly have an impact.

High-Impact Species Given the high overall rates of impact, species showed little variation in their likelihood of causing ecological change. Indeed, few species fell outside the 95% confidence limits around an overall probability of impact of 0.22 (binomial expectation, fig. 8.2; Lockwood 1999). The exceptions were Oreochromis niloticus (Nile tilapia), Ameiurus melas (black bullhead), and Pseudorasbora parva (stone moroko), which had higher than expected rates of impact; and Sander lucioperca (zander), which fell below. Because the confidence limits are large for the expected impacts of species introduced only a few times, it is worth noting that an additional 11 species established at least three times and produced effects in most places (proportion ⬎ 0.5, see fig. 8.2). Several species have been introduced so many times that even though their rate of reported impact is within the expected range, they have produced ecological effects in numerous places (see fig. 8.2). These species include Cyprinus carpio (common carp), Ctenopharyngodon idella (grass carp), Oreochromis mossambicus (tilapia), Oncorhynchus mykiss (rainbow trout), Gambusia affinis (mosquitofish), Micropterus salmoides (largemouth bass), and Carassius auratus (goldfish). In sum, based on the high rates or numbers of reported ecological effects, a few invaders in FishBase can be considered high-impact spe-

w h i c h i n va s i o n s m a t t e r ?

169

Figure 8.2 Impacts of established freshwater fishes by species. Each point represents a single fish species that is established in a specific number of countries (or regions). Impact is portrayed in terms of the number of places where ecological effects are reported, relative to the total number of places established (unknown effect ⳱ no effect). Lines represent 95% confidence limits for a binomial distribution around a probability of impact of 0.22, given different numbers of establishment events. Letters refer to fish species as presented in table 8.1.

cies (table 8.1). Many of the species on this list have been reported independently as problem invaders. For instance, Nile perch severely reduced a large proportion of endemic cichlid species in African lakes after its introduction to develop a fishery (Kitchell et al. 1997). Peacock bass altered the food web of Lake Gatun, Panama (Zaret and Paine 1973), although this species reportedly has no ecological effect and a positive socioeconomic effect in Florida’s canals (Shafland 1999). Ruffe and round goby have been introduced widely, largely unintentionally in ballast water, and reputedly eat and compete with fry of native fishes (Ricciardi and Rasmussen 1998; Wonham et al. 2000). North American ictalurids have been denied entry to suitable habitat in the southern hemisphere, in particular New Zealand, which harbors a highly endemic fish fauna (Townsend and Winterbourn 1992). The African walking catfish is on the list of prohibited species of several U.S. states (OTA 1993, p. 213). Interestingly, another catfish, Clarius gariepinus, appears to be a high-impact species yet was itself nega-

E North America Ponto-Caspian E North America SE Asia North Africa

East Africa Tropical South America East Africa

Centrarchidae Percidae Ictaluridae Clariidae Clariidae

Cichlidae Cichlidae Cichlidae

North Africa

Centropomidae

A. Lates niloticus (Nile perch) B. Micropterus salmoides (largemouth bass) C. Gymnocephalus cernuus (ruffe) D. Ameiurus melas (black bullhead) E. Clarias batrachus (walking catfish) F. Clarias gariepinus (North African catfish) G. Astatoreochromis alluaudi H. Cichla ocellaris (peacock cichlid) I. Oreochromis leucostictus

Origin

Family

Specie

I

I

I

I

I

I

U

I

I

Reason

2/3

3/6

3/3

4/7

4/6

8/17

3/3

13/46

4/5

Noted effect

0/3

0/6

0/3

1/7

0/6

0/17

0/3

1/46

0/5

No effect

23/SL

41/SL

19/SL

150/SL

40/SL

62/TL

25/TL

97/TL

200/TL

Size (cm)

Omnivore (plants)

Carnivore (nekton)

Carnivore (snails)

Carnivore (nekton)

Carnivore (zooben.)

Carnivore (zoobenthos) Carnivore (zooben.)

Carnivore (nekton)

Carnivore (nekton)

Diet

TABLE 8.1 Life history information for high-impact fish invaders Data from Fishbase (Froese and Pauly 1998; www.fishbase.org). These 21 species have been reported to have ecological effects in more than 50%, or in more than the expected (binomial) proportion, or in more than 8 of the countries where they have established. The table includes scientific and common name of species, family, region of origin, whether introduction has primarily been intentional (I) or unintentional (U), number of reports of ecological or no ecological effect, size (TL ⳱ total length, SL ⳱ standard length), and trophic mode.

J. Oreochromis mossambicus (Mozambique tilapia) K. Oreochromis niloticus (Nile tilapia) L. Neogobius melanostomus (round goby) M. Gambusia affinis (mosquitofish) N. Odontesthes bonariensis (pejerrey) O. Oncorhynchus tshawytscha (chinook salmon) P. Oncorhynchus mykiss (rainbow trout) Q. Carassius auratus (goldfish) R. Ctenopharyngodon idella (grass carp) S. Cyprinus carpio carpio (common carp) T. Pseudorasbora parva (stone moroko) U. Limnothrissa miodon (Lake Tanganyika sardine)

Southern Africa

North Africa Ponto-Caspian

E North and Central America South America W North America

W North America East Asia NE Asia Eurasia NE Asia East Africa

Cichlidae

Cichlidae Gobiidae

Poecilidae Atherinidae Salmonidae

Salmonidae Cyprinidae Cyprinidae Cyprinidae Cyprinidae Clupeidae

I

U

I

I

I

I

I

I

I

U

I

I

4/5

9/19

26/89

9/36

9/48

16/66

3/4

3/5

11/54

2/4

17/47

20/74

0/5

0/19

2/89

4/36

2/48

4/66

0/4

0/5

0/54

0/4

4/47

3/74

17/SL

11/TL

120/SL

150/TL

45/TL

120/TL

147/TL

23/SL

4/TL

25/TL

60/SL

39/SL

Carnivore (zooplankton)

Omnivore (zoobenthos) Carnivore (nekton)

Carnivore (zoobenthos) Omnivore (zoobenthos) Herbivore (plants)

Carnivore (zooplankton) Omnivore (zooplankton) Carnivore (nekton)

Herbivore (plants)

Herbivore (omnivore)

172

chapter 8

tively affected in its native range by the arrival of Nile perch (Goudswaard and Witte 1997). Several tilapias fit the high-impact criteria, although a recent review of tilapia introductions drew no firm conclusions about their role under novel conditions (Pullin et al. 1997). Although carps, catfish, salmonids, and tilapias figure prominently among high-impact species (see Table 8.1), the rates at which fish species were reported to have ecological effects did not vary among families (F13,184 ⳱ 1.5, P ⳱ 0.12; fig. 8.3). Essentially, many invaders that are considered to be a problem have been introduced many times, and impacts are reported often but not disproportionately.

Few Factors Predict Invasion Impact In multiple logistic regression (N ⳱ 690 cases), only two factors showed significant relationships with impact (table 8.2). Identical factors emerged as significant for binary (yes, no effect) and trinary (yes, no effect, no data) analyses. The likelihood of observing an ecological effect from an established fish increased with endemism of the invaded country and the number of years elapsed since the introduction. However, the regression model explained little of the variation in invasion impact, and different factors emerged as important when different subsets of cases were examined, as explained in the following paragraphs. Traits of exotic fishes (size, diet) did not help distinguish low-impact from high-impact invasions in multiple regression analysis (see table 8.2). In simple regression, size was not related to the probability of ecological effect whether or not phylogenetically independent contrasts were used (F1,253 ⬍ 0.1, P ⬎ 0.8 in all cases). Impact may be predicted poorly by life history traits because the expected relationship between each trait and impact can be argued both ways. For example, impact might be related negatively to size if small fish reach high density by reproducing early, but related positively if large fish reproduce prolifically. Similarly, impact might be highest from omnivores if effects are related to the number of direct trophic interactions, or it might be highest from carnivores if effects are primarily indirect via trophic cascades. Certainly, the introduction of predatory fishes, particularly to fishless lakes, has often resulted in ecosystem-level change (Brooks and Dodson 1965; Carpenter and Kitchell 1993; McIntosh and Townsend 1996). Previous summaries of fish invasions have attributed introduction success to a suite of traits, although these traits have not been examined quantitatively. The establishment and spread of fishes in the southeastern United States are thought to be

w h i c h i n va s i o n s m a t t e r ?

173

Figure 8.3 Impacts of established freshwater fishes by family. Means (Ⳳ S.E.) are presented for the proportion of invading fishes producing ecological effects in each fish family with more than 3 established species. Calculations were performed on arcsine-square root–transformed data, weighted by the number of cases per species.

TABLE 8.2 Multiple logistic regression of the invasion impact on year of first introduction and endemism of fishes in the country where established The following factors were removed from the model due to non-significance: fish size, diet, species richness by country, difference in latitude between origin and recipient countries, and reason for introduction. r2 ⳱ 0.01, N ⳱ 690. For entire model, Chi Square ⳱ 8.42, df ⳱ 2, P ⳱ 0.015.

Intercept Year Endemism

Estimate

SE

df

Chi Square

P

ⳮ4.88 0.0031 ⳮ0.890

2.71 0.0014 0.487

1 1

3.23 4.79 3.34

0.072 0.029 0.068

174

chapter 8

promoted by parental care, generalized diet, ability to breathe air, tolerance of wide ranges of temperature and water quality, and mediumto-large body size (Nico and Fuller 1999). More generally in the United States, successful fish invaders tend to be hardy, aggressive, ecologically novel, effective at reproducing and dispersing and prized by humans (Moyle 1986). Because of the scale of the analysis, it was difficult to examine the role of the abiotic environment in allowing fish invaders to have ecological effects. With regard to establishment, for instance, hydrological alteration at the level of the watershed, not country-level factors, appears to be important (Moyle and Light 1996). I examined the match between native and new environments in terms of the difference in latitude that a fish was moved. However, because this difference in latitude referred to the countries’ capitals, it was quite inaccurate, and few conclusions could be drawn from the fact that it was insignificant in multiple regression. Ecological effects rose with endemism in multiple regression (see table 8.2). Separately, I looked at the relationship between endemism and effect at the level of countries. The proportion of fishes with effects increased slightly across countries with increasing endemism (regression on arcsine-square root–transformed data, weighted by the number of established species: F ⳱ 4.2, P ⳱ 0.04, N ⳱ 159, r2 ⳱ 0.02; fig. 8.4A). The pattern is particularly striking for countries with at least 20 established fish invaders (fig. 8.4B). Endemism is a hallmark of isolation, and invaders may have larger impacts where the native fauna is distinct and insular (Loope and Mueller-Dombois 1989; Simberloff 1995; Casal et al. 1999). New fish species might play wholly new roles, and native species might have little evolutionary history with such species. For instance, native species might not be able to defend against or control an exotic species. In contrast to endemism, the diversity of the recipient country was not related to species’ impacts. Verbal models by some ecologists (Elton 1958; Chapin et al. 1998; but see Moyle and Light 1996) lead to predictions that diversity reduces the impacts of invasion. However, diverse plant assemblages harbor many (not necessarily high-impact) invasives (Stohlgren et al. 1999). Certainly, fish invaders have had substantial ecological impacts even in freshwater systems that are large (e.g., Great Lakes, Mills et al. 1994; Simberloff this volume) and diverse (e.g., African rift lakes, Kitchell et al. 1997). In the multiple regression model, earlier introductions were more likely to have impacts than recent introductions. This temporal pattern is consistent with a lag in ecological impact, particularly because there is likely to be a modern bias for detecting change with respect

w h i c h i n va s i o n s m a t t e r ?

175

Figure 8.4 Freshwater fish endemism and effects of invaders across countries. Each point represents a country (or region). A. All countries. B. Countries with at least 20 established invasive fishes. The proportion endemic is relative to total number of native species (endemic Ⳮ native). The proportion with effects is relative to the total number of established invaders (unknown effect ⳱ no effect). Letters refer to locations with particularly interesting values for endemism or effect: a, Malawi; b, Hungary; c, South Africa; d, Philippines; e, Italy; f, Hawai’i; g, Madagascar; h, India; i, United States; j, Mexico; k, Cuba; l, Australia; m, Chile; n, New Zealand.

176

chapter 8

Figure 8.5 Cumulative number of established freshwater fishes and those with noted ecological effects. The year of introduction was unknown for 353 cases (36 of which had effects); data for these were not included in this graph.

to recent baselines. However, the pattern disappeared when all possible cases were used to examine how reported impacts vary with the year of first introduction (logistic regression F ⳱ 1.5, P ⳱ 0.22, N ⳱ 1055). Because the cumulative number of established species has increased dramatically in the past half century, reported impacts have accelerated as well, but not disproportionately (see fig. 8.5). Conversely, the reason for introduction had no significant relationship with ecological effect in multiple regression, but it became important when all available data were used (N ⳱ 1190, Chi-square ⳱ 21.4, P ⬍ 0.001). Of the established species brought in for their effects (e.g., to provide forage, control pests), 35% of cases (21 of 60) had impacts. Of the established species released for themselves (e.g., for fisheries), 34% of cases (85 of 249) had impacts. In contrast, only 22% of other established cases with a known reason for introduction (881 imports and by-products) had ecological effects. In sum, fishes that were introduced directly and intentionally to a new area were 1.5 times more likely to have ecological effects reported than were those that entered by accident or escaped after import. This trend is different from reports of equal rates of ecological effects for intentionally and unintentionally introduced fishes in the United States (OTA 1993). One explanation is that fisheries managers are more likely to observe or study systems in which fish are purposefully released than in a

w h i c h i n va s i o n s m a t t e r ?

177

place where the invasion is a surprise. Another possibility is that humans tend to introduce particularly fast-growing, voracious species on purpose, and this selective transport contributes to the ecological effects from intentional introductions.

Summary A theme of this part of the book is that species additions can serve as a counterpart to extirpations for understanding the importance of species (Wonham this volume). Dramatic ecological change after particular introductions (and extinctions; Simberloff this volume) affirms that single species can be important (Simberloff 1991, 1999). Nevertheless, two thirds of introduced fish species were never reported to have ecological effects. Several explanations can be developed to account for these cases. First, fish may remain at low abundance or be redundant with species already in the system. In such circumstances, the ecological effects of invasion are too small to measure. The second possibility involves concomitant change: hydrological alteration facilitates invasion and also changes habitat for native species (Moyle and Light 1996), both of which are major causes of species endangerment (Foin et al. 1998; Wilcove et al. 1998; Froese and Torres 1999). Another possibility is that a species that causes an ecological effect is not studied in terms of that particular variable at the time of the invasion. Later, baseline conditions may be difficult to reconstruct. Such a scenario would imply that earlier impacts of invasions should be underreported, yet the proportion of fish introductions reported to have ecological effects stayed constant or perhaps even rose with the time since introduction. Fishes introduced more times were more likely to show mixed impacts rather than an all-or-none pattern (see fig. 8.2; Moulton and Sanderson 1997). Fully 23 species (and two-thirds of species established in at least 10 countries) were reported to have ecological effects in one country but no effect in another. These disparate results fit well with a developing theory of context dependence (Berlow 1997; Ruesink 1998), in which species can be important in some places but not others. For species established in more than 10 countries, all but one had effects at least once. Therefore, fish introductions imply that all species may be important in some contexts if distributed sufficiently widely. On the other hand, because of their mobility, and behavioral and trophic flexibility, fishes as a whole may constitute relatively risky introductions. Only one place-based factor (endemism) showed a weak and con-

178

chapter 8

sistent relationship with invasion impact in multiple regression, possibly because exotic species can play novel roles in isolated environments. Species size and trophic mode did not help predict impact, nor did latitudinal distance of the transfer or fish species richness in the recipient country. Species introduced on purpose might be more likely to change the recipient community than by-product introductions or imports. Are there really so few factors in common among ecologically important introductions? It is probably too soon to tell for freshwater fishes, since the database is incomplete and at an inappropriate scale for watershed-level phenomena. Current reports do not provide a predictive model of when introduced fishes will have ecological effects. In the future, these sorts of database analyses should become increasingly useful and interesting as information accumulates on the magnitudes of invasion impacts. In the short term, however, introduction decisions must be based on how closely a particular case matches the conditions in which a species previously proved damaging (Townsend and Winterbourn 1992). Such careful, controlled studies of species interactions in particular locations are part of the rich legacy of R. T. Paine.

Acknowledgments This chapter was inspired through conversations with D. Pauly and was greatly improved by P. Kareiva, D. Simberloff, and three anonymous reviewers. Ted Pietsch was instrumental in constructing the phylogeny of teleosts. Much of the conceptual development was supported by the Centre for Biodiversity Research, University of British Columbia.

Chapter 9

m Ecological Gambling: Expendable Extinctions Versus Acceptable Invasions Marjorie J. Wonham “The known patterns [of introductions] are fascinating since they reveal an ecological sword of Damocles: under certain circumstances the payoff can be substantial; under others a heavy price is exacted for poor ecological judgment.” Paine and Zaret 1975

Species extinctions and invasions are increasingly altering biological communities, generating growing concern about their ecological effects. The central question of this book is how we determine which species removals are most harmful. The complementary question—how we determine which species additions are most acceptable—presents an instructive contrast. Species removals and additions are natural processes, and under-

180

chapter 9

standing their dynamics and consequences has long kept ecologists busy (e.g., Darwin 1859; Clements 1936; Elton 1958; MacArthur and Wilson 1967; Paine 1966, 1969a; Connell and Slatyer 1977; Pimm 1991; Tilman et al. 1997). Today, however, this natural ebb and flow of species is being swamped by a tidal wave of human-induced extinctions and invasions, estimated to be orders of magnitude greater than natural background rates (OTA 1993; NRC 1995). These two agents of global change present compelling and complementary challenges to environmental protection. Biologically, extinctions and invasions share important features. Both involve dramatic changes in the density of a species: from present to absent (extinction) and vice versa (invasion). Both affect the entire web of additional species with which those species interact. For both, critical population dynamics at small population sizes can determine risk (extinction) or success (invasion). Conversely, management practices at this stage can effectively rescue (extinction) or eradicate (invasion) the species. Our legislative response to these two threats is telling. Generally speaking, we are loath to allow native species to go extinct. This ethic of zero expendability is embodied in the powerful wording of the U.S. Endangered Species Act (ESA). In contrast, we are far more likely to allow a non-native species to invade. Our default position is to accept most introductions until they are proven to be harmful, and we have no comprehensive federal legislation to set the tone for governing introductions. Instead, invasions are managed in piecemeal fashion, and many slip through the regulatory cracks. The opportunity is now at hand to develop a comprehensive listing and management process for introduced species. The National Invasive Species Act is soon due for reauthorization, and a recent Presidential Executive Order (13112, 2/1999) mandated the development of a National Invasive Species Management Plan. Given the similarities in the biology of extinctions and invasions, what lessons can we draw from three decades of active ESA implementation to help us develop effective, conservation-oriented legislation and policy on biological invasions? To answer this question, I first compare the nature and scale of extinctions and introductions in the United States, and contrast their federal legislative histories. I then illustrate how, under the current system, introduced species can be allowed to alter an ecosystem dramatically yet remain unregulated at the federal level. To make this discussion concrete, I focus on a case study of two introduced marine plants in the U.S. Pacific Northwest. Finally, drawing on lessons learned from ESA implementation about listing and managing species, I suggest key ingredients for a complementary legislative framework for

ecological gambling

181

invaders. Throughout the chapter, the terms introduced, invader, invasive, invasion, and non-native are used to refer only to species that have been transported by human activities beyond their native ranges.

Extinctions vs. Invasions Biology and Management The removal or addition of a species can have profound ecological ramifications. At their simplest, removals and additions are parallel phenomena: each extinction means the loss of a species and its interactions; each invasion means the arrival of a new player and the development of novel interactions. As they progress, both extinctions and invasions may alter ecological patterns and processes from the genetic to the ecosystem level (Elton 1958; Paine 1966; Zaret and Paine 1973; Paine and Zaret 1975; Duran and Castilla 1989; Lodge 1993b; Wootton et al. 1996; McClanahan et al. 1999; Parker et al. 1999; Simberloff 2001). The stylized population trajectories of an extinction and an invasion may be viewed as mirror images (fig. 9.1). An endangered species declines from a historic baseline population density towards zero. Extinction, by definition, is the end point of the decline after which the population cannot recover (see fig. 9.1a). A successful invasion shows the reverse pattern. Beginning with the inoculation of a small number of individuals, the population increases in size to a level that persists (fig. 9.1b). These trajectories are presented as reflections of each other, but they differ importantly in their boundary conditions: when population density is near zero, an extinction is nearly complete, whereas an invasion is only just beginning. Not only are extinctions and invasions inverse processes, but they also appear dynamically linked to each other in some terrestrial systems (e.g., Wilcove et al. 1998). Although the nature of their connection may differ somewhat in marine systems (Carlton et al. 1999; Carlton 2001; Bax et al. 2001), their basic population dynamics remain comparable. The inverse population trajectories of extinctions and invasions dictate certain differences and similarities in management options. The greatest difference occurs where a species is absent—after extinction or before invasion. After extinction, conservation options are nil (barring the unsatisfactory futuristic scenario of cloning and reintroduction). Prior to invasion, however, the option to prevent introductions is a powerful management tool. At the other ends of the trajectories, populations at high density are the easiest to protect from

182

chapter 9

Figure 9.1 Stylized population trajectories of (A) species extinction and (B) species invasion as mirror images through time (adapted in part from Hobbs and Humphries 1995). The difficulty of rescuing or eradicating a species also generally increases over time. Thus, an extinction is easiest to prevent while the population is still at high density and may become increasingly difficult to prevent—for logistic, population dynamic, and genetic reasons—as the population declines. Once extinct, the species cannot be restored (barring the development of cloning and re-creation techniques). A successful invasion begins with an inoculation event, followed by population growth and spread. Prevention may be easiest before inoculation; once introduced, the population may become harder to eradicate as it grows and establishes.

extinction (see fig. 9.1a) but can be the hardest to eradicate (depending somewhat on the case; Simberloff 2001) (see fig. 9.1b). In the middle stage, the challenges of detecting and managing small or rapidly changing populations are shared by extinctions and invasions. It is in this region that nascent legislation governing introduced species can most benefit from the lessons of the ESA.

Ecological and Economic Scale The scale of extinctions and invasions in the United States alone is impressive. The Fish and Wildlife Service (FWS) currently lists more

ecological gambling

183

than 1500 species as federally endangered, threatened, candidate, or proposed. Introduced species are not listed in a comparable way, but estimates suggest at least 6000 to 7000 established invaders are found in the country today (OTA 1993; W. Gregg, United States Geological Survey pers. comm.). The numbers of both continue to increase. In 1990, the direct cost of recovering all listed and candidate species in the United States was estimated at $4.6 billion over 10 years (Dwyer et al. 1995). At the same time, the cost of controlling one single invader, the zebra mussel (Dreissena polymorpha), was estimated at approximately $1 billion over 10 years (Carlton 2001). On an annual basis, the total estimated economic costs of introduced species in the United States run to over $100 billion (OTA 1993; Pimentel et al. 1999). Not surprisingly, given the ecological and economic ramifications of extinctions and invasions the scientific and legal interest in these topics is considerable. A survey of the biological literature of the last 10 years shows that roughly equivalent numbers of papers have been published on endangered and introduced species (fig. 9.2a). Within this literature, papers treating the legislation of these species are fewer and show a rather different pattern: approximately 100 times more papers discuss endangered species legislation than introduced species legislation (see fig. 9.2b). In the legal review literature, the citation pattern is similar, with a far larger discussion of endangered species legislation (fig. 9.2c). There is a striking disparity between extinctions and invasions in the volume of legislation-focused literature. Much of the extinction literature consists of evaluations of the ESA and suggestions for its refinement. I suggest that lessons from this extensive discussion can be brought to bear on the development of complementary legislation governing introductions. To illustrate the need for such legislation, the following sections briefly compare the evolution of legislation for endangered and introduced species and illustrate how the current introduced species legislation fails to deal adequately with unintentional introductions.

Legislation One hundred years ago, the auspiciously holistic Lacey Act of 1900 (Ch. 553 §1,31 Stat. 187) addressed extinctions and invasions from the perspective of wildlife management. Although species exploitation was not directly prohibited, trade in illegally killed game animals was, and further restrictions prohibited the import of any vertebrates, crustaceans, or molluscs that might harm “wildlife or wildlife resources” (NRC 1995; Bean and Rowland 1997; Corn et al. 1999). Fol-

184

chapter 9

Figure 9.2 Emphasis of biological and legal literatures compared between endangered species (ES) and introduced species (IS). From 1991 to 2000 (A), several thousand papers were published on endangered and introduced species in the biological literature. Very few of these papers treated both (ES & IS). In the same literature, an order of magnitude more papers dealt with the Endangered Species Act than with any of several acts treating introduced species (B). A similar pattern held in the legal literature (C). Literature searches were conducted for (A) and

ecological gambling

185

lowing this joint beginning, the pathways of extinction and invasion legislation diverged. Extinctions Taking its cue from the Lacey Act, early legislation regarding extinctions focused on protecting animal species in the Black Bass Act of 1926, the Migratory Bird Treaty Act of 1918, and the Migratory Bird Conservation Act of 1929 (16 U.S.C. §§851–856, §§703–712, and §§715– 715s, respectively) (NRC 1995). A more multispecies but still animalfocused approach was adopted in 1966 by the Endangered Species Preservation Act (P.L. 89–669, 80 Stat. 926), which provided protection for a select list of fish and wildlife. This protection was subsequently extended internationally by the Endangered Species Conservation Act of 1969, in the form of trade restrictions on endangered animals (P.L. 91–135, 83 Stat. 275). In 1973, the Endangered Species Act was passed, providing plants and animals with unprecedented and powerful protection based on their risk of extinction (16 U.S.C. §701, 3371–3378 and 18 U.S.C. §42) (NRC 1995). The primary agencies responsible for implementing the ESA are the FWS and the National Marine Fisheries Service. Since its enactment, the ESA has been significantly amended. In 1978, critical habitat designation based on ecological and economic concerns was mandated for each species, and the “God Committee” was established to allow economic interests to override these designations. In 1982, listings of critical habitat were separated from listings of species, and permits for incidental take of listed species were granted to non-federal landowners in exchange for habitat conservation plans. In 1988, emergency species listings were allowed, and the protection granted to plants was extended (it nonetheless remains less than that granted to animals) (Kohm 1991). Although the ESA remains in effect, its reauthorization has languished since the 103rd Congress (1990–

(B) in the databases Agricola, ASFA, BIOSIS, Life Sciences Abstracts, and Zoological Record; and for (C) in the LEXIS-NEXIS Legal Review database. Search terms for (A) were endangered species or threatened species, introduced species, non-native species, non-indigenous species, alien species, exotic species, and invasive species. Search terms for (B) and (C) were Endangered Species Act, Lacey Act, Plant Quarantine Act, Federal Plant Pest Act, Federal Noxious Weed Act, Plant Protection Act, National Aquatic Nuisance Prevention and Control Act, and National Invasive Species Act. For all searches, results were pared down to include only relevant references treating species and legislation in the United States.

186

chapter 9

1994) amid controversial discussions of the roles of science and economics in its implementation. Introductions Regarding introduced species, the Lacey Act as amended specifies a “dirty list” of invaders, such as the brown tree snake, the zebra mussel, and the Chinese mitten crab, that harm wildlife and are therefore prohibited from import. Subsequent to the Lacey Act, federal legislation governing introduced species focused less on protecting wildlife and more on protecting agricultural crops. The Plant Quarantine Act of 1912, the Federal Seed Act of 1939, the Federal Plant Pest Act of 1957, and the Federal Noxious Weed Act of 1974 all prohibited the import of plant and animal pests and diseases (7 U.S.C. §§151–164a, §§1551 et seq., §150aa–150jj , and §§2801–2814, respectively). The Plant Protection Act of 2000 continues in the same vein. Animal species, native or introduced, may also be controlled under the auspices of the Animal Damage Control Act of 1931 (7 U.S.C. §426), which allows the control of any species that damages agriculture, aquaculture, public health, or other enterprise. In the late 1970s, President Carter extended the scope of invasion prevention by restricting federal agency introductions of non-native species into “any natural ecosystem” (Executive Order 11987; May 24, 1977; 3 C.F.R. 116). Despite early agency attempts to implement this order with a clean-list approach, the dirty-list approach of prohibiting only known pests eventually became—and remains—federal policy (OTA 1993). In 1999, President Clinton’s Executive Order 13112 mandated the development of a National Invasive Species Management Plan to unify federal agency policy “. . . on those non-native species that cause or may cause significant negative impacts and do not provide an equivalent benefit to society.” In the last decade, marine and aquatic invasions have been targeted by a multispecies preventive approach in the National Aquatic Nuisance Prevention and Control Act of 1990 and its 1996 reauthorization, the National Invasive Species Act (16 U.S.C. §4701, et seq.; P.L. 104–332). Motivated largely by the devastating economic and ecological impacts of the zebra mussel invasion in the Great Lakes, these laws mandated the development of a National Aquatic Nuisance Species Task Force, and the implementation of guidelines for ballast water discharge from commercial ships. Among these laws, the multiple acts governing introduced species confer scattered and overlapping implementation and enforcement authority on a wide range of federal agencies in the various departments of agriculture (Animal Plant Health Inspection Service [APHIS],

ecological gambling

187

forest Service), commerce (National Oceanic and Atmospheric Administration), transportation (Coast Guard), defense (Army Corps of Engineers), and interior (FWS) (Corn et al. 1999). There remains no comprehensive federal legislation that addresses the prevention and control of introduced species. However, with the development of a National Invasive Species Management Plan, and with the pending reauthorization of the National Invasive Species Act, the time is ripe. Conservation Consequences Endangered species legislation evolved from early goals of protecting a few commercially valuable or publicly endearing animals to its current state of protecting all listed animals and plants, as well as providing for habitat- and ecosystem-level conservation. The listing and protection of a species under the ESA is based solely on the risk—and not on the consequences, either ecological or economic—of its extinction. In contrast, introduced species legislation departed from its focus on preventing harm to hunted species to preventing harm to agricultural species. Recent developments point to a more preventive, ecologically comprehensive approach, primarily for aquatic and marine communities. In general, the prevention and management of introductions, especially of those species not already known to be pests, remains fragmented. What are the consequences of this approach? In the following case study, I highlight the disparate legislative fates of two introduced marine vascular plants. Although both substantially alter the mudflat communities they invade, neither is listed or managed at the federal level. At the state level, one species is vigorously controlled, whereas the other is not.

Case Study: Invasion of the Pacific Northwest Tideflats The large estuarine systems of the Pacific Northwest comprise hundreds of square kilometers of tidal flats in Oregon, Washington, and British Columbia (fig. 9.3). These geologically young, dynamic estuaries are home to a unique mudflat community filled with rich invertebrate and vertebrate life (Ricketts et al. 1939; Kozloff 1973). In the last century, this mudflat ecosystem has been invaded rapidly by two introduced marine vascular plants: the smooth Atlantic cordgrass Spartina alterniflora Loisel (Poaceae) and the Asian eelgrass Zostera japonica Aschers. & Graebn. (Zosteraceae). Both grasses grow in perennial stands, reproducing vegetatively and by seed production. They both add three-dimensional structure above and below ground, altering the habitat and food web of the resident biota. Together, their

Figure 9.3 Location of major estuarine systems in the Pacific Northwest from British Columbia to Oregon. Although these estuaries are relatively small compared with the extensive bays and deltas of the Gulf Coast and East Coast of North America, they provide a unique habitat that is particularly important for juvenile fishes and crabs and for resident and migratory birds. The spread and impacts of two major invaders, the Atlantic cordgrass Spartina alterniflora and the Asian eelgrass Zostera japonica, have been studied in Coos Bay, OR; Willapa Bay, Grays Harbor, Puget Sound, and Padilla Bay, WA; and Boundary Bay and Roberts Bank, B.C.

ecological gambling

189

Figure 9.4 Historically, Pacific Northwest mudflat extended from the high native marshes (Carex, Distichlis, Scirpus, Salicornia) to the low intertidal and subtidal beds of the native eelgrass Zostera marina. Since the early 1900s, the introduced grasses Spartina alterniflora and Zostera japonica have invaded much of this habitat. The upper mudflat is spanned by the tidal limits of the two invaders; at the lower edge, especially on shallow-sloping flats, Z. japonica can encroach on the native Z. marina (which extends into the subtidal). Both introduced grasses modify the sediment; Spartina in particular raises the height of the substratum, an effect not shown here. Note exaggeration of the vertical scale.

vertical distribution spans the entire tideflat from the high native saltmarsh to the low native eelgrass beds (fig. 9.4). The natural history and impacts of these two Pacific Northwest invaders share many of the same features.

Natural History of the Invaders The smooth cordgrass Spartina alterniflora, a much-loved native of the East and Gulf Coasts of North America, arrived in Willapa Bay in the late 1800s (Sayce 1991; Daehler and Strong 1995; Simenstad and Thom 1995; Feist and Simenstad 2000). This accidental introduction, with oyster shipments or with solid ship ballast, spread during the next 90 years across some 6000 ha of Willapa Bay (Scheffer 1945; Cohen and Carlton 1995; Reeves 1998). Spartina alterniflora was also planted intentionally in Padilla Bay and at other Puget Sound sites to provide erosion control and duck habitat in the 1930s and 1940s (Parker and

190

chapter 9

Aberle 1979; Frenkel 1987; Frenkel and Kunze 1984). Locally, S. alterniflora spreads by seed and vegetative propagation, forming conspicuous clonal patches that eventually merge to become a cordgrass meadow (Frenkel 1987; Riggs 1992; S. Riggs pers. comm.). Three other Spartina species—S. anglica, S. patens, and S. densiflora—have also been introduced to the Pacific Northwest. Spartina anglica in particular may prove even more invasive and difficult to control than S. alterniflora, but all three are less well studied in this region, so I focus here primarily on S. alterniflora (hereafter, Spartina). Asian eelgrass, Zostera japonica, was introduced to the Pacific Northwest with Japanese oyster shipments (Harrison 1976). It was first reported in Washington in 1957 and has since been reported from British Columbia and Oregon (den Hartog 1970; Harrison 1976; Harrison and Bigley 1982; Bulthuis 1991). By the 1980s, over 300 ha of Z. japonica were reported in Padilla Bay, Washington, and over 3000 ha in Boundary Bay, British Columbia. This comparatively short, narrowbladed seagrass typically grows in a band above the native perennial Z. marina, although in some estuaries the two species grow together in mixed beds (Harrison 1982; Thom 1988; Nomme and Harrison 1991a). Zostera japonica (hereafter, Zostera) can behave as both an annual and a perennial, reproduces sexually and asexually, disperses by seed, and overwinters as seeds and shoots (Bigley and Harrison 1982).

Impacts of the Invaders Spartina and Zostera act in comparable ways to change the mudflat communities they invade. Both grasses add three-dimensional structure, altering habitat characteristics and the density and diversity of the mudflat biota. Both invaders also add new primary production, which is consumed by resident grazers and detritivores. These habitat and trophic alterations lead to fundamental changes in local speciesinteraction webs. Habitat Alteration Spartina and Zostera both convert mudflats to meadows. A Spartina meadow appears as a highly modified, semiterrestrial grassy area with shoot densities up to 519–638 mⳮ2 and considerably enhanced sediment deposition rates (Thom 1990; Bulthuis 1991; Thom et al. 1997). A Zostera meadow is recognized more readily as a vegetated tidal flat. Although shoot densities may number up to 1520–3200 mⳮ2, sediment deposition does not seem to be altered to the same extent (Harrison 1987; Posey 1988). In addition to creating extra three-di-

ecological gambling

191

mensional habitat, both invaders alter physical and organic sediment characteristics (Posey 1988; Bulthuis 1991; Zipperer 1996; Thom et al. 1997). These changes in habitat are associated with marked changes in the density and diversity of resident flora and fauna. Infaunal Density and Diversity The density, diversity, and composition of invertebrate species differs between mudflat and invaded areas for both grasses. Overall, Spartina in Willapa Bay appears to increase local taxon richness, primarily among native mud-dwelling invertebrate taxa (Zipperer 1996) (table 9.1). Individual taxa differ in density between Spartina patches and adjacent mudflat, but the magnitude and direction of the response varies (see table 9.1). For example, epifaunal cumaceans (non-native Nippoleucon hinumensis), amphipods (Corophium spp. and non-native Grandidierella japonica), and anemones generally decrease when cordgrass invades, whereas oligochaetes and dipterans tend to increase (Zipperer 1996). Within a taxon, responses may vary among species: thus, two of five polychaetes tend to be more abundant in Spartina patches, whereas three are more abundant on the mudflat (Zipperer 1996) (see table 9.1). Spartina affects not only surface-dwelling invertebrates but also mud-dwellers. Among infaunal bivalves, for example, the adult density of native clams (Macoma balthica) tends to be reduced by Spartina, whereas densities of Eastern clams, Mya arenaria, and Asian littleneck clams, Venerupis philippinarum, tend to increase (Ratchford 1995; Dumbauld et al. 1997) (see table 9.1). Like Spartina, Zostera alters the distribution and abundance of the resident biota (table 9.2). Where it grows in mixed beds, Zostera reduces the density and growth rate and alters the morphology of its native congener, Z. marina (Nomme and Harrison 1991a, b; Merrill 1995). The effects of the invading eelgrass on invertebrate density and species richness are variable. For small infaunal invertebrates, including polychaetes, amphipods, cumaceans, and insects, taxon density and richness tend to increase with the presence of Zostera (Posey 1988) (see table 9.2). In contrast, densities of larger burrowers, including the shrimp Neotrypaea californica and the tube worm Praxillella gracilis, tend to decrease (Harrison 1987) (see table 9.2). Among epifauna, invertebrate density and richness may decrease with the presence of eelgrass (during tidal submergence). Not all taxa respond in the same way, however; ostracods and nematodes tend to be substantially more abundant on the mudflat, whereas poecilostomatoid copepods and barnacle larvae tend to be more abundant amid the eelgrass (Simenstad et al. 1988) (see table 9.2). These patterns also vary according

192

chapter 9

TABLE 9.1 Habitat and trophic impacts of the introduced cordgrass Spartina alterniflora on resident biota in Willapa Bay, Washington The Change column represents significant differences in taxon attributes, tabulated as ↑ (increase), ↓ (decrease), or ns (no significant change). When significant differences were found at some but not all sites or times, the arrows are enclosed in parentheses. Taxon Plants Benthic microalgae

Attribute

Change

Abundance

ns

Adult density Adult density Growth rate Survival Adult density Recruitment 1 year survival Size Recruitment Recruitment 1 year survival Size Density

↓ (↑) ↓ ns (↑) (↑) ns (↓) ns ↓ ↓ ns ns

Annelids Polychaetes Manayunkia Capitella capitata Pygospio elegans Aphelochaeta sp. Streblospio benedicti Oligochaetes

Density Density Density Density Density Density

↑ (↑) (↓) ↓ ↓ (↑)

Arthropods Corophium spp. Grandidierella Nippoleucon Dipteran larvae

Density Density Density Density

↓ (↓) ↓ ↑

Cnidarians Actinarians

Density

(↓)

Taxon richness % Non-native

(↑) (↓)

Animals Molluscs Macoma spp. Venerupis

Mya arenaria

Macoma balthica

Mytilus sp.

Invertebrates

Source Thom et al. 1997

m Dumbauld et al. 1997

m Ratchford 1995

o

Zipperer 1996

ecological gambling

193

TABLE 9.2 Habitat and trophic impacts of the introduced eelgrass Zostera japonica on resident biota in Padilla Bay (WA), Coos Bay (OR), Boundary Bay (B.C.), and Roberts Bank (B.C.) The Change column represents significant differences in taxon attributes, tabulated as ↑ (increase), ↓ (decrease), ⌬ (significant change in multivariate analysis), or ns (no significant change). When significant differences were found at some but not all sites or times, the arrows are enclosed in parentheses. When results were not analyzed statistically by the original authors, changes in attribute are enclosed in square brackets: differences greater than twofold are indicated as increasing or decreasing arrows (e.g., [↓↑ ]); differences less than twofold are indicated as [⬍ 2x]. Taxa Plants Zostera marina

Animals Turbellarians Nematodes Annelids Polychaetes Pygospio elegans Streblospio benedicti Pseudopolydora Boccardia Eteone californica Capitella spp. Praxillella gracilis Oligochaetes

Arthropods Ostracods Decapods Neotrypaea

Attribute Leaf growth Shoot density Shoot density Morphology

Change ↓ ↓ ns ⌬

Density Biomass

[↓↑] [↓↑]

Density Biomass

[↓↑] [↓↑]

Density Biomass Density Density Density Density Density Density Density Density Density Biomass

[↓] [↓] ↑ (↑) (↑) ↑ ↑ ↑ (↓) (↑) [↓] [↓]

Density Biomass Density Biomass Density

[↓↑] [↓↑] [⬍2x] [⬍2x] (↓)

Source

其 其

Nomme and Harrison 1991a, b

m

Simenstad et al. 1988

Merrill 1995

m Posey 1988

Harrison 1987 Posey 1988

其

Simenstad et al. 1988

m

Simenstad et al. 1988 Harrison 1982

194

chapter 9

TABLE 9.2 continued Taxa Calanoids Harpacticoids Poecilostomatoids Cirripedes (larvae) Gammarids Corophium spp. Ampithoe valida Eobrolgus spinosus Caprellids Cumaceans Cumella vulgaris Tanaids Leptochelia dubia Insects

Spiders (Araneae) Invertebrates

Attribute Density Biomass Density Biomass Density Biomass Density Biomass Density Biomass Density Density Density Density Density Biomass Density Density Biomass Density Density Density Biomass Density Biomass Taxon richness Taxon richness Shannon-Weiner

Change [↓↑] [↓↑] [↓↑] [↓] [↑] [↓↑] [↑] [↑] [↑] [↓↑] ↓↑ (↑) ns [↑] [↑] [↓↑] ↑ [↓] [⬍2x] (↑) ↑ [↓] [↓↑] [↓] [↓] (↑) [↓↑] [↓↑]

Source

o

其 其 m

其

Simenstad et al. 1988

Posey 1988 Simenstad et al. 1988 Posey 1988

Simenstad et al. 1988

Posey 1988 Simenstad et al. 1988

to the tidal cycle. For the invader’s impacts to be understood more clearly, experimental designs with sufficient sampling power, which also account for differences in tidal height between mudflats and eelgrass beds, need to be employed (Simenstad et al. 1988; C. Simenstad and J. Cordell, University of Washington, pers. comm.). In summary, both Spartina and Zostera alter the richness and density of resident infaunal and epifaunal invertebrates. Not surprisingly, neither invader consistently enhances or depresses all other species; rather, the overall impression is one of altered community composition. These changes in invertebrate densities may translate into indirect invader impacts at higher trophic levels.

ecological gambling

195

Trophic Alterations The aboveground habitat provided by Spartina and Zostera adds new primary production to the system. This production can reach impressive levels: a dense Spartina patch may consist of 1335 g dry weight per m2 (gdw/m2) aboveground, plus 1137 gdw/m2 belowground, with an annual net primary productivity of 1520 gdw/m2/y (Riggs and Bulthuis 1994, Zipperer 1996). Zostera shoot biomass can reach 230 gdw/m2, with an annual productivity of 45 gdw/m2/y, or 66 gdw/m2/y with its associated epiphytes (Thom 1990). What is the fate of this new production? Thus far, we know that traces of Spartina-derived carbon are found in the tissue of assorted invertebrate and fish species, including, for example, the introduced cumacean Nippoleucon hinumensis, which in turn is eaten by Pacific salmon (Lubetkin 1997; S. Simenstad unpub. data). Zostera is eaten by isopods (Thom et al. 1991) and comprises over half the gut contents of the brant, mallard, northern pintail, and American wigeon foraging in Boundary Bay (Baldwin and Lovvorn 1994).

Interaction Webs Both these introduced plants clearly alter the habitat and trophic interactions of mudflat communities. At this point, it is impossible to integrate these alterations into an overall prediction of how the entire invaded ecosystem will change. Nevertheless, we can assemble three small interaction webs as examples of likely pathways of change (fig. 9.5). First, through habitat modification, Spartina reduces the mudflat foraging grounds available for juvenile chum salmon and English sole. At the same time, it probably increases the salt marsh nursery area available for juvenile chinook salmon (Simenstad and Thom 1995, Zipperer 1996) (see fig. 9.5a). Second, through trophic pathways, Zostera provides a major food source for waterfowl in Boundary Bay (Baldwin and Lovvorn 1994). However, its indirect effects may be negative, since it reduces the densities of certain harpacticoid copepods common in the diets of four ecologically and economically important fish species (Simenstad et al. 1988; Simenstad 1994) (see fig. 9.5b). Third, Spartina and Zostera appear to interact together in a complex web involving habitat alteration and competition with native burrowing shrimp, Neotrypaea californica, and non-native commercial oysters, Crassostrea gigas (Harrison 1987; Aberle 1993; Ebasco Environmental 1993; Pawlak and Olson 1995; Simenstad and Fresh 1995; Dumbauld et al. 1997; Thom and Rumrill 1999) (see fig. 9.5c). Although these interaction links have been identified, their consequences are, at

196

chapter 9

Figure 9.5 Interaction subwebs for the introduced cordgrass Spartina alterniflora and eelgrass Zostera japonica in a Pacific Northwest mudflat community. A. Habitat alteration. Spartina alterniflora reduces the extent of mudflat nursery grounds available for chum salmon (Oncorhyncus keta) and English sole (Pleuronectes vetulus) but increases the extent of salt marsh nursery grounds available for chinook salmon (O. tshawytscha). B. Trophic links. Zostera japonica is eaten by several species of waterfowl. Compared with the mudflat, Zostera japonica hosts lower densities of three copepods (Zaus sp., Tachidius sp., Tisbe sp.) that are important in the diet of four ecologically and economically important

ecological gambling

197

this point, speculative. Nonetheless, these three “sub-webs” are only small pieces of the ecosystem-scale jigsaw puzzle of potential alterations resulting from these two invaders. In summary, two habitat-engineering plants are currently invading the same mudflat communities. Both convert open, soft-bottom mud flats to rooted vegetation mats: Spartina creates salt marsh, and Zostera creates eelgrass bed. Both alter the community composition of resident infauna and epifauna, and both are linked through trophic and competitive interactions to commercially and recreationally important fishes (including salmon), birds, and shellfishes. Given these similarities in their ecological impacts, what are their similarities in management under federally introduced species legislation?

Legislation Governing Invading Species There is no automatic listing procedure for introduced species under U.S. federal legislation. Both Spartina and Zostera could be designated as federal noxious weeds, since they meet the Plant Protection Act criterion of “directly or indirectly caus[ing] damage to . . . natural resources.” However, since neither plant is in fact listed, regulation falls to the state level. In Washington, Spartina alterniflora is classified as a Class B noxious weed, which requires both state and private (but not tribal) landowners to eradicate the plant from their property. One component of this statewide program is the experimental introduction of an Atlantic leafhopper, Prokelisia marginata, as a biological control agent (Grevstad et al. 2000). Zostera, in contrast, is not listed. Not only is this invader not controlled, it receives de facto protection under a state policy of no net loss of eelgrass (Pawlak and Olson 1995). Thus, in the absence of clear federal criteria for the listing and management of invaders, one community-altering plant is targeted with a considerable control effort, whereas the other is not.

fishes. C. Competition. Spartina alterniflora and Zostera are linked together in a complex and ill-defined web of negative interactions that involve the native burrowing shrimp, Neotrypaea californica, and nonnative commercial oysters, Crassostrea gigas. In addition, Spartina may outcompete Zostera in some areas. Both plants can interfere with the development and maintenance of commercial oyster beds and are often chemically and mechanically removed during oyster culture and harvest. Burrowing shrimp, a pest in commercial oyster beds, grows in the same areas as Zostera; each can inhibit the other’s establishment. For sources, see text.

198

chapter 9

The inconsistent management of introduced species is not unique to plant invaders. For example, the Asian littleneck clam, Venerupis philippinarum, and the Asian varnish clam, Nuttallia obscurata, are both invaders on the coastal tideflats of the Pacific Northwest. Littlenecks were introduced with oyster shipments decades ago; varnish clams arrived more recently in ballast water (Mills in prep.; Wonham and Carlton in rev.) As with the plant invaders, there exists no federal provision for listing either species. The littleneck clam, but not the varnish clam, is listed in the United States Geological Survey (USGS) database of aquatic nuisance species, a listing that is informative but carries no legislative consequences. If demonstrated to cause harm, both species could in principle be listed as injurious wildlife under the Lacey Act and controlled under the Animal Damage Control Act. At the moment, however, V. philippinarum is grown in large-scale commercial operations and is seeded annually for recreational clam harvest by Washington State. Nuttallia obscurata is being evaluated for its potential as a market shellfish (Dinnel and Yates 2000; B. Brady, Washington Department of Fish and Wildlife, pers. comm.). Nor are these examples restricted to the Pacific Northwest. On the East Coast of the United States, the European green crab, Carcinus maenas, and the Asian veined whelk, Rapana venosa, are invasive invertebrate predators (Grosholz and Ruiz 1996; Harding and Mann 1999). Both probably arrived with shipping traffic, the green crab over a century ago and the whelk less than a decade ago. The crab, which produces planktonic larvae, eventually spread north along the coast, where it continues to cause concern among shellfish growers and public fisheries managers (Glude 1955; Walton and Walton 2001). The whelk, which produces crawl-away larvae and is thus more limited in its dispersal capacity, was first reported in 1999 as a small population in the lower Chesapeake Bay and was rapidly identified as a potential threat to commercial shellfish beds (Harding and Mann 1999). Like the clams, neither of these species is automatically listed as an introduced species under any federal legislation. Both are listed as USGS aquatic nuisance species. Both could be prohibited from import under the Lacey Act or controlled under the Animal Damage Control Act, but neither is. In practice, expensive municipal-funded programs trap and remove green crabs (Walton and Walton 2001), but no such program has been initiated for the whelk. These examples illustrate the consequences of our current ad hoc approach to managing species introductions. Some species are targeted vigorously for removal, whereas others are neglected, protected, or even enhanced. This default position of accepting any invader until it is shown to be harmful is not consistent with a precautionary, con-

ecological gambling

199

servation-oriented philosophy (e.g., Hobbs and Humphries 1995; Ruesink et al. 1995; Parker et al. 1999; Goodell et al. 2000). Clearly, there is room for more comprehensive and ecologically consistent tools for responding to introduced species. One such tool would be a federal Introduced Species Act (ISA) containing provisions for listing and managing all introductions. Based on the similarities between invasions and extinctions, we may find that lessons learned from implementing and revising the Endangered Species Act can provide guidance in developing an Introduced Species Act.

Framework for a Federal Introduced Species Act The case studies just presented highlight the need for a formal federal framework for listing and managing introduced species. A similar framework may be found in the ESA, in which species are first listed and then managed according to their listing status. In 25 years of implementation, the ESA has provoked considerable criticism and recommendation regarding both stages. Selected lessons from this discussion can guide us in fashioning a comprehensive ISA. Needless to say, to be fully effective both the ESA and an ISA share a need for the best use of available scientific data and tools as well as adequate funding (e.g., Tear et al. 1993; NRC 1995; Carroll et al. 1996; EasterPilcher 1996).

Listing Under the ESA, species are listed as candidate, threatened, or endangered, based on the severity and imminence of threat and the taxonomic distinctness of the species. Once listed, a species is automatically protected from further harm and is evaluated for subsequent management (specifically, the development of a recovery plan). In contrast, there is at present no legislated listing process for all introduced species. Several publicly available databases exist for certain taxa (e.g., the International Center for Living Aquatic Resources Management FishBase), pests (e.g., the APHIS federal noxious weed and biocontrol lists), and habitats (e.g., the USGS aquatic nuisance species database), but as yet there is no publicly available database with a comprehensive listing of all introduced species. Here, I propose a centralized legislative process to list all introduced species found in the United States. Listing would serve three primary purposes: (1) to record a critical component of our biotic in-

200

chapter 9

ventory, (2) to provide a database against which to assess biotic change, and (3) to facilitate the dissemination of information among scientists and managers. This database would also contribute to the development of an international invasion database (e.g., Ricciardi et al. 2000). To be maximally informative, the database would list all introduced species solely on the basis of their presence rather than on their risk of spread or impact. It would include both intentional and unintentional introductions. Many of the recommendations that have been made for refining the ESA listing process, such as objective listing criteria, taxonomic equality, and timely listing, would also enhance ISA listing. Objective Criteria To be listed under the ESA, a species must be identified and its population status assessed. These assessment criteria, however, have been criticized as ill defined and poorly documented (Rohlf 1991; Tear et al. 1993; Easter-Pilcher 1996; Carroll et al. 1996). For introduced species, the proposed listing process would be much simpler and would avoid many of these difficulties. Listing would simply be a matter of confirming the identification of a non-native species, regardless of how it got here, and would not require the evaluation of its population status. Unlike ESA listing, which confers automatic protection, ISA listing would carry no automatic prohibition on import. Listing categories would be essential for species transfers (i.e., introductions to non-native areas within the United States), introductions from other countries and continents, and cryptogenic species (sensu Carlton 1996). Current ESA listing criteria focus on the risk, rather the consequences, of a species’ extinction. Modified criteria have been proposed to prioritize more ecologically important species at the listing stage (Carroll et al. 1996). However, it may be one of the ESA’s greatest strengths that species are listed simply because they are threatened, rather than on the basis of their importance. Similarly, introduced species would be listed without consideration of their actual or potential impact. Taxonomic Equality Following the National Research Council’s recommendations for endangered species, ISA listing would treat plants and animals equally (NRC 1995). ISA would be extended further to include fungi (e.g., Dutch elm disease Ophiostoma spp. and Chestnut blight Cryphonectria parasitica), protozoans (e.g., benthic marine algae Caulerpa taxifolia and

ecological gambling

201

Sargassum muticum), bacteria and viruses (e.g., Vibrio cholerae pathogenic strains and West Nile virus), and other microorganisms. Pests and diseases are specifically excluded from ESA listing—a reasonable qualifier, since species listing confers automatic protection. For an ISA, the equivalent exclusion would be any unintentionally introduced species that generated economic or health benefit. However, given the proposed separation between ISA listing and management, such an exclusion would be unnecessary and inadvisable, since it would hinder the development of an accurate biotic inventory. Speed and Timing Some endangered species listings have occurred too late and too slowly for adequate protection (Rohlf 1991; Wilcove et al. 1993; Carroll et al. 1996). ESA amendments have attempted to speed up the process by separating the process of species listing and critical habitat designation, and by allowing emergency listing. For introduced species, the importance of speedy listing is equally if not more acute, since theory and experience tell us that a population may be more easily eradicated while still small (e.g., Allee 1931; Hobbs and Humphries 1995; Meinesz 1999). The simplicity of the proposed ISA framework would allow rapid listing and response, obviating the need for an emergency listing provision.

Management Not all endangered species listed under the ESA receive recovery plans. In a pragmatic, if not painless approach, plans are developed primarily for species that have reasonable recovery prospects; the poorest prospects are typically left to fend for themselves. Thus, like the listing process, species management is based on the likelihood of extinction rather than on the potential impacts of a species’ loss. An equivalent ISA approach would be to assign invasion management plans based on the potential for successful eradication. Aggressive invaders that were difficult to control would be left to spread unchecked. Clearly, this would not do as a blanket policy. The ESA model needs modification before it can be applied to an ISA. In briefly considering the management features of an ISA, I touch on three specific aspects of endangered species management that have been widely discussed: management plans, the separation of ecology from economics, and a comparison of single-species and ecosystem approaches.

202

chapter 9

Management Plans ISA management plans would necessarily be tailored to the type of invasion. Non-native species, on listing, would be classified as unintentional or intentional introductions, and as localized or widely established. Unintentional, localized introductions would be the simplest to handle: where possible, they would be targeted for immediate emergency eradication, in a system similar to the APHIS emergency response to disease outbreaks. Unintentional, established invaders would be evaluated more thoroughly and prioritized based on three essential considerations: (1) potential or observed population growth and spread, (2) potential or observed ecological impact, and (3) potential or observed economic impact of the invader. This is not a trivial proposition. Gathering and projecting these data would require some combination of population viability analysis, spatial modeling, observational and experimental studies, and estimation of direct and indirect interaction strengths and economic losses (e.g., Ruesink this volume). Characteristics associated with invasion success would also be accounted for (e.g., Baker and Stebbins 1965; Rejm´anek and Richardson 1996; Williamson and Fitter 1996a; Reichard and Hamilton 1997). This process would require considerable time and effort but, if done judiciously, would contribute to well-informed management plans with experimental components in an adaptive management framework. A prioritization scheme based on these criteria might also avoid the problems of unclear goals (Tear et al. 1993, 1995; NRC 1995; Smallwood et al. 1999) and taxonomic favoritism (Rohlf 1991; Clark et al. 1994; Dwyer et al. 1995; Shilling 1997) that have been identified in ESA implementation. Experience with intentional introductions tells us that careful monitoring of target and non-target impacts as well as transparent cost-benefit analysis (e.g., economic return vs. ecological change) are warranted. Those intentional introductions that are determined to be acceptable (e.g., wheat, lentils, cows, and horses would seem to be likely candidates) would remain listed and would continue to be monitored, especially as feral populations evolved. Intentional introductions determined to be unacceptable would fall under the same management framework as unintentional introductions. Determining the acceptability of these invaders would include essentially the same considerations as those outlined for prioritizing the eradication of unintentional invaders. Ecology vs. Economics From the ESA, we have learned that effective species management plans require a consideration of ecological and economic interests

ecological gambling

203

(Dwyer et al. 1995; NRC 1995; Tear et al. 1995; Carroll et al. 1996; Shilling 1997). Controlling or eradicating an invader comes neither easily nor cheaply and may involve physical habitat disruption, the application of chemicals, and the release of non-native biocontrol species. Such restoration projects cannot be assumed to be welcomed with open arms. Based on experience with the ESA, we can anticipate the need to define clearly the rights and responsibilities of private landowners and tribal, local, state, and federal governments in invader control and eradication. In Washington State, for example, a species classified as a Class B noxious weed must be eradicated from private property at the landowner’s expense—even though the landowner may not have released the invader. Incorporating such a provision into an ISA would provide a powerful management tool but would perhaps unfairly overburden the individual. At the same time, however, increased consumer appreciation of the consequences of invaders could help reduce the demand for novel species imports. To facilitate management, we can therefore anticipate the importance of creating appropriate market-based incentives to favor the prevention and control of introductions by industry, government, and private citizens. Species vs. Ecosystems Although the language of the ESA allows for ecosystem-level management, its implementation until recently focused on single species. This approach garnered extensive calls for a more ecosystem-oriented, proactive management (e.g., Clark et al. 1994; Dwyer et al. 1995; NRC 1995; Tear et al. 1995; Carroll et al. 1996). Official FWS policy now reflects a commitment to ecosystem-level protection, with a focus on habitats and important (e.g., keystone or umbrella) species. Invasion management can benefit from the comparable concepts of habitatlevel eradication and restoration plans and the eradication of highimpact invaders that profoundly alter community structure or facilitate further invasions (e.g., Simberloff 1998, 2001). A greater challenge facing ESA implementation is whether it can truly provide proactive protection. The idealized extinction trajectory illustrates a gradual decline in species density ending with an abrupt change when the species is lost (see fig. 9.1a). Although the ESA evolved as a speciesspecific reaction to that final transition, proactive protection would require targeting the earlier stage where a population appears healthy— a biologically easier but politically more challenging approach. In contrast, an invasion begins with the abrupt arrival of an invader (see fig. 9.1b). The immediate sources of invasions are thus more easily identified than the cumulative forces leading to extinctions, which perhaps

204

chapter 9

lends invasions more readily to preventive management (Campbell 2001; Czech et al. 2000; Shine et al. 2000). Such prevention in turn automatically confers ecosystem-level protection of natural communities. In this regard, it is possible that future ESA refinement can draw from our experience in preventing invasions. Even with the best possible preventive measures, extinctions and invasions will continue to accrue, and ultimately these two threats cannot effectively be managed independently. The global processes of urbanization, consumption, and trade that threaten endangered species serve simultaneously as pathways of species introduction. Further, invasions have been identified as a leading cause of extinctions (Wilcove et al. 1998), and extinctions may be associated with subsequent invasions (e.g., Elton 1958; Simberloff and Von Holle 1999). In the management sphere, habitat restoration often requires the recovery of endangered species as well as the eradication of invaders. The processes of extinction and invasion are thus more than mere mirror images: they are intimately connected processes that must, ultimately, be prevented and managed in concert (PCE 2000).

Summary Paine and Zaret (1975) commented that “[t]oday, increased opportunities for touring to isolated places, the rapidity of air transport, and the gradual disappearance of international, cultural, and trade barriers have all enhanced the level of unintentional ecological bombardment on previously safe communities.” A quarter century later, these anthropogenic processes are, if anything, accelerating, leaving behind them a wake of species extinctions and invasions. Extinctions and invasions are now recognized as major threats to biodiversity, community structure, and ecosystem function. Because they share parallel biologies and management challenges, they would appear to be amenable to comparable federal legislation. At present, however, our legislation governing species extinctions—the Endangered Species Act—is far more powerful and comprehensive than that governing species invasions. As illustrated in the case study of two invasive marine plants, the current piecemeal invasion legislation does not lend itself to the ecologically consistent management of introduced species. A broader, more effective Introduced Species Act could readily be based on the listing and management framework of the ESA. There is much we can learn from the ESA’s implementation that would facilitate the development of a conservation-based, comprehensive response to the threat of introduced species.

ecological gambling

205

Acknowledgments This chapter benefited greatly from discussion with J. Carlton, J. Cordell, S. Riggs, and S. Simenstad. R. T. Paine, D. Boersma, G. Orians, D. Simberloff, and two anonymous reviewers provided valuable comments on an earlier version. Thanks especially to P. Kareiva for editorial assistance. This chapter was written with support from the Estuarine Reserves Division, Office of Ocean and Coastal Management, National Ocean Service, National Oceanic and Atmospheric Administration (Fellowship No. NA77OR0250, Padilla Bay National Estuarine Research Reserve).

Chapter 10

m Rarity and Functional Importance in a Phytoplankton Community Daniel E. Schindler, Gary C. Chang, Susan Lubetkin, Sally E. B. Abella, and W. T. Edmondson

Heightened scientific and public interest in biodiversity has been driven by the accelerated rates of species extinctions at local and global scales. The question of which species are expendable derives from the humble realization that society is not able to prevent the extinctions of all species with which humans interact. The benefits of preventing extinctions are often in direct conflict with economic development in human societies. Asking scientists to estimate the expendability of specific species is a seductive solution to asking which species are worth “saving” versus which species will not be missed if they go extinct. The problem of determining the expendability of species requires that we place values on specific species. Humans place values on specific species in numerous ways. Humans value species for the commodities generated through their exploitation (e.g., fisheries and forestry), for the aesthetic appeal of hav-

phytoplankton community

207

ing certain species on landscapes, or simply by knowing that they exist. We also value species for the ecosystem services they provide (Daily 1997). For example, insects pollinate plants and wetland plants remove pollutants from water, to name some obvious services. Within the context of community ecology, the values of specific species may be based on their effects on the species with which they interact, their roles in maintaining the structure and dynamics of communities (Paine 1966, 1992), or their roles in maintaining critical system processes such as biomass accumulation, nutrient cycling, and energy flow. In this chapter we evaluate the relationship between the rarity of certain species and their expendability in a community context. We use a simple value system to estimate the expendability of phytoplankton from Lake Washington, in the United States. Although phytoplankton species are generally not considered high priorities in biodiversity conservation, the data set from Lake Washington is particularly long-term and of high taxonomic resolution, providing a unique opportunity to explore the functional importance of rare species over an ecologically relevant time period. For the purpose of this chapter, we estimated the value of certain species by their contribution to total phytoplankton community biomass. We were particularly interested in estimating the contributions of rare species to the maintenance of phytoplankton community biomass, to answer the question, “Are most species just rare and [ecologically] boring?” (R. T. Paine, pers. comm., 1999). Although phytoplankton have many characteristics that may determine their ecological value (e.g., grazability, ability to compete for nutrients, nutrient storage), we used their contribution to community biomass as our value metric because we were able to assign these independently to all species in the community. At present, we are unable to assign other value metrics independently to all species in the community.

What Are Rare Species? The ecological vernacular uses several concepts to refer to the commonness or rarity of species. Rabinowitz (1981) described a common classification scheme that is used for defining the rarity of species. This classification model assigns species to types of rarity based on their breadth of habitat requirements, their geographic range, and their abundance. Here, we measure rarity as the frequency of occurrence of specific species in time-series data (i.e., continuity or temporal rarity). In general, the likelihood of a species being detected in a monitoring study is related to the density of that species (Krebs 1999),

208

chapter 10

so temporally rare species may simply be species that exist at low abundances. In some cases, however, changes in life history strategies may cause certain species to remain undetected for a period of time simply because they have changed their behavior or habitat use. For example, some phytoplankton may be temporally rare in lakes because they only sporadically immigrate out of resting stages in the sediments into the water column (analogous to the seed bank in terrestrial plants; Sandgren 1988; Hansson 1993), and are usually detected only when they are suspended in the water column. However they are defined, temporally rare taxa may be especially prone to extinctions that result from human disturbances because ecologists (or society in general) are less likely to place high values on these species compared with those that are abundant and common. Thus, rare species might not always garner the political protection that temporally common species do. In addition, ecology may underrepresent the functional importance of temporally rare species over the long term. For example, the ecologist trying to estimate species interactions in a diverse community is faced with the logistical problem of being able to experimentally manipulate only a small number of species (Kareiva 1994). The obvious solution is to select species that are relatively common for these experiments. Cumulative experience and syntheses of these traditional ecological experiments are, therefore, likely to be highly biased in favor of understanding the ecological roles of common species. As a result, rare species are unlikely to work their way onto the ecologist’s list of nonexpendable species. The critical question is whether this bias poses a serious problem to identifying and protecting nonexpendable species. In this paper, we used a 36-year data set from an intensively studied ecosystem (Lake Washington, Washington State) to estimate the relative contribution of temporally rare species to the total species pool in a phytoplankton community. We ask whether the species that are temporally rare ever represent a functionally important component of the community. By asking this question, we are exploring the relationship between numerical rarity and temporal rarity (fig. 10.1). Our null expectation is that the numerical rarity of a species is correlated positively with its temporal rarity. However, we want to know whether temporally rare species can ever be functionally important (i.e., fall in the top left corner of fig. 10.1). Our analyses showed that the majority of species in this community are temporally rare and that a relatively small fraction of species is actually present continuously in the community. We then show that the vast majority of rare species are of little consequence for total community biomass, although temporally rare species can be substantial contributors during certain periods. This

209

low abundance

Numerical rarity

high abundance

phytoplankton community

rare

common Temporal rarity

Figure 10.1 Two components of species rarity are their abundance (numerical rarity) and their continuity (temporal rarity). Temporally rare species are those that are present (or detected) infrequently in a community. It is often assumed that these two components of rarity are linked, but the nature of their relationship is not understood.

result demonstrates that placing values on species (i.e., determining the expendability of species) is a strategy that is likely to fail in the long run—especially when long-term data are not available for assessing the potential importance of rare species. High species turnover rates will produce temporally rare species in communities and are probably common phenomena in many ecosystems when they are evaluated at ecologically relevant timescales. Priorities for research and conservation should be placed on understanding the effects of natural species turnover on system stability, identifying the natural processes that maintain species turnover, and protecting those processes that are probably an important component of ecological organization. Making estimates of the expendability of certain species requires that the value-placer has observed all possible states of the ecosystem in question or that the system of interest is static. Neither of these criteria can ever be fulfilled; therefore, measures of expendability undoubtedly underestimate the list of species that are important to ecosystems subject to future disturbances.

210

chapter 10

A History of Lake Washington Lake Washington is a large, urban lake located in the Greater Seattle area of Washington State (longitude: 122⬚W, latitude: 47⬚N). The ecology of Lake Washington plankton has been studied extensively by W. T. Edmondson and colleagues from 1962 to the present (Edmondson 1994). The lake has experienced several major perturbations resulting from substantial human activity in the watershed. These perturbations have been described in detail elsewhere (e.g., Edmondson and Lehman 1981; Edmondson and Litt 1982; Edmondson 1991) but are worth a brief overview here to provide a context for this chapter. Secondary sewage from Seattle was discharged directly into Lake Washington until the mid-1960s. The lake showed many of the symptoms of severe eutrophication in response to the increased nitrogen and phosphorus loads from the sewage. Throughout the 1960s, Lake Washington was characterized by low water transparency and extensive cyanobacterial blooms (Edmondson 1991). We refer to this period as the “sewage era.” By the late 1960s, sewage discharge into Lake Washington was halted, and the sewage discharge was diverted to Puget Sound. Water quality improved substantially within 5 years of sewage diversion (Edmondson 1991). Several cyanobacteria that became dominant during eutrophication (e.g., Oscillatoria spp.) disappeared or dropped to very low densities after sewage diversion. We refer to the period immediately following sewage diversion as the “recovery era.” Immediately after the water quality improvements resulting from sewage diversion, a major change in the fish community caused a concordant shift in the zooplankton community, from one dominated by small-bodied grazers to one dominated by large-bodied Daphnia spp. by 1976 (Edmondson and Litt 1982). This change in the grazing community was precipitated by increases in longfin smelt (Spirinchus thaleichthys) whose abundance increased substantially in Lake Washington after bottom dredging in the Cedar River was halted and their spawning habitat was no longer disturbed. Increases in smelt populations induced a trophic cascade (sensu Paine 1980) that further reduced phytoplankton biomass in the lake. Increased smelt predation on Neomysis mercedes (a large predatory crustacean) reduced predation on Daphnia, their preferred prey (Murtaugh 1981). As a result, Daphnia became the dominant grazer in the zooplankton community. This shift in the grazer community led to increased grazing pressure on phytoplankton and another substantial improvement in summer water quality (Edmondson and Abella 1988). We refer to the period from 1976 to the present as the “Daphnia era.”

phytoplankton community

211

The planktonic food web of Lake Washington has remained in a high smelt, low Neomysis, high Daphnia, low phytoplankton configuration from 1977 to 2002. Given the focus of this chapter, it is important to note that the data set used to evaluate the relationship between rarity and functional importance spans an enormous range in nutrient-driven changes in total phytoplankton biomass and in the possible range of grazer community structure. Although these changes were not planned experimentally, they span an informative range of ecosystem states over which to evaluate the commonness and rarity of phytoplankton species and how these species contribute to community biomass in an ecosystem that has been substantially altered by human activity. Detailed protocols of limnological sampling in Lake Washington and sample processing are described in Edmondson (1997). In brief, phytoplankton from Lake Washington were sampled weekly to monthly, year round since 1962. The average number of annual sampling events for phytoplankton was 33 (std. dev. 9.4); these events were spread out over the entire 12-month period with slightly higher sampling intensity during the summer months (April–October). This sampling intensity is as high as is realistically feasible in most limnological studies and has produced arguably the highest-resolution data set on phytoplankton species composition, spanning almost 4 decades and hundreds to thousands of generations of each phytoplankton species. All water samples were collected at a depth of 1 m using a Van Dorn bottle, and 100-mL subsamples were preserved in Lugol’s solution. Phytoplankton from these subsamples were settled into counting chambers and enumerated using an inverted microscope. Where possible, taxa were identified to species, counted, and measured for estimates of cell volume. Total biovolume (biomass) of each species was calculated as the product of the numerical density (噛/L) and the average volume per cell (mm3) of each species. For the analyses described in this paper, we have concentrated on 1962 to 1997, the period with the best temporal coverage in sampling, and have considered only phytoplankton species for which we have confirmed species identifications. All data were aggregated by calendar year.

Temporal Rarity in Phytoplankton We unambiguously identified 176 species of phytoplankton from the open water of Lake Washington from 1962 to 1997. Of these, 28 were species that occupy primarily benthic habitats of lakes and can be considered incidentals. We included these benthic species in all further analyses because occasional immigrations from suboptimal habi-

Figure 10.2 Temporal distribution for 176 species of Lake Washington phytoplankton during the 36-year period from 1962 to 1997 (bars). The cumulative temporal distribution is shown by the solid line.

phytoplankton community

213

tats represent an important component of temporal rarity in biological communities. In plankton (or all) communities, incidental species interact with other resident species in their suboptimal habitat, where they can contribute to community biomass and other ecological processes. Most of the species pool in the Lake Washington phytoplankton community is composed of temporally rare species. Fifty percent of all species were found in 6 or fewer years (17%) of the 36-year record (fig. 10.2). By contrast, only 6 species (3.4%) were found in all years of the study. The ubiquitous species were the diatoms Asterionella formosa, Aulacoseira subarctica, Fragilaria crotonensis, and Stephanodiscus niagarae and the cryptomonad Chroomonas minuta. All other species were observed in a fraction of the entire period from 1962 to 1997. The mean duration of presence of any one species in the community was 11 years, or 31% of this period. Approximately half of the cumulative species pool of phytoplankton was observed in any given year in Lake Washington. On average, 83 phytoplankton species were observed per year in Lake Washington (std. dev. 26). This result may imply that there is a high degree of interannual species turnover in the phytoplankton community that results from immigrations and extinctions. However, high apparent species turnover can result from sampling errors because the entirety of any community can never be detected with even the most intensive sampling programs (Arnott et al. 1999). At this point, we are unable to separate true species turnover from sampling biases. To answer the question “Are rare species boring?” we plotted the maximum proportion of total community biomass achieved by a species versus its temporal rarity (i.e., the number of years that a species was present in the long-term record). Ubiquitous species always made substantial (⬎5%) contributions to total community biomass (fig. 10.3). On average, the combination of ubiquitous species constituted about 50% of the community biomass in any given year. For relatively common species, however, there was a high degree of variation in the maximum proportion of community biomass contributed by any given species. Most of the species found in more than 50% of years made a maximum contribution of 0.1 to 10% of community biomass at some time during the study (see fig. 10.3). There was a general tendency for temporally rare species to contribute less than temporally common species to phytoplankton community biomass in Lake Washington (see fig. 10.3). However, there was considerable variation in this relationship, and several temporally rare species (i.e., found in less than 20% of the study period) constituted substantial fractions of the total community biomass at some time during the study period (see fig. 10.3). For example, Ana-

214

chapter 10

Figure 10.3 Relationship between the number of years certain phytoplankton species were present in Lake Washington and the maximum % of total community biomass attained by that species for the 176 phytoplankton sampled between 1962 and 1997. Histograms represent the distributions of species along the two axes representing numerical and temporal rarity. Overlapping points are jittered slightly to improve presentation.

baena utermolii was found in only 6% of the study period and contributed as much as 6% of total phytoplankton biomass. Similarly, A. oscillatoroides, Oscillatoria aghardii, and Coelosphaerium kutzingianum were found in 8, 17 and 19% of years, respectively, and contributed as much as 7, 5 and 9% of total biomass. Diatoma elongatum was detected in 28% of years and contributed as much as 36% of total phytoplankton community biomass. Thus, although most temporally rare species made only tiny contributions to the total community biomass, there

phytoplankton community

215

Figure 10.4 Average annual contributions to community biomass by 3 classes of phytoplankton based on the number of years that each species was detected in the 36-year data set.

were several exceptions in which rare species made short-term appearances in the phytoplankton community and made substantial contributions to total phytoplankton biomass. Hence, temporal rarity in the Lake Washington phytoplankton community does not imply that a species is always inconsequential. To explore further the relationship between temporal rarity and functional importance in the community, we grouped phytoplankton species into three groups based on their continuity: (1) species found in 1 to 9 years, (2) species found in 10 to 19 years, and (3) species found in at least 20 years. Contributions to annual averages of community biomass from each of these categories were plotted as a time series to evaluate the temporal changes in their importance between 1962 and 1997 (fig. 10.4). Total contributions by the most rare phytoplankton (considered as one group) were relatively small but were largest during the sewage era in Lake Washington. Species detected in 10 to 19 years contributed the largest fraction to the community during the sewage era and represented a substantial fraction of the community during the recovery era (see fig. 10.4). Species detected in fewer than 20 years contributed a very small fraction to annual biomass during the Daphnia era. Contributions by temporally common

216

chapter 10

Figure 10.5 Similarity between the phytoplankton community composition in 1962 and the community in each subsequent year of the data set. Similarities range from 0 to 1. Values of 1 refer to communities that have perfect overlap in the relative abundance of all of their species. Similarities of 0 refer to communities whose species compositions are mutually exclusive (Krebs 1999). These similarity indices were calculated for the phytoplankton community with all of the species (black diamonds) and for the community including only the temporally common species (detected in at least 20 years of the record; open squares). Species contributions to total community biomass were calculated from annual means. Sewage era, recovery era, and Daphnia era in Lake Washington are labeled for reference (see text).

species (i.e., those detected in ⱖ20 years) fluctuated through time but showed no trend across the 36-year study period (see fig. 10.4, least squares regression, p ⬎ 0.13). Thus, the increased phytoplankton community biomass during the sewage and recovery eras was accomplished by increases in abundances of temporally uncommon species whose contributions were added to the constant background biomass represented by the temporally common species. We also evaluated how temporal changes in the composition of the phytoplankton community were attributed to rearrangements in importance of species with different degrees of temporal rarity. To accomplish this, we calculated a similarity index between the phytoplankton community from the first year of the study (1962) and every subsequent year in the dataset (fig. 10.5). Similarity was calculated as: 兺 minimum(p1i, p2i), where p1i and p2i are the proportions of species i in the community at times 1 and 2 (Krebs 1999). Proportional contributions were estimated from annual mean biovolume for each spe-

phytoplankton community

217

cies. When all species were included in the similarity calculations, there were marked changes in community composition associated with changes in nutrient loading and with grazer community structure (see fig. 10.5). During the sewage era, similarity in phytoplankton composition among years was high (mean ⳱ 0.88). However, during the recovery and Daphnia eras, there were major changes in community composition such that mean similarity to 1962 conditions has been about 0.05 since 1976. To evaluate the effects of temporally rare species on changes in community composition, we recalculated the time series of similarities after excluding species detected in fewer than 5, 10, and 20 years. Excluding the most temporally rare species (i.e., those detected in less than 5 or 10 years) had no effect on the temporal changes in similarity of the phytoplankton community. In fact, the data were not visually distinguishable from the indices calculated with all the species included, so they are not shown in figure 10.5. However, excluding all species that were detected in less than 20 years of sampling resulted in a markedly different temporal pattern in community similarity through time. For only temporally common species (i.e., those detected in ⱖ20 years), the phytoplankton community composition was relatively stable (mean similarity with 1962 ⳱ 0.56) and showed no temporal trend over the study period (least squares regression, p ⬎ 0.6; see fig. 10.5). Taken together with the fact that the total contributions from temporally common species did not change through the major changes in ecosystem state associated with eutrophication (see fig. 10.4), this result provides a novel insight into phytoplankton community dynamics. These results imply that phytoplankton communities are organized around a set of core, temporally common species whose rank abundance and contributions to total community biomass are relatively stable in time, even during periods of substantial ecosystem change associated with nutrient status and grazer community structure. Community responses to ecosystem perturbations are realized by temporally rare species whose contributions to community biomass are additive to the stable contributions from core species.

Discussion Analysis of a 36-year, high-resolution data set of the species composition of a lake phytoplankton community showed that most phytoplankton species were rare with respect to both continuity and abundance. In fact, very few species were detected continuously or contributed more than 1% to total community biomass. Our analysis also showed

218

chapter 10

that most temporally rare species contributed very little to total community biomass. The bulk of community biomass was represented by common and ubiquitous species. This result can be interpreted as evidence that ecosystem function is largely independent of biodiversity in diverse phytoplankton communities. Our view of the functioning of this community is that a handful of common species provide the bulk of community biomass and that the majority of species drop in and out of the system on an irregular basis. In some instances, some of these temporally rare species account for a substantial component of community biomass. This conceptual model assumes that the community is not in equilibrium. Species turnover is probably a natural, stochastic, and important organizing feature of this—and probably all—communities. The prevalence of rare species in the phytoplankton community of Lake Washington is not unlike several other descriptions of diverse communities (Williams 1964). For example, McGowan and Walker (1993) showed that most species of marine plankton are numerically rare and contribute little to total community abundance. Similarly, R. B. Root and colleagues (Root and Cappuccino 1992; Root this volume) have demonstrated that terrestrial insect communities can be dominated by numerically rare species whose functions are highly redundant with other species. Plants in grassland systems show the same pattern for numerical rarity (e.g., Stohlgren et al. 1998). Examples of the distributions of temporally rare species and their contributions to ecological functions in communities are more elusive. We believe that the general patterns of temporal and numerical rarity observed in the Lake Washington phytoplankton community are probably common to many high-diversity communities—especially those containing many species that are vagile or have resting stages in their life cycles. Communities of zooplankton, phytoplankton, terrestrial plants, insects, and benthic invertebrates come to mind in this regard. From the perspective of expendability, the quick response to our result might be that most species are in fact rare and expendable, and that conserving the handful of common species will serve to maintain community structure and function. We think that this reaction is incorrect. In the last 4 decades in Lake Washington, several temporally rare species did represent an important fraction of phytoplankton community biomass at one time or another. Although a core group of common species maintained a substantial fraction of the community, independent of the nutrient status of the lake, temporally rare species contributed nearly all of the increased biomass in response to eutrophication. The short-lived nature of the eutrophic conditions in Lake Washington makes this argument somewhat circular. However, the

phytoplankton community

219

fact that the core species made a nearly constant contribution to community biomass, and that their rank abundance did not change with fluctuations in nutrient status, reinforces the argument for a community organized around (1) a group of core species that are relatively insensitive to perturbation and (2) a group of temporally unstable species that are highly responsive to environmental disturbances. In retrospect, the functional identities of the temporally rare species that did become important contributors to community biomass might have been predictable. For example, four species of cyanobacteria were important for a short period during the sewage era of Lake Washington and then disappeared from the community after its recovery from excessive nutrient loading. Given that increased populations of cyanobacteria are a common response to eutrophication, this result is not surprising. However, if we had access to only the last 15 years of phytoplankton data from Lake Washington and were asked to predict the species that would flourish during future eutrophication events, we would be hard pressed to identify these taxa to species-level resolution. Enough is known about functional roles of certain phytoplankton groups to predict that the importance of cyanobacteria is almost surely to increase during future eutrophication events in Lake Washington (Reynolds 1984; Cottingham and Carpenter 1998). However, we have little means with which to predict the exact identity of the species that will change. If we use contribution to community biomass as a simple metric of ecosystem function, the Lake Washington phytoplankton community appears to have a high degree of species redundancy. Other plankton communities also show high levels of species redundancy if this metric of ecosystem function is used (Frost et al. 1995; Cottingham and Carpenter 1998). In our long-term analysis of the Lake Washington phytoplankton community, we observed that some temporally rare species did constitute substantial fractions of community biomass during certain periods. However, what we do not know is whether a “redundant” species would have substituted for these rare species had they been absent from the species pool. In all likelihood, the degree of functional redundancy in diverse systems like phytoplankton communities is probably high, with the consequence that only a small subset of the species pool actually contributes anything substantial to ecosystem functions. Nevertheless, identifying these species is still a problem because rare species can occasionally be important. For decades, the high diversity of plankton communities has posed a paradox for aquatic ecologists (Hutchinson 1961). Equilibrium models of species interactions have perpetuated the debate about why resource-limited phytoplankton communities can be so diverse in apparently homogeneous environments (Hardin 1960). A more plausible

220

chapter 10

model of community organization is that the diversity of phytoplankton communities represents an ecological memory of the system (Padisak 1992; Reynolds et al. 1993) that is maintained by nonequilibrial dynamics among species as well as ecological processes that occur at regional spatial and temporal scales (Cornell and Karlson 1997). Species turnover rates are probably high in phytoplankton communities, even after accounting for the problems associated with sampling error and species detection (Arnott et al. 1999). Phytoplankton communities, and probably most plant communities, are composed of a subset of common and important species that dominate ecosystem processes, and by another larger subset of species that blink in and out of the system. These rare species are maintained in the species pool by invasions into the water column from resting stages in sediments (Sandgren 1988; Hansson 1993) and from immigration from other nearby systems. Although the majority of these species probably do not contribute substantially to community attributes and ecosystem processes, correctly identifying the few species that will be important in the future is essentially impossible—especially as ecosystems continue to change and respond to future natural and human disturbances. We believe that asking about the expendability of all species in diverse communities is the wrong question. Instead, priorities should be placed on (1) identifying and protecting those species that represent the core taxa in a community and (2) identifying and maintaining the processes and disturbances that produce species turnover and natural dispersal linkages among habitats and ecosystems that are important for sustaining diversity. This strategy is more likely to maintain the rare species that will respond to future environmental perturbations. Trying to identify which of these species will become important in future communities is probably futile, given the dynamic nature of ecosystems.

Acknowledgments We have known Bob Paine for various lengths of time (years to decades), and are deeply grateful for the impact he has had on all of us through his contributions to ecology and his collegiality and friendship. We thank Peter Kareiva and Simon Levin for inviting us to contribute to this volume. PK, SL, Peter Leavitt, and four anonymous reviewers commented on the manuscript. We are also thankful to the Andrew W. Mellon Foundation for their long-term support of this project.

Chapter 11

m Community and Ecosystem Impacts of Single-Species Extinctions Daniel Simberloff

Many authors have pointed out that species differ greatly in their importance to their ecosystem and community. Sometimes species matter to other species because they dominate the community in biomass; trees and coral, for example, provide physical structure. Other times, whether a species is common or scarce, some aspect of its activity is crucial to the persistence of the ecosystem. A species may be functionally unique in its ecosystem role; sometimes other species perform the same functions, or could perform them if that species were to disappear. These concepts—importance, function, uniqueness, and redundancy—have given rise to a variety of ideas that all lead to the hypothesis that the extinction of different species should have different impacts and that some of the impacts might be enormous and surprising, whereas others might be hard to detect. The most widely bruited of these concepts is that of the keystone species (Paine 1966, 1969; Power et al. 1996b). A keystone species was

222

chapter 11

originally defined to be a species whose importance to the rest of the community and the ecosystem is disproportionate given its abundance. It has come to be used casually to mean any species that has a very large impact on the ecosystem, whether or not it has small population size or abundance. The concept has been attacked (e.g., Mills et al. 1993; Hurlbert 1997) as a useless panchreston, largely because (1) there is no clear delineation about how important a species has to be in order to be a keystone, and (2) the actual measurements that would be required to assess importance are technically difficult. However, that there is an arbitrary cutoff point and that measuring importance is difficult do not render the concept unimportant. In fact, a better understanding of the keystone roles of particular species is probably crucial to better environmental management (Power et al. 1996; Simberloff 1998). However, it has proven difficult to identify which, if any, species in a community are keystones simply by observing the community. More commonly—and of particular interest to this paper—keystone species are recognized after the fact when a formerly prominent member of a community is removed, either experimentally or serendipitously (Power et al. 1996b). Discussion about keystone species today seeks to broaden the definition from one in which a species’ influence is exerted through trophic interactions (e.g., Paine 1966, 1969a) to one in which many other kinds of activities can qualify a species as a keystone. Thus the concept has come to subsume that of the ecosystem engineer (Jones et al. 1994, 1997), a species whose activities modify the availability of biotic or abiotic resources to other species. It also now encompasses the related concepts of transformer species (Wells et al. 1986), one whose activity or promotion of a phenomenon (e.g., fire) changes an entire ecosystem physically, and foundation species (Dayton 1972, 1975), “species which have roles in the maintenance of the community disproportionate to [their] abundance or biomass” (Dayton 1972, p. 84). The latter “usually include those species actually contributing most of the spatial structure of the community, the competitive dominants, and the disturbers preventing their domination” (Dayton 1972, p. 86). Whether a species is a keystone in a particular system may depend on substitutability or redundancy (Ehrlich and Mooney 1983; Walker 1991; Lawton and Brown 1993), the extent to which one or more other species will perform functions of a species lost from an ecosystem. Although generally addressed in the context of extinction, redundancy is equally applicable to the impacts of introduced species. The impacts of introduced species seem as unequally distributed as those caused by species extinctions, and the ones that tend to have high impact are those that perform a function not performed by any na-

single-species extinctions

223

tive. Thus, the enormous number of extinctions caused by vertebrate predators on small oceanic islands is due to the absence of such species among the native communities. Similarly, the great ecosystem impact of the zebra mussel in North America (Williamson 1996; Ricciardi et al. 1997, 1998) is due to its filling an “empty niche” (Johnson and Carlton 1996): no native adults have byssal threads that allow attachment to rocks or plants in the productive littoral zone, while the active pumping of the zebra mussel is better suited to exploit the resources of calmer waters than are the passive filtering systems of native species (e.g., insect larvae). In general, then, the impact of a species addition or removal would be expected to be a function of the extent to which it is a keystone and the degree to which there is redundancy in the ecosystem for the activities that make it a keystone. I have previously explored the range of impact sizes of introduced species in terms of this hypothesis (Simberloff 1991, 1999). Here, I ask whether the historical record of consequences of species extinctions supports the notion that particular species can be enormously important to entire communities and ecosystems. Three lines of evidence seem pertinent: (1) the consequences of recorded global extinctions, (2) the impacts of enormous declines in the abundance of species that might seem to be keystones, and (3) the impacts of near-extinctions or local extinctions. Of course, the definitions of impact and importance are usually implicit, and their quantitative measurement is rarely attempted (Hurlbert 1997; Parker et al. 1999). Thus, the literature is rife with subjective statements, with the same extinction being seen as enormously consequential by some workers and by others as negligible. For example, Dayton et al. (1998, p. 319) state that “. . . many pelagic systems have lost most of their tuna, cod, whales, turtles, elasmobranchs, etc., yet they have not collapsed and . . . there may be no known ecosystem consequences. The enigma is that . . . the systems continue to function without them.” What is the “system”? What would constitute a “collapse” or an “ecosystem consequence”? What does it mean to “function”?

Global Extinctions Of the thousands of extinctions recorded in historical times, the overwhelming majority are of species that one would not have expected, a priori, to be keystone species, because they were either extremely rare or very narrowly distributed. Rarity and restricted range are two traits that often predispose a species to extinction (Mace and Kershaw 1997), and most recently extinct species were not prominent members

224

chapter 11

of widespread communities. As was already pointed out, species that are relatively uncommon can be keystones; indeed, the original definition (Paine 1966, 1969a) demanded relative scarcity. However, identifying keystones beforehand has proven to be extremely difficult, and, for want of highly detailed natural historic study of most systems, much of the best evidence for the existence of keystones comes from observations of what happens when common species disappear. Unfortunately, even for our most recent extinctions, there is almost never a detailed study of the consequences for the surviving species. The passenger pigeon (Ectopistes migratorius), once perhaps the world’s most abundant bird, declined to extinction over the course of 300 years (Halliday 1980). It inhabited deciduous forests of central and eastern North America (Greenway 1967), darkening the sky and nesting as densely as 100,000,000 individuals per 1000 km2. One might guess that the disappearance of such a species would have enormous impacts on the remaining community—on its food plants and predators, for example. However, there is not even much sound speculation about such impacts. An indication of the difficulty in pointing to a substantial impact of this extinction is the recent discovery on extant birds of two lice thought to have gone extinct with this pigeon (Clayton and Price 1999; Price et al. 2000). Similarly, the Carolina parakeet (Conuropsis carolinensis) occupied deciduous forests of eastern and central North America, especially bottom lands, and was characterized by great flocks before its rapid decline (Greenway 1967; Forshaw 1977). One would imagine that there would be an impact of the total disappearance of the only North American psittacid at least on the plant species it ate, but there are no data on the impact of this extinction, and again, not even much speculation. As was suggested for its basic biology, the literature “is a mosaic of hearsay, second-guessing, and inference” (McKinley 1978, p. 223). If we have no solid information of the consequences of recent extinctions of enormously numerous birds, in a well-populated region of North America, it is unsurprising that we can say nothing definitive about the consequences of recent extinctions of less numerous vertebrates, such as the ivory-billed woodpecker (Campephilus principalis), many of which were even more poorly studied. The impacts of bird and mammal extinctions at times or in places where there were no scientists to study them will probably be forever shrouded in mystery. It would be surprising if the rapid extinction of such large, abundant species—the large Malagasay lemurs (Macdonald 1984), the moas, the giant rail, and many other birds of New Zealand (King 1984) as well as the dodos and solitaires of the Mascarenes (Green-

single-species extinctions

225

way 1967)—did not have major consequences for many members of their respective communities. This would be especially true of species that likely played unique ecological roles on their respective islands. But we are unlikely ever to know much about them. For example, regeneration failure in native conifers of New Zealand has been attributed to moa extinction (Wellman in Cooper et al. 1993), but it is difficult to know how to test this idea. Temple (1977) attributes the decline of a formerly common large endemic tree of Mauritius (Calvaria major) to the extinction of the dodo (Raphus cucullatus), which he speculated had to digest the seeds for germination to occur. Owadally (1979), however, casts grave doubt on this hypothesis. It is, of course, possible that comparative and other approaches will lead to clearer linkages between historic extinctions and their impacts (cf. Cooper et al. 1993), but it is difficult to be optimistic. Among mammals, the 18th-century extinction of Steller’s sea cow (Hydrodamalis gigas) in the northern Pacific would seem to have had major consequences. It was abundant in the Commander Islands in 1742 (Stejneger 1936, in Nowak 1991) and grazed on kelp and other marine algae. In the late Pleistocene, it occurred along the north Pacific rim from Japan to California (Nowak 1991) and was probably common in many areas. Reported by Steller (Stejneger 1936, in Nowak 1991) to be gluttonous and weighing up to 10,000 kg (Scheffer 1972, in Nowak 1991), these sea cows might be expected to have greatly influenced the kelp beds that provided the physical structure for much of the community. Likewise, their decline to extinction in a mere 26 years could have had an enormous impact. Estes et al. (1989) argue cogently that the impact of this extinction may indeed have been substantial in the warm-temperate waters of California, but they concede “that the global extinction of Steller’s sea cow took with it any possibility of obtaining direct evidence about its herbivorous role in kelp forest communities” (p. 260).

Great Reductions As with global extinctions, impacts resulting from a great decline in abundance of populations have been sought almost exclusively among species that were large and common. Thus, observers expected these species to be keystones in the general sense rather than in Paine’s original definition of relatively scarce species with strong per capita interaction strength. The impacts of declines of scarce species have rarely been studied in detail, unfortunately. The American chestnut (Castanea dentata) quickly became “func-

226

chapter 11

tionally extinct” (i.e., so uncommon that it would not serve its original ecosystem functions) after the Asian chestnut blight fungus (Cryphonectria parasitica) arrived in New York in the late 19th century. In less than 50 years, the fungus spread through 100 million ha of eastern North America from southern Ontario to northern Georgia and Alabama, killing almost all mature chestnuts (Anderson 1974; von Broembsen 1989). According to Harper (1977), “this is probably the largest single change in any natural plant population that has ever been recorded by man.” Chestnut had been a dominant tree in many areas, so one might expect its functional extinction to have had enormous impacts on native ecosystems. However, many authors (e.g., Vitousek 1986; Williamson 1996) assert that this near-extinction shows that the loss of even a dominant single species need not have a major impact on ecosystem structure. One problem, of course, is the one alluded to previously: how big does an impact have to be to qualify as major? A bigger problem, however, is that there are simply insufficient data to do more than guess at the full impact. There are hints that the impact may have been major. For example, several moths host-specific to chestnut went extinct (Opler 1979), and nutrient cycling was probably decelerated, since chestnut decomposes rapidly (K. Cromack, pers. comm.). Other effects might have been less direct and very subtle but nonetheless consequential. For instance, the oak wilt disease (Ceratocystis fagacearum) has increased on many species because the susceptible red oak, Quercus rubus, increased when chestnut disappeared (Quimby 1982). But the sum of the observations and speculation about this case is that there is simply insufficient evidence to know the impact of this near-extinction. American elm (Ulmus americana) ranges throughout the eastern United States and much of the Midwest and was used widely as a shade tree elsewhere in North America. Ophiostoma ulmi, a fungus causing Dutch elm disease, first appeared in Europe in the early 1900s, then spread to North America in 1930 (von Broembsen 1989; Campbell and Schlarbaum 1994). By 1977, the disease had spread to most of the contiguous states and had killed 75% of all elm trees in the Northeast alone. The disease returned in more virulent form to Europe in the 1960s, where it eliminated mature elms of all species from much of England (Richens 1983); the earlier epidemic had killed 10 to 20% of elms in England (Wilkinson 1978). Although not a dominant, elm was a common tree in many forests and was a common urban shade tree in both Europe and North America. As with chestnut, pre-disease data are woefully inadequate to assess the impact of this loss, although many organisms are (or were) obligate associates of elm (Richens 1983), and the structural changes of many urban and

single-species extinctions

227

some natural and seminatural landscapes are striking (Wilkinson 1978; Richens 1983; Campbell and Schlarbaum 1994). The destruction of the massive bison (Bison bison) herds of central North America in the 19th century surely had major ecosystem consequences, and the recovery of a few herds that are now abundant (albeit over much smaller areas) permits some estimate of the impact. Before European settlement, North American prairies were a sea of grass regulated by fire and large ungulates, primarily bison in the recent past (Axelrod 1985). The bison were probably particularly crucial in eliminating isolated trees on the plains as well as trees and large shrubs in river valleys, swales, and other lower, moister areas. Thus, for example, Moss (1932) and Campbell et al. (1994) argue that the near extinction of bison in the northern plains was a key reason for the expansion of quaking aspen (Populus tremuloides) woodland into prairies. Even the loss of bison feces may have had substantial consequences, not only for insects and other species that would have inhabited and consumed them. Bison feces burn far hotter than the grassland fires that ignite them, and they probably would have increased habitat heterogeneity and thus local species richness (Crockett and Engle 1999). Further, bison selectively use certain habitats (e.g., prairie dog [Cynomys ludovicianus] colonies; Vinton et al. 1993), and they favor certain plant species for food (Vinton et al. 1993) and for rubbing and horning (Coppedge and Shaw 1997). Controlled studies of the impact of resurrected bison herds in the Tallgrass Prairie Reserve in Oklahoma (Coppedge and Shaw 1997; Crockett and Engle 1999) and the Konza Prairie Research Natural Area in Kansas (Vinton et al. 1993) show a variety of effects. However, even these large preserves with their small, fenced herds are too small to allow a full understanding of the impact of the almost complete removal of bison from hundreds of thousands of square kilometers over which they roamed freely and patchily. The 90% decline in populations of the North American beaver (Castor canadensis) owing to trapping after European settlement (Naiman et al. 1986) is believed to have caused enormous landscape changes; shallow, aquatic communities and even upland communities were replaced by swamps, flooding increased, and major changes were generally effected in hydrological regimes (Naiman et al. 1986; Hackney and Adams 1992; Pollock et al. 1995). Beavers are the paradigmatic ecosystem engineers (Jones et al. 1994, 1997; Pollock et al. 1995). Their dams generate many impacts that are often reflected in increased habitat diversity and changed biogeochemical cycling (but some are far subtler; see, e.g., Martinsen et al. 1998). Data on the precipitous decline of the sea otter (Enhydra lutris) in

228

chapter 11

the northern Pacific owing to hunting from the mid-18th century through the early 20th century, allow an estimate of the ecosystem impact of their loss (Estes et al. 1989, 1998). The initial decline of the sea otter was followed by partial recovery in only some parts of their original range and then followed by a rapid decline again in some of these areas because of predation by killer whales. The most dramatic impact of the decline was the deforestation of kelp beds owing to rapid increases in sea urchin populations released from predation by the otter. That this deforestation was due to the absence of otters was confirmed by regrowth of kelp forests after recolonization by the otters. The kelp deforestation, in turn, led to the decline of large nearshore fish species. In many other regions originally dominated by kelps, the release of sea urchins has transformed an area into an “urchin barren” (Dayton 1984). In the Gulf of Maine, for example, overfishing of large predatory fishes led to a dramatic increase in populations of the green sea urchin (Strongylocentrotus droebachiensis), transforming large areas into urchin barrens (Harris and Tyrrell 2001). Subsequent commercial harvesting of the urchins led not to a regrowth of the kelp beds but to an entirely new set of communities dominated by several introduced species (including a recently arrived exotic alga) and opportunistic natives. Many commercial fisheries have experienced drastically reduced populations of particular species, sometimes to the point of functional extinction (Parsons 1996; Dayton et al. 1998). Although, for the most part, the ecosystem consequences of these losses are studied poorly and are confounded by other factors, in some cases correlated changes in many other populations, in aggregate, seem to indicate a major change in the ecosystem. For example, the removal of 90% of Antarctic blue whales (Balaenoptera musculus) is associated with great increases in populations of several other krill eaters, while the removal of 5 million tons per year of pollock (Theragra chalcogramma) in the Bering Sea has caused drastic declines in mammals and birds that fed on them (references in Parsons 1996). Overfishing and the invasion of the sea lamprey (Petromyzon marinus) led to the local extinction of lake trout (Salvelinus namaycush) in Lake Michigan. This loss allowed the alewife (Alosa pseudoharengus) to invade, leading to reductions or losses of several planktivores, reduction in zooplankton abundance, and enhanced algal production (Crowder et al. 1996).

Local Extinctions or Near-Extinctions The removal of livestock from Santa Cruz Island, California, in 1989 resulted in an explosive, unexpected increase in exotic weeds, espe-

single-species extinctions

229

cially fennel (Foeniculum vulgare), which had been introduced in the 1850s and was not previously a prominent weed before the livestock operation (Dash and Gliessman 1994; R. Klinger pers. comm., 2001). The fennel, in turn, replaced native plant species on much of the island. Other impacts, such as those of herbivores on fennel and the plants it replaces, or on the fire regime, have not been studied, but large parts of the island that were probably a mixed, diverse prairie of native plants before European settlement are now a sea of tall fennel. Surely the total ecosystem consequences of the removal of cattle would be measured as major. Similarly, feral pigs and goats have recently been eradicated from Sarigan Island, a 5-km2 island in the Mariana Islands (Kessler 2002), with the unexpected subsequent explosive growth of an introduced vine, Operculina ventricosa. The impact of Operculina on the regeneration and expansion of native forest is as yet unquantified. However, vast stretches of ground and trees are now thickly carpeted with this vine, a formerly minor component of the flora, and it is hard to believe that this vegetational change has not greatly affected individual population abundances as well as various ecosystem traits. These examples suggest that the extinction of single species of livestock, and perhaps of ungulates generally, can have enormous ecosystem consequences. A similar decline of mammalian grazers, but one for which one might have expected functional redundancy to have dampened the impact of near-extinction, was caused by the eruption of rinderpest, a disease introduced with Indian cattle to Africa in the 1880s (Sinclair 1979, 1995). This introduction led to an epizootic in the 1890s that raged through much of the continent until the 1960s, when cattle vaccination largely stemmed the disease in wildlife. Although many ungulates were affected by rinderpest, some species (e.g., buffalo and wildebeest) suffered as much as 90% mortality, while the impact on others was much less severe. The disappearance of all these animals, followed by the disappearance of the disease and the dramatic resurgence of some of the ungulates, constitutes a double example of the impact of near-extinctions. The effects in both instances were farreaching and dramatic (Sinclair 1979, 1995). The earlier decline led not only to great changes in relative abundances of the ungulates but also to major shifts in the distribution and abundance of their food plants, their predators, and various ecosystem processes (e.g., fire). With respect to functional redundancy, the fact that many ungulates were much less affected by the virus did not prevent a huge impact from the tremendous decline experienced by certain species. Of course, the various species are not exact functional equivalents, but the main evidence for this statement is probably the suite of impacts of the nearextinctions of the buffalo and the wildebeest. In other words, the

230

chapter 11

whole notion of functional equivalence, like that of keystone species, may prove difficult to apply beforehand and may be understood primarily in the wake of a species’ disappearance. Introduced rabbits (Oryctilagus cuniculus) have been completely eradicated from Phillip Island in the Norfolk Island group (Hermes et al. 1986), Round Island off Mauritius (Atkinson 1988), and several small New Zealand islands (Veitch and Bell 1990). Further, the spread of myxomatosis in 1954 led to drastic declines in introduced rabbit populations of Great Britain and many of its offshore islands (Harper 1969). The results of these eradications and near-eradications (Harper 1969; Atkinson 1988; Bullock et al. 2002; Towns 2002) were farreaching and sometimes surprising. Generally, vegetation changed greatly. In the British Isles, for example, Harper (1969) observed that the vegetation of Skokholm (where rabbits were not afflicted with myxomatosis) was unchanged from a grazed lawn-like appearance, whereas the neighboring island of Skomer, with heavy rabbit mortality, became a “tussocky hay field” (p. 51). On several islands rid of rabbits, native plant species were resurrected that had been extremely rare or even believed to be extinct. In some instances, the impact spread from the changed vegetation to animals. On Korapuki Island, New Zealand, for example, the eradication of rabbits led to a great increase in two native plant species, which in turn led to increases in a honeydew-producing scale insect and then in a gecko population (Towns 2001). Many introduced plant populations have been locally or regionally reduced to levels at which they are functionally absent. For instance, in Florida, alligatorweed (Alternanthera philoxeroides) once completely blanketed about 1000 ha of Florida waters but has been controlled since the 1970s at low levels by an introduced flea beetle (USDA 1977, Center et al. 1997). Similarly, water hyacinth (Eichhornia crassipes) covered more than 50,000 ha of Florida waters in the 1950s but has been kept below 1600 ha since 1988 by a combination of chemical and mechanical control (Schardt 1997; Schmitz et al. 1997). In both instances, the presence of a thick canopy over formerly open water had many impacts. For water hyacinth, light intensities and oxygen levels were lowered, water temperatures and evapotranspiration were raised, sedimentation was increased (through leaf decay), native submersed vegetation was smothered, and the fish community changed (Schmitz et al. 1997), although a quantitative estimate of ecosystem impact was never made. The nearly complete elimination of these weeds might have been expected to produce great changes throughout the ecosystem, but there is no accounting of this impact, probably because the primary goals of these weed control projects were navigational and aesthetic.

single-species extinctions

231

Perhaps the best-known success of the biological control of an introduced plant concerns the prickly pear (Opuntia stricta) in eastern Australia. Imported in 1839, it spread quickly and infested about 250,000 km2 by 1925. Half of this area was dense, high growth completely covering the ground (Dodd 1959). After the introduction of the moth Cactoblastis cactorum in 1926, prickly pear was quickly rendered a minor component of the flora in most areas, and, though much of the area is grazed by livestock, some has reverted to a semblance of the original native community (White 1981). There is no study of the ecosystem- or even community-level consequences of the disappearance of prickly pear, but most other species of plants and animals were likely affected substantially.

Conclusions The examples presented, which purport to show major communityor ecosystem-wide impacts of the extinction of individual species, are the result of largely reasonable speculation based on a few observations. For the recent historical extinctions, it is evident that no one made the necessary measurements to enable strong statements about such impacts; probably no one thought to do so. The overwhelming impression one gets from reading contemporary accounts is that the visible, physical changes in the biota were so large that it would have seemed obvious that the ecosystem impact was enormous, had the authors understood the concept of ecosystem. Even many observers of current declines seem to take large physical changes almost as ipso facto evidence of a major impact. Of course others, as noted earlier with respect to the chestnut decline and the loss of large pelagic vertebrates, are quite skeptical that such changes signal a major impact— although “major impact” remains undefined. Furthermore, the new biodiversity-ecosystem function paradigm (Chapin et al. 1998; Naeem et al. 1999; Naeem 2000, pers. comm. 2001) will probably not incline researchers to quantify the measurements that would be needed to determine just what the impact of the loss of a particular single species might be. The impacts of keystone species (sensu latuo) are squarely in the camp that sees the main drivers of ecosystem properties as traits of dominant species rather than number of species, and this camp sees itself as unjustly disenfranchised in the rush to show that biodiversity per se matters (Wardle et al. 2000). Of course, the fact that I can point to some examples in which the loss of a single species appears to have had a substantial community or ecosystem impact does not mean that the losses of most species have similar impacts. Whether or not there is some statistical ten-

232

chapter 11

dency for ecosystem functions to decline with the loss of each species, it is unlikely that most species, if extinguished, would contribute much to that loss. Just as with introduced species, in which a small minority (say, 10%) seem to generate impacts that affect an ecosystem substantially (Williamson 1996; Williamson and Fitter 1996b, Lockwood et al. 2001), most species, if eliminated alone, would probably not greatly affect the remaining community and ecosystem. Extinction and introduction are two sides of the same coin, and the frequencies with which these two processes lead to substantial community and ecosystem impacts are bound to be similar, as both reflect the percentage of all species that can act as keystones in certain systems. At the outset, I stated that I would examine the ecosystem impacts of extinction among the small minority of species that one might expect to be keystones in the broad sense—prominent, common, usually large species. In the long list of species recently gone extinct that were not particularly prominent members of their communities, there is scant evidence of substantial ecosystem-wide impact. This fact should not console us as we continue to extinguish species after species. It simply suggests that our justification for wanting to save them all should not derive primarily from whatever roles they play in maintaining ecosystems that provide services to humans. Callicott (1980) and Taylor (1986) have argued that species have their own inherent worth, independent of human needs, welfare, or knowledge, and that we should not eradicate them on those grounds. Even on utilitarian grounds, we should recall that ecologists have not yet proven adept at predicting the impacts of the losses or additions of particular species. The question arises of whether the extinction of any keystones by Paine’s original definition (relatively scarce, with strong per capita interaction strength) has led to substantial community or ecosystem effects. I have been unable to find such cases, but this failure is not strong evidence that such events have not occurred. One has only to realize that, even among the extinctions, near-extinctions, and local extinctions of common species cited here, there is almost never sufficient empirical evidence to demonstrate the full impact of the loss on the community or ecosystem. Further, as observed previously, the very fact that a species is a keystone species (sensu stricto) is often recognized only after its removal. Paine’s classic experiment pointing to the keystone concept, entailing the very local removal of the sea star Pisaster ochraceus (Paine 1966), is an excellent example. Surely this research indicates that, if Pisaster were to become regionally extinguished, strong reverberations would permeate the entire community. The only reason we can make such a statement is because of the intensive experimental work Paine conducted

single-species extinctions

233

on the system. For the vast majority of scarce species that have gone extinct, even recently, there is no research on their role in their respective ecosystems, much less the detailed work that would allow us to know the impacts of their loss. Finally, it is not a coincidence that my survey of single-species extinctions mirrors surveys of single-species introductions. Too often, the necessary measurements were not made before (or even after) the event (introduction or extinction), and we are forced to infer community and ecosystem impact from visible, physical change. Further, just as invasion biology has largely been a catalog of more or less wellstudied specific cases (Williamson 1996), so the study of single-species extinction currently emphasizes the details of individual cases or the resultant rates (e.g., Lawton and May 1995). Both fields would profit from an integration of empirical research with a more comprehensive, theoretical treatment—probably one that would deal with both phenomena together. Finally, virtually all introductions and recent extinctions were caused by the activities of one keystone species, Homo sapiens.

Acknowledgments I thank Todd Campbell, Paul Dayton, Peter Kareiva, Nathan Sanders, Mary Tebo, and Marjorie Wonham for constructive suggestions on this paper.

Pa r t I I I

m LINKAGES AND EXTERNALITIES

Can we ever really conclude that a species is expendable? While it is straightforward to identify a research protocol that provides evidence for the value of a species or of biodiversity in general, using science to conclude that a species is expendable seems unbearable to many ecologists and evolutionary biologists. This final section reveals the difficulty that leading ecologists experience when asked to consider a judgment that some species might in fact be expendable. First, Leigh points out that our current understanding of the interdependencies among species is so limited, and falls so short of the deep understanding required, that questions about “expendability” are staggeringly na¨ıve. He examines the web of influences among tropical species, uncovering numerous surprises that could be detected only with the use of long-term and intensive field studies. Leigh scolds modern ecology for its under-appreciation of detailed natural history, which in his view dooms ecology to shallow and erroneous assessments of species expendability. Morris focuses his attention on the value of the pollination services provided by various species that visit flowering plants. He uses data

236

pa r t i i i

from a wide range of studies on patterns of pollinator visitation to quantify the consequences to the plant of losing any given pollinator species. Specifically, Morris asks whether infrequent pollinators might be expendable from the plant’s point of view. Even this more narrowly defined question proves difficult to answer, and Morris’s analyses reveal that even species that are infrequent visitors to plants have value. Root goes further toward a possible verdict of “expendability” in recounting his studies of rare herbivorous insects. He points out that after lengthy study (and the sort of detailed natural history that Leigh champions), some goldenrod insects would seem to be truly ecologically expendable. Root backs away from this potentially shocking conclusion, however, by reminding us that studies of plant-herbivore interactions are replete with examples of host shifts and evolutionary change. He points out that if a taxon represents a proper species with a unique evolutionary lineage and hence a unique (and unpredictable) future, then evolutionary (rather than ecological) considerations imply that the taxon is not expendable. Palumbi arrives at this same conclusion but starts from the perspective of an evolutionary biologist rather than a community ecologist. Specifically, Palumbi extends Root’s caution about evolutionary potential by describing several examples of rapid species evolution that in turn altered the nature of species interactions and hence a species’ ecological role. Once we admit that ecologically significant traits of species can evolve rapidly, then a species’ value is not so easy to dismiss—especially given the dramatic anthropogenic disturbances that currently impact our planet. Today’s ecologically trivial species could be tomorrow’s keystone species. Perhaps it is not surprising that the contribution that most seriously grapples with the issue of setting priorities for conservation comes from a group of ecologists who work on salmon recovery for the National Marine Fisheries Service. Federal agencies and other “on-the-ground” conservation efforts consistently face the reality of limited funding, and hence the need for rational guidelines to be used in making choices about what to save. Ruckelshaus, McElhany, and Ford in this section ask which populations of endangered salmonids might be given lower priority for protection (or recovery). Although populations are one step below the species level, their approach is germane to questions of “expendability” for any unit. A quantitative constraint (minimum “services” that must be provided) is specified, then the contribution of different collections of populations is assessed. Academic researchers would probably be troubled by the willingness of these conservation practitioners to establish priorities for protection, but if science does not establish those priorities for sal-

linkages and externalities

237

mon, economics will. Ruckelshaus et al. find that the greatest obstacle to developing rational priorities for population protection is an inadequate understanding of the frequency, magnitude, and impact of catastrophes and how this impact is mediated by evolutionary responses. The same obstacle would probably be true of species expendability, since so many of the arguments for biodiversity ultimately rest on the value of “buffering,” “redundancy,” and “insurance” in the face of dramatic change. Finally, Power and Flecker analyze a remarkable data set that describes the specificity of insect-vectored plant diseases for the species they infect, and for the host insects that vector these diseases from plant to plant. We forget sometimes that pathogens are species as well, and that any discussions of biodiversity and the value of species need to consider pathogens. What makes Power and Flecker’s chapter so interesting is the completeness of the data set and the number of species it involves (1673 viruses). For most taxonomic groups or associations of species, it would be impossible to compile such a definitive list of direct interactions. One interesting result of Power and Flecker’s analyses is the notion that diseases may be vulnerable to the loss of their vectors (because they are carried by few vectors), which in turn could have broader implications for plant population dynamics in the wild. The most striking feature of the chapters in this section is the general reluctance of scientists to deem any species to be more or less important than another. If ecological studies fail to find an important role for a species, for instance, then evolutionary arguments are invoked. The problem is that “importance,” and conversely “expendability,” are value-laden terms that have implications for environmental policy. The discomfort we feel in using a word such as “expendable” or “redundant” with regard to any species suggests that policy regarding biodiversity (or extinctions) must be founded on general principles rather than on a case-by-case justification for each species. Or, perhaps ecologists need to admit to themselves, as well as to others, that there are ethical dimensions to the discussion that contribute to their unwillingness to deem any species expendable. At the same time, it is essential that ecology develop the ability to predict the consequences of particular species additions or losses, so that such gains and losses can be managed for the best possible sustainable future. Reactive conservation, which emphasizes the maintenance of all biodiversity and of “natural pristine systems,” can decide on its course of action without regard to the detailed roles of species. However, proactive conservation, which emphasizes providing for a future natural world that is likely to differ from any historical vision of pristine, will require the sorts of insights developed in this volume, and inspired by Bob Paine’s career.

Chapter 12

m Social Conflict, Biological Ignorance, and Trying to Agree Which Species Are Expendable Egbert Giles Leigh Jr.

Deciding which species are expendable is a singularly contentious question, at three levels. First, posing it looks very like sitting in judgment on God’s Creation. One need not be religious to be wary of this sort of pride. Second, if we do choose to sit in a judgment seat that belongs to Another, how do we decide what aspects or functions of ecosystems should be preserved or enhanced? Finally, if we are not put off by the orgy of self-centeredness involved in deciding what we want from ecosystems, how do we discern whether a particular extinction will bring on consequences that we deem unacceptable? A look at the fossil record tells us that any number of species have gone extinct. Except for major crises such as those that ended the Paleozoic and Mesozoic, we can rarely see how these extinctions threatened their ecosystems. The extinction of the American megafauna at the end of the Pleistocene must have caused major changes in at least

240

chapter 12

some ecosystems (Janzen and Martin 1982), but human beings responded to these changes quite readily, even creatively. The invention of agriculture seems to be among these responses (Piperno and Pearsall 1998). At the moment, our ecological understanding is usually far too crude to allow precise judgments of which extinctions would matter. To decide what species, if any, are expendable, we must consider two issues: 1. Suppose we knew our biology. What criteria are appropriate for deciding which extinctions are unacceptable? In other words, what features of the world ecosystem must we preserve? 2. What kinds of biology do we need to know to preserve the essential features of our ecosystem? To know how the extinction of different species affects their ecosystems requires knowing how ecosystems are organized and the many ways we depend on them.

How Do We Decide What Features of the World Ecosystem to Preserve? What are our responsibilities toward our environment and the other species in it? Because traditional religions have played a crucial role in maintaining harmony between human beings and the order of nature, in societies as different as the Achuar of Amazonian Ecuador (Descola 1993), the Amuesha of Amazonian Peru (Santos-Granero 1991) and traditional China, let us begin with two traditional answers. The scriptures of Judaism, Christianity, and Islam proclaim us lords and stewards of God’s creation, responsible to God for its integrity (Nasr 1996). Moreover, God made wonderful wild animals that are living praises of their Creator, although they are without any conceivable use to human beings (Psalm 104; Job, chapter 38:36 through chapter 41). Theoretically, this view would not allow us to proclaim any species expendable. Nonetheless, although God is concerned for young lions, who are said to seek their food from Him (Psalm 104:21), the Bible recommends no particular human provision for their survival or welfare. Instead, good stewardship means preserving harmony between human beings and the order of nature. Preserving this harmony requires harmony between human beings and God. Because Adam rebelled against God, he was told “cursed is the ground because of you; in toil you shall eat of it all the days of your life; thorns and thistles shall it bring forth to you . . .” (Genesis 3:17–19). Harmony between human beings and nature centers on the responsible

social conflict

241

use of land, with due regard to the interests of poor neighbors and even of wild animals (cf. Exodus 23:10, Leviticus 25:1–7). Indeed, harmony between humans and nature requires a modicum of social justice: social injustice leads to ruin of the land (cf. Isaiah 5:8–10). Let us now turn to a very different traditional voice. The Makuna, hunter gatherers and swidden agriculturalists of Amazonian Colom˚ bia (Arhem 1996), are much more dependent on the integrity of wild nature than were the Hebrews of the Torah. Perhaps as a result, the Makuna consider themselves parts of nature—or nature as an integral part of society. They view hunting as a carefully regulated exchange among humans, animals, and the animals’ Spirit Owners. Before an animal is hunted, the shaman must negotiate consent from its Spirit Owner. When the animal is killed, the shaman must despatch the spirit of the slain animal to its species’s “birth house” so that a new young of the species may be born. Overhunting is punished by disease or death in the hunter’s community. It is the Makuna’s job to maintain harmony with the forest animals and their Spirit Owners, which are members of the society of which humans are only a part. Maintaining this harmony involves the avoidance of overhunting. Here again, the maintenance of harmony among humans and other forest denizens is the primary goal: the avoidance of extinctions would simply be a by-product. As the impact of the ecological crisis on the rural poor of Third World countries increases and becomes increasingly evident, the ideal of human beings as stewards of God’s creation, working for the common good of the creation as a whole, is being put forward as a fundamental principle to govern relations between humans and their environment (Hall 1990; Nasr 1996; Northcott 1996). Of course, answers reeking of theism or the spiritual value of harmony with nature will not satisfy everyone. Secular answers about our responsibilities come in two basic kinds. First, a person who, like Paine, has devoted a lifetime to understanding the ecology of a setting of striking natural beauty, does not wish this beauty to disappear from the face of the Earth. Many ecologists, and other lovers of nature, would consider an extinction unacceptable if it threatens the integrity or beauty of some natural system. To what extent these systems can or should be restored to their pristine condition is, however, a contentious issue. Not everyone who wishes to keep developers out of Yellowstone wants to see it filled with grizzly bears. Not everyone who wishes to keep the central California coast “unspoiled” welcomes the return of abalone-eating sea otters. Even the biologists of Barro Colorado Island might have misgivings about repopulating that island with bushmasters. Suppose that we could recreate the Pleis-

242

chapter 12

tocene megafauna a` la Jurassic Park: how many farmers would welcome the opportunity to share their crops with marauding mammoths? Agreeing on what beauties of nature to defend, or restore, is not easy. Second, others would manage the world ecosystem to further human well-being. This goal is very vague. People disagree on how much human well-being depends on access to natural beauty, how much natural ecosystems can tell us about the management of artificial ones, and what sorts of economic sacrifice are appropriate to maintain clean rivers or clean air. Even when there is agreement in principle, disputes occur. Although no one wants global warming, some do not or will not see the connection between fuel use or forest destruction and global warming; others ask whether local attempts to reduce carbon dioxide emissions will merely subsidize cheaters elsewhere in the world, or wonder how much one should sacrifice to reduce the suffering that global warming may cause others. (How much do citizens of Denver care about the flooding of London or Bangladesh?) No one wants to see the world fisheries ruined by overexploitation or destructive fishing techniques. Yet ruinous fishing techniques are used, on the assumption that competitors will use them anyway. Dysfunctional forms of competition, often excused by expressions of despair over the prospect of fair regulation, are a major obstacle to intelligent conservation planning. Where there is a sufficient sense of community, intelligent conservation is possible. The exceptional sense of community among the inhabitants of Monteverde, Costa Rica, enabled them to preserve over 10,000 ha of nearby rainforest and its fauna—one of tropical America’s most striking conservation successes (Nadkarni and Wheelwright 2000; Leigh 2001). This success stemmed from several factors: (1) the infectious example of sustainable, profitable land use by immigrant Quakers; (2) the Quakers’ effort to transform Monteverde into a community meant to seek the good of each member, farmer, or town dweller, where community decisions were reached by consensus after all views had been heard and considered; (3) a communal reserve of ridgetop forest set aside by the Quakers to protect the area’s soil and water supply, which became a tourist attraction; (4) a Quaker cheese factory that paid fair prices for all comers’ milk, providing the basis for a widely shared prosperity—which allowed community members to profit from the increasing numbers of visitors to the reserve by building hotels and eateries for them, a circumstance that increased support for more reserves. Unfortunately, few societies approach the Quaker ideal of social harmony. Indeed, in most tropical countries, as in the United States, social conflict, latent or blatant, is the biggest single obstacle to con-

social conflict

243

servation (Bruenig 1996). Western society now seems particularly prone to conflict. Capitalists are promoting an aggressive, often heedless, economic competition; Marxists are promoting conflict among classes; and other groups are responding with violence. These conflicts are spreading all over the world. A tragedy of the commons (Hardin 1968) is most likely when social harmony does not reign among its users (Netting 1976, 1993, pp. 172–178). A community of interest in protecting a commons is weakest, and least effective, when there is high turnover among its users. Yet the free market capitalism currently in fashion increases turnover within neighborhoods by seeking a mobile labor force that is willing to move where the money is. A worldwide tragedy of the commons becomes inevitable when the poor feel that they are paying the expenses of conservation with no prospect of sharing in its benefits, and when the rich and their governments are indifferent to their plight. Communities cannot decide how to maintain harmony with nature when they are riven by such conflicts: it is like trying to safeguard the good of one’s country when fighting a desperate civil war.

How Can We Decide Which Extinctions Will Compromise Our Chosen Goals? Currently, the most effective, and appropriate, argument for preventing extinctions is Aldo Leopold’s: we should not discard any part of an ecosystem before we know its function. Academics might question the implied analogy between an ecosystem and a machine designed for a purpose. Nonetheless, we are just beginning to learn the many ways we depend on natural ecosystems, the variety of their aspects on which we depend, and the range of services natural ecosystems provide (Daily et al. 1997). Natural ecosystems provide models for forest managers. For example, disturbances that occur normally in natural forests suggest appropriate cutting regimes for managed ones (Bruenig 1996; Kohm and Franklin 1997). Natural ecosystems also provide us with medicines, and our crops with genes that enhance pest resistance and organisms that control pest populations. They provide small farmers living near forests with protein, firewood, and construction materials. They moderate regional climates and regulate global temperature. All too often, we learn the importance of some ecosystem service only when human activities begin to compromise it. Resources for conservation are limited, however, so we will inevitably be confronted with choices of what to conserve. Ideally, such

244

chapter 12

decisions presuppose an understanding of how the extinction of a species affects its ecosystem and how biodiversity influences ecosystem function. How can we answer such questions?

What Can We Learn from Treating Ecosystems Holistically? The most important lesson of holistic ecology is that ecosystems are organized—adapted—for functions that enhance their productivity and the diversity of their species (Leigh and Vermeij 2002). This principle will not help us identify which species are expendable, but it strengthens Leopold’s argument for conservation and helps us see through some of the methods proposed for demonstrating the importance of biodiversity. How can we recognize whether ecosystems are adapted? Knowing nothing of how adaptation evolves, Aristotle argued that organisms are adapted—organized—to grow and reproduce because visibly mutant organisms are usually less functional than their normal counterparts (Physics 199b 1–4: see Barnes 1984, p. 340). Fisher (1930, p. 38) expressed a similar view: An organism is regarded as adapted to a particular situation . . . only in so far as we can imagine an assemblage of slightly different situations . . . to which the animal would on the whole be less well adapted; and equally in so far as we can imagine an assemblage of slightly different organic forms, which would be less well adapted to that environment. Aristotle’s remark and Fisher’s definition provide a criterion that allows us to decide whether ecosystems are adapted. We imply the “adaptedness” of ecosystems when we say that disturbance usually injures them. Is disturbance injurious? Here, I can only sketch relevant kinds of evidence. One mark of ecosystem adaptedness is the naturally occurring ecosystem properties that cultivators desire but must strive to obtain. For example, a good soil embodies a host of seeming contradictions. It is soft enough for roots to penetrate yet cohesive enough to stay put. It keeps the nutrients and an appropriate amount of the water entering it from leaching or draining away yet allows plants to suck them out when needed. Even when holding abundant water, it allows air and carbon dioxide to circulate through it (Bruenig 1996; Marshall et al. 1996). A natural forest protects its soil and usually improves it. Deforestation usually increases erosion rate and decreases the soil’s thickness, fertility, and ability to hold water (Stallard et al. 1999). Careful cultivation of deforested land can pre-

social conflict

245

serve the soil and its quality (Bruijnzeel 1990). Such care, however, presupposes an understanding of the land that usually reflects generations of experience with the land, and a long-ripened traditional knowledge of it. For humans to preserve the integrity and quality of soil as well as natural forest does is an achievement that does not happen by chance. Another mark of ecosystem adaptedness is the radical degradation that often results from human disturbance (Leigh and Vermeij 2002). In many tropical areas, for example, careless or incompetent land use has led to the replacement of diverse forest by depauperate monodominant grasslands (Jacobs 1988, p. 252; D’Antonio and Vitousek 1992). Similarly, killing off the megafauna in eastern Siberia led to the replacement of grassland by much less productive moss tundra, whose limited transpiration causes waterlogging of the soil (Zimov et al. 1995). These marks suggest not that ecosystems are optimally designed but that they are sufficiently adapted that major disturbance usually compromises their productivity or diversity. Ecosystems are not units of selection. Although Leigh (1999, chapter 9) and Leigh and Vermeij (2002) consider various mechanisms that may adapt ecosystems, how ecosystems adapt is one of the great mysteries of biology. Here, I demonstrate the adaptedness of ecosystems without explaining how this adaptedness evolves. Trying to assess the effects of extinction or the relevance of biodiversity by treating ecosystems holistically, however, has been unfruitful. Comparative studies of productivity, biomass, and nutrient cycling provide no general criteria for identifying those species whose extinction matters most. In terrestrial ecosystems, different species affect soil quality and nutrient cycling very differently, thanks to the differing chemistry of their fallen leaves and twigs, the presence or absence of nitrogen-fixing nodules on their roots, and the like (Hobbie 1992; Bruenig 1996; Silver et al. 1996). It is clear that some species are more expendable than others; the problem is finding general criteria to identify which species matter most. Holistic studies have been particularly unsuccessful in assessing how diversity affects ecosystem function. Wright (1996) concluded that in tropical forest the ecological roles and functions of different plant species overlap so completely that the extinctions of a few plant species would have no impact on productivity, biomass, or nutrient cycling. Vitousek and Hooper (1994) found that, beyond a relatively low threshold value, further increases of plant diversity had no effect on a soil’s content of nitrogen or organic matter. All this must be true, yet such results are a grossly inadequate basis for making decisions about conservation.

246

chapter 12

Many holistic studies fail because they are too “schematic”; they neglect the many and various demands on natural ecosystems. Those who have studied natural ecosystems in fluctuating environments have learned something about the relevance of biodiversity. Diverse grasslands recover more quickly from drought or unusual grazing pressure—and maintain steadier production in the face of environmental variation or normal grazing levels—than do species-poor grasslands (McNaughton 1985, 1994; Tilman 1996).

To Predict the Effects of Extinction, We Must Know How Ecosystems Work To assess the effect of an extinction, we must know how the relevant ecosystem is organized. This means knowing how its populations are regulated or controlled, how its species can coexist, how its species depend on each other, and how the ecosystem responds to invading exotics or severe fluctuations in its physical environment. These criteria sound obvious and elementary. Yet some have tried to assess the importance of biodiversity by studying artificial, simplified ecosystems (experimental or theoretical) whose development was not shaped by natural selection. However, in so far as natural communities are adapted systems whose species are mutually adjusted, these studies are irrelevant. Moreover, they sometimes overlook crucial factors. One such study of how biodiversity affects ecosystem properties (Tilman et al. 1997) considers only communities of primary producers. This study cannot tell us how, for example, tree diversity defends tropical forest against its enemies (Ridley 1930, p. xvi: Regal 1977). On the other hand, mechanistic ecological studies, designed in light of how natural selection works, can provide a much clearer idea of how extinctions affect ecosystems, and can sometimes even suggest which extinctions matter most. Such studies should also consider which features allow ecosystems to resist invaders, because these same features render communities less susceptible to the extinctions such invaders can cause, and because invasions can tell us how an ecological community is organized. Paine (1974, 1977) learned much about the ecological organization of the rocky intertidal at Tatoosh by causing “extinctions” of selected species in experimental plots. Paine was also fascinated by the often radically disruptive consequences of species introductions (Zaret and Paine 1973; Paine and Zaret 1975). Indeed, invasions by exotics are also informative probes of the ecological organization of natural ecosystems. Next, consider how a knowledge of how populations are regulated, how species coexist, how they

social conflict

247

depend on each other, and what factors influence an ecosystem’s ability to resist invaders sharpen our vision of what extinctions are most damaging and why biodiversity matters. I focus primarily on tropical forest, but the lessons of Tatoosh, like the lessons from the extinctions on newly created islets, are never far from my mind.

How Understanding Population Regulation Can Identify Indispensable Keystone Species Paine (1984) knew as well as anyone that many features of ecological systems are governed by competition either for food or for a means (e.g., space) to procure it. In Neotropical forests, most vertebrate herbivores are limited by seasonal shortages of fruit and new leaves (Leigh 1975, 1999; Leigh and Windsor 1982; Smythe 1986). Many of these herbivore species depend on a few “keystone” sources of fruit to survive these seasons of shortage. The extinction of one or more of these keystone species would be especially disruptive of their ecosystems (Terborgh 1986). The competition of predator and prey (or herbivore and food plant) for resources in the prey’s body, a process much studied by Paine (1966, 1969b, 1971, 1974, 1976, 1980, 1992; Paine and Vadas 1969) plays a crucial role in the regulation of many populations. A keystone predator is a species whose extinction would cause major changes in the structure and ecological organization of its community (Paine 1969a). Paine (1966, 1974) showed experimentally that on the weather coasts of the northeastern Pacific, the sea star Pisaster keeps beds of the mussel Mytilus californianus from spreading into the lower intertidal. By limiting the spread of mussels, Pisaster makes space for a diversity of algae and sessile animals in the mid-intertidal. Because its activities maintain species diversity in the rocky intertidal, Pisaster qualifies as a keystone predator. Over the long term, keystone predators can affect the evolution of ecosystem properties. In the northeastern Pacific, sea otters limit subtidal populations of sea urchins (Strongylocentrotus spp). Where sea otters are absent, sea urchins devastate kelp beds. Where sea otters limit urchin populations, luxuriant kelp beds develop (Estes and Palmisano 1974: Estes and Duggins 1995). The resulting increase in nearshore productivity supports animals as different as harbor seals and bald eagles (Palmisano and Estes 1977). By making this increased productivity possible, sea otters qualify as keystone predators. Before human beings nearly finished them off, northeastern Pacific sea otters had long consumed enough sea urchins to reduce the kelps’ need for

248

chapter 12

antiherbivore defense. In New Zealand, however, kelp-eating herbivores are food limited, not predator limited. Thus, New Zealand kelps contain much higher levels of antiherbivore toxins, and New Zealand herbivores are much more tolerant of these toxins than are their counterparts of the northeastern Pacific (Steinberg et al. 1995). There is a fairly strict tradeoff between a plant’s growth rate and the effectiveness of its antiherbivore defense (Coley 1988; Kitajima 1994). Thus, sea otters have presumably favored the evolution of faster growing, more productive kelps. Tropical forests need the help of birds and other insectivores to limit populations of insect herbivores (Leigh 1975, 1999). Leigh and Windsor (1982) originally inferred this need by comparing insect consumption by birds, as inferred from the numbers, weights, and diets of Barro Colorado’s birds, with leaf consumption by insects. It takes about 250 kg dry weight of foliage to feed the 25Ⳮ kg dry weight of insects eaten by birds in tree canopies per ha per year (Leigh 1999). This amount represents up to half the foliage eaten by insects other than leafcutter ants (which birds do not eat). Other researchers are now becoming interested in the extent to which predators limit populations of insect herbivores (Coley and Barone 1996; Bernays 1998; Letourneau and Dyer 1998). How important is this “third trophic level” to the maintenance of the luxuriance of tropical forest? Are specific keystone predators involved in protecting the forest?

Forest Fragmentation as a Tool for Understanding How Forest Populations Are Regulated Forest fragmentation, especially the creation of forested islands by new reservoirs, provides a new tool for understanding how different terrestrial populations are regulated. The smaller the island, the more of its predator populations go extinct. Populations that explode when these islands are cut off from the mainland must lack the factor that limits them on the mainland, often a predator that died out when the island was created. By causing extinctions of different sets of predators on different islands, forest fragmentation provides terrestrial ecologists with their nearest counterparts to the exclusion experiments of the intertidal ecologist. Forest fragmentation can suggest the existence of previously unsuspected keystone species. On newly created 1-ha islands in Lake Guri, Venezuela, leafcutter ant colonies are extraordinarily abundant. These islands average over 2 mature colonies per ha, 20 times their density on the mainland (Terborgh et al. 1997). On these islands, leaf-

social conflict

249

cutter ants severely limit the density and diversity of seedling recruitment (Rao et al. 2001; Terborgh et al. 2001). Something other than food supply, presumably some predator, must limit the abundance of leafcutter ants on the mainland. This predator would, accordingly, be a keystone species protecting forest diversity (Rao 2000). What might this keystone species be? Terborgh et al. (1997) note that on the mainland, army ants, Ecitonini, limit populations of many litter arthropods and forest floor ant species. A 1-ha or even a 10-ha island, however, is too small to maintain even a single colony of army ants. Even though most species of army ant cannot harm an established leafcutter colony (Swartz 1998), they do destroy young colonies (C. Ziegler, pers. comm.). Can they limit the recruitment of new colonies? This question is testable. Herbivore populations can explode on small islands for other reasons. In Venezuela, capybaras are common on certain islands of Lago Guri (Terborgh et al. 1997), perhaps because the nearness of water makes the habitat favorable. Some 1-ha islands in Lago Guri have one or more resident howler monkeys, Alouatta seniculus (Terborgh et al. 1997). Monkey populations normally regulate their numbers socially, through the dispersal of young, to levels that allow a troop to compete effectively with others for food when it is scarce. These islets offer no opportunity for dispersal, but they do protect the monkeys from competing troops. These monkeys only need to survive food shortages: they do not have to be strong enough to repel competing troops. Thus, they maintain unusually high population densities on these islets. In sum, herbivore pressure must be several times higher on the Guri islands than on the surrounding mainland (Terborgh et al. 2001). Does this heavy herbivore pressure favor better defended, slower growing plants? Studies are on foot to find out. Work in another set of forest fragments suggests that forest fragmentation does favor slower growing, more herbivore-resistant plants. Small forest fragments surrounding Malaysia’s Pasoh Reserve are being taken over by plants with low photosynthetic capacity and explosively dispersed seeds that do not travel far from their parents (Thomas 2002). Low photosynthetic capacity presumably reflects low nitrogen content in the leaves (Zotz and Winter 1994), which makes the leaves less attractive to herbivores (Coley 1983). Moreover, poorly dispersed plants, which are likely to grow close to conspecifics and are thus easily found by their pests (Ridley 1930, p. xvi), must therefore be better defended against herbivores (see below). By eliminating other limits on herbivore populations, fragmentation may favor a less productive vegetation.

250

chapter 12

We Must Know How Species Coexist to Protect Biodiversity and Predict the Effects of Extinction Competition favors biodiversity insofar as the “jack of all trades is master of none” (MacArthur 1961)–that is to say, insofar as there are trade-offs between the ability to exploit different foods or habitats (Fisher 1930, p. 126). Among plants, a famous trade-off allows lightdemanding pioneers of clearings and shade-tolerant trees of mature forest to coexist (Pacala and Rees 1998). To learn which management strategies best protect biodiversity, one must know what trade-offs maintain biodiversity. For example, if pioneers and mature forest species coexist because there is a trade-off between the ability to colonize new clearings and the ability to oust competitors (Skellam 1951), then devoting even small patches of forest to agriculture can cause the extinction of those mature forest species whose balance between colonization (recruitment) and mortality is so delicate that it will be overset by wasting some of the seeds on permanently cleared fields (Tilman et al. 1994). On the other hand, if coexistence between pioneers and persistents is driven by the tradeoff between growing fast in high light and surviving in shade (Kitajima 1994; Pacala et al. 1996), limited agricultural disturbance is much less threatening. Pests and pathogens appear to play an integral role in allowing species of tropical tree to coexist (Janzen 1970). First, I review evidence for this proposition, then I show how this tree diversity affects ecosystem properties. Monocultures everywhere are liable to devastating pest outbreaks (Dethier 1976), particularly in the tropics (Ridley 1930). In natural settings, too, pest pressure is much more intense in the tropics. Insect activity is spread much more evenly through the year in the tropics than further north (Wolda 1983). The dry season reduces tropical pest populations less sharply and predictably than winter sets back their high-latitude counterparts. Accordingly, young tropical leaves are eaten much more rapidly, despite being much more poisonous, than are the young dicot leaves of temperate-zone forests (Coley and Barone 1996; Coley and Kursar 1996). Moreover, in most plant species, the greatest damage is inflicted by specialist pests (Barone 1998). In the tropical forest of Barro Colorado Island, Panama, saplings of most tree species survive better and recruit more abundantly per conspecific adult, the greater the proportion of the trees within 10 m, young and adult, belonging to other species (Wills et al. 1997). A plant species appears to suffer most from pest pressure where it is most abundant (Wills and Condit 1999). Therefore, “pest pres-

social conflict

251

sure” is presumed to promote diversity among tropical trees (Ridley 1930; Janzen 1970; Connell 1971). In everwet forests, no dry season lowers pest pressure (Wolda 1983), and the proportion of tree species with animal-dispersed seeds is highest. Ceaseless pest pressure appears to create a high premium on dispersing seeds away from the mother tree and its associated pests. Tree diversity is highest in these everwet forests (Gentry 1982, Leigh 1999), as expected if pest pressure maintains it. How does the role of pests in maintaining tropical tree diversity relate to the properties of tropical forest ecosystems? Cushman (1995), echoing a widespread doubt, questioned whether studies in population biology enhance understanding of ecosystem function. I explore the intellectual bases for concluding that pest pressure maintains tree diversity and that tree diversity is an essential condition for the productivity and luxuriance of most tropical forests. A crucial assumption of Ridley’s (1930) proposal—that pest pressure enhances plant diversity—is that when a species is rare enough, the abundance of a consumer dependent on it declines. Together with its counterpart assumption—that when the consumer species is abundant, abundance of the victim species declines—this assumption is fundamental to the theory of predator-prey cycles (Volterra 1926, 1931; Rosenzweig and MacArthur 1963; Bulmer 1976). Experiments have abundantly verified that the numbers of a consumer dependent on a single victim species decline when the victims are rare enough (see, for example, Gause 1935; Maly 1969, 1978). Another assumption implicit in the pest pressure hypothesis is that, to increase its density, a herbivore-limited species must strengthen its antiherbivore defenses. This proposition seems too obvious for serious attention. Nevertheless, in a simple Volterra model of two victim species limited by the same consumer, the victim species that can support the most consumers without declining prevails. Improved defenses are often favored, but only if the abundance of consumers increases as a result of the increased victim abundance (Appendix 12.1). This theory is good enough to predict that in rivers and streams with two or four trophic levels, primary producers are much rarer than those with one or three trophic levels (Wootton and Power 1993). The theory may also apply to grasslands protected by their grazers from encroaching trees. It makes no sense, however, for forests, which are adapted to reduce the herbivore load. A plant population does benefit from improved antiherbivore defenses if its numbers are regulated in part by plant density. Improved defenses presumably reduce the herbivore’s ability to find its host, the proportion of host plants it can use, or its capacity to digest what it

252

chapter 12

eats. Any of these changes increases the density of host plants required to maintain a specialized herbivore population in equilibrium. Competition with other plants limits the increase a plant population derives from improved antiherbivore defenses. Instead, improved defenses reduce the per capita mortality inflicted by the herbivores (see appendix 12.1). If the defenses are good enough that the herbivores cannot maintain themselves on the density of host plants supported by the environment, the new defenses will put the herbivores out of business. The advantage of rarity as an escape from specialist herbivores and the role of defense in maintaining a host plant’s density in the face of its pests jointly imply that the rarer a plant species, the lower the proportion of its productivity lost to pests specializing on that species, or the lower the proportion of its resources this plant must devote to defense against specialist pests to ensure its survival. Thus, the more diverse a forest, and the rarer each of its component species, the fewer resources need be devoted by this forest’s plants to defense against specialized pests. Since, even in diverse tropical forest, most damage is inflicted by specialist pests (Barone 1998), rarity must allow most plant species to reduce their total investment in antiherbivore defenses. The pest pressure hypothesis has ecosystem implications, because reduced investment in antiherbivore defenses allows faster growth (Coley et al. 1985). There is abundant evidence for Coley’s thesis. Comparisons among plant species in the wild show that saplings of fast-growing species are eaten more and invest less in antiherbivore defenses such as tannins, other phenols, and lignins than saplings of slower growing species (Coley 1988). In the laboratory, Cecropia peltata seedlings with tannin-rich leaves are less vulnerable to armyworms (Spodoptera sp.) but where herbivores are absent, seedlings with tannin-poor leaves produce more foliage (Coley 1986). In a field experiment with Psychotria horizontalis on Barro Colorado, cuttings with tannin-rich leaves were eaten less than low-tannin counterparts when exposed to the full range of the island’s herbivores. Tannin-poor cuttings, however, had faster intrinsic growth, so a cutting’s total weight gain was not correlated with the tannin content of its leaves (Sagers and Coley 1995). Among clones of these cuttings in a nearby, protected enclosure, tannin-poor plants grew faster than their tannin-rich counterparts (Sagers and Coley 1995). Intense antiherbivore defenses not only slow plant growth, they impose other costs. For example, plant species that cannot escape their enemies by being rare are more likely to have defenses that damage the soil. Better defended plants have longer lived leaves that

social conflict

253

are defended by long-lasting compounds such as tannins and other phenols, which break down slowly (Coley 1988). These toxins hinder decomposition and nutrient recycling (Waring and Schlesinger 1985). High concentrations of these toxins in leaves and wood can injure the soil (Whitmore and Burnham 1969; Reich et al. 1995); although the extent of this damage has yet to be quantified. Nonetheless, it is becoming increasingly apparent that the tree diversity created by pest pressure makes a forest more luxuriant and productive (Corner 1964; Leigh 1994, 1999).

How Species Depend on Each Other: Keystone Mutualists Ecosystems are webs of interdependence. We must know these interdependences to assess the consequences of an extinction, since a species whose extinction entails the extinctions of many dependent species is not expendable. Moreover, to preserve a species, we must know those species and habitats on which it depends. Discovering the interdependences that maintain the integrity of tropical forests and other ecological communities presupposes an interest in natural history that is second nature to Paine, but which most research ecologists dismiss as not science. These interdependences are surprising, complex, and varied (Bond 1994b). Near Gothic, Colorado, for example, red-naped sapsuckers excavate cavities in fungus-infected aspens that two species of swallow require as nest sites. They also drill holes in willows, the sap from which supplies abundant nourishment to several bird species and a few mammals (Daily et al. 1993). Moreover, interdependences no one ever thought about can be crucial. Vast oyster reefs in Chesapeake Bay previously filtered the equivalent of one third of the bay’s water every day. Although from 1750 onward, farming was increasing the amount of nitrogen and phosphorus draining into the bay, these oyster reefs kept the bay’s water clear. Mechanical harvest dredges destroyed these oyster reefs between 1875 and 1910. The resulting “ecological extinction” of oysters was followed by sharp increases in phytoplankton, sedimentation of organic matter, incidence of anoxia on the bay bottom, and crashes in the benthic fauna (Jackson et al. 2001). Tropical forests depend on animal pollinators to maintain tree diversity, which in turn preserves their productivity and luxuriance. Plant species that use animals to convey their pollen to conspecifics can survive when made rare by pests; they can accordingly divert resources from anti-herbivore defense to faster growth. Conifers and other wind-pollinated species, however, are reliably pollinated only

254

chapter 12

when conspecifics are nearby. Insofar as wind-pollinated species must be common where they occur they must invest heavily in antiherbivore defense. The natural pesticides in the fallen leaves of windpollinated conifers injure the soil (Waring and Schlesinger 1985). Similarly, seed dispersers help maintain tree diversity in mature tropical forests by allowing large-seeded tree species to disperse far enough from their parents to escape their pests (Leigh 1994; Tiffney and Mazer 1995). Some pollinators and seed dispersers are replaceable. The extinction of others, such as elephants (Alexandre 1978), leads to the extinction of some of the plants they serve. Indeed, the extinction of a few animal species favored by hunters could greatly reduce the diversity of tropical trees by eliminating crucial seed dispersers (Emmons 1989). Moreover, the multiplicity of relationships involving pollinators and seed dispersers ensures that no tropical forest, and certainly no single tree species, is an island sufficient unto itself. Most pollinators and seed dispersers rely on a diversity of plants to maintain themselves. Some depend on access to several different habitats to avoid seasonal shortages in each one: they migrate seasonally from one to another, according to where the resources are (Loiselle and Blake 1991). The mutualism through which each species of fig tree has its own species of pollinating wasp transforms fig trees into keystones for their forests (Corner 1940, Herre 1996). These minute pollinating wasps, each of which matures within the confines of a single fig seed, are truly remarkable. Several fig species maintain extraordinarily high genetic diversity because their trees can attract pollinators from more than 10 km away (Nason et al. 1996, 1998). To maintain these pollinators, however, some trees of each fig species must be coming into fruit all year round. The year-round availability of their fruit makes neotropical figs a keystone resource for all sorts and conditions of animals (Terborgh 1986), including a guild with 10 species of fruit-eating bats (Kalko et al. 1996). Their pollinators allow some fig species to survive when very rare indeed. Perhaps this is why fig trees grow so fast, produce such productive and edible foliage (Zotz et al. 1995), and rot so fast when dead: “By leaf, fruit, and easily rotted wood fig plants supply an abundance of surplus produce” (Corner 1967, p. 24). Indeed, fig trees support many species other than frugivores (Zeh and Zeh 1992a, b; 1994). Indeed, the evolution of pollinators capable of seeking out flowers of a particular species triggered the explosive diversification of angiosperms by allowing the evolution of a great variety of relatively rare, presumably fast-growing, angiosperm weeds (Crepet 1984). These plants occupied scattered openings and disturbed sites (Wing et al.

social conflict

255

1993), crowding out slower growing gymnosperm seedlings. Similarly, “domesticating” mammals and birds to serve as seed dispersers enabled large-seeded angiosperm trees to invade mature coniferous forests, replacing a depauperate, slow-growing vegetation by a fastgrowing, diverse one (Tiffney 1984).

Forest Fragmentation, Mutualism, and the Diversity and Productivity of Tropical Forest Forest fragmentation can reveal the variety of mutualisms that maintains tree diversity and forest productivity. Creating Gatun Lake in central Panama by damming the Chagres River severed many islands from the surrounding mainland by 1914. Islands of less than 1 ha have lost agoutis and other mammals (Adler and Seamon 1991). On those ⬍1 ha islands that have been forested ever since their severance from the mainland, tree diversity has dropped precipitously since 1914 (Leigh et al. 1993). Four tree species are spreading on these islands, all with large seeds. Seeds of three of these species, and latefalling seeds of the fourth, are not attacked by insects. On Barro Colorado, seeds of some other large-seeded species, now absent from or not regenerating on these islands, escape destruction by insects only if buried by agoutis (Smythe 1989; Forget and Milleron 1991). Fragmentation into such small islands is clearly catastrophic for tree diversity. Does this happen because agoutis are keystone animals for maintaining Neotropical tree diversity? This thesis is in urgent need of further testing. Fragmenting tropical forests may reverse the direction of angiosperm evolution. Many pollinators and seed dispersers no longer visit small islands (Cosson et al. 1999), thereby reducing these islands’ plant diversity. In the area surrounding Malaysia’s Pasoh Forest Reserve, fragmentation has favored the tree species that do not need animals to disperse their seeds or whose leaves have low photosynthetic capacity (Thomas 2002). Disrupting mutualisms with pollinators and seed dispersers degrades the diversity and productivity of tropical forest. Natural tropical forest ecosystems are adapted to maintain diversity and productivity. To preserve these qualities, pollinators, seed dispersers, and other mutualists must be protected.

What Invasions by Exotics Reveal about Natural Ecosystems Natural ecosystems that are exposed to repeated invasions become resistant to invasion (Elton 1958), presumably because they no longer

256

chapter 12

offer unused or poorly used sources of energy for invaders to exploit (Leigh 1971). Ecosystems that were invaded rarely in the past, such as oceanic islands, are notoriously susceptible to introduced species (MacDonald et al. 1991). In part, this is because their species were not exposed to effective competition. Exotics are spreading in Mauritius and Reunion, even though native forests, with their strong wood and even canopy, readily resist cyclone winds that devastate the exotics. Apparently, the reproductive capacity of the exotics, whose seeds are often dispersed by introduced birds, far outweighs the superior survival of the natives. An abundance of unexploited resources, or a totally open niche, can make an isolated ecosystem catastrophically vulnerable to invasion. Guam lacked nocturnal carnivores until the brown tree snake was introduced in 1950. This snake has become abundant, causing many extinctions and utterly transforming Guam’s ecosystem in the process (Fritts and Rodda 1998). Diversity in itself provides no immunity against invasion. The astonishing diversity of Lake Victoria’s species flock of haplochromine fishes did not block the initial onrush of predatory Nile perch, but it appears to have provided the fish community sufficient resilience to avoid utter extinction (Goldschmidt 1996). The diversity of South Africa’s fynbos has not protected it from invasion by more productive nitrogen-fixing acacias from Australia (Macdonald 1984; Witkowski 1991). Exotic invaders are relatively rare in mature tropical forests on continents (Rejm´anek 1996). Of the 42 species of exotics listed by Rejm´anek (1996) as successfully invading mature tropical forest, 21 are invading forests on oceanic islands, 14 are invading forests of East Africa, which are both somewhat isolated and rather degraded, and 2 are invading forest fragments in Singapore that are less diverse than the extensive forests elsewhere in Malaysia. Even in the continental tropics, however, land abuse offers footholds for invaders that indirectly threaten natural forest. During the Vietnam War, seeds of the aggressive southeast Asian grass Saccharum spontaneum established in Panama. This grass has occupied untended, open places, where annual fires late in the dry season have transformed it into a low-diversity “dysclimax” inhospitable to most wildlife. This grass now occupies fields that formerly would have been colonized by pioneer trees and undergone forest succession (Dalling and Denslow 1998). It takes great labor, and some luck, to replace these grasslands by forest. Open areas elsewhere in the world have also been occupied by aggressive grasses whose annual fires exclude trees (D’Antonio and Vitousek 1992). These grasses transform ecosystem restoration from something that happens naturally to an achievement demanding abundant human labor.

social conflict

257

Can extinctions open ecological communities to destructive takeovers by aggressive species? In Indonesia, the grass Imperata cylindrica has taken over thousands of hectares of misused land, forming sterile monospecific grassland (Jacobs 1988). In Nepal’s Royal Bardia National Park, however, deer, rhinoceros, and elephant maintain diverse, nutritious grazing lawns in the midst of a grassland dominated by Imperata cylindica but including other grasses (Karki et al. 2000). These grazing lawns include I. cylindrica and Saccharum spontaneum, the aggressive pest grasses of Indonesia and central Panama. In Indonesia, the extinction of large herbivores has helped transform I. cylindrica from a member of a diverse ecosystem into a pest.

Concluding Remarks This essay has two aims. One is to remind biologists that social conflict, whether latent or blatant, is the biggest single obstacle to sound conservation. Suppose for a moment that we were Laplacean demons, able to assess the ecological consequences of each and every extinction. Would this circumstance really cause society to close ranks behind conservation? As the world seems organized to destroy any sense of community among human beings, a greater knowledge of biology is not likely to make it easier for society to agree on conservation policy. The essay’s second aim is to show that, like any other aspect of conservation planning, deciding which species, if any, are expendable depends on understanding the ecological organization of natural communities and how the balance of nature works (Ehrlich 1994). This is a tall order: can we learn enough to assess the effects of eliminating a species? Just as field experiments were crucial tools for understanding the ecological organization of rocky intertidal communities and identifying some of its keystone species, so the pseudoexperimental approach of studying the effects of forest fragmentation is revealing the ecological organization of intact tropical forests and identifying keystone species essential to their integrity. An interest in natural history is crucial to the assessment of expendability. Natural history is unpopular nowadays. The student who showed that some tree species need agoutis to bury their seeds out of reach of insect pests (Forget 1994) came to Barro Colorado only because political events in Panama frightened away higher ranking applicants. In ecology, the “stone the builders rejected” often “becomes the head of the corner” (Ps 118, v. 22). In other words, preconceived

258

chapter 12

notions of what is important often prove to be deceptive. Whether we can learn enough natural history to predict the effects of most extinctions is, however, an open question.

Appendix to Chapter 12 Let N1 and N2 be the population densities of two species, 1 and 2, limited by the same species of consumer, whose density is C. In the simple theory of Volterra (1931), dN1 dt dN2 dt

⳱ r1N1 ⳮ a1C N1 C ⳱ a1C N1 ⳱ r2N2 ⳮ a2C N2 C ⳱ a2C N2

( (

r1 ⳮC a1C

) )

r2 ⳮC, a2C

where ri is the logarithmic rate of increase of species i in the absence of consumers, aiCNiNC is the rate at which the consumer species eats victims of species i, and ri/aiC is the number of consumers victim species i can sustainably support. The consumer’s abundance increases to the level the more tolerant victim species can support (i.e., the larger of the two quantities r1/a1C and r2/a2C). The victim with smaller ri/aiC is forced to decline to extinction as the consumer’s abundance increases beyond what it can sustainably support. A victim species benefits from improved antiherbivore defenses that reduce aiC, only if this change increases the number of consumers it can support. Now assume that the victims are limited in part by their own density. Suppose that only the first victim species is present, and that dlnN1 dt dlnC dt

⳱ r1 ⳮ b1 N1 ⳮ a1CC

⳱ ␭a1C N1 ⳮ m,

where b1 represents the impact of a unit increase of victim density on its per capita rate of increase, ␭ represents the increase in consumer population per unit of victim consumed, and m is the per capita death rate of consumers in the absence of victims. If there are no consumers, N1 equilibrates at r1/b1. On the other hand, if consumers are present, their population just maintains itself (i.e., dlnC/dt ⳱ 0) when N1 ⳱

social conflict

259

m/␭a1C). The smaller ␭a1C (i.e., the less effectively the consumer finds suitable victims or the less effectively it digests them), the larger the density N1 of victims required to sustain the consumer population in equilibrium, and therefore the smaller the per capita mortality a1C C ⳱ r1 –b1N1 inflicted on victims by consumers. If antiherbivore defenses are so effective that m/␭a1C ⬎ r1/b1, competition limits victims at a level too low to support the consumer species.

Chapter 13

m Which Mutualists Are Most Essential? Buffering of Plant Reproduction against the Extinction of Pollinators William F. Morris

By ensuring that plant reproduction occurs, animal pollinators act as essential components of both agricultural and natural ecosystems; hence, human societies depend indirectly on pollinators for both food and “ecosystem services” (Nabhan and Buchmann 1997). To the extent that such services can even be quantified, one estimate puts the value of the benefits humans receive from biotic pollination at about $120 billion annually (Costanza et al. 1997). The magnitude of the dependence of humans and other species on pollinator activities raises an important question: what aspects of the pollinator community are key to its ability to maintain plant populations? One such aspect is the number of pollinator species that visit the flowers of a given plant species at a given location. Although we are all familiar with textbook examples of plants (e.g., yuccas and figs) that depend for reproduction on a single species of pollinator, the vast majority of

plants and pollinators

261

plants have multiple floral visitors (Howe 1984; Herrera 1996; Waser et al. 1996). From the perspective of the plant population, are all of these multiple visitor species essential to ensure adequate reproduction? To put the question another way, how well are plant populations “buffered” against the loss of species from the pollinator community? In fact, we can ask a more specific question: are certain kinds of floral visitors more essential than others in maintaining populations of plants that have generalized pollination systems? Typically, different species of visitors to a single plant species vary widely in both their abundance and their ability to effect fertilization (Schemske 1983; Schemske and Horvitz 1984; Herrera 1987, 1988). Here, I refer to effectiveness as the amount of “service” provided by an individual visitor of a particular taxon during a single visit to a previously unvisited flower. In empirical studies, pollinator service has been measured in a variety of currencies, including the number of pollen grains deposited on stigmas or removed from anthers, the probability of fruit set, the number of seeds fertilized, and indirect components of fertilization such as stigma tripping in flowers that have specialized morphologies. Such studies have often found that there is no relationship, or even a negative relationship, between the frequency of visits by a particular taxon and its per-visit pollination effectiveness (Schemske and Horvitz 1984; Herrera 1996). This observation raises the question: are plant populations typically more buffered against the loss of less frequent visitors or the loss of less effective visitors? Knowing whether visitation rate or effectiveness more frequently identifies essential visitors might help us design better monitoring and management strategies for rare plants whose pollination biology is poorly known. For one thing, if it is the former then we may be able to get away with simple counts of visitors rather than having to perform the detailed experimental studies needed to measure effectiveness. Another common observation in the pollination literature is that the frequency of visits by a particular visitor taxon varies widely from year to year at a single location (Herrera 1988; Schemske and Horvitz 1989; Pettersson 1991). Consider a plant that has two equally effective visitor species with the same mean abundance, one of which is “reliable” from year to year and the other highly variable (throughout, I use the term “reliable” to label taxa whose visitation frequency varies little over time, rather than “constant,” which in the pollination biology literature refers to visitors that are likely to move to another plant of the same species). We would expect that the more reliable visitors would be more essential for the plant population, because if they were eliminated, the remaining visitors might be so variable that reproductive failure could occur in some years. Those years might have

262

chapter 13

a devastating effect on population persistence if the plant is an annual or a short-lived perennial and has a low rate of selfing. Hence, another important question is whether reliability is a better indicator of essential visitors than is visitation frequency or pollination effectiveness. Thus, we would predict that the value of the services provided to a plant population by a potentially mutualistic visitor taxon should be enhanced by high visitation frequency, high per-visit effectiveness, and high reliability. To assess which of these features best identifies the most essential visitors, I compiled data from studies that measured, for one or more years, the abundance and effectiveness of multiple visitors to a single plant species at a single location. In the following sections, I describe these data and then present the results of a graphical approach used to assess the consequences of deleting taxa from the visitor pool in order of increasing abundance, effectiveness, or reliability. I discuss the implications of these results for plant conservation and also review important caveats of the approach I have adopted.

Methods Description of the Data I found 24 data sets that included quantitative estimates of visitation frequency and the effectiveness of multiple visitors to a single plant species (table 13.1). I was interested in patterns across multiple visitors, so I only included studies that provided both types of data for three or more visitor taxa (range: 3–26 taxa). Most of the plants in table 13.1 are wild species, although two (Phaseolus coccineus and Helianthus annuus) are cultivated crops. Because Schemske and Horvitz (1984, 1989) reported visitation rates for all visitor taxa in 1983 but quantified variation in visitation by hymenopteran visitors over 4 years, I separated all taxa in 1983 and hymenopterans over all years into separate data sets. Although I would have liked to account for all species of visitors, this was often impossible because the original studies lumped visitors into higher taxonomic categories (e.g., “Bombus spp.” or “Hymenoptera”). My focus in this chapter is the contributions different visitor taxa make to the maintenance of plant populations, not to the fitness of individual plants. Consequently, I was interested primarily in studies that quantified aspects of seed production rather than male reproductive success. Most of the studies in table 13.1 measured visitor effectiveness in term of direct correlates of female reproductive success, such as the number of pollen grains deposited onto stigmas, or fruit

plants and pollinators

263

or seed set resulting from single visits to virgin flowers; however, two studies (Montalvo and Ackerman 1986; Davis 1987) used more indirect measures. For comparison, I also included the only two data sets I could find (Thøstesen and Olesen 1996; Fishbein and Venable 1996) that measured effectiveness in terms of components of male reproductive success (pollen and pollinia removal, respectively). I used the mean result (e.g., mean number of pollen grains deposited) as the index of effectiveness, but because effectiveness typically varies considerably among visits, I discarded visitor taxa whose effectiveness had been calculated on the basis of fewer than five visits. Visitation frequency was usually measured by calculating the mean number of individuals arriving at a focal plant or patch of plants over repeated observation periods. For seven data sets in which no such observations were performed (see table 13.1), I used the sample sizes for the effectiveness measurements as an (admittedly imperfect) index of relative abundance; I include a discussion of the potential consequences of doing so.

A Graphical Tool for Identifying the Most Essential Pollinators I now present a simple graphical method that I use to assess visitor contributions and to evaluate the extent of pollination buffering in the empirical data sets (for a similar approach used to quantify the effect on cumulative biodiversity of adding habitat reserves to a conservation portfolio, see Quinn and Harrison 1988). I start by calculating the total pollination service provided to the plant by the entire visitor pool. Total pollination service is the product of the visitation frequency of each visitor taxon multiplied by its pollination effectiveness, summed over all taxa. Paine (1992) used a similar measure to gauge the possible community-level effects of a suite of consumers of a competitively dominant species. If effectiveness is measured as the number of pollen grains deposited per visit, then total pollination service is proportional to the total number of pollen grains an average plant receives from all its visitors over the flowering period; if effectiveness is measured as fruit set per visit, total service is an indication of the plant’s maximum potential fruit set. A plant’s actual reproductive success is not necessarily related linearly to total pollination service. Instead, resources may limit reproduction at a level lower than that set by pollination alone, as many studies have shown (for reviews, see Young and Young 1992; Kearns and Inouye 1993; Burd 1994). However, even if total pollination service exceeds actual reproduction, it nevertheless serves as a useful measure of the degree of

Davis 1987 Young 1988 Arizmendi et al. 1996 Montalvo and Ackerman 1986 Thøstesen and Olesen 1996 Parker 1981 Snow and Roubik 1987

THGE DILO FUMI SPFR

HEAN CARE

ACSE

Kendall and Smith 1976 Bertin 1982 Snow and Roubik 1987

Source

PHCO CARA COVI

Data Set

TABLE 13.1 Empirical studies used in figure 13.2

Helianthus annuus Cassia reticulata

Phaseolus coccineus Campsis radicans Cochlospermum vitifolium Thalia geniculata Dieffenbachia longispatha Fuchsia microphylla Spathiphyllum friedrichsthalii Aconitum septentrionale

Plant Species

5 5

4

3 3 3 4

3 3 3

Number of Visitor Taxa

1 1c

1c

3 3 1 1

2b,c 1 1c

Years of Visitation Frequency Data

Seed set Pollen deposition

Stigma tripping Fruit set Seed set Rank correlation: fruit set vs. visits Pollen removal

Fruit set Pollen deposition Pollen deposition

Measure of Single-Visit Pollination Effectiveness

0.175 0.718

0.4

1 ⳮ1 0.5 0.8

ⳮ1 ⳮ0.5 ⳮ0.5

Rank Correlation between Visitation Frequency and Pollination Effectivenessa

Ashman and Stanton 1991 Keys et al. 1995 Arizmendi et al. 1996 Schemske and Horvitz 1984, 1989 Devall and Thien 1989 Thompson and Pellmyr 1992 Gomez and Zamora 1999 Kearns and Inouye 1994 Fishbein and Venable 1996 Fishbein and Venable 1996 Olsen 1997 Schemske and Horvitz 1984 Pettersson 1991 Herrera 1987, 1988 Heterotheca subaxillaris Calathea ovandensis (All visitors in 1983) Silene vulgaris Lavandula latifolia

Asclepias tuberosa

Linum lewisii Asclepias tuberosa

Hormathophylla spinosa

Prosopis velutina Salvia mexicana Calathea ovandensis (Hymenoptera only) Ipomoea pescaprae Lithophragma parviflorum

Sidalcea oregana

11 26

8 9

7

7 7

6

6 6

5 5 6

15

4 6

1 1

2

1c 2

4b

1c 2

1c 1 4

1

Pollen deposition Pollen deposition

Fruit set Fruit set

Pollinia insertion

Pollen deposition Pollinia removal

Pollen deposition

Fruit set Seed set

Fruit set Seed set Fruit set

Pollen deposition

b

Calculated using the arithmetic mean visitation frequency for studies that lasted more than one year. Significantly more buffering than expected at random. c Sample sizes in effectiveness measurements were used as an index of visitation frequency (see text).

a

SIVU LALA

HESU CAOV

ASTU-I

LILE ASTU-R

HOSP

IPPE LIPA

PRVE SAME CAOV-H

SIOR

0.482 ⳮ0.152

0.145 0.119

0.505

ⳮ0.029 0.180

ⳮ0.829

ⳮ0.2 0.771

0.0 0.3 ⳮ0.429

0.3

266

chapter 13

Figure 13.1 Three basic patterns for the effect of visitor deletions on total pollination service. If visitors are deleted at random, the average total pollination service declines linearly (curve A). Curve B implies that taxa early in the deletion sequence are relatively nonessential and that total pollination is “buffered” against those deletions, whereas curve C implies that early-deleted taxa are essential. To facilitate comparison among studies, total pollination service is calculated relative to its value when all visitor taxa are present (hence a maximum of 1 for the y-axis).

“safety” built into the pollination process itself, which is the main focus of our attention here. Moreover, as total pollination service declines, reproduction, at least for predominantly outcrossing species, must eventually become pollen limited. I proceed by deleting species iteratively from the visitor pool, each time recalculating the total pollination service provided by the remaining visitors and graphing the result against the number of species deleted. Three patterns are particularly informative (fig. 13.1). First, if n species are deleted at random, then total pollination service averaged over all possible combinations of n deletions declines linearly with n (see fig. 13.1A); each increment in n results in the subtraction of one average per-taxon contribution from the total pollination service. Thus, if a particular sequence of visitor deletions results in a linear decline in total pollination service, each visitor taxon contributes roughly equally to the total pollination such that one is no more (or less) essential than another. Alternatively, total pollination service might be relatively insensitive to the deletion of the first few species from the visitor pool, followed by a sharp decline as deletions approach the total number of taxa in the pool (see fig. 13.1B). This

plants and pollinators

267

pattern indicates that the pollination service received by the plant is to some extent “buffered” against the loss of those visitor taxa that appear early in the deletion sequence, and that those visitors are less essential from the plant’s perspective. Finally, the opposite pattern—a sharp decline in total pollination early in the deletion sequence (see fig. 13.1C)—implies that taxa involved in the initial deletions are valuable contributors to plant reproduction. Using this graphical approach, I compared different scenarios of visitor deletion to the random deletion case. For all data sets, I deleted visitor taxa according to two different sequences: (1) from the least frequent to the most frequent visitor, and (2) from the least effective to the most effective visitor. Rather than simply compare the curves visually, I used the empirical data to perform randomization tests to determine whether pollination is significantly more buffered against the deletion of visitors in a particular order than would be expected if deletions were performed in random order. Buffering means that there is little reduction in total pollination early in the deletion sequence. Therefore, the more buffered the total pollination, the greater the area under the curve of total pollination versus the number of taxa deleted (see fig. 13.1). To test for significant buffering, I performed the following test. I deleted visitors in random order and calculated the area under the curve. I then asked whether this area was greater than the observed area under the curve for a particular deletion sequence of interest (e.g., deletion in order of increasing visitation frequency). I repeated this process many times (typically 10,000) and calculated the fraction of trials in which random deletion led to an area greater than the observed area. If this fraction was small (e.g., less than 0.05), I determined that it was unlikely that the observed degree of buffering would occur at random.

Assessing Whether Reliable Visitors Are More Essential Most pollination studies shown in table 13.1 were of short duration; only eight presented visitation data for more than 1 year, and three of these lasted only 2 years. Despite the sparsity of long-term studies, I used the three studies with data from 4 or more years (Herrera 1988; Schemske and Horvitz 1989; Pettersson 1991) to perform a preliminary assessment of how well a visitor’s reliability predicts its value to a plant population. To quantify reliability, I calculated the coefficient of variation (standard deviation divided by the mean) in visitation frequency across years, and ranked taxa with the lowest coefficients

268

chapter 13

of variation as the most reliable. My intent in using the coefficient of variation was to obtain a reliability index that was independent of the mean visitation frequency; as I show later, this did not always succeed. The question “are more reliable visitors more valuable to a plant population?” is worth asking only if our measure of the service a plant receives from a particular visitor taxon assigns different weights to years in which that visitor is rare versus abundant. If I were to calculate pollination service as a function of the arithmetic mean visitation frequency across years, those visitors that are more frequent on average would always be identified as the most valuable, regardless of their reliability (assuming equal effectiveness). Instead, we need to account for the fact that the detrimental effect of years in which a visitor taxon is scarce or absent may be greater than the beneficial effect of years in which that taxon is abundant. For example, a single year without visitation by a particular taxon would spell extinction for populations of a self-incompatible annual plant with no seed bank, were that the only visitor taxon. To give greater weight to years of low visitation frequency, I chose to calculate each taxon’s pollination service as the product of its effectiveness and its geometric mean visitation frequency over time, rather than the arithmetic mean. The geometric mean, and hence overall pollination service, is zero for any taxon that fails to visit the plant in one or more years. Not assigning any value to the service provided in some years by visitors that are periodically absent is probably too extreme, at least for perennial plants that can ride out years of no reproduction. A better measure of pollination service for perennial plants or annuals with a seed bank is the long-term growth rate of a population when visited only by a single pollinator, given its estimated variance in abundance (that is, one might adopt a stochastic variant of the projection matrix methods used by Ehrl´en and Eriksson [1995] and Parker [1997]). Such a measure would require information about other demographic processes (e.g., survival and growth of aboveground individuals and dynamics of the seed bank) that is unavailable for most of the species examined here; hence, I opted for the simpler geometric mean approach. Because the geometric mean is an increasing function of the arithmetic mean visitation frequency and a decreasing function of the temporal variance in visitation, a measure of pollination service based on the geometric mean visitation allows a meaningful comparison of the relative effects of average visitation frequency and reliability. I used the same randomization test described earlier to test for significant buffering against the deletion of visitors in order of average visitation frequency, effectiveness, and reliability. I calculated pollination service

plants and pollinators

269

using the arithmetic mean visitation frequency when comparing these multiyear studies to those that lasted only one or a few years.

Results Table 13.1 shows the Spearman rank correlation coefficients between visitation frequency and pollination effectiveness for all visitor taxa in each data set. None of these correlations differs significantly from zero, although the sample sizes (i.e., numbers of taxa) are admittedly small in most cases. But even with seven or more taxa, most of the correlations are not far from zero. Thus, there is no simple “rule-ofthumb” relationship between visitation frequency and effectiveness that applies across data sets, and it is clearly worth asking which factor better predicts the overall pollination service a visitor taxon provides to a plant population. The curves obtained by deleting visitors in sequence are shown in figure 13.2, in which the graphs are ordered by the number of taxa in the visitor pool. First, consider the effect of deleting visitors from the least frequent to the most frequent (solid lines). Although a few of the graphs (e.g., DILO, COVI; see data set abbreviations in table 13.1) resemble the random deletions case (curve A in fig. 13.1), the most common pattern is more similar to curve B, especially when the number of taxa in the pool is large. In fact, for 15 of the 24 data sets, deletion of the least frequent half of the visitor taxa results in a decline in total pollination service of only 20% or less relative to the intact visitor community. In 13 of the 18 data sets that included 4 or more visitor taxa, total pollination was significantly more buffered against the deletion of visitors in order of frequency than against random deletion (table 13.2). Among these 13 data sets were the two data sets that measured effectiveness in terms of male reproductive success (ACSE and ASTU-R) and five of the data sets that used sample size as an index of visitation frequency (see table 13.2). In summary, the majority of plant species in the analysis can tolerate the loss of a substantial fraction of their less common visitors with little adverse effect on pollination. In about half of the data sets, deletions performed in order from the least to the most effective (dotted lines in fig. 13.2) show a pattern quite similar to deletions in order of increasing visitation. These data sets also tend to be those cases in which the most frequent visitors are also the most effective (see rank correlations in table 13.1), as would be expected. When the correlation is negative (i.e., the rarer visitors tend to be more effective), the most common pattern is for pollination

270

chapter 13

Figure 13.2 Decline in total pollination service as taxa are deleted from the visitor pool. Deletions were performed in order of increasing visitation frequency (solid lines) or increasing per-visit effectiveness (dotted lines). Lines overlap for THGE, SPFR, and PRVE. Species codes as in table 13.1.

plants and pollinators

Figure 13.2 continued

271

272

chapter 13

Figure 13.2 continued

to be more buffered against the loss of less frequent visitors than against the loss of less effective visitors (see table 13.2). Overall, significant buffering against deletion by effectiveness occurred less often (6 of 18 data sets with 4 or more taxa) than for deletion by frequency, and only 1 case (SIOR) showed the former but not the latter. The difference in the proportion of data sets showing significant buffering between the deletion by frequency and deletion by effectiveness scenarios was itself statistically significant (G-test; likelihood ratio ChiSquare ⳱ 5.61, p ⳱ 0.018). These results imply that when there is a

plants and pollinators

273

TABLE 13.2 Randomization tests for buffering in empirical data sets See Methods section for a description of the test. Asterisks indicate significantly more buffering than expected at random. Because there are only 6 possible random deletion sequences with 3 visitor taxa, randomization tests were not performed for the first 6 data sets in Table 13.1. Fraction of Random Removals with Greater Area Compared to:

Data Set

No. of Randomizations Performed

Removal According to Visitation Frequency

Removal According to Pollination Effectiveness

SPFR ACSEa HEAN CAREa SIOR PRVE SAME CAOV-H IPPEa LIPA HOSP LILEa ASTU-R ASTU-I HESU CAOV SIVUa LALA

1000 1000 1000 1000 1000 1000 1000 10,000 10,000 10,000 10,000 10,000 10,000 10,000 10,000 10,000 10,000 10,000

0.042* 0.000* 0.036* 0.013* 0.107 0.066 0.054 0.0529 0.0059* 0.0011* 0.0000* 0.0189* 0.0075* 0.0000* 0.0003* 0.0860 0.0006* 0.0000*

0.042* 0.497 0.114 0.065 0.015* 0.138 0.079 0.3565 0.3043 0.0137* 0.8996 0.1487 0.0444* 0.1390 0.1956 0.0727 0.0058* 0.0100*

a Sample sizes in effectiveness measurements were used as an index of visitation frequency (see text). *Significantly more buffering than expected at random.

tradeoff between visitation frequency and effectiveness, the more common visitor taxa are usually more valuable to a plant population than are the more effective taxa. For the subset of studies that quantified variability in visitation frequency over 4 or more years, deleting visitors in order of increasing average (i.e. arithmetic mean) visitation frequency yielded strongly significant buffering in all three cases (table 13.3, fig. 13.3). The data set CAOV-H also showed significant buffering against deletion of vis-

274

chapter 13

TABLE 13.3 Randomization tests for buffering in the three data sets for which visitation frequency was measured for 4 or more years See Methods section for a description of the test. Asterisks indicate significantly more buffering than expected at random. In all cases, 10,000 randomizations were performed. Fraction of Random Removals with Greater Area Compared to:

Data Set

Removal According to Mean Visitation Frequency

Removal According to Pollination Effectiveness

Removal According to Visitation Reliabilitya

CAOV-H SIVU LALA

0.0077 0.0002 0.0001

0.4206 0.1489 0.0175

.0326 0.1583 0.0711

a

Inverse of the coefficient of variation.

itors in order of increasing reliability (table 13.3), and the curves for the two deletion scenarios are very similar (see fig. 13.3). Not surprisingly, the rank correlation between visitation frequency and reliability in this case is quite positive (0.714), making it difficult to assess which factor is more closely associated with the value of a visitor’s service. Neither of the other data sets showed significant buffering against deletion by reliability, and only one of the three data sets exhibited buffering against deletion by effectiveness. Thus, although few studies were carried out long enough to give even a preliminary assessment of visitor reliability, the results here are consonant with those reported for the shorter term studies: namely, that average visitation frequency serves to identify essential visitors more often than does either effectiveness or reliability.

Discussion Buffering of Pollination against the Loss of Different Types of Visitors Despite much recent alarm about “forgotten pollinators” (Buchmann and Nabhan 1996) and the “impending pollination crisis” (Nabhan and Buchmann 1997), the analysis I have presented here suggests that populations of many plant species may be able to withstand the loss of a substantial fraction of the animal taxa visiting their flowers with-

plants and pollinators

275

Figure 13.3 Effects on total pollination service of deleting taxa in order of increasing mean visitation frequency (solid lines), increasing pollination effectiveness (dotted lines), or increasing reliability of visitation (dashed lines). Pollination service was calculated using the geometric mean visitation frequency across years; see methods section of text.

out suffering a large decrease in pollination service. Low visitation rate to a particular plant species may indicate visitor taxa with low population densities (but this need not necessarily be so; see later). If it does, those taxa may be particularly prone to local extinction driven by habitat fragmentation, and we would have good reason to expect that visitors would be lost from the pollinator community roughly in order of increasing visitation frequency. Thus, it is interesting that in the data sets I examined, total pollination is especially well buffered

276

chapter 13

Figure 13.4 For many data sets in table 13.1, most visitor taxa contribute proportionately little to total visitation. To ease comparison, visitation frequencies are scaled to the most common visitor, and data sets are grouped by the number of visitor taxa.

against the deletion-by-frequency scenario. Less frequent visitor taxa are less essential than more frequent ones because their absolute visitation rate is proportionally even smaller than their visitation rank would suggest. Specifically, visitation rate falls off exponentially with rank, most notably in data sets with higher numbers of visitor taxa (fig. 13.4). Such log-normal abundance curves have been pointed out in other pollinator assemblages by Herrera (1989). The plants in this analysis thus appear to be typical in that a large proportion of their multiple visitor taxa have quite low visitation rates and are therefore likely to contribute little to total pollination. Interestingly, visitation frequency was generally a better predictor of a visitor taxon’s value to a plant population than its per-visit effectiveness. Hence, for frequent visitors sheer numbers of individuals or high numbers of flowers visited per individual per unit time may

plants and pollinators

277

often more than compensate for a pollinator’s relative “ineptness” at each visit. Herrera (1987) has argued that effective pollinators may simply be those that are large and therefore carry more pollen, and that this may be the underlying reason for a tradeoff between effectiveness and visitation frequency (because larger animals are likely to have lower population densities). However, the weak and positive rank correlations in table 13.1 suggest that such a tradeoff is not inevitable. A second potential reason why some visitors are more effective is that they may have morphological or behavioral features that bring them into greater contact with reproductive structures while visiting a flower. Close correspondence between the traits of flowers and their visitors may be more likely to occur for visitor taxa that are specialists on a particular plant species or group of related species. The idea that such specialist visitors should be the primary targets of management efforts aimed at sustaining plant reproduction has intuitive appeal. However, this intuition may be incorrect. It is difficult to determine from the original studies whether effectiveness and specialization are correlated in the pollination systems examined here. If they are, however, the results in figure 13.2 imply that a taxon’s visitation frequency may be a more appropriate indicator of its management priority (from the plant’s perspective) than its degree of specialization per se. In the beginning of the chapter, I argued that, all else being equal, more reliable visitors should be more valuable. However, only one of the three longest term data sets in table 13.1 (CAOV-H) showed significant buffering against the deletion of increasingly reliable visitors. For this data set, there was a strongly positive rank correlation between reliability and average visitation frequency. That is, use of the coefficient of variation in the reliability index was not sufficient to disentangle variation in visitation from the mean visitation, making it difficult to ask which independent factor better identifies essential visitors. Infrequent visitors contribute little pollination service on average, but because their abundances are proportionately more variable, they are also more likely to be absent in some years, causing their geometric mean visitation frequency—and hence their contribution to total pollination service—to be zero. For the other two longterm data sets, the rank correlations between average visitation frequency and reliability were also positive, but less strongly so (0.1 and 0.376 for SIVU and LALA, respectively). In these cases, reliability helped little to identify essential visitors, in contrast with visitation frequency (see table 13.3). Obviously, the conclusion that average visitation frequency is a better indicator of the most essential visitors than is reliability must be viewed as highly tentative until more studies have measured variation in visitation over multiple years.

278

chapter 13

Caveats In interpreting the results I have presented here, four caveats must be kept in mind. First, because visitor species were often lumped into higher taxonomic categories in the original studies, both visitation frequency and effectiveness may vary among the species that were combined (Herrera 1987). Accounting for such interspecific differences might alter the relative effects of deleting species in order of visitation frequency versus effectiveness. However, the species that were most likely to be lumped were the rarer visitors. Therefore, splitting up multispecies taxa would most likely result in even more classes of rare visitors whose contribution to total pollination is minor. Thus, the conclusion that common visitors are more essential may be robust to the lumping of visitor species. The second caveat is that the sample size in effectiveness measurements may be a biased estimator of visitation frequency. It seems likely that investigators attempting to compare the effectiveness of multiple visitors would undersample the most common visitors to achieve a more even representation of rarer visitors in the data set. As a result, the sample sizes for common visitors are likely to underestimate their true relative visitation frequency. However, this bias is once again likely to render conservative the conclusion that infrequent visitors are less essential, because the relative contribution of more frequent taxa is likely to be even greater than my estimate of total pollination service would indicate. The third caveat is that total pollination service (e.g., the number of pollen grains deposited or number of fruits sets) may not capture differences among visitors in the “quality” of their pollination service (Schemske 1983; Herrera 1987). For example, two visitors may deposit equal amounts of pollen yet differ in the fraction of self-pollen that they carry, and thus have different effects on the outcrossing rate, seed quality, and realized plant reproduction. Detailed information on the quality of pollination service provided by multiple taxa visiting a single plant species is collected only rarely but would allow a more accurate assessment of pollination effectiveness. The final, and most important, caveat is that my assessment of the effects of visitor deletions does not account for changes in the visitation rate or effectiveness of the remaining visitor species. If visitor taxa compete with one another, the elimination of one visitor may result in a compensatory increase in visitation by one or more of the remaining species, due to numerical or behavioral responses. In addition, the pollen that would have been collected by the deleted visitor

plants and pollinators

279

would remain available to other visitors, possibly altering average pollen loads and thereby the per-visit effectiveness of the remaining taxa. These changes might increase or decrease the summed contribution of the remaining taxa to the total pollination service (which my method assumes does not change). If changes in visitation or pollen loads result in more of the pollen being carried by visitors that deposit onto stigmas a high proportion of the pollen they acquire, the summed contribution of remaining taxa would be expected to increase. If so, the decline in total pollination following the deletion of infrequent visitors would be even less than implied by the curves in figure 13.2, again supporting the notion that frequent visitors are more essential. On the other hand, deletion of a visitor may result in more pollen being acquired by competing visitors that deposit little of the pollen they collect (Wilson and Thomson 1991; Thomson and Thomson 1992). The loss of that pollen from the plant population might actually cause the summed pollination contribution of the remaining taxa to decrease. In this sense, infrequent visitors might provide an important indirect service to the plant population by inhibiting more frequent but less mutualistic visitors. Moreover, environmental impacts such as habitat fragmentation may result in both the wholesale elimination of some visitor taxa and changes (either increases or decreases) in the densities (and therefore visitation frequencies) of the remaining taxa (Aizen and Feinsinger 1994a, b; Kearns et al. 1998). Our currently poor understanding of competitive interactions among floral visitors and of asymmetries between the acquisition and deposition of pollen make it difficult to predict whether these positive and negative compensatory effects of visitor deletion on the residual pollination service will balance out. The assumption I have made here— that they do balance out—can be seen as a null expectation that, in the absence of more detailed information, provides some initial insight into which kind of visitors are likely to be the most essential to plant reproduction.

Conclusions and Conservation Implications In comparing across 24 data sets, I found few cases in which the multiple visitors to a plant’s flowers contribute equally to the total pollination service (as would be indicated by curve A in figure 13.1). Thus, from the perspective of a single rare plant species, some visitors may have more conservation “value” than others. In this review, the most frequent visitors were typically identified as the most important contributors to plant reproduction. This result suggests two implica-

280

chapter 13

tions for plant conservation. First, in the absence of detailed knowledge of a rare plant’s pollination ecology, a sensible default strategy to maintain viable populations of the plant may be to ensure that the most common visitor taxa are retained in the community. Of course, there will likely be cases in which more effective or more reliable visitors contribute more to plant reproduction than do more frequent visitors, but it is striking that in only one of the studies examined here did a factor other than visitation frequency show a significantly stronger statistical association with total pollination (SIOR, see table 13.2). Second, it may also be important to ensure that even if the most frequent visitor taxa are retained, their densities do not undergo substantial declines. This review also highlights the need for a better understanding of competitive relationships among multiple visitors to a single plant species, and the compensatory changes in pollination that may result from visitor deletion.

Acknowledgments Thanks to J. Bronstein for ideas, comments, and directions to useful data sets, and to C. Damiani, K. Feeley, T. Feldman, B. Hudgens, J. Stinchcombe, J. Thomson, and an anonymous reviewer for helpful suggestions on the draft. This work was supported by National Science Foundation grants DEB-9509563 and DEB-9806818.

Chapter 14

m The Expendability of Species: A Test Case Based on the Caterpillars on Goldenrods Richard B. Root

Any decision that would cause the extinction of a species must be judged on both ethical and scientific grounds. In this chapter I consider only the scientific issues that are raised by such a decision. Specifically, I examine the issue of the expendability of species by taking a well-studied natural community and asking, “Does it contain any species that are functionally so insignificant that there would be no appreciable effects if the species were exterminated?” In other words, are the effects of such a loss so small that they cannot be distinguished from the normal variance in the properties of the community? The protocol I suggest for addressing this question starts with compiling a list of the traits that a species must possess if it is truly expendable. This list then forms the basis for identifying species that provide us with a “best case” for the existence of expendable species. If the candidate species passes the listed criteria (shown in italics

282

chapter 14

throughout), the evaluation moves on to take up further concerns, such as likely responses to rare events and changes in the future status of other species. To maintain the focus of this inquiry, I restrict my discussion to the expendability of herbivorous insects and to situations in which the subject species is likely to become globally extinct.

Traits that Contribute to Functional Insignificance in Herbivorous Insects Species deemed to be inconsequential have a low impact on the production and fitness of their host plants because they are: (1) rare throughout their geographic range, (2) distributed narrowly so that they inhabit only a small portion of the hosts’ habitats, (3) small so that they have limited requirements. Furthermore, species that have low impact rarely cause extensive collateral damage because they have thrifty feeding habits; they do not induce extensive lesions, vector harmful pathogens, or attack valuable tissues during a critical stage of development. These same traits also serve to reduce the impacts of an herbivore on its competitors, mutualists, and enemies. A species that is of little consequence does not provide the sole resource that sustains other species during periods of dearth. To detect this trait, we need to consider the web of interactions that links the candidate species—the species that is under consideration for being designated “insignificant”—with all the species that use it for food, shelter, and mutualistic services. For a species to be truly inconsequential, we expect to find that all of its users are generalists that can easily persist by switching to alternative resources. These alternative resources must be accessible throughout the seasons when the candidate species is available, for otherwise the user would be at risk of also going extinct because of the gap in supplies created by the loss of the sole provider. To be assured that a species is expendable requires the presence of other species that are capable of performing all of its significant functions. To provide the necessary backup of functions requires the existence of species that (1) employ similar feeding styles (e.g., sap tappers, leaf miners); (2) devour the same plant species and tissues; (3) sustain the same enemies; (4) participate in the same mutualisms; (5) provide the same dispersal services for pollen, seeds, pathogens, etc.; and (6) build similar structures, such as webs, tunnels, leaf rolls, and galls that provide shelter for other organisms. For these functions to be replaced fully requires that other species perform the candidate’s activities in the same habitats throughout the same geographical range and during the same seasons. In the interest of completeness, I should note that mammalian

test case of species expendability

283

herbivores have important indirect impacts on physical conditions and ecosystem processes as a consequence of their trampling, wallowing, defecating, etc. Such functions are not so obviously performed by insects although further study of the uses of honeydew and insect feces might be revealing. Predicting a biota’s ability to cover the functions of an extinct species raises important questions. How similar must the activities of a substitute species be to effectively provide the equivalent function? Do substitute species that share a large combination of traits with the candidate species provide higher “quality” backup? These questions have never been investigated fully. It seems obvious, however, that closely related species, as a consequence of their shared history, are more likely to hold traits in common and to perform activities in a similar manner. Therefore, I am including the presence of closely related species on the list of traits that contribute to a species’ functional insignificance.

The Case for Concluding that Dichomeris leuconotella, a Leaf-Rolling Caterpillar, Is Expendable in the Short Term Based on my long-term experience with the insect fauna associated with the tall goldenrod, Solidago altissima, I have chosen the gelechiid moth, Dichomeris leuconotella, as the member of the association that is most likely to be functionally insignificant. The goldenrod system is well suited to this test because the host plant is a native perennial that is a widespread dominant in midsuccessional communities throughout the Northeastern United States and adjacent Canada. As a consequence of the host’s abundance and apparent age, there has been ample opportunity for different lineages of herbivores to evolve a capacity for life on goldenrods. There are at least 101 species of phytophagous insects that feed and develop on Solidago altissima; of these, 42 species, drawn from 17 different families and representing a diversity of feeding styles, are specialists that feed only on goldenrods. Thus, the community appears to have a relatively mature functional structure. Dichomeris leuconotella is a small caterpillar (the larva in its final instar is 3 mg) that builds itself a protective lair by using silk to pull the edges of pliant leaves together to create a leaf roll. To measure the impact of this species, we censused 22 different sites throughout the Finger Lakes region of New York for 6 years (Root and Cappuccino 1992). In addition, we monitored caterpillar populations at 23 additional sites in New York and Pennsylvania (Root et al. unpub. ms.).

284

chapter 14

Over a wide region, D. leuconotella was never abundant enough to cause significant damage to its host. The biomass of D. leuconotella never exceeded 0.2% of the goldenrod leaf biomass over the 6-year period. In another investigation, I showed that such herbivore loads have no measurable impact on the production and fitness of goldenrods (Root 1996). The caterpillars have a thrifty style of eating that also contributes to their low impact. The younger instars rasp tiny pits on one surface of the leaf at the entrance of their leaf roll; only during the final stage are they able to eat indentations into the leaf margins. There are no necrotic areas surrounding the feeding scars as we would expect to see if the insects’ salivary secretions were incompatible or if pathogens were gaining access through the feeding lesions. The functions of D. leuconotella are highly redundant due to the presence of six congeners that are sympatric with each other throughout the Northeastern United States, the upper Midwest, and adjacent Canada (table 14.1). All these species construct leaf rolls that protect the caterpillars from generalist predators and severe physical conditions (Loeffler 1993, 1996). Typically, caterpillars build five leaf rolls over the course of their development. Small spiders and mites often use these abandoned rolls as a refuge (the ecology of secondary invaders of leaf rolls is reviewed in Cappuccino 1993). There are slight differences in the tightness of the rolls fashioned by different species: there is a tendency for caterpillars in tight rolls to rasp pits at the ends of the rolls, and for those in loose rolls to rasp pits along the margins (Loeffler 1994). These differences, however, are minor and appear to have little influence on the caterpillar’s impact on the plant or the activities of the commensals that live in old rolls. If D. leuconotella were to be removed, five congeneric species would remain active in meadows over a broad region (see table 14.1). Of these meadow species, three have the same phenologies as D. leuconotella (fig. 14.1). Host ranges also overlap broadly (table 14.2) so that every plant species eaten by D. leuconotella is also consumed by at least one of the three caterpillar species that share D. leuconotella’s phenology and habitat requirements. Indeed, many of D. leuconotella’s hosts are eaten by two or three coexisting congeners. The chief predators of all the Dichomeris species are spiders, ants, and predatory hemipterans that attach a wide variety of prey (Loeffler 1996). All the parasitoids that are known to assail D. leuconotella have relatively broad host ranges. As a consequence, every species of parasitoid that attacks D. leuconotella can also attack at least one other Dichomeris species that occurs in the same habitats at the same stage of development (table 14.3). Because of this extensive overlap, espe-

test case of species expendability

285

TABLE 14.1 Selected traits of Dichomeris caterpillars Dichomeris Species D. bilobella

D. flavocostello

D. leuconotella

D. levisella

D. nonstrigella

D. ochripalpella

D. purpureofusca

Geographic Rangea Nova Scotia south to Maryland, west to Kansas Quebec south to Florida, west to Arkansas and Manitoba Nova Scotia south to Maryland, west to Colorado and Washington Nova Scotia south to New Jersey, west to Minnesota Nova Scotia south to Maryland, west to Arkansas Quebec south to North Carolina, west to Arkansas Nova Scotia south to New Jersey, west to Alberta

Habitatb

Type of Leaf Roll Tight

Forests Loose Meadows Loose

Meadows Loose Openings, meadows ? Openings, meadows Tight Forests, meadows ? Openings, meadows

a

Data from Loeffler 1994. Data from Hodges 1986.

b

cially in the close match in phenologies, it seems highly unlikely that the disappearance of D. leuconotella would create a significant gap in the supply of resources that support its predators or parasites. The extensive overlap in the list of traits that Dichomeris spp. hold in common raises an important concern, however. The species are so similar that it seems likely that the forces responsible for the demise of an expendable species would also endanger some of its congeners. Based on the available evidence, D. leuconotella appears to be an excellent candidate for designation as an expendable species over the short term. There are gaps in our knowledge, however, that must be filled before we can reach a responsible decision. Specifically, we need to know more about the status of D. leuconotella in Colorado and Washington, where it encounters entirely different host plants, enemies, congeners, and physical conditions. Furthermore, our six years

286

chapter 14

Figure 14.1 Phenology of leaf-folding Dichomeris species on Solidago in central New York. Figure adapted from Loeffler 1994.

of census results are insufficient to detect population outbreaks such as those observed in the spruce budworm (Choristoneura fumiferana) in which the species remains very rare for decades between devastating outbreaks that cause widespread tree mortality (Morris 1963; see also Hunter 1995 and Cappuccino et al. 1995 concerning the traits of outbreak species). Finally, any attempt to evaluate a species’ functional status would ideally include removal experiments to verify that the subject species does indeed have inconsequential influences. In many circumstances, however, such experiments are difficult to mount because the subject species is rare or threatened. In the case of D. leuconotella, such an experiment would be akin to removing a needle from a haystack.

Difficulties of Predicting the Long-Term Status of Species To this point I have been concerned with the functions of species in relatively steady environments. This view is a useful place to begin judging the functional significance of a species, but it is not adequate

test case of species expendability

287

TABLE 14.2 Host ranges of Dichomeris species that are abundant in central New York Caterpillar Species D. D. D. D. D.

bilobella flavocostella leuconotella levisella ochripalpella

Plant Species S1

S2

S3

S4

S5

S6

S7

✔ ✔ ✔ ✔ ✔

✔ ✔ ✔

✔ ✔

✔

✔

✔ ✔ ✔

✔ ✔ ✔ ✔ ✔

✔ ✔ ✔

✔

✔

E1

✔ ✔ ✔

A1

A2

✔ ✔

✔

✔

✔

A3 ✔ ✔ ✔ ✔

S1, Solidago altissima; S2, S. arguta; S3, S. bicolor; S4, S. caesia; S5, S. gigantea; S6, S. juncea; S7, S. rugosa; E1, Euthania gramnifolia; A1, Aster novae-angliae; A2, A. sagittifolius; A3, A. lineolatus. Data from Loeffler 1994 and Root, Loeffler, and Rawlins (manuscript)

for drawing final conclusions because it fails to consider how a species’ status might shift in response to changes elsewhere in the system. Certainly the goldenrod system, which supports and limits D. leuconotella, has undergone profound changes. Prior to the relatively recent clearing of the North American forests by European settlers, the meadow species of goldenrods used by D. leuconotella grew as widely scattered, small populations where there were natural breaks in the canopy, such as along gravel bars in rivers, on cliff faces, and atop beaver dams (Marks 1983). Today, the same goldenrods form a conspicuous part of the landscape in the Northeastern United States, where they dominate disturbed sites such as old fields and roadside ditches. Many forces can come into play when such significant changes TABLE 14.3 Frequently encountered parasitoids that attack Dichomeris caterpillars Parasitoid Species Caterpillar Species D. D. D. D. D.

bilobella flavocostella leuconotella levisella ochripalpella

B1

✔ ✔

B2

B3

B4

E1

✔ ✔ ✔

✔ ✔ ✔ ✔

✔ ✔ ✔ ✔

✔ ✔ ✔ ✔

H1 ✔

I1

I2

✔ ✔

✔ ✔ ✔ ✔

✔

Braconidae: B1, Apanteles 1; B2, Apanteles 3; B3, Meteorus sp.; B4, Orgilus sp. Eulophidae: E1, Elachertus sp. Encyrtidae: H1, Paralitomastix Ichneumondae: I1, Campoplex; I2, Isomeris Tachinidae: T1, Actia

T1 ✔ ✔ ✔

288

chapter 14

occur in the biomass and dispersion of host plants (Root 1973; Kareiva 1983; Bach 1988; Huntley 1991; Kareiva and Wennergren 1995; Cappuccino and Martin 1997; Golden and Crist 1999.) In responding to these forces and the opportunities provided by the changing landscape, the goldenrod fauna must have undergone shifts in geographical ranges, species interactions, demographic rates, dispersal rates, and selective regimes. Furthermore, given that these changes have occurred during the past 200 generations of leaf rollers, the insects’ performances may still be in a state of evolutionary flux. Based on recent shifts in land use, it seems likely that the dispersion of food plants will change again with meadow goldenrods declining in abundance as old fields are replaced by housing developments or abandoned to forest succession. Evidence for ongoing evolutionary change in the goldenrod system is provided by observations in Europe, where Solidago altissima has become a serious invasive weed in natural habitats following its escape from cultivation around 1850 (Weber 1998). When plants from different regions in Europe are grown in common gardens, they display genotypic differences in several adaptive traits, suggesting that significant evolution is possible within a couple of centuries (Weber and Schmid 1998). Furthermore, preliminary data from common gardens reveal that the European plants are more robust and more susceptible to plant pathogens than their North American relatives (Gretchen Meyer, pers. comm.). Other investigations have shown that the insect fauna on European goldenrods is composed of generalists that are relatively rare, and that none of the taxa responsible for seriously damaging goldenrods in North America occurs in Europe (Root and Cappuccino 1992; Jobin et al. 1996). These results suggest that there has been a relaxation of the selective forces associated with herbivores in Europe, creating a situation that favors larger plants that may be superior competitors. (Hypotheses concerning trade-offs involved in the evolution of resistance to enemies and plant competitive ability are discussed in Blossey and Notzold ¨ 1995.) As a consequence, the impact of D. leuconotella, which is normally quite rare, could greatly increase if it were to colonize a region, such as Europe, where the host’s defenses have been lost. The functional significance of an insect can also shift as a direct consequence of evolution. For instance, the apple maggot, Rhagoletis pomonella, which is native on hawthorn, became a serious pest after it evolved a host race on apple (around 1864) and another host race on cherry (around 1950) (Bush 1975). Similarly, the Colorado potato beetle, Leptinotarsa decemlineata, became a worldwide pest after it broadened its host range from feeding only on nightshades to eating potato fo-

test case of species expendability

289

liage (Borror et al. 1989). The rhopalid bug, Jadera haematoloma, provides a convincing demonstration that such host shifts are the outcome of evolutionary changes. During the past century, these bugs colonized three recently introduced plants belonging to the Sapindaceae and, in each case, underwent adaptive changes in the size of their feeding mouthparts (Carroll and Boyd 1992). The status of a species can also be influenced by changes in the status of its enemies, competitors, and partners in mutualisms. Such changes can be generated by the colonization of an ecosystem by species that evolved elsewhere (the “allochthonous route”) or by the emergence of new characters that evolved locally (the “autochthonous route”). The wide-ranging and unexpected consequences of these new interactions are documented in the extensive literature on classic biological control (several reviews are listed in Louda et al. 1997) and, of course, in the many investigations that have been inspired by R. T. Paine’s writings on keystone species and the proper interpretation of ecological webs.

Conclusions Our effort to judge if D. leuconotella is expendable illustrates issues that are fundamental to all such trials. How do we deal with the fact that most species have the capacity to change in response to everchanging conditions? How do we take into account that these changes are inevitable but almost impossible to predict? A species that appears to be expendable may change its status in a variety of ways depending on the exact nature of any changes that may arise in its physical environment, hosts, enemies, or mutualists. These shifts in the functional importance of species do not necessarily require evolution, however; species can assume new roles by simply expressing their innate capacities within new contexts, such as those that might be created by the extinction of a potent enemy or a change in climate. Evidence is mounting to show that the functional status of species can also be modified by rapid evolutionary changes (see the previous section and the chapter by Palumbi, this volume). Species are, by definition, the vessels in which mixtures of heritable traits are integrated. As such, a species is an independent evolutionary lineage that might contain the genetic variants that enable it to take advantage of changed conditions. To discard a species as expendable, therefore, is to discard a unique opportunity to meet a new environmental challenge. Thus, a judgment of “ecologically insignificant” might be logically impossible for evolutionary reasons; as long

290

chapter 14

as we are dealing with true species, which have an independent evolutionary history, we are dealing with entities that have unpredictable and limitless potential for further contributions. The case of D. leuconotella also illustrates how judging the expendability of species solely on the basis of their functional traits can be misleading. As is the case with D. leuconotella, most species are rare (Root and Cappuccino 1992) and perform highly redundant functions; most would produce no appreciable effects if they were removed (see Boersma this volume). Although current efforts to document the importance of biodiversity for maintaining ecosystem functions and services are essential for understanding large-scale processes, I suspect that many of the species encountered in such investigations will be determined to play only negligible roles. To appreciate why these species with little current significance are not expendable requires that we consider the evolutionary dimension. To use Hutchinsons metaphor (1965), it may be that our most compelling reasons for retaining species will be found in the evolutionary play rather than on the ecological stage.

Summary In this chapter, I suggest a protocol for judging the expendability of species that begins by listing the traits that we expect to be associated with functional insignificance. I then apply these criteria to a wellstudied case, the leaf-rolling caterpillars that feed on goldenrods, to show that at least one species has no appreciable impacts on its community and all of its functions are performed by other closely related species. To be sure that such a species is truly expendable requires that we move on to the next steps in the protocol by conducting removal experiments, observing if the species’ role shifts in different regions, and monitoring populations for sufficient periods to detect if outbreaks occur. The members of the goldenrod system have undergone major shifts in distribution, abundance, associates, and tolerances during the past 200 years. These changes illustrate a fundamental issue that must be taken into account in any discussion of expendability. Environmental and evolutionary changes are commonplace and their effects cascade throughout the system in such complex ways that it is almost impossible to predict the role that a species might play in the future. As a consequence, the very notion that expendable species exist may not be a testable proposition.

test case of species expendability

291

Acknowledgments Carol Loeffler’s stamina, painstaking attention to detail, and special talents for finding and following tiny insects were essential for taking the measure of Dichomeris, a taxon that is “typically rare” and, therefore, more representative of most situations in nature. Bernd Blossey, Jim Brown, Tim Carr, Harry Greene, Ruth Hufbauer, Peter Marks, Gretchen Meyer, and Maria Uriarte were helpful in discussing various points of the study. I was deeply flattered to find that Peter Kareiva and Simon Levin felt that I might have something to contribute. Of course, Bob Paine set the stage that drew me to think about a topic that, frankly, I still find rather daunting in its implications. Finally, I thank Bob for a special friendship that has remained stimulating and strong across so many years and miles. I fondly remember those days in Ann Arbor when we were setting out; little did we suspect that our paths would be so fulfilling.

Chapter 15

m An Evolutionary Perspective on the Importance of Species: Why Ecologists Care about Evolution Stephen R. Palumbi

From one point of view, no species in a community can be considered absolutely essential because all species eventually go extinct, and communities tend to persist longer than the life spans of individual species. The average life span of species, measured by the fraction of species in fossil assemblages that can be identified in modern communities, is on the order of 2 million years for mammals and up to 6 million years for marine bivalves (Stanley 1979). Over this time span, communities experience many species extinctions and replacements, the balance of which has long been thought to determine some of the basic patterns of island biogeography and latitudinal gradients in species diversity (MacArthur and Wilson 1967; Brown and Gibson 1983; Pianka 1988). During the past 2 million years, for example, there have been such strong changes in the species composition of Atlantic bivalves and Caribbean coral communities that dominant

evolutionary perspective

293

species have been replaced (Johnson et al. 1995). The lesson from these observations is that the continuous evolution of new species can provide ecological replacements for those that have become extinct. If evolution proceeds over only this extended timescale, such a perspective is not particularly useful for providing ecological or conservation insights about which species are critical to community structure. Although species replacements over a million years may be rapid on evolutionary timescales, they are slow from an ecological point of view and irrelevant from a contemporary management perspective. The fact that speciation over the next 5 million years may be able to replace some of the thousands of species that have gone extinct in the past century (Wilson 1990) does not mean that species conservation efforts should be slowed. This difference in intrinsic timescales is the major reason why ecological investigations frequently ignore evolutionary issues, and why the evolutionary genesis of communities has played such a minor role in the conduct of ecological experiments. However, evolution more rapid than this million-year timeframe is becoming appreciated much more widely (Endler 1986; Palumbi 2001, 2002), which may lead to important ecological and management implications. Rapid evolution of ecological relationships may result in much more intricate relationships among coexisting species, even those species that have come into contact only recently or that encounter new ecological conditions. The extinction of a species that plays a key role in the life cycle of other species may have little community impact if the role of that species can be replaced by others in the community. But if coevolution has crafted a unique role for a species, then its extinction may be a harbinger of broader ecological damage. For example, plants that have flexible pollinator systems or multiple reproductive strategies are less sensitive to pollinator extinction than plants that are tightly coevolved with specialist pollinators (Bond 1994a). From this perspective, we can shift the debate from a focus on the importance of species (which has many moral, scientific, and aesthetic elements) to a consideration of the severity of community-level ecological change that results from the extinction of a particular species. Which species will have the greatest ecological impact on extinction? If evolution is slow compared with species invasions and extinctions, then communities might tend to be assembled by those that happen to survive (Diamond and Case 1986). In this scenario of limited coevolution, the species that have the least impact might be those that contribute little to the energy flow within the community or provide few community resources, such as like physical structure. But if local evolution is fast compared with invasions and extinctions, most spe-

294

chapter 15

cies will be linked together in interacting communities. Under these conditions, even rare species—or those with little contribution to community energy flow—may nevertheless provide crucial interactions. How rapid does the evolution of ecological interactions occur? In the past, such interactions have been considered to change only slowly— a current keystone species is likely to have always been a keystone species. Over the past decade, however, a greater appreciation for the rapidity of fundamental evolutionary change has emerged from studies of the impact of environmental variation on the morphology and behavior of a wide variety of species. In particular, the rampant ecological changes that humans are causing throughout the world are creating a dramatic impact on the evolution of species within communities. This evolution is now known to extend to the evolution of species interactions as well. Species that have invaded new habitats not only evolve rapidly, they also cause rapid change in species at other trophic levels. If contemporary ecological shifts caused by human impact are often widely associated with rapid evolutionary change, we may be more confident that evolution has also rapidly affected relationships within existing communities. This evolutionary perspective is one in which ecological and evolutionary timescales are closer than previously considered: species are continuously evolving within their ecological communities. In this essay, I describe examples of rapid change in morphology, life history, and behavior that are associated with recent ecological change in well-studied communities. Examples are drawn from marine, aquatic, and terrestrial settings and involve insects, vertebrates, and marine invertebrates. Although acclimatory change may produce some of the effects described, these examples show that complex interrelationships among species in modern settings can shift quickly when ecological conditions change.

Rapid Ecological Evolution Hermit Choice Along the coasts of New England, the hermit crab Pagurus longicarpus occupies shells derived from a variety of local gastropods, including shells from the introduced species Littorina littorea. Having invaded into New England from a 19th-century range expansion out of Nova Scotia (Berger 1977; Vermeij 1982), L. littorea provides most of the shells used by medium to large individuals of P. longicarpus in New England (Blackstone and Joslyn 1984). By contrast, L. littorea is absent

evolutionary perspective

295

along the middle Atlantic and southeastern coasts of North America, and crab species of these areas inhabit other shells (Blackstone 1985). This gradient in shell availability parallels a gradient in chela shape, size, and asymmetry along the Atlantic coast; southern crabs have relatively longer claws and greater right-left asymmetry than in the north (Blackstone 1985). Although there are no tests of the relative fitness of crabs with differing of chela shapes, Blackstone (1985) showed through common garden experiments that many of these shape changes, especially in females, derive from both plastic morphological responses to different shells and to nonplastic, presumably genetically determined, differences among crab populations. The implication is that, against a background of acclimation of individuals to different shells, the morphological evolution of hermit crabs in the presence of an abundant new source of large shells has taken place. Crab behavior has also evolved. Hermit crabs from the mouth of Long Island Sound that were given a choice of shells from local and introduced species showed a clear preference for the shells of the introduced species L. littorea. This preference intensified in medium and large crabs (Blackstone and Joslyn 1984), resulting in a strong bias for the use of the introduced shells. Again, experiments in which shells were switched showed that these preferences were ingrained rather than induced by prior occupation of L. littorea shells (Blackstone and Joslyn 1984). These experiments suggest shell preferences evolved in the northern populations of P. longicarpus after the snail invasion.

Plastic Crabs The introduction of other species into New England habitats has been associated with rapid morphological change. A complex example involves the change in shell strength of a common shoreline snail along the rocky, wave-swept coasts. The snail, Littorina obtusata, is a small herbivore common to the middle intertidal zone (Lubchenco 1980). Predation by the crab Carcinus maenas has been shown to be a powerful mortality source in some populations (Seeley 1986), but the crab is a recent invader, having been introduced to the mid-Atlantic coast of North America in the 1800s, and spreading by about 1900 north of Cape Cod. Seeley (1986) documented a strong morphological shift in L. obtusata during this same period. In particular, shells were thicker and spire height was lower for populations collected from 1871 to 1893 compared with those collected from 1982 to 1984. Seeley (1986) interpreted this change as a rapid evolutionary response to increased crab predation. There may be another aspect to the story, however. Snails are known

296

chapter 15

to alter shell characteristics in different environments, particularly increasing shell thickness in areas of high wave exposure and with close proximity to predators (Appleton and Palmer 1998). Littorina obtusata has been shown to respond similarly, exhibiting a strong cline in shell thickness with wave exposure and in response to predator effluent (Trussell 1996; Trussell and Smith 2000). Although the historic differences documented by Seeley (1986) clearly reflect a major shift in the phenotype of L. obtusata after ecological invasion by the green crab, the basis of this response may not be strict Darwinian evolution due to the selection of snails with thicker shells; rather, it may involve an ingrained phenotypic plasticity. Currently, L. obtusata occurs on both coasts of the northern Atlantic, showing differences in gene frequency that are characteristic of populations without much gene flow (Berger 1977). Presumably, L. obtusata was eliminated from the rocky shores of New England by glaciers that covered most of this habitat (Vermeij 1982). Although the subsequent arrival of L. obtusata on North American shores was unheralded, the invading stock probably came to New England from populations in Europe where the green crab is commonly found. Interestingly, the species description for populations in northwest Europe emphasizes a low spire (⬍10% of the major body whorl) and a thickened shell (Hayward and Ryland 1995)—characteristics that result from the common presence of crabs in Europe and were not apparent in the North American populations from 1872 and 1898 (Seeley 1986). Thus, the European snails, when they recolonized American coasts, probably brought with them the ability to respond morphologically to the presence of green crabs but did not need this ability until the arrival of the crabs in the 20th century. This view suggests that the ecophenotypic plasticity expressed by L. obtusata was not selected against after its colonization of America and so has persisted until the present day. Perhaps the ability to respond phenotypically to crabs was advantageous because of the presence of the native shore crab Cancer borealis, or perhaps selection against this plastic ability was too weak for it to disappear over a short period. In either case, relationships between the predator and prey are ecologically and evolutionarily complex, belying the recent invasion of crabs to New England.

Salmon Size Intense predation has often been documented as a major agent affecting both ecological change (Zaret and Paine 1973) and evolutionary

evolutionary perspective

297

change (Endler 1986). Humans are now the world’s greatest marine predators, each year extracting from the ocean, about 70 million tons of fish and invertebrate species. Overfishing has reduced populations of most of the world’s fisheries—many to critical levels—and many species sustain strong fishing mortality among adults. Among salmon, an estimated 80% of returning adults have been captured by net or hook-and-line fishing (Ricker 1981). Moreover, individuals captured by the fishery tend to be large, particularly when size-selective gear such as gill nets are used. Because this intense fishing mortality is not random with respect to size, strong evolutionary force has been exerted on fish populations, leading to dramatic changes in the size of some species. Pink salmon follow a typical life cycle and spend their teenage years at sea, growing from fingerling to several kg in mass. Their life cycle is much more regular than in most salmon, and they return to natal streams to spawn during their second year. The pink salmon clock runs so precisely that there are really two types of pink salmon: the odd-year fishes that return on odd years (e.g., 1999) and produce young that return to the stream in the next odd year; and the evenyear fishes that use the same streams during the summers of even years. These odd-year and even-year fish classes are genetically different and have different population sizes, showing that pink salmon regulate very precisely their return to the streams during the second year. This precise punctuality means that all fishes returning to streams to spawn are the same age and that any differences in size can be ascribed to differences in growth rate. Data collected over the decades of the most intense fishing pressure on pink salmon have been tested for any impact of fishing on size (Ricker 1981). The data show a clear and steady decline in growth rate so that fish caught in 1974 were 20 to 30% smaller than the fish caught in 1951 (fig. 15.1). Visible in both even- and odd-year classes, these declines do not represent the simple elimination of the biggest, oldest fishes because all returning fishes are the same age. Instead, declining size represents the evolution of salmon with lower growth rates—fishes that become smaller spawning adults that more easily escape capture. In natural circumstances, larger fishes have higher fecundities than smaller ones, and natural selection often favors the former. But when oceanic predation is high, large fishes have a lower life expectancy, and survival is highest for those individuals that have an slower than usual intrinsic growth rate. Exactly the same scenario has played out in a completely different environment—the freshwater lakes of northern North America. In

298

chapter 15

Figure 15.1 Size of 2-year-old pink salmon caught at Bella Coola (for odd and even year classes). Between 1951 and 1974, size shows a steady and significant decline (redrawn from Ricker 1981).

Lesser Slave Lake, Alberta, commercial fishing has been intense since the early 1900s. Lake trout were fished out by the 1920s, and lake whitefish declined until the spawning stock finally collapsed in 1965. Between 1941 and 1965, the average size for lake whitefish was declining, but this was long thought to be due to the removal of older, bigger fishes. However, careful analysis of growth data revealed a surprisingly different reason: that the fishes were growing much more slowly than before (Handford et al. 1977). In the 1940s a 9-year-old fish weighed about 2 kg. By 1960, few 9-year-olds weighed this much, with the average weight at this age only 1.5 kg. By 1970, the average 9-yearold lake whitefish weighed only 1.2 kg—barely the size of a 4-yearold fish from 1945. In addition, fish “condition”—a measure of weight per length—declined 30% during this period, and the variance in weight/length measures also dropped (Handford et al. 1977). Very young fishes still grew just as quickly in 1970 as they did in 1945, suggesting that the conditions for growth within the lake were not declining. Instead, the fishes had evolved a new life strategy in which most fishes were thin and slow growing.

evolutionary perspective

299

In Lesser Slave Lake, gill nets of 5–5.5⬙ mesh width were the major source of fishing mortality. Nets of this size readily capture fishes weighing more than 700 g with a condition factor of 1.3 or greater (Handford et al. 1977). In 1947, virtually all fishes older than 3 years were vulnerable to such nets. By 1970, only males 6 years and older were vulnerable because of the combination of smaller size and thinner morphotype. Such patterns of phenotypic change are difficult to interpret in many fishes because of their plastic growth trajectories. Although condition and growth rate are heritable in fishes, it is possible that some of the morphological change recorded in lake whitefish in Lesser Slave Lake was derived from environmental change. Nevertheless, shifts in form and size among lake whitefish show the power of ecological change—in this case, the introduction of a human top carnivore—on the evolution of community members. Other analyses suggest similar effects on marine fishes. In North Sea plaice, the length at sexual maturity has declined by 10 to 20% since 1910, associated with approximately 50% fishing mortality for fishes older than over 2 years (Rijnsdorp 1993). In this case, growth rates of younger fishes are much faster and may have produced the observed shifts in length at maturity. However, an analysis of the growth data suggests that a substantial fraction of the decline in length at sexual maturity is not explainable by differences in juvenile growth rate. Furthermore, the appearance of many more reproductively mature fishes at smaller size can not be explained by changes in ecological condition (Rijnsdorp 1993). Like the cases of pink salmon and lake whitefish, life history features of plaice appear to have changed during the last century.

Guppy Love and Death Although these examples of changes in fish life history are consistent with evolution caused by fishing pressure, there is also direct evidence for the role of predation in selection for life history features, and for the rapid evolution of prey populations. Such evidence derives abundantly from a careful study of life history evolution in stream fishes particularly guppies. Along the north coast of Venezuela, rivers and streams descend from the highlands in waterfalls to meander along the coastal lowlands. Predatory fishes tend to be restricted to areas below waterfalls, and populations of guppies in lowland portions of streams experience high predation rates (Endler 1995). Below waterfalls, high predation is associated with a 25% faster rate

300

chapter 15

of maturation and a 20 to 25% smaller size at maturity (Reznick and Endler 1982). Experimental translocations of fishes from predator-rich to predator-free environments were used to show how rapid evolution could occur after an ecological shift. These experiments showed rapid evolution toward this same late-maturing life history strategy seen in upstream habitats (Reznick et al. 1990). Similar evolutionary shifts have also been observed for male coloration. Brightly colored males are preferred by females but experience much higher rates of predation in lowland streams (Endler 1980; Houde 1997). As a consequence, male coloration patterns are balanced by sexual selection based on female choice and predation rate. When predation rate changes, male coloration evolves quickly, becoming brighter when predation declines (Houde 1997). Such evolutionary changes are so fast that the population of guppies in Trinidad is a mosaic of different morphologies and life history strategies, each stream and river drainage effectively acting as a different ecological and evolutionary unit (Houde 1997).

Shrimp Sex Changes in fish growth patterns are a reflection of the new evolutionary pressures exerted on fishes by humans. Other changes are occurring in other fisheries as well, reflecting massive differences in animal lifestyles that are successful in the modern world of stainless steel hooks and factory ships. In at least one case, hunting has led to a fundamental change in sex determination for the hunted species. The cold-water rock shrimp, Pandalus borealis, lives in the northern Atlantic, where Swedish and Danish fisheries increased production by at least threefold during the late 1950s and the 1960s. As in many fisheries, shrimpers target the largest individuals, and the proportion of shrimp longer than 80 mm declined from 44% in 1950 to 14% in 1962. Unfortunately for Pandalus borealis, almost all the large shrimp are female, because in this species each individual undergoes a sex change sometime after the first year of life. Born male, every shrimp functions as a male until it reaches 75 or 80 mm in length, at which point it changes sex to become a mature female. The reason for this shift lies in the method by which these shrimp guard eggs—females hold eggs under their abdomen, using pleopods to ventilate the egg mass with oxygenated water. Only large females can carry large batches of eggs, whereas even small males can fertilize a large clutch. For this reason, life history theory (Ghiselin 1969) suggests that sequential sex change should occur from male to female. Evolutionary models fur-

evolutionary perspective

301

ther suggest that individual fitness is most sensitive to the size at which sex change occurs (Charnov 1981). The success of this sex-changing strategy depends on the survival rate of large shrimps. If all large animals are taken by a fishery, few females will remain to provide any eggs; thus, there will be few advantages to waiting until after the male years to become female. A better evolutionary strategy would be for some individuals to reproduce as females as early as possible, before the fishery claims them. Between-population variation in sex at first reproduction is common in Pandalid shrimp, varying from 13 to 65% female among different years in populations off the California coast (Charnov et al. 1978). In the Danish fishery, this variation has been visible as a strong shift in the size at the time of sex change from male to female as well as an increase in the numbers of individuals that breed as females during their first reproductive season. In the European fishery, there were no females smaller than 75 mm before 1954 (Charnov 1981). Starting soon after 1954, however, coincident with the increase in fishing pressure and the decline in numbers of large shrimp, smaller females became more common. By 1954, 8% of the females were smaller than 75 mm, and by 1962, this figure had jumped to 30%. These values probably represent the filtering of preexisting life history variation by selection to reproduce early. The few individuals that converted to females early—or were never male— had greater reproductive success than individuals undergoing the normal pattern of sex change. In this case, a complex life cycle was converted into a simpler one by intense selection against large individuals. Like the case of shell morphology of intertidal snails, however, changes in sex determination could have an environmental component. The age at sex transition is younger in less dense populations of Alaskan Pandalus borealis (Anderson 1991), largely because growth rate is higher at low density, allowing animals to attain a large size earlier. Thus, fishing could accelerate the timing of sex change by increasing the growth rates of surviving populations. This change would not alter the size at sex change, however (Charnov 1981). A well-known determinant of sex change in protogynous fishes and crustaceans is the social setting—when dominant males are removed from the population, large females undergo sex change to become males (Robertson 1972) Whether small males of protandrous species can similarly detect the relative lack of large females after heavy fishing mortality—and change sex earlier than usual—is unknown. But such a mechanism could explain the abundance of small females in heavily fished populations without invoking strong natural selection or genetic changes derived from the fishery.

302

chapter 15

Invasion, Extinction, and the Evolution of Ecology Ecological conditions change dramatically when new species are introduced to a region or previously common species are driven to extinction. Evolutionary changes in species in such altered communities are common and demonstrate that morphology, behavior, and ecological interactions can evolve during community disruption. Insects have evolved both morphologically and in ovoposition preferences in response to the invasion of new food plants. Soapberry bugs have evolved mouthpart features to take advantage of introduced plants (Carroll and Boyd 1992), and butterflies have evolved ovoposition preferences in the presence of new larval food plants (Thompson 1996) by a process similar to the evolution of shell preference in hermit crabs cited earlier (Blackstone and Joslyn 1984). The weevil Rhinocyllus conicus was introduced in western North America to control invasive thistles of the genus Carduus. Although the original weevil population from Europe was shown to have a preference for Carduus and higher larval development on this host genus, after introduction in North American it increased its host range dramatically. Between 1992 and 1996, the weevil became a serious pest of native thistles (Louda et al. 1997). The extent to which this host shift was accompanied by evolutionary shifts in the weevil is unknown, but it would be extremely interesting to know if adaptation to native thistles has occurred in this example of biological control gone awry. Extinctions also generate strong evolutionary pressures. The i’iwi, a Hawaiian honeycreeper (Vestiaria coccinea) currently feeds on the short, clustered flowers of the ohia tree, but its long curved bill is actually adapted to tubular flowers of the nearly extinct Hawaiian lobelioids. Since the early 1900s, the upper mandibles of i’iwi bills have become shorter, largely through the elimination of individuals with the longest bills, perhaps as a consequence of dietary shift to nontubular flowers (Smith et al. 1995). Behaviors are also quick to shift when ecological settings change. Nest parasites, birds that lay their eggs in the nests of other species, leave their offspring to be raised by foster parents, often at the expense of the nest owners’ young. A behavioral response to nest parasitism in birds is egg heaving—when nest owners remove a foreign egg from their nest. This behavior may save the owners from raising a fledgling that is not their own offspring, but egg heaving is costly because birds sometimes make mistakes and destroy their own eggs

evolutionary perspective

303

(Cruz and Wiley 1989). In European forests, for example, spotted cuckoos frequently lay eggs in nests of the common magpie. Magpies regularly inspect eggs in their nests and eliminate any that seem the wrong size or color. Studies of magpie behavior in Europe have shown that the likelihood that a bird recognizes and heaves a foreign egg changes from forest to forest. Where nest parasites such as cuckoos abound, egg inspection dominates the magpie’s behavioral repertoire. Where parasites are rare, the local birds do not inspect and heave eggs, presumably to avoid the cost of mistakenly dumping the wrong egg (Soler et al. 1999). Such behavior evolves quickly after introductions. Small birds called village weavers show strong “egg-inspection” behaviors in their native Africa, where nest parasites abound. In the 18th century, they were introduced to the Caribbean island of Hispaniola, where no nest parasites exist naturally (Cruz and Wiley 1989), and quickly lost their egg-inspection behaviors. As a result, they became an easy target when the egg-dumping shiny cowbird invaded Hispaniola in 1972. By the 1980s, these nest parasites were shown to have a devastating effect on weaver nest success (Cruz and Wiley 1989). But by 1999, the Weavers had reevolved their egg-inspection and parasite-heaving behaviors, rejecting 89% of cowbird eggs and reducing the impact of cowbird parasites on Weaver reproduction (Magali and Sorci 1999). These observations and experiments demonstrate the rapidity with which behaviors can evolve in response to changes in nest parasite abundance, suggesting that interactions among species that have previously never encountered one another can quickly shift.

Conclusion These examples of rapid change in communities suggest that species are not immutable ecological units that are assembled like legos into a functioning ecosystem. Instead, at least some interactions among species appear to evolve quickly. In certain cases, such as that of the village weavers, the evolution of new behaviors in historical timeframes seems to have had a dramatic impact on population size. No ecologist can be blind to the massive changes in natural habitats that are occurring as the human population expands and our ecological impact increases. The rapid evolution of species within communities suggests that human ecological impact has strong evolutionary correlates, and that this evolutionary change will be reflected in careful studies of modern ecological communities. These examples suggest

304

chapter 15

that the integration of species into coevolved ecological communities is far faster than the slow turnover of extinction and diversification seen in the fossil record, and that even recently added community members are potential candidates for “important” status. Because species interactions are dynamic and evolution can be rapid, strong interactions may evolve over short timescales.

Chapter 16

m Recovering Species of Conservation Concern—Are Populations Expendable? Mary Ruckelshaus, Paul McElhany, and Michael J. Ford

Simply mentioning the notion of species expendability may seem reprehensible, but it is a question raised often when conservation biology meets conservation practice, as in this volume. In particular, when arguing for the value of biodiversity, biologists are faced with the challenge of examining the roles that particular species might play in community function and whether there are redundancies in the functions of certain species. Does every species need to be conserved for our ecosystems to function? The question of expendability also is important in conservation planning that occurs at the species level: does every population need to be saved to maintain species viability? This question contains the same challenge aimed at community and ecosystem ecologists—do populations serve redundant roles in species viability, and if so, are some expendable? To address the question “Are populations or species expendable?”

306

chapter 16

it is first necessary to specify a biological or management context for the question. For example, any particular population in an abundant, widespread species might well be expendable with respect to the species’ viability but could be nonexpendable with respect to maintaining the current attributes of its local community or ecosystem. The same population may or may not be deemed expendable with respect to resource management goals such as recreational or commercial harvest. In this chapter, we explain approaches that we—as biologists with the National Marine Fisheries Service (NMFS)—are developing to address the issue of expendability in conserving and managing of anadromous Pacific salmonids, a species group that raises difficult issues with regard to population protection. The stakes for addressing such questions are high: if some populations are deemed expendable in their contribution to species viability, it is unlikely that they will be targeted for conservation or recovery efforts when political, social, and economic concerns enter into planning decisions. Conversely, if the expendable populations are identified as essential, limited resources may be spent on populations that contribute little to the viability of a species. Among other functions, the NMFS is the agency in charge of administering the Endangered Species Act (ESA) for six anadromous species of Pacific salmon (Oncorynchus spp.) found on the West Coast of North America. Pacific salmon spawn in rivers and streams all around the northern Pacific rim, from southern California to Korea (see Groot and Margolis 1991 for a thorough review). After hatching, the juvenile salmon spend weeks to years living in fresh water before migrating to the ocean. Ocean residency lasts several months to several years depending on the species, population, and individual, after which the fish return with generally high fidelity to reproduce in their natal stream. Several of the species also have life history forms that spend their entire lives in fresh water. Most of the species exhibit high levels of life history variability within and among populations, and there is evidence that much of this diversity is adaptive (reviewed by Taylor 1991). Since the early 1900s, most Pacific salmon species have experienced considerable declines in both abundance and diversity (Nehlsen et al. 1991), and since the early 1990s, the NMFS has listed populations from five of the six Oncorhynchus species under its jurisdiction as threatened or endangered under the ESA (table 16.1). The ESA, therefore, provides much of our context for determining which populations are important and, indirectly, which are less important and perhaps expendable. In addition, most of the Pacific salmon species are also managed heavily as a natural resource by state, tribal, and

recovering species

307

TABLE 16.1 Conservation status of listed Evolutionarily Significant Units (ESUs) of Pacific salmonids under NMFS jurisdiction as of January 2002. The listing status of each ESU under the Endangered Species Act is specified—ESUs are listed separately as “species” under the Act as distinct population segments. Species

ESU

Listing Status

Chinook salmon

Sacramento River winter run Upper Columbia River spring run Snake River fall run Snake River spring/summer run Puget Sound Lower Columbia River Upper Willamette River Central Valley spring run California coast

Endangered Endangered

Coho salmon

387

Threatened Threatened Threatened Threatened Threatened Threatened Threatened

Central California Southern Oregon/northern California coasts Oregon coast

Threatened Threatened Threatened

Chum salmon

Hood Canal summer run Columbia River

Threatened Threatened

Sockeye salmon

Snake River Ozette Lake

Endangered Threatened

Steelhead

Southern California Upper Columbia River South-central California coast Central California coast Snake River Basin Lower Columbia River California central valley Upper Willamette Middle Columbia River

Endangered Endangered Threatened Threatened Threatened Threatened Threatened Threatened Threatened

387

federal governments; the desire to manage these species for human consumption plays a large role in determining which populations are important. The primary goal of the ESA (as amended in 1978; 16 U.S.C. §§ 1532[16]) is to prevent the extinction of species, subspecies, and (for vertebrates only) “distinct population segments.” The biological context in which we ask the question “Which populations are important?” is, therefore, one of preventing the extinction of a listed

308

chapter 16

group of fish and recovering the group to a level of viability for which it no longer needs the direct protection of the ESA. For the purposes of ESA listing, the NMFS determined that a salmon population or group of populations will be considered a “distinct population segment” if it is an Evolutionarily Significant Unit (ESU), which Waples (1991) defined as a population or group of populations that is substantially isolated demographically from other populations and contains an important component of the evolutionary legacy of the species. Since the early 1990s, the NMFS has subdivided the seven Pacific salmon species into 57 ESUs (e.g. fig. 16.1), and has listed 27 of these as either threatened or endangered under the ESA (see http://www.nwr.noaa.gov/). The development and application of the ESU concept to Pacific salmon touches on the issue of how expendable major subgroups are to the viability of a species as a whole (see Waples 1995); in the context of the ESA, however, no ESU is considered legally expendable. To our knowledge, there has been no thorough attempt to determine the biological expendability of entire ESUs, and we do not attempt to address that issue here. In developing recovery plans, the context of the ESA requires rather that we address the question on a smaller scale and determine how many and which populations are necessary for the long-term viability of a listed ESU. The biological analyses that we conduct are therefore defined within this context, and the questions we ask are focused at the level of within-ESU population structure: (1) how many populations are necessary for ESU persistence? and (2) which combined set of population characteristics constitutes a viable ESU? In the remainder of this chapter, we outline our approach to both of these questions in turn.

How Many Populations Are Necessary for ESU Persistence? Population Number and Persistence: Guidance from Existing Conservation Frameworks Simply identifying the numbers of individuals necessary for species or ESU viability is not a sufficient conservation goal alone, because the population structure of a threatened or endangered species can have a significant effect on the likelihood that the species persists (Hanski and Gilpin 1997). In spite of the clear effect of the distribution and number of populations on species persistence, it is surprising to note that several of broad-ranging conservation recovery documents do not include explicit targets for the numbers of populations needed

recovering species

309

Figure 16.1 Geographic boundaries of Evolutionarily Significant Units (ESUs) of chinook salmon (Oncorhynchus tshawytscha) in the Northwestern United States.

for species viability to occur. For example, most of the recovery plans developed by the U.S. Fish and Wildlife Service for species listed as threatened or endangered under the ESA do not specify the numbers of populations needed for being taken off the list (Tear et al. 1993; Schemske et al. 1994; Tear et al. 1995). Of those plans completed before 1993 that did include recovery goals for numbers of populations, 37% (of 163 plans) had population number targets that were lower

310

chapter 16

than the existing number of populations at the time of listing (Tear et al. 1993). The ratio of target number of populations to the extant number of populations ranged from 1.2 for threatened plants to 3.0 for endangered plants (mean ⳱ 2 Ⳳ 1.9; range ⳱ 0.08–10; Schemske et al. 1994). Recovery goals that were specified for threatened and endangered animals resulted in ratios of 1.3 to 2.0 (data from Tear et al. 1995). None of the few plans that did specify population number targets provided a biological rationale for the numbers provided, so it is difficult to evaluate whether those numbers are expected to be sufficient for species viability. At least two widely used conservation risk assessment protocols do include population number targets. The International Union for the Conservation of Nature (IUCN) Species Survival Commission states in their Red List categories that the number of “locations” in which a species occurs (for most species, a location encompasses a good portion of a population or an entire population) must be five or greater to avoid being assigned to even the lowest risk category of “vulnerable” (IUCN 1994). A species occurring in a single “subpopulation” (corresponds roughly to a population for most species, according to the IUCN guidebook) is automatically assigned to at least the vulnerable risk category, according to the IUCN guidelines. The Nature Conservancy and NatureServe have a protocol for evaluating conservation risk that includes guidelines for the number of “element occurrences” (often corresponding to a local population but in some cases a subpopulation) in which a species with 5 or fewer occurrences is considered to be “critically imperiled,” with 6 to 20 occurrences is “imperiled,” and with 21 to 100 occurrences is considered to “vulnerable” (L. Master, NatureServe unpub. ms.). The biological justification for these numbers is not apparent, which makes applying the protocols to a particular species with a particular life history a challenge. It is clear that, in applying these protocols, defining what constitutes a “population” and how populations, subpopulations, element occurrences, and locations are related for species of interest will strongly affect the ultimate results of any risk evaluation.

Population Number and Persistence: Theory and Applications to Salmon There is considerable theoretical work on the expected viability of metapopulations, and this body of work could be used to determine how many salmon populations are necessary for the persistence of an ESU. Metapopulation theory explores the dynamics of groups of populations located in discrete habitat patches. How many patches con-

recovering species

311

tain populations at any given time is a function of the rate at which individual populations go extinct, the rate at which suitable patches are colonized, and the dynamics of the habitat patches themselves (reviewed by Hanski and Gilpin 1997). From a consideration of these factors, it may be possible to estimate the number of populations or the number of habitat patches that are required for the entire system to persist, where persistence is defined as the existence of at least one population at some time in the future. Application of this approach demands an understanding of (1) contributions of within-population dynamics and catastrophe rates to extinction risks, (2) dispersal patterns and colonization rates, and (3) the physical and biological processes that control habitat dynamics. The predictive capability of multipopulation viability models is likely to be low, given the scarcity of information needed for the development of such models (Groom and Pascual 1998; Morris et al. 1999); the dearth of information on salmon is no exception. Because a fully developed salmon metapopulation model based on empirically derived parameter estimates is currently not feasible, we are working to develop general guidelines by exploring simplifications of the metapopulation theory grounded in salmon biology. The first task in determining how many populations are necessary is to define a population. McElhany et al. (2000) addressed this issue while developing the concept of a Viable Salmonid Population to guide salmon recovery planning. They defined a salmon population as a reproductively isolated group of fish that is demographically quasi-independent of other groups over a 100-year period. Based on this definition and given the dispersal patterns of Pacific salmon, population boundaries are likely to encompass relatively large watersheds. In the Puget Sound chinook salmon ESU, for example, 21 populations of chinook have been identified, occupying watersheds averaging 122,000 ha (range: 48,000–260,000 ha) (PSTRT 2001). McElhany et al. (2000) further define a viable population as one that has a negligible risk of extinction within 100 years due to intrinsic processes and “normal” levels of environmental variation. Explicitly excluded from this definition is the consideration of catastrophic events, which are considered the most likely cause of extinction for a population that is large enough and stable enough to qualify as viable. By defining populations in this way, we reduce the problem of determining how many populations are necessary for ESU persistence to an analysis of risks from catastrophic events (e.g., Ralls et al. 1996). Through this approach, questions of within-population dynamics and determining if an individual population is viable are addressed with separate analyses. We can simplify the problem further by making informed assump-

312

chapter 16

tions about recolonization rates and patch (entire watershed) dynamics. At one extreme, we might consider catastrophic events as those that permanently destroy watersheds and from which populations can never recover. Such large-scale, permanent damage might be caused, for example, by volcanic eruptions or massive chemical spills. If we further consider a scenario in which no new occupiable watersheds are created, we can determine the probability that no population will remain extant after a given period and with given initial metapopulation size. If we assume that populations experience independent, identical catastrophic extinction risks, this probability is given as n

probextinct ⳱ (1 ⳮ eⳮ␭*t)

(1)

where t is the number of years of conservation concern, ␭ is the annual rate of catastrophes, and n is the initial number of populations in the metapopulation (fig. 16.2). The question is not if the ESU will go extinct but rather when the ESU will go extinct. If this period is sufficiently far in the future and the probability of ESU extinction is sufficiently low, the risk may be acceptable and the initial number of populations could be a suitable approximation of the minimal number of populations needed for ESU persistence. To develop a guideline using equation 1, we can look at the probability that an ESU will persist for as long as the average time between catastrophes. This requires estimating the average time between catastrophes and determining the level of acceptable risk. Catastrophic events that wipe out entire populations tend to be quite rare and unpredictable in their rate of occurrence. In addition, many potential catastrophic events (e.g., potential dam failure, landslides from clear cuts) are of recent human creation, and we have a limited historic record over which to evaluate these contemporary risks. Nevertheless, if we assume that the types of major events that permanently destroy the ability of an entire watershed to support a viable population are extremely rare—occurring at timescales of hundreds of years—equation 1 suggests that 5 to 10 viable populations would be sufficient for ESU persistence for hundreds of years. Choosing such a timescale is supported by the fact there is no indication of the catastrophic extinction of any of the 21 demographically quasiindependent chinook populations identified in Puget Sound over the last several hundred years (PSTRT 2001). Five to ten viable populations are a plausible minimum number for ESU persistence if catastrophes are independent and permanently damage populations. Using equation 1, the population numbers needed for

recovering species

313

Figure 16.2 The probability of ESU extinction (i.e., zero populations) after 100 years (A) or 500 years (B) given different initial numbers of populations and different catastrophe rates. The model assumes that populations have equal risks of catastrophe and that catastrophes occur randomly. Once extinct as a result of catastrophe, a population does not recover.

ESU persistence could be over estimated or underestimated. On one hand, this approximation could provide an upper bound for the minimum number of required populations because the scenario is quite pessimistic—many types of events considered catastrophes would not lead to permanent loss of watersheds or to irrecoverable population extinction. On the other hand, such estimates could be low because of the impact of spatially correlated catastrophic events on metapopulation persistence. An analysis of Puget Sound chinook populations shows that trends in abundance are correlated among quasi-independent populations, indicating that those populations

314

chapter 16

likely experience common environmental conditions (PSTRT 2001). Such correlations in population dynamics indicate the potential for correlated catastrophic events, which would tend to increase the number of populations needed for ESU persistence. In the next section, we address concerns about spatially correlated threats to populations through a consideration of ESU-wide spatial structure and diversity, since both these variables affect the likelihood that a single event will affect multiple populations. Equation 1 considers the scenario in which watersheds are removed permanently as suitable salmon habitat. What if the salmon in a watershed are extirpated but the habitat is still suitable for recolonization? Such a scenario could occur, for example, as the result of an extreme weather event or landslide that temporarily prevented access to a watershed. Data on the behavior of fish populations after the eruption of Mt. St. Helens suggest that salmon may switch to an adjacent river system habitat after such a dramatic disturbance, then recolonize the historic area once the disturbance is reduced (Leider 1989). This behavior indicates that although salmon have high homing fidelity, they exhibit some plasticity and can respond adaptively to largescale disturbance. To obtain one bound on the minimum number of populations necessary for ESU persistence, we can explore the hypothesis that at least one healthy population in an ESU will allow any suitable but empty watershed to be recolonized quickly. The question then reduces to the probability that all the populations in an ESU will go extinct simultaneously within a single year. The probability equation is ␣

probESU ⳱ 1 ⳮ (1 ⳮ (1 ⳮ eⳮ␭)nPops)

(2)

where probESU is the ESU extinction probability, ␣ is the number of years of conservation concern (i.e., how many years we want the ESU to persist), ␭ is the rate of catastrophes (i.e., 1/mean time between catastrophes), and nPops is the number of initial populations in the ESU. As can be seen in figure 16.3, the probability of ESU extinction becomes extremely small as the number of populations exceeds two. Again, this equation does not consider correlated catastrophes, which would tend to increase the risk of ESU extinction. In considering the scenarios for permanent population loss (eq. 1) and simultaneous population loss (eq. 2), concern about permanent loss yields a higher estimate of the minimum number of identical independent populations required for ESU persistence.

recovering species

315

Figure 16.3 Probability of simultaneous extinction of all the populations in an ESU as a function of the number of populations over a 500-year period. Different lines indicate different mean times between extinction events.

Which Population Combinations Constitute a Viable ESU? As illustrated in the previous section, demographic models can provide some help in estimating the minimum number of populations necessary to avoid a particular risk of ESU extinction. ESU viability per se is one conservation goal we consider in developing recovery plans for federally listed salmon. Conserving the diversity of fish in the historical ESU is another goal for recovery planning. Therefore, in addition to estimating the numbers of populations needed for species viability, providing guidance for the characteristics of populations and their locations also is important for conservation planning. The examples in this section and the next indicate that it is likely that efforts to preserve spatial and life history diversity will require more populations per viable ESU than a simple consideration of independent extinction risks. The susceptibility of a population to local extinc-

316

chapter 16

tion and the propensity of an area to be recolonized after extinctions both can be affected by the attributes of the individuals within a group and the habitat features in the local area. For example, behavioral, life history, or morphological traits can affect the response of a local population to an environmental perturbation that could lead to extinction. For highly mobile salmon species that spend their lives in more than one habitat type, some population locations may be key to providing nursery areas, migratory stopovers, or corridors (e.g., Groot and Margolis 1991). Determining the biological significance of differences in population attributes is an important step in identifying population characteristics that might be useful for setting conservation priorities. The concept of exchangeability has been introduced by evolutionary biologists to focus questions of population distinctiveness on adaptive differences and their underlying genetic variation (Templeton 1989, 1994; Crandall et al. 2000). Populations are exchangeable if rates of gene flow, natural selection, or genetic drift do not limit the spread of new genetic variants between populations.

Population Characteristics and Persistence in Salmonids Population diversity is important to ESU persistence for several reasons. First, the diversity of life history and other traits allows members of a species to use a wider array of environments than they could without it, allowing for a more effective use of resources and greater overall production. For example, varying the timing of adult returns to the river and spawning allows several salmonid species to use a greater variety of spawning habitats (Groot and Margolis 1991). Second, diversity buffers a species from short-term spatial and temporal changes in the environment. Fishes that have different characteristics have different likelihoods of persisting, depending on local environmental conditions. Therefore, the more diverse a population is, the more likely it is that some individuals will survive and reproduce in the face of environmental variation. For example, all the Pacific salmonid species except pink salmon contain within- and amongpopulation diversity in age at maturity. This life history diversity has the effect of spreading the population’s productivity out over several years, thus buffering the populations from poor environmental conditions or catastrophic losses in any particular year. Third, genetic diversity provides the raw material for surviving long-term environmental changes. Salmonids regularly face cyclic or directional changes in their freshwater, estuarine, and ocean environ-

recovering species

317

ments due to natural and human causes, and genetic diversity allows them to adapt to these changes. For example, it has been hypothesized that river-type sockeye salmon are essential for species survival during times of glacial advance—when the more highly adapted (and currently more abundant) lake forms go extinct in areas covered by ice (Wood 1995). Pacific salmonids generally home to their natal spawning streams, and there is considerable evidence that this homing behavior has facilitated the evolution of local adaptations (reviewed by Taylor 1991). Conserving locally adapted populations may be particularly important for promoting species-level viability, because a locally adapted population may be difficult to replace once lost. For example, Zinn et al. (1977) examined the susceptibility of four chinook salmon populations to the freshwater myxosporean parasite Ceratomyxa shasta. Three of the populations originated from the Columbia River Basin, where the infectious stage of the parasite is present, and the fourth originated from the Trask River on the Oregon coast, where the parasite is absent. The three Columbia River populations were all resistant to the parasite, whereas the coastal population was highly susceptible. Differential resistance to disease provides a clear example of how the nonexchangeability of populations needs to be taken into account in setting recovery goals.

Population Features and Regional Conservation Planning for Viable Salmonid ESUs For salmon recovery planning, determining how many populations are necessary for ESU persistence is difficult enough because of the lack of information with which to describe parameters for quantitative models. Incorporating additional conservation goals, such as diversity and spatial structure, into quantitative ESU viability analyses is even less likely to be fruitful because of large gaps in information. Instead, we are developing an approach that generates a range of options for salmon recovery by choosing sets of populations that achieve ESU-wide conservation targets. In particular, we are modifying what are known as “reserve siting algorithms” to help prioritize among populations for inclusion in a viable ESU. Siting algorithms traditionally have been used to assign priorities for the protection of sites aimed at maximizing biodiversity in terrestrial and marine systems (Kirkpatrick 1983; Margules et al. 1994; Dinerstein et al. 1995; Sullivan and Bustamante 1997; Ward et al. 1999; Beck et al. 2000; Leslie et al. 2002). Our within-species application is based on the same

318

chapter 16

principles, but the units in this case are populations (instead of species or habitat types) and the conservation goal is species viability (instead of biodiversity). The siting algorithm we apply uses a relatively new and flexible optimization tool to select populations whose collective characteristics achieve conservation targets we specify at the ESU level (Kirkpatrick et al. 1983; Ball 1999; Possingham et al. 2000; Leslie et al. 2002). The usefulness of siting algorithms in ranking populations can be demonstrated using information we are gathering for recovery planning in the Puget Sound chinook ESU in Washington State (see fig. 16.4). Chinook salmon in Puget Sound were listed as threatened under the ESA in 1999 (NMFS 1999e). In 2000, the NMFS convened a recovery team to develop de-listing criteria for the ESU. As mentioned previously, the recovery team has identified 21 demographically quasi-independent populations of chinook within the ESU (PSTRT 2001). An important question for designing ESU recovery goals is, which populations should be given highest priority for protection or restoration efforts? In other words, what combination of population attributes will result in a viable ESU? In the example presented here, we use information from five population attributes to select sets of populations that satisfy ESU-wide target levels of those attributes. The targets themselves are chosen through a combination of biological and policy criteria, and in this example we do not attempt to quantify how the targets affect the viability of the ESU. Rather, the purpose of this example is to illustrate how considering a number of conservation goals for the ESU affects the number of populations necessary for ESU recovery. Simple demographic models suggest that 2 to 10 populations are necessary to achieve goals for ESU persistence, assuming that populations experience identical and independent extinction risks. We know that populations are not identical (populations are diverse in genetic and life history traits), nor are they likely to experience independent extinction risks (because of correlated population dynamics). The example in this section illustrates how siting algorithms can be used to ask how many more populations are needed to account for diversity and spatial structure goals at the ESU level. We used the reserve design package MARXAN v2.1 (Ball 1999) to select populations within the geographic region that contains the Puget Sound chinook ESU. MARXAN is designed to choose a set of sites (in this case, populations) from a larger array of potential sites; site selection is based on site attributes, specified values associated with particular attributes, and the costs associated with not achieving regionwide targets. The user specifies a region-wide (ESU-wide) tar-

recovering species

319

get for each population attribute that must be represented in the final set of populations chosen, and MARXAN selects the smallest set of populations that achieves the ESU-wide target at the lowest cost. We used the 21 populations within the Puget Sound ESU as our sites and the presence/absence or value of each of the five attributes we were interested in conserving in a recovered ESU (table 16.2). The five population attributes are (1) estimated abundance needed for population viability, (2) the proportion of juveniles that emigrate as subyearlings and yearlings, (3) the time of year adults return to fresh water, (4) the genetic composition of the population (based on 21 polymorphic allozyme loci), and (5) the geographic region within the ESU in which the population occurs. As we obtain more data, we expect to add other biological attributes to the analysis, such as population-specific productivity and growth rates that result in persistence, the likelihood and intensity of threats, and the expected connectivity of populations in a watershed. We established conservation goals for each population attribute based on its estimated contribution to ESU viability or diversity goals. Because targets ultimately involve a combination of biological and policy choices, the values we use in this example are meant to bracket a range of possible conservation goals whose consequences for population selection can be explored using this approach. In addition to higher ESU-wide goals, we explore target minima to ask how many population combinations can achieve ESU conservation goals when the number of populations needed for ESU viability is close to the minimum as estimated from simple demographic models. We estimated the population abundance necessary for viability using a simple population viability analysis that incorporates information on population size, trend in abundance, and variation in abundance for each of the 21 populations (Dennis et al. 1991; Holmes 2001). The recovery team for Puget Sound chinook is currently in the process of exploring the best sets of parameter values for the extinction risk model that estimates viable population sizes. For this example, we chose to use results from model runs using a quasiextinction threshold that varies with the size of the watershed in which the population occurs. The model predicts the number of fishes needed to avoid population extinction within 100 years (McElhany and Payne in prep.). The acceptable level of risk used in these runs was a 95% probability of not reaching the quasiextinction threshold. One thousand simulated population trajectories were generated for each population to estimate the minimum viable size. We assigned each population the minimum viable abundance estimated from the quantitative extinction model and then set ESU-wide targets based on summed contri-

320

chapter 16

TABLE 16.2 Evolutionarily Significant Units and Listing Status of Independent populations of chinook salmon in Puget Sound N refers to the number of naturally produced adults spawning in the wild that are necessary for the population to avoid extinction; proportion subyearling emigrants refers to the proportion of juveniles that migrate to sea as subyearlings (sy) as opposed to yearlings (y); river entry refers to the time of year adults enter the river to spawn—S/S are adults that return in spring and summer, S/F are adults that return in summer and fall. Targets for each characteristic are ESU-wide values that form the basis for population selection in the reserve siting algorithm. Targets span relatively stringent requirements for ESU recovery and minimal ESU-wide goals.

Population

N

Proportion Subyearling Emigrants

N Fork Nooksack S Fork Nooksack Suiattle Upper Cascade Upper Sauk Lower Sauk Lower Skagit Upper Skagit S Fork Stillaguamish N Fork Stillaguamish Snoqualmie Skykomish Cedar N Lake Washington Duwamish-Green Puyallup White Nisqually Skokomish Dungeness Elwha Total Targets

4000 2000 4000 2000 4000 4000 3000 3000

0.6 0.4 0.4 0.4 0.4 0.6 0.7 0.4

S/S2 S/S S/S S/S S/S S/F S/F S/F

1 2 3 3 3 3 3 3

1 1 1 1 1 1 1

3000

0.9

S/F

3

1

4000 2000 3000 3000 2000 4000 4000 4000 4000 5000 3000 4000 71,000 50,000 30,000

0.9 0.7 0.4 0.7 1 0.8 0.9 0.4 0.9 0.9 0.9 0.6 14 sy, 7 y 10 sy, 5 y 2 sy, 2 y

S/F S/F S/F S/F S/F S/F S/F S/S S/F S/F S/S S/F 14 S/F, 7 S/S 7 S/F, 6 S/S 2 S/F, 2 S/S

3 3 3 4 4 4 4 5 4 4 1 1 2, 2, 10, 6, 1 1, 1, 4, 3, 1 1, 1, 2, 2, 1

1 1 1 2 2 2 2 2 2 3 3 3 12, 6, 3 4, 3, 2 2, 2, 2

River Entry

Genetic Group

Geographic Region

recovering species

321

butions from each population at its viable population size. In other words, we assumed that if a population was selected for ESU recovery goals, we could manage it to achieve viable abundance levels. The total estimated abundance of the Puget Sound ESU if all populations contain enough fishes to have a negligible risk of extinction is 71,000 naturally spawning adults (see table 16.2). For this example, we explored two different ESU abundance targets: 50,000 fishes and 30,000 fishes, which are approximately 10% and 4% of the estimated historical abundance of chinook in the Puget Sound area, respectively (Myers et al. 1998). These ESU abundance targets are consistent with the previously estimated range of the number of populations required in a viable ESU (see previous section). If achieving a numerical target for the ESU were the only conservation goal, no additional tools would be needed for choosing sets of populations that meet such a goal, since combinations of populations that add up to the ESU-wide goal can be generated directly from table 16.2. However, the siting algorithm allows us also to include diversity and spatial distribution goals into criteria for population prioritization. To prioritize populations for protection or restoration, we used three different indicators of chinook population diversity: the age of juvenile emigration, the timing of river entry, and the genetic composition of each population. We chose to focus on these traits because data were readily available for many populations, and each trait was expected to have adaptive significance. Most chinook in Puget Sound streams emigrate to saltwater habitats during their first year of life (i.e., as subyearlings), but some streams have a fraction of yearling emigrants (Marshall et al. 1995; Myers et al. 1998). Fishes that exhibit different ages at emigration typically spend different amounts of time in freshwater and estuarine rearing habitats, and this observed variation appears to have both genetic and environmental components (Randall et al. 1987; Clarke et al. 1992). The fitness consequences of these alternative life histories are not well understood, but differences in growth rates, morphology, and behavior of the different life history types have been documented (Carl and Healey 1984; Taylor and Larkin 1986; Cheng et al. 1987; Taylor 1990a, b). We set ESU-wide targets for age at emigration by adding the “equivalent” number of populations of each life history type for the entire ESU. We treated the proportion of each emigrant type in a population as a proportion of an “equivalent” population in the ESU with that emigrant age, then added the fractions of subyearling and yearling migrants in each population to generate an ESU-wide number of populations with each life history type. For the ESU, there are 14 subyearling and 7 yearling migrant-equivalent populations. Targets for the

322

chapter 16

recovered ESU were set at (1) a stringent target of 10 subyearling migrant-equivalent populations and 5 yearling migrant-equivalent populations and (2) a minimum of 2 populations of each emigrant type. The timing of river entry varies considerably within and among populations of the Puget Sound chinook ESU (WDF et al. 1993; Myers et al. 1998). With some exceptions, chinook salmon that enter the river in summer and fall tend to occupy the lower parts of watersheds, whereas spring and spring/summer runs occupy the upper reaches. Differences in run timing among populations are believed to be influenced genetically and are often adaptive (Miller and Brannon 1981; Groot and Margolis 1991). Therefore, it may be particularly important to focus conservation efforts on the few spring-run populations that still remain in the ESU, because if these populations are lost the adaptive characteristics and habitats they currently occupy might be lost to the ESU for a considerable period. We classified populations into two run-timing categories—summer/fall and spring/summer (see table 16.2)—and tallied the total number of populations of each: 14 and 7, respectively. We set two different ESU-wide targets for river entry: (1) a stringent target of seven populations of summer/fall and six populations of spring/ summer and (2) a minimum of two populations of each run-timing type. The higher proportion of spring/summer populations in the target relative to extant spring/summer populations accounts for the likely reduction in spring/summer runs relative to historical characteristics of the ESU (Myers et al. 1998). Chinook in Puget Sound can be grouped according to similarities in genetic composition at 21 polymorphic allozyme loci (Marshall et al. 1995). Distinct groups emerge consistently from analyses of genetic data using several genetic distance measures and clustering algorithms (A. Marshall and C. Busack, WDFW unpub. data; PSTRT 2001). Targets for the genetic groupings are based on the total number of populations contained within each genetic class (see table 16.2). The final population characteristic that we considered was the geographic region in which the population occurs. The rationale for this characteristic was twofold: populations more spread out in space are less likely to fall victim to spatially correlated threats, and a greater diversity of selective environments (and therefore phenotypic diversity) is likely to be represented in a broader geographic area. We divided the geographic region encompassing the Puget Sound chinook ESU into three areas: north Sound, mid-south Sound, and the Hood Canal–Strait of Juan de Fuca region (fig. 16.4). There are 12 populations in the north Sound, 6 in the mid-south Sound, and 3 in the

recovering species

323

Figure 16.4 Geographic distribution of the Puget Sound chinook salmon ESU. Twenty-one demographically quasi-independent populations of chinook have been identified in the ESU (PSTRT 2001).

Hood Canal-Strait region. Targets for each region were (1) four, three, and two populations, respectively, and (2) two populations in each geographic region (see table 16.2). Because target values for each attribute greatly influence the outcome of the algorithm’s selection process, we performed two analyses

324

chapter 16

to select populations within the ESU. First, we asked the algorithm to choose the “best” solution under a given set of ESU-wide targets. The best set of populations was defined as the solution with the lowest cost in terms of the number of populations and any penalties for not achieving the target value for each attribute. The second set of analyses bracketed a range of target values for each attribute and asked which populations were chosen most frequently under a wide range of ESU-wide conservation targets. In this second analysis, we tallied the proportion of times a population was chosen under a variety of target values and expressed that proportion as a summed “irreplaceability score,” whereby populations with higher scores are more critical to the success of the ESU in attaining its conservation targets (e.g., Leslie et al. 2002). In other words, the higher a population’s summed irreplaceability score, the more likely it is to be a high priority site (and hence not expendable), regardless of specific conservation goals. We ran an irreplaceability analysis by recording the number of times particular populations were chosen out of 1000 runs for each of several ESU-wide target values. To evaluate the effectiveness of the siting algorithm, we compared the sets of populations chosen by the siting algorithm to randomly selected sets in their ability to meet the most stringent conservation goals (see table 16.2; target 1) and the minimum goals (target 2). We randomly selected 1000 sets of 15 populations or 8 populations for comparison to targets 1 and 2, respectively (these were the sizes of the best sets found by the algorithm under the two scenarios). The performance of a random set relative to each of the five criteria specified in the target was measured as a scaled deviation from the target value. The product of these scaled deviations gave an “effectiveness” score with values between 0 (population sets that failed to include any representatives of the required population types) and 1 (population sets that met or exceeded all five criteria). The “best” sets found by the algorithm achieved all five criteria under both target scenarios and so had effectiveness scores of 1.

Results of Siting Algorithms—Ranking Chinook Populations for Protection Between 8 and 15 populations within the Puget Sound chinook ESU are needed to achieve the conservation targets we explored in this example. The populations whose collective characteristics best satisfied our ESU-wide targets depended on the conservation scenario explored (table 16.3).

recovering species

325

TABLE 16.3 The “best” set of populations chosen under alternative ESU recovery scenarios explored using MARXAN. The scenarios contrast the ESU-wide abundance of naturally produced spawners (i.e., N ⳱ 50,000 and 30,000) and the number of populations with different life history types. “Stringent” refers to targets for life history types that are relatively high. “Minima” refers to targets for life history types that require only 1–2 populations per type (see table 16.2 and text for actual target values). 50K, Stringent

50K, Minima

30K, Stringent

30K, Minima

NF Nooksack Upper Sauk Skokomish White Elwha Puyallup NF Stillaguamish SF Stillaguamish Nisqually Lower Sauk Suiattle Lower Skagit Cascade Snoqualmie Upper Skagit

NF Nooksack Upper Sauk Skokomish White Elwha Puyallup NF Stillaguamish Cedar Nisqually Lower Sauk Suiattle Duwamish-Green North Lake Wash. none none

NF Nooksack Upper Sauk Skokomish White Elwha Puyallup NF Stillaguamish Cedar Nisqually Lower Sauk Suiattle Lower Skagit SF Nooksack Snoqualmie Dungeness

NF Nooksack Upper Sauk Skokomish White Elwha Puyallup NF Stillaguamish SF Stillaguamish none none none none none none none

The “best” population sets are those that achieve the ESU-wide targets with the fewest number of populations. The first seven populations listed in table 16.3 were chosen in every scenario, but the Cascade, north Lake Washington, Duwamish-Green, South Fork Nooksack, upper Skagit, and Dungeness populations were chosen only in one out of four conservation scenarios explored. The effectiveness of the siting algorithm is greater than random selection of populations (fig. 16.5). Random population sets achieved the ESU-wide conservation target fewer times than did those population sets chosen by the siting algorithm, especially for the targets involving minimum goals for ESU recovery. Because it is often difficult to come up with a biological rationale for distinguishing among conservation targets or even agreeing on how best to characterize populations, it is most informative to examine those populations that are chosen most frequently under all conservation target scenarios explored. Not surprisingly, the scenarios with more stringent requirements for ESU recovery result in more populations with high summed irreplaceability scores than those sce-

326

chapter 16

Figure 16.5 Effectiveness of MARXAN, the siting algorithm, in selecting populations that achieve ESU-wide target conservation values. The ability of 1000 randomly chosen sets of populations to achieve conservation targets is compared with those population sets chosen by the siting algorithm. By definition, the 1000 population sets chosen by the siting algorithm achieved conservation targets 100% of the time. Target 1 represents stringent ESU-wide recovery criteria; target 2 contains minimal criteria.

narios requiring fewer numbers of fishes and fewer populations with particular life history types (fig. 16.6). Irreplaceability results tallied over all four conservation scenarios suggest that of the 21 populations that comprise the Puget Sound ESU, 7 specific populations stand out consistently as necessary for recovery (fig. 16.7). Of course, as we refine our conservation goals and consider more population characteristics, the details of this answer may change. Even at this early stage of analysis, however, it is informative to identify the North Fork Nooksack, White, Dungeness, Suiattle, and Skokomish populations as among the top seven in terms of their consistent presence in ESU recovery scenarios.

Summary Establishing goals for species viability requires that we know how many and which populations are critical to species persistence. In de-

recovering species

327

Figure 16.6 Distribution of summed irreplaceability scores for populations in 4 ESU-wide conservation scenarios explored with MARXAN. Summed irreplaceability scores are the number of times a population is chosen out of 1000 iterations of the simulated annealing siting algorithm for each set of conservation targets. Results are depicted from two extremes of ESU abundance and life history conservation targets.

veloping recovery criteria for Pacific salmon ESUs listed under the Endangered Species Act, our task in part is to determine how many and which populations are necessary for the long-term viability of the ESU. As we have shown here, at this time we do not have enough information to quantitatively model the relationship between ESU viability and the number and diversity of populations in the ESU with

328

chapter 16

Figure 16.7 Irreplaceable populations of chinook salmon in the Puget Sound ESU. The primary spawning areas of each population are enclosed within ovals on the map. The 7 highlighted populations indicate those chosen most frequently in all conservation scenarios explored using the siting algorithm. (The highlighted populations represent those that most commonly were chosen to achieve the conservation targets we specified in this example; they do not necessarily reflect those that may ultimately be included in ESU-wide recovery scenarios.)

recovering species

329

much confidence. Instead, we are developing quantitative recovery goals for individual populations (which we can model with some assurance), as well as ESU-wide recovery goals in terms of how many and which of those populations must meet their numerical recovery goals. In effect, such an evaluation is equivalent to asking whether some populations are expendable in their contributions to species viability. Whether some populations are eventually determined to be truly expendable with respect to the goals of the ESA will probably vary widely among ESUs. Some threatened ESUs are currently broadly distributed and abundant; these ESUs are listed not because they are in immediate danger of extinction but because they will reach that state if present trends of habitat loss or other factors continue. For these ESUs, it will not be surprising if a recovery team determines that some populations are less important for ESU viability than others. In the often zero-sum game of conservation planning, these populations may in effect be considered expendable. In contrast, some endangered ESUs currently exist in only one or a few populations. For these ESUs, it seems highly unlikely that any existing population could be considered expendable for recovery purposes; and in fact, some of these ESUs may require the establishment of additional populations to be considered viable. In the end, choices will be made; the question is whether scientific concerns will play a role in any of these choices. By providing general guidance from population biology and irreplaceability conclusions in the form of multiple, essentially biologically equivalent scenarios, conservation decisions are less likely to be determined solely by politics and convenience.

Acknowledgments We thank Eric Buhle and Lisa Holsinger for help with analyses and figures. Hugh Possingham, Ian Ball, Heather Leslie, and Sandy Andelman have helped and continue to work with us in developing siting algorithms for salmon conservation applications. Jeff Hard, Lisa Holsinger, Peter Kareiva, Larry Crowder, and Robin Waples provided useful comments on a previous version of this chapter.

Chapter 17

m Virus Specificity in Disease Systems: Are Species Redundant? Alison G. Power and Alexander S. Flecker

Despite our increasing recognition of the key role that microbes play in ecosystem processes (e.g., Giller et al. 1997; Mills and Bever 1998; van der Heijden et al. 1998), concern over the loss of biodiversity in ecological communities rarely focuses on microbes— particularly pathogenic microbes which are more often targeted for eradication than preservation (but see Windsor 1995). Yet it is becoming clear that disease-causing microbes may be important in shaping biotic communities (Real 1996a), and their diversity significantly influences populations, communities, and ecosystems. We are surrounded by examples of the influence of disease on the landscape. One of the most dramatic examples is the loss of mature chestnut and elm trees in European and North American deciduous forests following the introductions of chestnut blight (Cryphonectria parasitica) and Dutch elm disease (Ophiostoma ulmi/O. novo-ulmi), respectively (for reviews, see Jarosz and Davelos 1995; Real 1996b;

virus specificity in disease

331

Hubbes 1999; Griffin 2000). Most pathogens probably do not have such catastrophic impacts on their host species or landscape but instead have much more subtle and therefore less easily recognized effects. We may be most likely to detect these effects when species are already under stress from other factors. For example, an epidemic of canine distemper in black-footed ferrets has been identified as the proximate cause of their extinction in the wild (Williams et al. 1988). Similarly, extinctions of native birds in Hawaii have been attributed to avian malaria, although current patterns of species distributions suggest that disease probably interacted with habitat loss and other stress factors (Pimm 1991). Our understanding of the consequences of fungal pathogens for natural plant communities has grown considerably over the past decade or two (Jarosz and Davelos 1995), but we know less about the effects of plant viruses despite their ubiquitous distribution in plants. Several recent studies have stressed the prevalence of viruses in natural plant populations (e.g., Power and Remold 1996; Ooi et al. 1997; Cheng and Jones 1999; Ooi and Yahara 1999; Raybould et al. 1999). Moreover, Maskell et al. (1999) emphasize that the failure to detect viruses in wild plants could be due to high mortality of infected individuals. But what are the consequences of viruses for plants in natural communities? Greenhouse and field experiments have demonstrated that virus infection can significantly reduce the growth, survivorship, and reproduction of nondomesticated plants (Friess and Maillet 1996, 1997; Funayama et al. 1997; Maskell et al. 1999; Funayama et al. 2001; Funayama-Noguchi 2001; Power 2002). For example, Maskell et al. (1999) found reduced seed output and higher mortality in wild cabbage (Brassica oleracea) infected with turnip mosaic potyvirus or turnip yellow mosaic tymovirus in the field. Competitive abilities are also affected by infection. In greenhouse studies, the fitness and competitive abilities of infected purslane (Portulaca oleracea) (Friess and Maillet 1996) and chickweed (Stellaria media Vill.) (Friess and Maillet 1997) were reduced significantly by infection with cucumber mosaic virus. Similarly, the ability of wild oats (Avena fatua) to compete effectively against other grasses was reduced significantly by infection with barley yellow dwarf virus (Power 2002). Although data from field studies are still scarce, there is every reason to think that viruses can have important effects on plant populations and communities. In this chapter, we use plant viruses as a model system for analyzing species dependencies and redundancy. Evaluating the dependence of one species on another is often difficult, yet it is essential for predicting the consequences of species losses to community structure and function. Assessing dependence may be straightforward for some

332

chapter 17

cases of pairwise interactions, such as host-specific insect parasitoids. For more complex interactions, however, documenting dependencies can take years of careful observation and field experiments. In contrast, determining the specificity of a plant virus is relatively straightforward, and such viruses depend quite strongly on their host plants and on their vectors. Moreover, we have reliable data on such dependencies for hundreds of species, allowing us to ask questions that are difficult to ask in many other systems.

Plant Viruses and Specificity Plant viruses are obligate symbionts that depend on their host plants for reproduction. To colonize new hosts, most plant viruses also require vectors, which may include insects, mites, nematodes, or fungi. Although some viruses can be transmitted intergenerationally from parent to offspring in the seed, transmission by vectors clearly plays an important role in determining fitness for most viruses, and it can result in extremely rapid spread of a virus. Transmission within a host generation, such as that carried out by vectors, is known as horizontal transmission, whereas transmission between host generations is known as vertical transmission. Viruses vary in the number of possible hosts that they can infect, whether by horizontal or vertical transmission, and in the number of vectors that can transmit them effectively. Given the important role that viruses may play in plant communities, how redundant are these disease systems? What patterns of host and vector specificity should we expect for plant viruses? There are several plausible scenarios of specificity that have considerably different implications for patterns of redundancy in disease systems. We present two straightforward scenarios as alternative hypotheses of specifity. First, if the biological barriers to virus invasions of plants or vectors are great, we might predict that most viruses would show host and vector specificity, as depicted in figure 17.1a. To replicate, viruses must take over the reproductive machinery of a cell, a process that requires specific receptors. Similarly, recent research has demonstrated that vector transmission requires highly specific viral transport mechanisms in the vector (Gray and Banerjee 1999; van den Heuvel et al. 1999). These biological constraints would point to low redundancy, such that viruses would likely depend on one or a few species of hosts and vectors. On the other hand, because viruses are completely dependent on their symbionts for reproduction, we might expect that it would be advantageous for viruses to have broad host ranges and an abun-

virus specificity in disease

333

Figure 17.1 Predicted proportion of plant viruses that would fall into 9 categories of host and vector range (A) when host and vector range are highly constrained by biological barriers to infection; and (B) when biological barriers to infection are not severe.

334

chapter 17

dance of potential vectors. Because viruses show rapid evolution and substantial genetic diversity generated by high mutation rates and frequent recombination (Domingo and Holland 1997; Roosinck 1997), they may adapt quickly to new hosts and vectors. In this case, we would predict that most viruses would be host and vector generalists, as shown in figure 17.1b. This scenario would result in high functional redundancy in disease systems, such that diversity of hosts and vectors could act as “insurance” against spatial and temporal fluctuations that might influence virus epidemiology. This idea is consistent with recent theory, which supports the general hypothesis that species diversity and functional redundancy act as “insurance” against environmental fluctuations that may affect community stability and productivity (Doak et al. 1998; Yachi and Loreau 1999; Ives et al. 2000). Under this scenario, we would also expect that relatively few viruses would be either host specialists or vector specialists, and that specialism for both hosts and vectors would be extremely rare (see figure 17.1b). To evaluate redundancy in plant virus systems, we analyzed the degree of diversity in hosts and vectors of plant viruses, addressing the following questions: 1. From the point of view of the virus, is there more redundancy in available hosts or available vectors? That is, are viruses more likely to be specific to hosts or to vectors? 2. Do taxonomic relations between vectors and viruses influence system redundancy? 3. Does the degree of intimacy between virus and host or between virus and vector influence patterns of specificity and redundancy?

Sources of Data We used the VIDE database on plant viruses (Brunt et al. 1996) to address the questions just outlined. This database is the most current listing of plant virus species and contains 1673 total viruses, including 910 unique virus species. For each of the 910 unique viruses in the database, we recorded taxonomic information (species, genus, family) and type of transmission (vector, seed, mechanical, graft). The mechanism of transmission is known for only 59% of these 910 species; of the 59%, 89% are transmitted by vectors. For the 474 viruses that are known to be transmitted by vectors, we compiled data on the type of vector (insect, mite, nematode, fungus); whether the virus is retained by the vector (persistent, semipersistent, nonpersistent); whether the

virus specificity in disease

335

virus propagates in the vector; the number of vector species, genera, and families; and the number of host species, genera, and families. For each virus, the range of host species or vector species was categorized as specific (single species), narrow (2–5 species), or broad (ⱖ6 species). Equivalent categories for genera and families used the same numerical boundaries. For each of the questions listed earlier, we used contingency tables and chi square (␹2) tests to assess differences among host and vector ranges at the level of species, genus, and family. There are several potential biases in this database that may influence the implications of our analysis. First, the majority of data on plant viruses comes from economically important viruses—in most cases, viruses that infect crop plants. However, the plants that comprise the host range are primarily non-crop plants. Second, reports for hosts and vectors do not all come from natural infections in field populations. Experimental inoculations of feasible hosts and transmission trials with a range of possible vectors may provide information about the potential for associations, but they do not guarantee that such associations actually occur in the field under natural conditions. Thus, the host ranges and vector ranges reported in the database may be overestimates. On the other hand, the logistics of carrying out such trials restricts the number of hosts or vectors that can be tested, and therefore host or vector diversity may be underestimated for some species, especially those with less economic importance. Finally, recent advances in virus molecular genetics are enhancing our understanding of taxonomic relationships among viruses, but species determinations are still somewhat in flux for some virus groups.

Vector Specificity versus Host Specificity Neither of our alternative predictions about patterns of host and vector specificity (see fig. 17.1) were fully supported by the data. Instead, we found high levels of vector specificity accompanied by low host specificity. In general, viruses have few vectors and many hosts, and the largest proportion of viruses fall into the category of “host generalist/vector specialist” (fig. 17.2). Viruses have more specialized relations with their vectors than with their hosts at the level of species (fig. 17.3a, ␹2 ⳱ 40.44, P ⬍ 0.0001), genus (fig. 17.3b, ␹2 ⳱ 37.44, P ⬍ 0.0001), and family (fig. 17.3c, ␹2 not obtainable because of empty cells in contingency table). A majority of vector-transmitted viruses (58.4%) have a single known vector species, whereas only 9.9% of such viruses have a single known host plant. In contrast, the proportion of host specialists among viruses that lack vectors (18.0%) is sig-

336

chapter 17

Figure 17.2 Proportion of 474 plant viruses that fall into the 9 categories of host and vector ranges.

nificantly higher (␹2 ⳱ 6.59, P ⬍ 0.05). Although many viruses have a narrow range of vectors but a large host range, viruses with many vectors never have a narrow host range. That is, no viruses fall into the category of “host specialist/vector generalist” (see fig. 17.2). Viruses with a narrow host range inevitably have a restricted set of vector species. Patterns of host and vector specificity differ significantly among types of vectors (host specificity: ␹2 ⳱ 17.09, P ⬍ 0.01, vector specificity: ␹2 ⳱ 45.64, P ⬍ 0.0001). Of the 474 viruses transmitted by vectors, 86.7% are transmitted by insects, 5.5% by fungi, 4.0% by nematodes, and 1.8% by mites. Because insects are the most common vectors, the patterns of specificity for the subset of insect-transmitted viruses are similar to those for all vectored viruses (fig. 17.4a). Again, viruses have much more specialized relations with their insect vectors than with their hosts at the level of species (␹2 ⳱ 41.82, P ⬍ 0.0001), genus (␹2 ⳱ 40.25, P ⬍ 0.0001), and family (␹2 not obtainable because of empty cells in contingency table). Among insect-transmitted viruses, 54.9% have a single vector species, and only 10.8% have a single host plant. Viruses that have more intimate associations with their insect vectors might be expected to be the most vector specific. The transmission modes of insect-vectored plant viruses have been categorized according to the intimacy of their association with the vector, ranging

virus specificity in disease

337

Figure 17.3 Relationship between number of hosts and number of vectors for 474 plant viruses. Each point represents one virus species. A. Host species vs. vector species. B. Host genera vs. vector genera. C. Host familes vs. vector familes.

338

chapter 17

Figure 17.4 Relationship between number of host species and number of vector species for (A) insect vectors, (B) fungal vectors, (C) mite vectors, and (D) nematode vectors. Each point represents one virus species.

from stylet-borne viruses (low intimacy) to propagative viruses (high intimacy). Stylet-borne viruses are carried on the mouthparts of the vectors and are known as “nonpersistent” because they are lost once a vector has fed on a host. Viruses borne in the foregut of the vector are “semipersistent” in their vectors. Circulative, “persistent” viruses pass through the insect gut into the hemolymph and then into the salivary glands via highly specific transport mechanisms and can be transmitted repeatedly to new plants. Propagative viruses are circulative viruses that replicate in the insect vector as well as the plant host, via relations with vectors that are thought to be highly specific. Our analysis indicates that circulative and foregut-borne viruses are somewhat more likely to be vector specialists than are stylet-borne viruses (61% and 59% vs. 44%; ␹2 ⳱ 10.64, P ⬍ 0.05). Circulative viruses that propagate in their vectors are not more likely than nonpropagative viruses to be absolute vector specialists, but they are more likely to have a narrow range of vectors (ⱕ5 species) and are less likely to be vector generalists (␹2 ⳱ 7.88, P ⬍ 0.05). Viruses transmitted by fungal vectors are even more specialized

virus specificity in disease

339

with respect to their vector relations than insect-transmitted viruses, since each of the 25 viruses in the VIDE database is transmitted by a single species of fungus; but only 12% of these viruses are host specialists (see fig. 17.4b). Mite-transmitted viruses show a similar pattern (fig. 17.4c): these 9 viruses are all vector specialists, but none are host specialists. Nematode-transmitted viruses are less likely to be vector specialists—50.0% have a single vector—and they are exclusively host generalists (fig. 17.4d). Host and vector specificities differ significantly among virus families (␹2 ⳱ 180.00, P ⬍ 0.0001 and ␹2 ⳱ 98.16, P ⬍ 0. 0001, respectively; fig. 17.5), despite the fact that only 53% of the viruses in this database have been assigned to a family. Vector type or mode of transmission cannot account fully for these differences among families. Although the viruses in some of these families are transmitted exclusively by a single vector type using a single transmission mode— such as with the insect-transmitted, circulative Geminiviridae (see fig. 17.5a)—viruses in other families are transmitted by several vector types, sometimes with different transmission modes. The Rhabdoviridae (see fig. 17.5b), for example, are transmitted propagatively by insects and mites. The Potyviridae (see fig. 17.5c) are transmitted by aphids, mites, and fungi, and the transmission mechanisms are distinct for the different vector types. There is no evidence, however, that virus families transmitted by more vector types are likely to have significantly more species of hosts or vectors (P ⬎⬎ 0.10). These patterns of vector and host specificity suggest that virus distribution is constrained more by the specificity of virus-vector relations than by the specificity of virus-host plant relations. Many viruses have a very narrow range of vectors but a large host range. In contrast, no viruses have a narrow range of host plants if they have many vector species (see fig. 17.2). The feeding range of the vector determines in large part the host range of the virus, suggesting that viruses can adapt to new hosts fairly readily. Moreover, in a number of well-known cases, expansion of the host range of insect vectors has been shown to increase the host range of the viruses that these vectors transmit (e.g., Goldbach and Peters 1994; Harrison and Robinson 1999). Recent studies of insect-transmitted plant viruses demonstrate highly conserved molecular motifs in viral genomes that regulate the specificity of insect transmission (Power 2000). In contrast, advances in our understanding of host plant response to virus infection reveal generalized patterns of host defense against a diverse array of viruses (Carrington and Whitham 1998; Waterhouse et al. 1999; Escaler et al. 2000).

340

chapter 17

Figure 17.5 Relationship between number of host species and number of vector species for different families of viruses: (A) Geminiviridae, (B) Rhadoviridae, and (C) Potyviridae. Each point represents one virus species.

As an extreme example, geminiviruses are typically transmitted by a single species of vector, yet many have extremely large host ranges (see fig. 17.5a). The range of diseases incited by geminiviruses has increased dramatically since the early 1990s, largely due to the introduction to the Americas of the Old World B-biotype Bemesia tabaci

virus specificity in disease

341

whitefly. B-biotype B. tabaci have an unusually broad host range and can transmit viruses among host plants that did not previously share insect vectors (Brown et al. 1995). Apparently, the introduction of this vector to the New World provided a novel opportunity for preexisting viruses to be transmitted from their wild hosts to a variety of crops. Recent molecular analyses of the whitefly-transmitted geminiviruses indicate striking levels of genetic diversity within virus species (Ooi et al. 1997; Sanz et al. 1999), diversity that presumably allows rapid adaptation to new hosts. In contrast, regions of the genome that encode the proteins responsible for insect transmission are much less variable. At the level of virus genera, there are no documented examples of different members of a virus genus having different insect transmission modes (i.e., one species of virus being transmitted in a styletborne manner and another being transmitted in a circulative mode, although some stylet-borne viruses may also be seed-transmitted). This pattern appears to result from a consistent evolutionary constraint, such that virus genera can be assigned to a particular insect transmission mode (Nault 1997). The consistency of transmission mode within a virus genus, and the generally greater specificity of vector relations compared with host relations, suggests that selection imposed by a requirement for efficient vectors may be more severe than that imposed by host plant defenses (Power 2000). This implies that the rate-limiting step for virus spread is attaining access to new hosts rather than overcoming the defenses of these hosts. Our analysis suggests that most plant viruses have substantial redundancy in the hosts that are available to them. In contrast, there is little redundancy in potential vectors for most viruses. Because the majority of viruses depend on vectors for transmission, virus epidemiology is delimited largely by the life history and behavior of the vectors. In particular, the host range and feeding preference of the vectors are likely to play a major role in determining the realized host range of the virus.

Seed Transmission and Host Specificity The low host specificity of most vector-transmitted viruses is less apparent when we look at seed-transmitted viruses. In addition to horizontal transmission by vectors, the other major biotic mechanism of transmission is via seed, whereby viruses are transmitted vertically from parent to offspring in the seed. Seed transmission requires that the virus enter the germ line, and only 131 viruses (14.4%) are known to be capable of this. Moreover, among viruses capable of seed trans-

342

chapter 17

mission, actual rates of transmission are often extremely low, suggesting that infection of the germ line is difficult to achieve or that there is strong selection against seed transmission. Ecological theory and laboratory experiments indicate that only those viruses with relatively low virulence should be restricted to vertical transmission modes such as seed transmission, since virus transmission requires that infected hosts survive until reproduction (Power 1992; Bull 1994; Messenger et al. 1999). Because seed transmission would seem to require a more intimate association between virus and host than would vector transmission, we tested the hypothesis that viruses with seed transmission have smaller host ranges than those without. Transmission by seed and vector are not mutually exclusive; certain viruses can be transmitted both by vectors and by seed. Therefore, we included in our analysis three categories of viruses: (1) viruses with seed transmission only, (2) viruses with vector transmission only, and (3) viruses with both seed and vector transmission. The hypothesis that seed-transmitted viruses should have greater host specificity than viruses that are not transmitted by seed was only partially supported by the data. Our analysis indicated that viruses that depend on seed transmission alone are significantly more likely to be host specialists than viruses with vector transmission (36.4% vs. 6.3%, ␹2 ⳱ 30.33, P ⬍ 0.0001; fig. 17.6), whether those viruses with vector transmission were also seed-transmitted (3.4%) or not (7.0%). Contrary to our expectations, seed-transmitted viruses that were also transmitted by vectors were more likely to have broad host ranges (93.2%) than viruses that depended exclusively on vectors for transmission (73.1%, ␹2 ⳱ 14.05, P ⬍ 0.001; see fig. 17.6). This pattern of host specificity may be due in part to the type of vector most likely to transmit seed-transmitted viruses. A significantly higher proportion of nematode-transmitted viruses were also seedtransmitted compared with viruses transmitted by insects, fungi, or mites (␹2 ⳱ 30.095, P ⬍ 0.0001; fig. 17.7). Because nematode-transmitted viruses are exclusively host generalists (see fig. 17.4d), these viruses may influence the overall pattern of host specificity of viruses transmitted by both seeds and vectors. In addition, among insect-transmitted viruses, stylet-borne viruses are more likely to also be capable of seed transmission than are viruses with other transmission modes (␹2 ⳱ 31.77, P ⬍ 0.0001), and stylet-borne viruses are also more likely to be host generalists (78.6%) than viruses with other transmission modes. Viruses with extremely intimate relations with their vectors, such as circulative and propagative viruses, are never transmitted by seed.

virus specificity in disease

343

Figure 17.6 Percentage of viruses with specialized, narrow, and broad host ranges, according to transmission mode.

Figure 17.7 Percentage of viruses that are seed transmitted, according to vector type.

344

chapter 17

Implications of Specificity for Species Losses, Community Structure, and Conservation Disease agents have been shown to be strong ecological interactors under certain conditions, and they can play important roles in structuring communities. What are the potential consequences of the loss of plant viruses for plant communities? The loss of viruses could release host plants from suppression and significantly affect their competitive relations with other members of the plant community. Because virus infection can have strong effects on the populations of other plant pathogens and herbivores (Power 1992), the loss of viruses could also alter the remaining community of parasites on these host plants (Esch et al. 1990), leading to second-order effects on hosts. The consequences of the loss of plant viruses are likely to be complex; the case of wild oats infected with barley yellow dwarf virus may serve as an illustration. Wild oats are highly invasive weeds in the western United States in both natural and agricultural systems (Holm 1977). Introduced wild oats compete heavily with native grasses in natural grasslands, causing declines in native grass populations and hampering current efforts at grassland restoration (Barbour et al. 1993; Dyer and Rice 1997). Our studies show that the growth and reproduction of wild oats are reduced substantially by virus infection and that infected plants are less successful competitors against other plant species (Power 2002). Virus infection is extremely common in wild oat populations, suggesting that the fitness and competitive abilities of these grasses are currently being suppressed by infection in the field. Loss of the virus, or a significant decline in its abundance, could result in the release of wild oats from ecological constraints normally imposed by virus infection. In turn, this could result in a greater impact of wild oats on native plant populations, potentially modifying the diversity and structure of native grassland communities. Because viruses are obligate symbionts that lack a free-living stage, the loss of virus species is likely to occur through the loss or reduction of host or vector populations. The likelihood of species loss depends on the specificity of relations between viruses, their hosts, and their vectors. Host-specific plant viruses may be lost as a result of the extinction or decline in abundance of the host plant. In this case, the consequences of virus loss for plant communities might be minimal, since the major impact would be due to the loss of the host plant species itself. On the other hand, for host-generalist viruses, the likelihood of losing the virus species via host loss is probably low, but any losses in host plant species might significantly change the dynamics

virus specificity in disease

345

of virus infection in surviving host species, potentially affecting plant competition and the structure of plant communities. The pattern of extreme vector specificity revealed by our analysis suggests that virus species are more likely to be lost through extinctions (or declines in the abundance) of vector species, particularly insects, than through host plant extinctions or declines. This conclusion is disconcerting because in general we have a limited ability to detect extinctions of insects or threats to their survival, with the possible exception of some well-studied butterfly species. One of the problems is the “taxonomic impediment” (Wilson 1985), since the vast majority of insect species are undescribed, and even described species may be very difficult for field biologists to identify. Certainly, our knowledge of the major families of insect vectors (aphid, leafhoppers, planthoppers, and whiteflies) is largely concentrated on species of economic importance, such as crop pests, which typically are common, abundant, and distributed widely. How effective are current strategies for insect conservation? The conventional approach to biodiversity conservation is based on habitat and relies almost exclusively on vegetation characteristics. The proposition that we can best conserve populations of insect herbivores by conserving their host plants is well established in conservation programs around the world (Collins and Thomas 1991; Scott et al. 1993; Panzer and Schwartz 1998). Entomologists have sometimes challenged the efficacy of this approach for insect conservation. Surprisingly, there are relatively few data to test the assumption that a vegetation-based approach is effective for insects. In the only study to date that included insect vectors, Panzer and Schwartz (1998) found that plant community characteristics would be useful indicators of leafhopper species diversity in the tallgrass prairie. Their results suggest that a vegetation-based approach to conservation that uses plant species richness and plant community richness as biodiversity indicators is likely to be relatively effective in conserving leafhopper species (Panzer and Schwartz 1998). Although our analysis suggests that most virus species would be lost through losses or declines in vector populations, highly host-specific viruses, such as viruses restricted to seed transmission, might be lost through the extinction of plant species. In general, the extinction rate of host-specific parasites is probably similar to the extinction rate of their hosts (Stork and Lyal 1993). For example, host-specific parasites of the black-footed ferret appear to have gone extinct about the time that the single surviving population of ferrets was brought into captivity (Gompper and Williams 1998). There were deliberate efforts to reduce parasite loads in captive ferrets, since disease was one of

346

chapter 17

the proximate causes of their demise in the wild (Williams et al. 1988). Although there are no well-documented examples of plant extinctions leading to pathogen extinctions, vegetation-based conservation may be effective for the direct conservation of plant viruses as well as for indirect conservation by means of the conservation of insect vectors, as discussed earlier.

Conclusions We have argued here that microbial pathogens such as plant viruses can have powerful, though sometimes cryptic, effects on natural plant populations and communities. The loss of such pathogens has the potential to alter community structure and function dramatically. Our analysis suggests that viruses are more likely to become extinct through the loss or decline of their vectors than through the loss of their host plants, because typically they are highly dependent on a small number of vector species and because strategies for conserving these vectors are not well developed. At this point, vegetation-based conservation approaches are likely to be the most effective, both for the direct conservation of plant viruses in their hosts and for the conservation of their insect vectors.

Conclusion

m Bob Paine’s Contributions to the Science of Assessing Species Importance: Past, Present, and Future?

When community ecology has completed its job, we will be better informed about the likely consequences, both long-term and short-term, of any particular extinction. Of course, there will always be surprises, and prediction will never be perfect; some core of unknowability will linger in any effort at prediction. Currently, however, we have much work to do in developing both the theory and the empirical foundations necessary for accurately predicting the aftermath of species losses. At times, our uncertainty regarding the consequences of losing species can seem overwhelming, even paralyzing. As a cure for the gloom cast by our ignorance, it is useful to remind ourselves of the progress we have made in the last several decades. Because this volume grew out of a symposium honoring Robert T. Paine’s career, there is no better way of reviewing recent progress than to recount some of his most noteworthy contributions to the field of ecology. One well-supported generalization regarding extinction is that the

348

conclusion

disappearance of a dominant top predator is likely to have profound consequences for an ecosystem. This concept can be traced directly back to Paine’s classic sea star removal experiment (Paine 1966), which demonstrated that the removal of Pisaster ochraceus converted a diverse rocky intertidal shore into a monotonous coverage of mussels. Bob coined the concept “keystone predator” to describe this effect, and top predators that mediate competition among prey species stand out as predictably important species. Bob and his students have extended their thinking about “species importance” well beyond their early emphasis on top predators. Using field manipulations of species abundance, Bob has championed the measurement of interaction strengths as a device for quantifying the importance of any species, regardless of trophic position (Paine 1992). The idea is simple, but powerful: measure densities under control conditions (with all species present), then delete one species and again measure the densities of all remaining species. Coefficients that quantify pairwise interaction strengths between the removed taxa and all other species are calculated as the difference between control and manipulated densities, divided by the original density of the removed species (to obtain a per capita effect). There are subtleties and problems with this approach (e.g., how long should one wait? how should one deal with variability in space and time?), but the summation of a species’ interaction strengths across all associated species provides a valuable starting point for assessing the importance of a species. It is possible that simply by accumulation of suites of interaction strengths for a variety of communities and contexts, generalizations about species importance may start to emerge. A second theme of Bob’s research career is a distrust of patterns for which there has been insufficient examination of the processes creating those patterns. Early contributions in this vein involved studies by Bob and his collaborators on the role of disturbance in natural communities (Paine and Levin 1981; Paine et al. 1985). These are among the first studies to emphasize that disturbance is such an integral aspect of communities that one would get a very incomplete picture of communities and ecosystems without accounting for disturbance. In turn, when assessing the importance of a species, it is key to explore the species’ contribution to recovery after disturbance, including disturbances that might be extremely infrequent (e.g., hurricanes, forest fires, or even volcanic eruptions). During the 1990s, ecology increasingly turned its attention to environmental problems and the challenge of protecting biodiversity. In 1991, for example, the Sustainable Biosphere Initiative laid out an ambitious research program aimed at providing answers to critical ques-

conclusion

349

tions regarding environmental management (Lubchenco et al. 1991). The first issue of Conservation Biology appeared in 1986, and Ecological Applications was founded as a journal in 1991, launching two of the most successful new journals of the last 30 years. Bob has been among the leaders in providing the impetus for uniting basic and applied work. It was Bob’s unquestioned excellence as a basic scientist that allowed him to elevate applied studies to their proper status. In the monograph he wrote on receiving the Ecology Institute Prize (Paine 1994), Bob noted that ecology began with fisheries and pest management, and he championed a melding of insights from these applied research areas with more traditional basic ecology. The connection between basic and applied research has emerged in several other aspects of Bob’s recent scientific endeavors. For example, his involvement in the scientific review of the Exxon Valdez spill prodded him to critique the monitoring program that followed the spill, and to suggest ways in which environmental monitoring should be improved (Paine et al. 1996). Now, as chair of a blue-ribbon panel of scientists who are reviewing research and management aimed at salmon recovery in western North America, Bob is taking on one of the most daunting contemporary conservation quandaries (www. nwfsc.noaa.gov/recovery). West Coast salmon, once legendary for their productivity, are now threatened and endangered in locations from California to Oregon and Washington—spanning an area, a diversity of ecosystems and habitats, and a range of human impacts that makes successful conservation seem impossible to many. But Bob’s optimism, combined with his love of fishing for (and eating) salmon, have persuaded him to accept this scientific challenge “in retirement.” Inspired by Bob’s unflagging optimism, we conclude this book with our own thoughts on some directions in which ecology and evolutionary biology should move if we are to provide satisfying guidance regarding conservation priorities. Clearly, if the processes that maintain ecosystems are to be protected, we will need a sufficiently thorough understanding of nature that the consequences of species losses and invasions can be predicted. Three of the biggest challenges in this quest are (1) identifying measurements that can serve as appropriate indicators of “ecosystem processes,” (2) learning how to scale local measurements of effect up to large-scale systems that span thousands of square miles, and (3) taking a long-range temporal view that could include evolutionary capacity and global climate shifts. All these needs are hinted at in this volume, but we expand a bit on each here. The maintenance of ecosystem services and processes is a goal that

350

conclusion

is hard to argue against but even harder to assess in a quantitative manner. How will we know whether we have succeeded or failed? Is it merely a matter of ensuring that total productivity (measured as carbon fixed) does not decline due to species losses? Or do we need to worry about “stability,” a favored concept among community ecologists, or the more recent notions of “resiliency?” If so, how are these attributes to be assessed? Identifying the measurements that can best assess ecosystem functioning is both a theoretical and practical quandary. For example, should we conclude that a wetland in which the invasive non-native Phragmites replaces Spartina is somehow a handicapped ecosystem, even though the Phragmites wetland is just as productive as the native Spartina wetland (e.g., Fell et al. 1998)? In fact, knowing how to measure “success” when it comes to protecting ecosystems is germane to the entire environmental movement, which has recently been chastised for lacking adequate performance measures (Knudson 2001; see http://www.sacbee.com/news/projects/environment/ 20010422.html). A second major challenge is scale. Numerous recent articles and books have asked how ecological processes or parameters change as a function of the scale at which they are measured (e.g., Levin 1992). The reason this issue is especially important for studies of species importance is that the answer can vary with scale. For instance, when analyzed at large geographic scales, systems with richer diversity appear to be more easily invaded than less diverse systems (Levine and D’Antonio 1999; Stohlgren et al. 1999). In contrast, when studied at small scales, the opposite pattern emerges—systems with richer diversity are less easily invaded (Levine and D’Antonio 1999; Naeem et al. 2000). This example is just one of many demonstrating how the consequences of depleting biodiversity or losing particular species might differ depending on the scale of analysis. A third challenge, mentioned by several contributors to this volume, is the value of adopting an evolutionary perspective on the importance of species. The tendency among ecologists is to emphasize ecosystem services such as productivity, nutrient cycling, and water purification; but species also represent an evolutionary potential that is difficult to assess in terms of the standard metrics associated with ecosystem services. For example, those primordial mammals that coexisted with dinosaurs in the earth’s distant past may have provided little in the way of ecosystem services during the Cretaceous, but there is no arguing that these species were very important as reservoirs of an evolutionary future. We do not know which of today’s species will be the crucial reservoirs of future evolutionary responses to our ever-changing global environment. Although speculation about

conclusion

351

evolutionary futures may seem fanciful, it is perhaps because of their evolutionary potential that species may be least expendable. Finally, a topic not broached in this volume is the melding of social science and ecology. Here, too, Bob Paine has been a leader. Bob has recently entered into the social science arena of probing how decisions regarding salmon recovery are made and whom they affect. As we increasingly come to grips with the fact that no ecosystem on the planet escapes large human impact, the value of the social sciences to explorations of species importance and biodiversity will become widely accepted. Indeed, humans are such central players in the world ecosystems that we cannot be separated from any study of community and ecosystem dynamics. We are part of ecosystem dynamics; thus, changes in ecosystems, including the loss of species, can alter patterns of human settlement, employment, and agriculture. After the next half of Bob’s career, we hope that the second edition of this book will be replete with studies that intertwine the fate of species with human demographic, social, and economic processes. In that way, instead of simply responding to threats of extinction, ecology can play a major role in shaping visions for alternative natural futures (Gallopin et al. 1997).

References

m

Aarssen, L. W. 1997. High productivity in grassland ecosystems: Effected by species diversity or productive species? Oikos 80: 183–184. Aberle, B. 1993. The biology and control of introduced Spartina (cordgrass) worldwide and recommendations for its control in Washington. Master’s thesis, Evergreen State College. Abrams, P. A., B. A. Menge, G. G. Mittelbach, D. A. Spiller, and P. Yodzis. 1996. The role of indirect effects in food webs. In G. A. Polis and K. Winemiller, eds., Food Webs: Integration of Pattern and Dynamics, 371– 395. New York, NY: Chapman & Hall. Adler, G. H., and J. O. Seamon, 1991. Distribution and abundance of a tropical rodent, the spiny rat, on islands in Panama. J. Trop. Ecol. 7: 349–360. Aizen, M., and P. Feinsinger. 1994a. Habitat fragmentation, native insect pollinators, and feral honey bees in Argentine “Chaco Serrano.” Ecol. App. 4: 378–392. . 1994b. Forest fragmentation, pollination, and plant reproduction in a Chaco dry forest, Argentina. Ecology 75: 330–351. Aizen, M. A., and W. A. Patterson. 1995. Leaf phenology and herbivory along a temperature gradient: a spatial test of the phenological window hypothesis. J. Veget. Sci. 6: 543–550. Alexandre, D.-Y. 1978. Le role disseminateur des e´ l´ephants en forˆet de Ta¨ı, ´ Cote-d’Ivoire. ˆ La Terre et la Vie (Revue d’Ecologie) 32: 47–72. Allee, W. C. 1931. Animal Aggregations: A Study in General Sociology. Chicago, IL: University of Chicago. Allison, G. W. 1999. The implications of experimental design for biodiversity manipulations. Amer. Nat. 153: 26–45.

354

references

Anderson, P. J. 1991. Age, growth and mortality of the northern shrimp Pandalus borealis Kroyer in Pavlof Bay, Alaska. U.S. Nat. Mar. Fish. Service Bull. 89: 541–554. Anderson, T. W. 1974. The chestnut pollen decline as a time horizon in lake sediments in eastern North America. Can. J. Earth Sci. 11: 678–685. Andres, L. L., and R. D. Goeden. 1971. The biological control of weeds by introduced natural enemies. In C. B. Huffaker, ed., Biological Control, 143–164. New York, NY: Plenum Press. Andrewartha, H. G. 1961. Introduction to the Study of Animal Populations. Chicago, IL: University of Chicago Press. Andrewartha, H. G., and L. C. Birch. 1984. The Ecological Web. Chicago, IL: University of Chicago Press. Appleton, R. D., and A. R. Palmer. 1998. Water-borne stimuli released by crabs and damaged prey induce more predator-resistant shells in a marine gastropd. Proc. Nat. Acad. Sci USA 85: 4387–4391 ˚ Arhem, K. 1996. The cosmic food web: human-nature relatedness in the Northwest Amazon. In P. Descola and G. P´alsson, eds., Nature and Society, 185–204. London: Routledge. Arizmendi, M. C., C. A. Dominguez and R. Dirzo. 1996. The role of an avian nectar robber and of hummingbird pollinators in the reproduction of two plant species. Func. Ecol. 10: 119–127. Arnott, S. E., N. D. Yan, J. J. Magnuson, and T. M. Frost, 1999. Interannual variability and species turnover of crustacean zooplankton in Shield lakes. Can. J. Fish. Aquat. Sci. 56: 162–172. Ashman, T. L., and M. Stanton. 1991. Seasonal variation in pollination dynamics of sexually dimorphic Sidalcea oregana ssp. spicata (Malvaceae). Ecology 72: 993–1003. Atkinson, I.A.E. 1988. Presidential address: Opportunities for ecological restoration. N. Z. J. Ecol. 11: 1–12. Atlegrim, O. 1989. Exclusion of birds from bilberry stands: impact on insect larval density and damage to the bilberry. Oecologia 79: 136–139. Austin, M. P., R. H. Groves, L.M.F. Fresco, and P. E. Kaye. 1985. Relative growth of six thistle species along a nutrient gradient with multispecies competition. J. Ecol. 73: 667–684. Axelrod, D. I. 1985. Rise of the grassland biome, central North America. Bot. Rev. 51:163–201. Bach, C. E. 1988. Effects of host patch size on herbivore density. Ecology 69: 1090–1102. Baker, H. G., and G. L. Stebbins. 1965. The Genetics of Colonizing Species. London: Academic Press. Balciunas, D., and S. P. Lawler. 1995. Effects of basal resources, predation, and alternative prey in microcosm food chains. Ecology 76: 1327–1336. Baldwin J. R., and J. R. Lovvorn. 1994. Habitats and tidal accessibility of the marine foods of dabbling ducks and brant in Boundary Bay, British Columbia. Mar. Biol. 120: 627–638. Ball, I. R. 1999. Mathematical applications for conservation ecology: The dynamics of tree hollows and the design of nature reserves. Ph.D. diss., Adelaide University, Australia.

references

355

Ball, I. R., A. Smith, J. R. Day, R. L. Pressey, and H. P. Possingham. 2001. Comparison of mathematical algorithms for the design of a reserve system for nature conservation: an application of genetic algorithms and simulated annealing. J. Env. Man. Submitted. Barbour, M., B. Pavlik, F. Drysdale, and S. Lindstrom. 1993. California’s Changing Landscapes: Diversity and Conservation of California Vegetation. Sacramento, CA: California Native Plant Society. Barnes, J., ed. 1984 The Complete Works of Aristotle, Vol. 1. Princeton, NJ: Princeton University Press. Barnes, W. J. and E. Dibble. 1988. The effects of beaver in riverbank forest succession. Can. J. Bot. 66: 40–44. Barone, J. A. 1998. Host-specificity of folivorous insects in a moist tropical forest. J. Anim. Ecol. 67: 400–409. Baskin, Y. 1997. The Work of Nature: How the Diversity of Life Sustains Us. Washington, D.C.: Island Press. Bax, N., J. T. Carlton, A. Mathews-Amos, R. L. Haedrich, F. G. Howarth, J. E. Purcell, A. Rieser, and A. Gray. 2001. The control of biological invasions in the world’s oceans. Cons. Biol. 15(5): 1234–1246. Bean, M. J., and M. J. Rowland. 1997. The Evolution of National Wildlife Law. Westport, CT: Praeger. Beck, M. W., M. Odaya, J. J. Bachant, J. Bergan, B. Keller, R. Martin, R. Mathews, C. Porter, and G. Ramseur. 2000. Identification of Priority Sites for Conservation in the Northern Gulf of Mexico: An Ecoregional Plan. Arlington, VA: The Nature Conservancy. Beirne, B. 1975. Biological control attempts by introductions against pest insects in the field in Canada. Can. Entomol. 107: 225–236. Belovsky, G. E., and J. B. Slade. 1993. The role of vertebrate and invertebrate predators in a grasshopper community. Oikos 68: 193–201. Belyea, L. R., and J. Lancaster, 1999. Assembly rules within a contingent ecology. Oikos 86: 402–416. Berenbaum, M. 1981. Patterns of furanocoumarin production and insect herbivory in a population of wild parsnip (Pastinaca sativa L.). Oecologia 49: 236–244. Berger, E. M. 1977. Gene-enzyme variation in three species of Littorina. II. The Roscoff population with a note on the origin of North American Littorina. Biol. Bull. 153: 255–264. Berlow, E. L. 1997. From canalization to contingency: historical effects in a successional rocky intertidal community. Ecol. Monogr. 67: 435–460. . 1999. Strong effects of weak interactors in ecological communities. Nature 398: 330–334. Berlow, E. L., S. A. Navarrete, C. J. Briggs, M. E. Power, and B. A. Menge. 1999. Quantifying variation in the strengths of species interactions. Ecology 80: 2206–2224. Berman-Frank, I., and Z. Dubinsky. 1999. Balanced growth in aquatic plants: Myth or reality? BioScience 49: 29–37. Bernays, E. A. 1998. The value of being a resource specialist: behavioral support for a neural hypothesis. Amer. Nat. 151: 451–464. Bertin, R. I. 1982. Floral biology, hummingbird pollination and fruit produc-

356

references

tion of trumpet creeper (Campsis radicans, Bignoniaceae). Amer. J. Bot. 69: 122–134. Bertness, M. D., and G. H. Leonard. 1997. The role of positive interactions in communities: lessons from intertidal habitats. Ecology 78: 1976–1989. Bertness, M. D., G. H. Leonard, J. M. Levine, P. R. Schmidt, and A. O. Ingraham. 1999. Testing the relative contribution of positive and negative interactions in rocky intertidal communities. Ecology 80: 2711– 2726. Bevill, R. L. 1998. The study and management of rare plant species with special attention to the threatened pitcher’s thistle and insect herbivory. Master’s thesis, University of Nebraska. Bevill, R. L., and S. M. Louda. 1999. Comparisons of related rare and common species in the study of plant rarity. Cons. Biol. 13: 493–498. Bevill, R. L., S. M. Louda, and L. M. Stanforth. 1999. Protection from natural enemies in managing rare plant species. Cons. Biol. 13: 1323–1331. Bigley, R. E., and P. G. Harrison. 1982. The recent introduction of the seagrass Zostera japonica Aschers. and Graebn to the Pacific coast of North America. Can. J. Fish. Aquat. Sci. 39(12): 1642–1648. Bisevac, L., and J. D. Majer. 1999. Comparative study of ant communities of rehabilitated mineral sand mines and heathland, Western Australia. Restor. Ecol. 7: 117–126. Blackstone, N. W. 1985. The effects of shell size and shape on growth and form in the hermit crab Pagurus longicarpus. Biol. Bull. 168: 75–90. Blackstone, N. W., and A. R. Joslyn. 1984. Utilization and preference for the introduced gatropod Littorina littorea (L.) by the hermit crab Pagurus longicarpus (Say) at Guilford, Connecticut. J. Exp. Mar. Biol. Ecol. 80: 1–9. Blossey, B., and R. Notzold. ¨ 1995. Evolution of increased competitive ability in invasive non-indigenous plants: a hypothesis. J. Ecol. 83: 887–889. Bock, C. E., J. H. Bock, and M. C. Grant. 1992. Effects of bird predation on grasshopper densities in an Arizona grassland. Ecology 73: 1706– 1717. Boldt, P. 1997. Response of a Rhinocyllus researcher. Biocontr. News Inform. 18: 100N. Bond, W. J. 1994a. Do mutualisms matter—Assessing the impact of pollinator and disperser disruption on plant extinction. Phil. Trans. Roy. Soc. London B 344: 83–90. . 1994b. Keystone species. In E.-D. Schulze and H. A. Mooney, eds., Biodiversity and Ecosystem Function, 237–253. Berlin: Springer-Verlag. Borror, D. J., C. A. Triplehorn, and N. F. Johnson. 1989. An Introduction to the Study of Insects, 6th ed. Philadelphia: Saunders College Publishing. Bosman, A. L. and P.A.R. Hockey. 1986. Seabird guano as a determinant of rocky intertidal community structure. Mar. Ecol. Prog. Ser. 32: 247– 257. Bosman, A. L., J. T. DuToit, P.A.R. Hockey, and G. M. Branch. 1986. A field experiment demonstrating the influence of seabird guano on intertidal primary production. Est. Coast. Shelf Sci. 23: 283–294.

references

357

Branch, G. M., A. Barkai, P.A.R. Hockey, and L. Hutchings. 1987. Biological interactions: causes or effects of variability in the Benguela ecosystem? South Afr. J. Mar. Sci. 5: 425–445. Bremer, K. 1994. Asteraceae: Cladistics and Classification. Portland, OR: Timber Press. Bronstein, J. L. 1994. Conditional outcomes in mutualistic interactions. Trends Ecol. Evol. 9: 214–217. Brooks, J. L., and S. I. Dodson. 1965. Predation, body size and composition of plankton. Science 150: 28–35. Brown, J., and A. C. Gibson. 1983. Biogeography. St. Louis, MO: Mosby. Brown, J. K., D. R. Frolich, and R. C. Rosell. 1995. The sweetpotato or silverleaf whiteflies: biotypes of Bemisia tabaci or a species complex? Ann. Rev. Entomol. 40: 511–534. Bruenig, E. F. 1996. Conservation and Management of Tropical Rainforests. Wallingford, U.K.: CAB International. Bruijnzeel, L. A. 1990. Hydrology of Moist Tropical Forests and Effects of Conversion: A State of Knowledge Review. Amsterdam: Free University. Brunt, A. A., K. Crabtree, M. J. Dallwitz, A. J. Gibbs, L. Watson, and E. J. Zurcher. 1996 onward. Plant Viruses Online: Descriptions and Lists from the VIDE Database. Version: 16th January 1997; biology.ann. edu.au/Groups/MES/vide/. Bryan, G. W., P. E. Gibbs, R. J. Huggett, L. G. Hummerstone, and G. R. Burt. 1986. The decline of the gastropod Nucella lapillus around south-west England: evidence for the effect of tributyltin from antifouling paints. J. Mar. Biol. Assoc. U.K. 66: 611–640. Buchmann, S. L., and G. P. Nabhan. 1996. The Forgotten Pollinators. Washington, D.C.: Island Press. Bull, J. J. 1994. Perspective: virulence. Evolution 48: 1423–1437. Bullock, D. J., S. G. North, M. E. Dulloo, and M. Thorsen. 2002. The impact of rabbit and goat eradication on the ecology of Round Island, Mauritius. In C. R. Veitch and M. Clout, eds., Turning the Tide: The Eradication of Invasive Species. Auckland, N.Z.: World Conservation Union. Bulmer, M. G. 1976. The theory of predator-prey oscillations. Theor. Pop. Biol. 9: 137–150. Bulthuis, D. A. 1991. Distribution of habitats and summer standing crop of seagrasses and macroalgae in Padilla Bay, Washington, 1989. Mount Vernon, WA: Padilla Bay National Estuarine Research Reserve Technical Report 2. Burd, M. 1994. Bateman’s principle and plant reproduction: the role of pollen limitation in fruit and seed set. Bot. Rev. 60: 83–139. Bush, G. L. 1975. Sympatric speciation in phytophagous parasitic insects. In P. W. Price, ed. Evolutionary Strategies of Parasitic Insects and Mites, 187–206. New York, NY: Plenum Press. Bustamante, R. H., and G. M. Branch. 1996. The dependence of intertidal consumers on kelp-derived organic matter on the west coast of South Africa. J. Exp. Mar. Biol. Ecol. 196: 1–28.

358

references

Bustamante, R. H., G. M. Branch, and S. Eekhout. 1995. Maintenance of an exceptional intertidal grazer biomass in South Africa: subsidy by subtidal kelps. Ecology 76: 2314–2329. Bustamante, R. H., G. M. Branch. S. Eekhout, B. Robertson, P. Zoutendyk, M. Schleyer, A. Dye, N. Hanekom, D. Keats, M. Jurd, and C. McQuaid, 1995. Gradients of intertidal primary productivity around the coast of South Africa and their relationships with consumer biomass. Oecologia 102: 189–201. Callicott, J. B. 1980. Animal liberation: a triangular affair. Envir. Ethics 2:311– 318. Camerano, L. 1880. Dell’equilibrio dei viventi merc`e la reciproca distruzione. Atti dell Reale Accademia delle Scienze di Torino 15: 393–414. Translated 1994: On the equilibrium of living beings by means of reciprocal destruction. In S. A. Levin, ed., Frontiers in Mathematical Biology, 360– 380. New York, NY: Springer-Verlag. Campbell, C., I. D. Campbell, C. B. Blyth, and J. H. McAndrews. 1994. Bison extirpation may have caused aspen expansion in western Canada. Ecography 17:360–362. Campbell, F. T. 2001. The science of risk assessment for phytosanitary regulation and the impact of changing trade regulations. BioScience 51(2): 148–153. Campbell, F. T., and S. E. Schlarbaum. 1994. Fading Forests. North American Trees and the Threat of Exotic Pests. New York, NY: Natural Resources Defense Council. Cappuccino, N. 1993. Mutual use of leaf-shelters by lepidopteran larvae on paper birch. Ecol. Entomol. 18: 287–292. Cappuccino, N., and M. A. Martin, 1997. The birch-tube maker, Acrobasis betwulella, in a fragmented habitat. The importance of patch isolation and edges. Oecologia 110: 69–76. Cappuccino, N., H. Damman, and J. Dubuc. 1995. Spatial behavior and temporal dynamics of outbreak and nonoutbreak species. In N. Cappuccino and P. W. Price, eds. Population Dynamics: New Approaches and Synthesis, 65–82. San Diego, CA: Academic Press. Carl, L. M., and M. C. Healey. 1984. Differences in enzyme frequency and body morphology among three juvenile life history types of chinook salmon (Oncorhynchus tshawytscha) in the Nanaimo River, British Columbia. Can. J. Fish. Aquat. Sci. 41: 1070–1077. Carlton, J. T. 1996. Biological invasions and cryptogenic species. Ecology 77: 1653–1655. . 2001. Endangered marine invertebrates. In Encyclopedia of Biodiversity 2, S. A. Levin (ed.), 455–464. San Diego, CA: Academic Press. Carlton, J. T., J. B. Geller, M. L. Reaka-Kudla, and E. A. Norse. 1999. Historical extinctions in the sea. Ann. Rev. Ecol. Syst. 30: 515–538. Carpenter, S. R., and J. F. Kitchell. 1993. The Trophic Cascade in Lakes. Cambridge, U.K.: Cambridge University Press. Carpenter, S. R., T. M. Frost, J. F. Kitchell, and T. K. Kratz. 1993. Species dynamics and global environmental change: a perspective from ecosys-

references

359

tem experiments. In P. M. Kareiva, J. G. Kingsolver, and R. B. Huey, eds., Biotic Interactions and Global Change, 267–279. Sunderland, MA: Sinauer Associates Inc. Carrington, J. C., and S. A. Whitham. 1998. Viral invasion and host defense: strategies and counter-strategies. Curr. Opinion Plant Biol. 1: 336–341. Carroll, M. L., and R. C. Highsmith. 1996. Role of catastrophic disturbance in mediating Nucella-Mytilus interactions in the Alaskan rocky intertidal. Marine Ecological Progress Series 138: 125–133. Carroll, R., C. Augspurger, A. Dobson, J. Franklin, G. Orians, W. Reid, R. Tracy, D. Wilcove, and J. Wilson. 1996. Strengthening the use of science in achieving the goals of the Endangered Species Act: an assessment by the Ecological Society of America. Ecol. App. 6: 1–11. Carroll, S. P., and C. Boyd, 1992. Host plant radiation in the soapberry bug: natural history with the history. Evolution 46: 1052–1069. Casal, C.M.V., D. M. Bartley, R. Froese, P. Sa-A, and M.L.D. Palomares. 1999. Documenting the status of freshwater fish introductions in Oceania. In B. S´eret and J.-Y. Sire, eds., Proceedings of the 5th Indo-Pacific Fish Conference, Noumea New Caledonia, 1997, 385–392. Paris: Soci´et´e Fran¸caise d’Ichtyologie and Institut de Recherche pour le D´eveloppement, Paris. Castilla, J. C., and L. R. Duran. 1985. Human exclusion from the rocky intertidal zone of central Chile: the effects on Concholepas concholepas (Gastropoda). Oikos 45: 391–399. Center, T. D., J. H. Frank, and F. A. Dray, Jr. 1997. Biological control. In D. Simberloff, D. C. Schmitz, and T. C. Brown, eds., Strangers in Paradise. Impact and Management of Nonindigenous Species in Florida, 245– 263. Washington, D.C.: Island Press. Chapin, F. S., III, O. E. Sala, I. C. Burke, J. P. Grime, D. U. Hooper, W. K. Lauenroth, A. Lombard, H. A. Mooney, A. R. Mosier, S. Naeem, S. W. Pacala, J. Roy, W. L. Steffan, and D. Tilman. 1998. Ecosystem consequences of changing biodiversity. BioScience 48: 45–52. Charnov, E. L. 1981. Sex reversal in Pandalus borealis: effect of a shrimp fishery? Mar. Biol. Lettr. 2: 53–57. Charnov, E. L., D. W. Gorshall, and J. G. Robinson. 1978. Sex ratio: adaptive response to population fluctuations in Pandalid shrimp. Science 200: 204–206. Cheng, K. M., I. M. McCallum, R. I. McKay, and B. E. March. 1987. A comparison of survival and growth of two strains of chinook salmon (Oncorhynchus tshawytscha) and their crosses reared in confinement. Aquaculture 67: 301–311. Cheng, Y., and R.A.C. Jones. 1999. Distribution and incidence of necrotic and non-necrotic strains of bean yellow mosaic virus in wild and crop lupins. Austr. J. Agr. Res. 50: 589–599. Chew, F. S. 1977. Coevolution of pierid butterfiles and their cruciferous host plants. II. The distribution of eggs on potential food plants. Evolution 31: 568–579. Clark, C. W., and M. L. Rosenzweig. 1994. Extinction and colonization pro-

360

references

cesses: Parameter estimates from sporadic surveys. Amer. Nat. 4: 583–596. Clark, T. W., R. P. Reading, and A. L. Clarke. 1994. Endangered Species Recovery. Finding the Lessons, Improving the Process. Washington, D.C.: Island Press. Clarke, W. C., R. E. Withler, and J. E. Shelbourn. 1992. Genetic control of juvenile life-history pattern in chinook salmon (Oncorhynchus tshawytscha). Can. J. Fish. Aquat. Sci. 49: 2300–2306. Clayton, D. H., and R. D. Price. 1999. Taxonomy of New World Columbicola (Phthiraptera: Philopteridae) from the Columbiformes (Aves), with descriptions of five new species. Ann. Entomol. Soc. Amer. 92: 675– 685. Clements, F. E. 1936. Nature and structure of the climax. J. Ecol. 24: 252–284. Cohen, A. N., and J. T. Carlton. 1995. Biological Study. Nonindigenous aquatic species in a United States estuary: a case study of the biological invasions of the San Francisco Bay and Delta. Washington, D.C.: United States Fish and Wildlife Service and The National Sea Grant College Program, NTIS Report No. PB96-166525. . 1998. Accelerating invasion rate in a highly invaded estuary. Science 279: 555–557. Cohen, J. E., F. Briand, and C. M. Newman. 1990. Community Food Webs: Data and Theory. New York, NY: Springer-Verlag. Coley, P. D. 1983. Herbivory and defensive characteristics of tree species in a lowland tropical forest. Ecol. Monogr. 53: 209–233. . 1986. Costs and benefits of defense by tannins in a neotropical tree. Oecologia 70: 238–241. . 1988. Effects of plant growth rate and leaf lifetime on the amount and type of anti-herbivore defense. Oecologia 74: 531–536. Coley, P. D., and J. A. Barone. 1996. Herbivory and plant defenses in tropical forests. Ann. Rev. Ecol. Syst. 27: 305–335. Coley, P. D., J. P. Bryant, and F. S. Chapin, III. 1985. Resource availability and plant antiherbivore defense. Science 230: 895–899. Coley, P. D., and T. A. Kursar. 1996. Anti-herbivore defenses of young tropical leaves: physiological constraints and ecological trade-offs. In S. S. Mulkey, R. L. Chazdon, and A. P. Smith, eds., Tropical Forest Plant Ecophysiology, 305–336. New York, NY: Chapman & Hall. Collinge, S. K., and S. M. Louda. 1988. Herbivory by leaf miners in response to experimental shading of a native crucifer. Oecologia (Berlin) 75: 559–566. Collins, N. M., and J. A. Thomas. 1991. The Conservation of Insects and Their Habitats. London: Academic Press. Collins, S. L. 2000. Disturbance frequency and community stability in native tallgrass prairie. Amer. Nat. 155: 311–325. Collins, S. L., A. K. Knapp, J. M. Briggs, J. M. Blair, and E. M. Steinauer. 1998. Modulation of diversity by grazing and mowing in native tallgrass prairie. Science 280: 745–747. Collins, S. L., F. Micheli, and L. Hartt. 2000. A method to determine rates and patterns of variability in ecological communities. Oikos 91: 285–293.

references

361

Connell, J. H. 1961. The influence of interspecific competition and other factors on the distribution of the barnacle Chthamalus stellatus. Ecology 42: 710–723. . 1970. A predator-prey system in the marine intertidal region. I. Balanus glandula and several predatory species of Thais. Ecol. Monogr. 40: 49–78. . 1971. On the role of natural enemies in preventing competitive exclusion in some marine animals and in rain forest trees. In P. J. den Boer and G. R. Gradwell, eds. Dynamics of Populations, 298–310. Wageningen, The Netherlands: Centre for Agricultural Publication and Documentation. . 1975. Some mechanisms producing structure in natural communities: a model and evidence from field experiments. In M. L. Cody and J. M. Diamond, eds., Ecology and Evolution of Communities, 460–490. Cambridge, MA: Belknap. . 1983. On the prevalence and relative importance of interspecific competition: evidence from field experiments. Amer. Nat. 122: 661–696. Connell, J. H., and E. Orias. 1964. The ecological regulation of species diversity. Amer. Nat. 98: 399–414. Connell, J. H., and R. O. Slayter. 1977. Mechanisms of succession in natural communities and their role in community stability and organisation. Amer. Nat. 111: 1119–1144. Connolly, S. R., and J. Roughgarden. 1999. Theory of marine communities: competition, predation, and recruitment-dependent interaction strength. Ecol. Monogr. 69: 277–296. Cooper, A., I.A.E. Atkinson, W. G. Lee, and T. H. Worthy. 1993. Evolution of the moa and their effect on the New Zealand flora. Trends Ecol. Evol. 8: 433–437. Coppedge, B. R., and J. H. Shaw. 1997. Effects of horning and rubbing behavior by bison (Bison bison) on woody vegetation in a tallgrass prairie landscape. Amer. Midl. Nat. 138: 189–196. Corn, M. L., E. H. Buck, J. Rawson, and E. Fischer. 1999. Harmful Non-native Species: Issues for Congress. Washington, D.C.: Congressional Research Service, U.S. Library of Congress. Cornell, H. V., and R. H. Karlson. 1997. Local and regional processes as controls of species richness. In D. Tilman and P. Kareiva, eds., Spatial Ecology: The Role of Space in Population Dynamics and Interspecific Interactions, 250–268. Princeton, NJ: Princeton University Press. Corner, E.J.H. 1940. Wayside Trees of Malaya. Singapore: Government Printer. . 1964. The Life of Plants. Cleveland, OH: World Publishing. . 1967. Ficus in the Solomon Islands and its bearing on the post-Jurassic history of Melanesia. Phil. Trans. Roy. Soc. London B, 253: 23–159. Cosson, J.-F., J.-M. Pons, and D. Masson. 1999. Effects of forest fragmentation on frugivorous and nectarivorous bats in French Guiana. J. Trop. Ecol. 15: 515–534. Costanza, R., R. C. d’Arge, R. de Groot, S. Farber, M. Grasso, B. Hannon, K. Limburg, R. V. O’Neill, J. Paruelo, R. G. Raskin, S. Naeem, P. Sutton,

362

references

and M. Van Den Belt. 1997. The value of the world’s ecosystem services and natural capital. Nature 387: 253–261. Cottingham, K. L., and S. R. Carpenter. 1998. Population, community, and ecosystem variates as ecological indicators: phytoplankton responses to whole-lake enrichment. Ecol. App. 8: 508–530. Cowling, R. M., P. J. Mustart, H. Laurie, and M. B. Richards. 1994. Species diversity; functional diversity and functional redundancy in fynbos communities. South Afr. J. Sci. 90: 333–337. Cox, G. W. 1999. Alien species in North America and Hawaii: impacts on natural ecosystems. Washington, D.C.: Island Press. Crandall, K. A., O.R.P. Bininda-Emonds, G. M. Mace, and R. K. Wayne, 2000. Considering evolutionary processes in conservation biology. Trends Ecol. Evol. 15: 290–295. Crawley, M. J. 1986. The population biology of invaders. Phil. Trans. Roy. Soc. London B 314: 711–731. . 1987. What makes a community invasible? In A. J. Gray, M. J. Crawley, and P. J. Edwards, eds., Colonization, Succession and Stability, 429–453. Oxford, U.K.: Blackwell Scientific Publications. Crawley, M. J., P. J. Harvey, and A. Purvis. 1997. Comparative ecology of the native and alien floras of the British Isles. In J. Silvertown, M. Franco, and J. L. Harper, eds., Plant Life Histories, 36–56. Cambridge, U.K.: University Press. Crepet, W. L. 1984. Advanced (constant) insect pollination mechanisms: pattern of evolution and implications vis-a-vis angiosperm diversity. Ann. Missouri Bot. Garden 71: 607–630. Crockett, J. S., and D. M. Engle. 1999. Combustion characteristics of bison (Bison bison) fecal pats ignited by grassland fires. Amer. Midl. Nat. 141:12–18. Crooks, K. R., A. W. Suarez, D. T. Bolger, and M. Soul´e. 2001. Extinction and colonization of birds on habitat islands. Cons. Biol. 15: 159–172. Crowder, L. B., D. P. Reagan, and D. W. Freckman. 1996. Food web dynamics and applied problems. In G. A. Polis and K. O. Winemiller, eds., Food Webs. Integration of Patterns and Dynamics, 327–336. New York, NY: Chapman & Hall. Cruz, A., and J. W. Wiley. 1989. The decline of an adaptation in the absence of a presumed selection pressure. Evolution 43: 55–62. Cushman, J. H. 1995. Ecosystem-level consequences of species additions and deletions on islands. In P. M. Vitousek, L. L. Loope, and H. Adersen, eds., Islands: Biological Diversity and Ecosystem Function, 135–147. Berlin: Springer-Verlag. Cushman, J. H., and T. G. Whitham. 1989. Conditional mutualism in a membracid-ant association: temporal, age-specific, and density-dependent effects. Ecology 70: 1040–1047. Czech, B., P. R. Krausman, and P. K. Devers. 2000. Economic associations among causes of species endangerment in the United States. BioScience 50(7): 593–601. Daehler, C. C. 1998. The taxonomic distribution of invasive angiosperm

references

363

plants: ecological insights and comparisons to agricultural weeds. Biol. Cons. 84: 167–180. Daehler, C. C., and D. R. Strong. 1995. Status, prediction and prevention of introduced cordgrass Spartina spp. invasions in Pacific estuaries, USA. Biol. Cons. 78: 51–58. Dahlhoff, E. P., B. A. Buckley, and B. A. Menge. 2001. Physiology of the rocky intertidal predator Nucella ostrina along an environmental stress gradient. Ecology 82: 2816–2829. Dahlhoff, E. P., and B. A. Menge. 1996. Influence of phytoplankton concentration and wave exposure on the ecophysiology of Mytilus californianus. Mar. Ecol. Prog. Ser. 144: 97–107. Daily, G. C., ed. 1997. Nature’s Services: Societal Dependence on Natural Ecosystems. Washington, D.C.: Island Press. Daily, G. C., S. Alexander, P. R. Ehrlich, L. Goulder, J. Lubchenco, P. A. Matson, H. A. Mooney, S. Postel, S. H. Schneider, D. Tilman, and G. M. Woodwell. 1997. Ecosystem services: benefits supplied to human societies by natural ecosystems. Iss. Ecol. 2: 1–16. Daily, G. C., P. R. Ehrlich, and N. M. Haddad. 1993. Double keystone bird in a keystone species complex. Proc. Nat. Acad. Sci. USA 90: 592–594. Dalling, J. W., and J. S. Denslow. 1998. Soil seed bank composition along a forest chronosequence in seasonally moist tropical forest, Panama. J. Veget. Sci. 9: 669–678. Daly, G. C., D. Bachelet, J. M. Lenihan, R. P. Neilson, W. Parton, and D. Ojima. 2000. Dynamic simulation of tree-grass interactions for global change studies. Ecol. App. 10: 449–469. D’Antonio, C. M., and P. M. Vitousek. 1992. Biological invasions by exotic grasses, the grass/fire cycle, and global change. Ann. Rev. Ecol. Syst. 23: 63–87. Darwin, C. R. 1859. The Origin of Species by Means of Natural Selection. London: John Murray. Reprinted 1976. New York, NY: Macmillan. Dash, B. A., and S. R. Gliessman. 1994. Nonnative species eradication and native species enhancement: fennel on Santa Cruz Island. In W. L. Halvorson and G. J. Maender, eds., The Fourth California Islands Symposium: Update on the Status of Resources, 505–512. Santa Barbara, CA: Santa Barbara Museum of Natural History. Davidson, C. B., K. W. Gottschalk, and J. E. Johnson. 1999. Tree mortality following defoliation by the European gypsy moth (Lymantria dispar L.) in the United States: A review. For. Sci. 45: 74–84. Davis, M. A. 1987. The role of flower visitors in the explosive pollination of Thalia geniculata (Marantaceae), a Costa Rican mark plant. Bull. Torrey Bot. Club 114: 134–138. Dayton, P. K. 1971. Competition, disturbance, and community organization: the provision and subsequent utilization of space in a rocky intertidal community. Ecol. Monogr. 41: 351–389. . 1972. Toward an understanding of community resilience and the potential effects of enrichments to the benthos at McMurdo Sound, Antarctica. In B. C. Parker, ed., Proceedings of the Colloquium on

364

references

Conservation Problems in Antarctica, 81–95. Lawrence, KS: Allen Press. . 1975. Experimental evaluation of ecological dominance in a rocky intertidal algal community. Ecol. Monogr. 45: 137–159. . 1984. Processes structuring some marine communities: Are they general? In D. R. Strong Jr., D. Simberloff, L. G. Abele, and A. B. Thistle, eds., Ecological Communities. Conceptual Issues and the Evidence, 181– 197. Princeton, NJ: Princeton University Press. Dayton, P. K., M. J. Tegner, P. B. Edwards, and K. L. Riser. 1998. Sliding baselines, ghosts, and reduced expectations in kelp forest communities. Ecol. App. 8: 309–322. De Angelis, D. L. 1992. Dynamics of Nutrient Cycling and Food Webs. London, U.K.: Chapman & Hall. de Ruiter, P. C., A.-M. Neutel, and J. C. Moore. 1995. Energetics, patterns of interaction strengths, and stability in real ecosystems. Science 269: 1257–1260. de Jong, T. J., and P.G.L. Klinkhamer. 1986. Population Ecology of the Biennials Cirsium vulgare and Cynoglossum officinale: An Experimental and Theoretical Approach. Utrecht: Drukkerij Elinkwijk BV. den Hartog, C. 1970. The Sea-grasses of the World. Amsterdam: North Holland Publishing Company. Dennis, B., P. L. Munholland, and J. M. Scott. 1991. Estimation of growth and extinction parameters for endangered species. Ecol. Monogr. 61: 115– 143. Denny, M. W. 1988. Biology and the mechanics of the wave-swept environment. Princeton, NJ: Princeton University Press. Descola, P. 1993. Les Lances du Crepuscule. Paris: Plon. Dethier, V. G. 1976. Man’s Plague? Insects and Agriculture. Princeton, NJ: Darwin Press. Devall, B., and G. Sessions. 1985. Deep Ecology. Salt Lake City, UT: Gibbs Smith. Devall, M. S., and L. B. Thien. 1989. Factors influencing the reproductive success of Ipomoea pescaprae (Convolvulaceae) around the Gulf of Mexico. Amer. J. Bot. 76: 1821–1831. Dial, R., and J. Roughgarden. 1995. Experimental removal of insectivores from rain forest canopy: direct and indirect effects. Ecology 76: 1821–1834. Diamond, J., and T. J. Case. 1986. Overview: introductions, extinctions, exterminations, and invasions. In J. Diamond and T. J. Case, eds. Community Ecology, 65–79. New York, NY: Harper & Row. Diehl, S., and Eklov, P. 1995. Effects of piscivore-mediated habitat use on resources, diet, and growth of perch. Ecology 76: 1712–1726. Dinerstein, E., D. M. Olson, D. J. Graham, A. L. Webster, S. A. Pimm, M. A. Bookbinder, and G. Ledec. 1995. A Conservation Assessment of the Terrestrial Ecoregions of Latin America and the Caribbean. Washington, D.C.: The World Bank. Dinnel, P. A., and E. Yates. 2000. Biological and ecological assessments of Nuttallia obscurata in North Puget Sound. J. Shellfish Res. 19(1): 630.

references

365

Doak, D. F., D. Bigger, E. K. Harding, M. A. Marvier, R. E. O’Malley, and D. Thomson. 1998. The statistical inevitability of stability-diversity relationships in community ecology. Amer. Nat. 151: 264–276. Doak, D. F., P. Kareiva, and B. Klepetka. 1994. Modeling population viability for the desert tortoise in the Mojave Desert. Ecol. App. 4: 446–460. Dodd, A. P. 1959. The biological control of prickly pear in Australia. In A. Keast, R. L. Crocker, and C. S. Christian, eds., Biogeography and Ecology in Australia (Monographiae Biologicae, Vol. 8), 565–577. The Hague: Dr. W. Junk. Domingo, E., and J. J. Holland. 1997. RNA virus mutations and fitness for survival. Ann. Rev. Microbiol. 51: 151–178. Downing, A. L. 2001. The role of biological diversity for the functioning and stability of pond ecosystems. Ph.D. diss., University of Chicago. Downing, A. L., and M. A. Leibold. 2002. Ecosystem consequences of species richness and composition in pond food webs. Nature 416: 837–841. Downs, B. J., P. S. Lake, E.S.G. Schreiber, and A. Flaister. 1998. Habitat structure and regulation of local species diversity in a stony, upland stream. Ecol. Monogr. 68: 237–257. Drake, J. A. 1989. Communities as assembled structures: do rules govern pattern? Trends Ecol. Evol. 5: 159–163. . 1991. Community assembly mechanics and the structure of an experimental species ensemble. Amer. Nat. 137: 1–26. Drake, J. A., H. A. Mooney, F. di Castri, R. H. Groves, F. J. Kruger, M. Rejmanek, and M. Williamson. eds. 1989. Biological Invasions: A Global Perspective. Chichester, U.K.: Wiley. Dudgeon, S. R., R. S. Steneck, I. R. Davison, and R. L. Vadas. 1999. Coexistence of similar species in a space-limited intertidal zone. Ecol. Monogr. 69: 331–352. Dudt, J. F., and D. J. Shure. 1994. The influence of light and nutrients on foliar phenolics and insect herbivory. Ecology 75: 86–98. Duggins, D. O., C. A. Simenstad, and J. A. Estes. 1989. Magnification of secondary production by kelp detritus in coastal marine ecosystems. Science 245: 170–173. Dumbauld, B. R., M. Peoples, and L. Holcomb. 1997. The potential influence of the aquatic weed Spartina alterniflora and control practices on clam resources in Willapa Bay, Washington. In Proceedings of the Second International Spartina Conference, Olympia, WA; March 20–21, 1997; 51–57. Duncan, D. P. 1997. The role of competition and introduction effort in the success of passeriform birds introduced to New Zealand. Amer. Nat. 149: 903–915. Dunson, W. A., and J. Travis. 1991. The role of abiotic factors in community organization. Amer. Nat. 138: 1067–1091. Duran, L. R., and J. C. Castilla. 1989. Variation and persistence of the middle rocky intertidal community of central Chile, with and without human harvesting. Mar. Biol. 103: 555–562. Dwyer, L. E., D. D. Murphy, and P. R. Ehrlich. 1995. Property rights case law

366

references

and the challenges to the Endangered Species Act. Cons. Biol. 9(4): 725–741. Dyer, A. R., and K. J. Rice. 1997. Intraspecific and diffuse competition: The response of Nassella pulchra in a California grassland. Ecol. App. 7: 484–492. Easter-Pilcher, A. 1996. Implementing the Endangered Species Act. BioScience 46(5): 355–363. Ebasco Environmental. 1993. Noxious Emergent Plant Management Environmental Impact Statement. Olympia, WA: Washington State Departments of Agriculture, Ecology, Natural Resources, Fisheries, Wildlife, and the State Noxious Weed Control Board. Edmondson, W. T. 1991. The Uses of Ecology: Lake Washington and Beyond. Seattle, WA: University of Washington Press. . 1994. Sixty years of Lake Washington: a curriculum vitae. Lake Reserv. Manage. 10: 75–84. Edmondson, W. T. 1997. Aphanizomenon in Lake Washington. Arch. Hydrobiol. Suppl. 107: 409–446. Edmondson, W. T., and S.E.B. Abella. 1988. Unplanned biomanipulation in Lake Washington. Limnologica 19: 73–79. Edmondson, W. T., and J. T. Lehman. 1981. The effects of changes in the nutrient income on the condition of Lake Washington. Limnol. Oceanogr. 26: 1–29. Edmondson, W. T., and A. H. Litt. 1982. Daphnia in Lake Washington. Limnol. Oceanogr. 27: 272–293. Ehrl´en, J., and O. Eriksson. 1995. Pollen limitation and population growth in a herbaceous perennial legume. Ecology 76: 652–656. Ehrlich, P. R. 1994. Foreword. Biodiversity and ecosystem function: need we know more? In E.-D. Schulze and H. A. Mooney, eds., Biodiversity and Ecosystem Function, vii–xi. Berlin: Springer-Verlag. Ehrlich, P. R., and L. C. Birch. 1967. The “balance of nature” and “population control.” Amer. Nat. 101: 97–107. Ehrlich, P. R., and H. A. Mooney. 1983. Extinction, substitution, and ecosystem services. BioScience 33: 248–254. Eisner, T., A. Alsop, and G. Ettershank. 1964. Adhesiveness of spider silk. Science 146: 1058–1061. Elton, C. S. 1927. Animal Ecology. London: Sedgwick and Jackson, Ltd. . 1958. The Ecology of Invasions by Animals and Plants. London: Methuen and Co. Emmons, L. H. 1989. Tropical forests: why they have so many species and how we may lose this biodiversity without cutting a single tree. Orion Nat. Quart. 8(3): 8–14. Endler, J. A. 1980. Natural selection on color patterns in Poecilia reticulata. Evolution 34: 76–91. . 1986. Natural Selection in the Wild. Princeton, NJ: Princeton University Press. . 1995. Multiple-trait coevolution and environmental gradients in guppies. Trends Ecol. Evol. 10: 22–29.

references

367

Ernst, W. G. 2000. The Earth’s place in the solar system. In W. G. Ernst, ed., Earth Systems: Processes and Issues. 45–58. Cambridge: Cambridge University Press. Escaler, M., M. A. Aranda, C. L. Thomas, and A. J. Maule. 2000. Pea embryonic tissues show common responses to the replication of a wide range of viruses. Virology 267: 318–325. Esch, G. W., A. O. Bush, and J. M. Aho. 1990. Parasite Communities: Patterns and Processes. London: Chapman & Hall. Estes, J. A., and D. O. Duggins. 1995. Sea otters and kelp forests in Alaska: generality and variation in a community ecological paradigm. Ecol. Monogr. 65: 75–100. Estes, J. A., D. O. Duggins, and G. B. Rathbun. 1989. The ecology of extinctions in kelp forest communities. Cons. Biol. 3: 252–264. Estes, J. A., and J. F. Palmisano, 1974. Sea otters: their role in structuring nearshore communities. Science 185: 1058–1060. Estes, J. A., N. S. Smith, and J. F. Palmisano. 1978. Sea otter predation and community organization in the western Aleutian Islands, Alaska. Ecology 59: 822–833. Estes, J. A., M. T. Tinker, T. M. Williams, and D. F. Doak. 1998. Killer whale predation on sea otters linking oceanic and nearshore ecosystems. Science 282: 473–476. Evans, A. A., B.I.P. Barratt, and C. M. Ferguson. 1994. Susceptibility of legume and Hieracium seedlings to feeding by native broad-nosed weevils (Coleoptera: Curculionidae). In A. J. Popay, ed., Proceedings, 47th New Zealand Plant Protection Conference, 9–11 August 1994. New Zealand Plant Protection Society. Ewel, J. J., M. J. Mazzarino, and C. W. Berish. 1991. Tropical soil fertility changes under monocultures and successional communities of different structure. Ecol. App. 1: 289–302. Fagan, W. F., and L. E. Hurd. 1994. Hatch density variation of a generalist arthropod predator: population consequences and community impact. Ecology 75: 2022–2032. Fairweather, P. G., and A. J. Underwood. 1991. Experimental removals of a rocky intertidal predator: variations within two habitats in the effects of the prey. J. Exp. Mar. Biol. Ecol. 154: 29–75. Farrell, T. M. 1991. Models and mechanisms of succession: an example from a rocky intertidal community. Ecol. Monogr. 61: 95–113. Fauth, J. E. 1999. Identifying potential keystone species from field data—an example from temporary ponds. Ecol. Lett. 2: 36–43. Feist, B. E., and C. A. Simenstad. 2000. Expansion rates and recruitment frequency of exotic smooth cordgrass, Spartina alterniflora (Loisel), colonizing unvegetated littoral flats in Willapa Bay, Washington. Estuaries 23: 267–274. Fell, P. F., S. P. Weissback, D. A. Jones, M. A. Fallon, J. A. Zipperi, E. K. Faison, K. A. Lennon, K. J. Newberry, and L. K. Reddington. 1998. Does invasion of oligohaline tidal marshes by reed grass, Phragmites australis (Cav.) Tin. Ex Steud., affect the availability of prey resources for

368

references

the mummichog, Fundulus heteroclitus L? J. Exp. Mar. Bio. Ecol. 222: 59–77. Felsenstein, J. 1985. Phylogenies and the comparative method. Amer. Nat. 125: 1–15. Ferguson, K. I., and P. Stiling. 1996. Non-additive effects of multiple natural enemies on aphid populations. Oecologia 108: 375–379. Fishbein, M., and D. L. Venable. 1996. Diversity and temporal change in the effective pollinators of Asclepias tuberosa. Ecology 77: 1061–1073. Fisher, R. A. 1930, The Genetical Theory of Natural Selection. Oxford, U.K.: Clarendon Press. Foin, T. C., S.P.D. Riley, A. L. Pawley, D. R. Ayres, T. M. Carlsen, P. J. Hodum, and P. V. Switzer. 1998. Improving recovery planning for threatened and endangered species. BioScience 48: 177–184. Folke, C., C. S. Holling, and C. Perrings. 1996. Biological diversity, ecosystems and the human scale. Ecol. App. 6: 1018–1024. Forbes, S. A. 1887. The lake as a microcosm. Bulletin of the Peoria Scientific Association, 77–87. Reprinted 1991: In L. A. Real and J. H. Brown, eds., Foundations of Ecology, 14–27. Chicago, IL: University of Chicago Press. Forget, P. M. 1994. Recruitment pattern of Vouacapoua americana (Caesalpiniaceae), a rodent-dispersed tree species in French Guiana. Biotropica 26: 409–419. Forget, P. M., and T. Milleron. 1991. Evidence for secondary seed dispersal by rodents in Panama. Oecologia 87: 596–599. Forshaw, J. M. 1977. Parrots of the World. Neptune, NJ: T.F.H. Publications. Fowler, A. C., R. L. Knight, T. L. George, and L. C. McEwen. 1991. Effects of avian predation on grasshopper populations in North Dakota grasslands. Ecology 72: 1775–1781. Fownes, J. H. 1995. Effects of diversity on productivity: quantitative distributions of traits. In P. M. Vitousek, L. L. Loope, and H. Adsersen, eds., Islands: Biological Diversity and Ecosystem Function, 177–186. New York: Springer-Verlag. Frank, D. A., and S. J. McNaughton. 1991. Stability increases with diversity in plant communities: empirical evidence from the 1988 Yellowstone drought. Oikos 62: 360–362. Fraser, S. M., and J. H. Lawton. 1994. Host range expansion by British moths onto introduced conifers. Ecol. Entomol. 19: 127–137. Frenkel, R. E. 1987. Introduction and spread of cordgrass (Spartina) into the Pacific Northwest. Northwest Envir. J. 3: 152–154. Frenkel, R. E., and L. M. Kunze. 1984. Introduction and spread of three Spartina species in the Pacific Northwest. Annual Meeting of the Association of American Geographers, Washington, D.C.; April 22–25, 1984. Unpublished paper available from the authors. Fretwell, S. D. 1977. The regulation of plant communities by food chains exploiting them. Persp. Biol. Med. 20: 169–185. Friess, N., and J. Maillet. 1996. Influence of cucumber mosaic virus infection on the intraspecific competitive ability and fitness of purslane (Portulaca oleracea). New Phytol. 132: 103–111.

references

369

. 1997. Influence of cucumber mosaic virus infection on the competitive ability and reproduction of chickweed (Stellaria media). New Phytologist 135: 667–674. Fritts, T. H., and G. H. Rodda. 1998. The role of introduced species in the degradation of island ecosystems: a case history of Guam. Ann. Rev. Ecol. Syst. 29: 113–140. Froese, R., and D. Pauly. 1998. FishBase 98: Concepts, Design and Data Sources. Manila, Philippines: ICLARM (International Center for Living Aquatic Resource Management). Froese, R., and A. Torres. 1999. Fishes under threat: an analysis of the fishes in the 1996 IUCN (International Union for Conservation of Nature) Red List. In R.S.V. Pullin, D. M. Bartley, and J. Kooiman, eds., Towards Policies for Conservation and Sustainable Use of Aquatic Genetic Resources, 131–144. ICLARM (International Center for Living Aquatic Resource Management) Conf. Proc. No. 59. Frost, T. M., S. R. Carpenter, A. R. Ives, and T. K. Kratz. 1995. Species compensation and complementarity in ecosystem function. In C. Jones and J. Lawton, eds., Linking Species and Ecosystems, 224–239. New York, NY: Chapman and Hall. Fuller, P. L., L. G. Nico, and J. D. Williams. 1999. Nonindigenous Fishes Introduced into Inland Waters of the United States. Special Publication No. 27. Bethesda, MD: American Fisheries Society. Funayama, S., K. Hikosaka, and T. Yahara. 1997. Effects of virus infection and growth irradiance on fitness components and photosynthetic properties of Eupatorium makinoi (Compositae). Amer. J. Bot. 84: 823–829. Funayama, S., I. Terashima, and T. Yahara. 2001. Effects of virus infection and light environment on population dynamics of Eupatorium makinoi (Asteraceae). Amer. J. Bot. 88: 616–622. Funayama-Noguchi, S. 2001. Ecophysiology of virus-infected plants: A case study of Eupatorium makinoi infected by geminivirus. Plant Biol. 3: 251–262. Gallopin, G., A. Hammond, P. Raskin, and R. Swart. 1997. Branch Points: Global Scenarios and Human Choice. Stockholm: PoleStar Series, Swedish Environmental Institute. Garnier, E., M.-L. Navas, M. P. Austin, J. M. Lilley, and R. M. Gifford. 1997. A problem for biodiversity-productivity studies: How to compare the productivity of multispecific plant mixtures to that of monocultures? Acta Oecol. 18: 657–670. Garrity, S. D. 1984. Some adaptations of gastropods to physical stress on a tropical rocky shore. Ecology 65: 559–574. Garrity, S. D., and S. C. Levings. 1981. A predator-prey interaction between two physically and biologically constrained tropical rocky shore gastropods: Direct, indirect, and community effects. Ecol. Monogr. 51: 267–286. Gassmann, A., and S. M. Louda. 2001. Rhinocyllus conicus: Initial evaluation and subsequent ecological impacts in North America. In E. Wajnberg, J. K. Scott, and P. C. Quimby, eds., Evaluating Indirect Ecological Effects of Biological Control, 147–183. Wallingford, U.K.: CABI Publishing.

370

references

Gause, G. F. 1935. V´erifications Exp´erimentales de la Th´eorie Math´ematique de la Lutte pour la Vie. Paris: Hermann et Cie. Gentry, A. H. 1982. Patterns of neotropical plant species diversity. Evol. Biol. 15: 1–84. Ghiselin, M. T. 1969. The evolution of hermaphroditism among animals. Q. Rev. Biol. 44: 189–208. Giller, K. E., M. H. Beare, P. Lavelle, A.-M. N. Izac, and M. J. Swift. 1997. Agricultural intensification, soil biodiversity, and agroecosystem function. App. Soil Ecol. 6: 3–16. Gillett, J. B. 1962. Pest pressure, an underestimated factor in evolution. Syst. Assoc. Spec. Pub. 4: 37–46. Gitay, H., J. B. Wilson, and W. G. Lee. 1996. Species redundancy: a redundant concept? J. Ecol. 84: 121–124. Glude, J. B. 1955. The effects of temperature and predators on the abundance of the soft-shell clam, Mya arenaria, in New England. Trans. Amer. Fish. Soc. 84: 13–26. Goeden, R. D. 1971. The phytophagous insect community of milk thistle in Southern California. J. Econ. Entomol. 64: 1101–1104. . 1974. Comparative survey of the phytophagous insect faunas of Italian thistle, Carduus pycnocephalus, in southern California and southern Europe relative to biological weed control. Envir. Entomol. 3: 464–474. Goeden, R. D., and S. M. Louda. 1976. Biotic interference with insects imported for weed control. Ann. Rev. Entomol. 21: 325–342. Goldbach, R., and D. Peters. 1994. Possible causes of the emergence of tospovirus diseases. Sem. Virol. 5: 113–120. Goldberg, D. E., and A. M. Barton. 1992. Patterns and consequences of interspecific competition in natural communities: a review of field experiments with plants. Amer. Nat. 139: 771–801. Golden, D. M., and T. O. Crist. 1999. Experimental effects of habitat fragmentation on old-field canopy insects. Oecologia 118: 371–380. Goldschmidt, T. 1996. Darwin’s Dreampond. Cambridge, MA: MIT Press. Gomez, J. M., and R. Zamora. 1999. Generalization vs. specialization in the pollination system of Hormathophylla spinosa (Cruciferae). Ecology 80: 796–805. Gompper, M. E., and E. S. Williams. 1998. Parasite conservation and the blackfooted ferret recovery program. Cons. Biol. 12: 730–732. Goodell, K., I. M. Parker, and G. S. Gilbert. 2000. Biological impacts of species invasions: implications for policy makers. In National Research Council, Incorporating Science, Economics, and Sociology in Developing Sanitary and Phytosanitary Standards in International Trade, 87–117. Washington, D.C.: National Academy Press. Goodwin, B. J., A. J. McAllister, and L. Fahrig, 1999. Predicting invasiveness of plant species based on biological information. Cons. Biol. 13: 422–426. Goudswaard, P. C., and F. Witte, 1997. The catfish fauna of Lake Victoria after the Nile Perch upsurge. Envir. Biol. Fishes 49: 21–43. Gradwohl, J., and R. Greenberg. 1982. The effect of a single species of avian predator on the arthropods of aerial leaf litter. Ecology 63: 581–583.

references

371

Gray, S. M., and N. Banerjee. 1999. Mechanisms of arthropod transmission of plant and animal viruses. Microbiol. Molec. Biol. Rev. 63:128–148. Green, P. T., D. J. O’Dowd, and P. S. Lake. 1997. Control of seedling recruitment by land crabs in rain forest on a remote oceanic island. Ecology 78: 2474–2486. Green, R. E. 1997. The influence of numbers released on the outcome of attempts to introduce exotic bird species to New Zealand. J. Anim. Ecol. 66: 25–35. Greenway, J.C., Jr. 1967. Extinct and Vanishing Birds of the World, 2nd ed. New York, NY: Dover. Grevstad, F., M. Wecker, D. Strong, A. Hastings, J. Civille, D. Garcia-Rossi, H. Davis, and K. Sayce. 2000. A Petition to Import Prokelisia marginata to Willapa Bay, Washington from California for the Biological Control of Spartina alterniflora. Long Beach, WA: Spartina Biocontrol Program, Olympic Natural Resources Center. Griffin, G. J. 2000. Blight control and restoration of the American chestnut. J. Forestry 98: 22–27. Grime, J. P. 1977. Evidence for the existence of three primary strategies in plants and its relevance to ecological and evolutionary theory. Amer. Nat. 111: 1169–1194. . 1997. Biodiversity and ecosystem function: the debate deepens. Science 277: 1260–1261. Grimm, N. B. 1995. Why link species and ecosystems? A perspective from ecosystem ecology. In C. G. Jones and J. H. Lawton, eds., Linking Species and Ecosystems, 5–15. New York: International Thomson Publishing. Groom, M., and M. Pascual. 1998. The analysis of population persistence: An outlook on the practice of viability analysis. In P. Fiedler and P. Kareiva, eds., Conservation Biology for the Coming Decade, 4–27. New York, NY: Chapman & Hall. Groot, C., and L. Margolis. 1991. Pacific Salmon Life Histories. Vancouver, B.C.: University of British Columbia Press. Grosholz, E. D., and G. M. Ruiz. 1996. Predicting the impact of introduced species: lessons from the multiple invasions of the European green crab. Biol. Cons. 78: 59–66. Grover, J. P., and M. Loreau. 1996. Linking communities and ecosystems: trophic interactions as nutrient cycling pathways. In M. E. Hochberg, J. Clobert and R. Barbault, eds., Aspects of the Genesis and Maintenance of Biological Diversity, 180–193. Oxford, U.K.: Oxford University Press. Guretzky, J. A., and S. M. Louda. 1997. Evidence for natural biological control: insects decrease the survival and growth of a native thistle. Ecol. Appl. 7: 1330–1340. Gutierrez, J. R., P. L. Meserve, S. Herrera, L. C. Contreras, and F. M. Jaksic. 1997. Effects of small mammals and vertebrate predators on vegetation in the Chilean semiarid zone. Oecologia 109: 398–406. Hackney, C. T., and S. M. Adams. 1992. Aquatic communities of the southeastern United States: Past, present, and future. In C. T. Hackney,

372

references

S. M. Adams, and W. H. Martin, eds., Biodiversity of the Southeastern United States. Aquat. Comm. 747–760. New York, NY: Wiley. Hairston, N. G. 1980. Evolution under interspecific competition: field experiments on terrestrial salamanders. Evolution 34: 409–420. Hairston, N. G., F. E. Smith, and L. B. Slobodkin. 1960. Community structure, population control and competition. Amer. Nat. 94: 421–425. Hall, D. J. 1990. The Steward: A Biblical Symbol Come of Age. Grand Rapids, MI: Eerdmans. Halliday, T. R. 1980. The extinction of the passenger pigeon Ectopistes migratorius and its relevance to contemporary conservation. Biol. Cons. 17: 157–162. Halpern, C. B., J. A. Antos, M. A. Geyer, and A. A. Olson. 1997. Species replacement during early secondary succession: the abrupt decline of a winter annual. Ecology 78: 621–631. Hamrick, J. L. 1983. Botanical Studies. In E. A. Martinko and J. L. Hamrick, eds., Pesticide Use Reduction Through Integrated Control Procedures on Musk Thistle, 8–48. Final Report, Grant R-805455. Washington, D.C.: Environmental Protection Agency, Office of Research & Development. Hamrick, J. L., and J. M. Lee. 1987. Effect of soil surface topography and litter cover on the germination, survival, and growth of musk thistle (Carduus nutans). Amer. J. Bot. 74: 451–457. Handford, P., G. Bell, and T. Reimchen. 1977. A gillnet fishery considered as an experiment in artifical selection. J. Fish Res. Bd. Can. 34: 954–961. Hanski, I., and M. E. Gilpin, 1997. Metapopulation Biology: Ecology, Genetics and Evolution. San Diego, CA: Academic Press. Hansson, L. A. 1993. Factors initiating algal life-form shift from sediment to water. Oecologia 94: 286–294. Hardin, G. 1960. The competitive exclusion principle. Science 131: 1292–1297. . 1968. The tragedy of the commons. Science 162: 1243–1248. Harding, J. M., and R. Mann. 1999. Observations on the biology of the Veined Rapa whelk, Rapana venosa (Valenciennes, 1846) in the Chesapeake Bay. J. Shellfish Res. 18: 9–17. Hare, S. R., J. N. Mantua, and R. C. Francis. 1999. Inverse production regimes: Alaska and west coast Pacific salmon. Fisheries 24: 6–14. Harley, C.D.G. 2001. The modification of biological interactions by environmental variables: comparisons across scales. Ph.D. dissertation, University of Washington. Harley, C.D.G., and J. P. Lopez. 2002. The natural history, physiology, and ecological impacts of the intertidal mesopredators, Oedoparena spp. (Diptera, Dryomyzidae). Invert. Biol. In press. Harper, J. L. 1969. The role of predation in vegetational diversity. In G. M. Woodwell and H. H. Smith, eds., Diversity and Stability in Ecological Systems (Brookhaven Symposia in Biology, No. 22), 48–62. Upton, NY: Brookhaven National Laboratory. . 1977. Population Biology of Plants. New York, NY: Academic Press. Harris, L. G., and M. Tyrrell, 2001. Changing community states in the Gulf of

references

373

Maine: synergism between invaders, overfishing, and climate change. Biol. Inv. 3: 9–21. Harrison, B. D., and D. J. Robinson. 1999. Natural genomic and antigenic variation in whitefly-transmitted geminiviruses (Begomoviruses). Ann. Rev. Phytopathol. 37: 369–398. Harrison, P. G. 1976. Zostera japonica (Aschers. & Graebn.) in British Columbia, Canada. Syesis 9: 359–360. . 1982. Spatial and temporal patterns in abundance of two intertidal seagrasses, Zostera americana den Hartog and Zostera marina L. Aquat. Bot. 12: 305–320. . 1987. Natural expansion and experimental manipulation of seagrass (Zostera spp.) abundance and the response of infaunal invertebrates. Est. Coast. Shelf Sci. 24: 799–812. Harrison, P. G., and R. E. Bigley. 1982. The recent introduction of the seagrass Zostera japonica Aschers. and Graebn. to the Pacific coast of North America. Can. J. Fish. Aquat. Sci. 39: 1642–1648. Harte, J., and A. P. Kinzig. 1993. Mutualism and competition between plants and decomposers: implications for nutrient allocation in ecosystems. Amer. Nat. 141: 829–846. Hartnett, D. C., and G.W.T. Wilson. 1999. Mycorrhizae influence plant community structure and diversity in tallgrass prairie. Ecology 80: 1187–1195. Hayward, P. J., and J. S. Ryland. 1995. Handbook of the Marine Fauna of NorthWest Europe. London: Oxford University Press. Hector, A. 1998. The effects of diversity on productivity: detecting the role of species complementarity. Oikos 82: 597–599. Hector, A., J. Joshi, S. Lawler, E. Spehn, and A. Wilby, 2001. Conservation implications of the link between biodiversity and ecosystem functioning. Oecologia. 129: 624–628. Hector, A., B. Schmid, C. Beierkuhnlein, M. C. Caldeira, M. Diemer, P. G. Dimitrakopoulos, J. A. Finn, H. Freitas, P. S. Giller, J. Good, R. Harris, P. Hogberg, ¨ K. Huss-Danell, J. Joshi, A. Jumpponen, C. Korner, ¨ P. W. Leadley, M. Loreau, A. Minns, C.P.H. Mulder, G. O’Donovan, S. J. Otway, J. S. Pereira, A. Prinz, D. J. Read, M. Scherer-Lorenzen, E.-D. Schulze, A.-S. D. Siamantziouras, E. M. Spehn, A. C. Terry, A. Y. Troumbis, F. I. Woodward, S. Yachi, and J. H. Lawton. 1999. Plant diversity and productivity experiments in European grasslands. Science 286: 1123–1127. Helfman, G. S., B. B. Collette, and D. E. Facey. 1997. The Diversity of Fishes. Malden, MA: Blackwell Science. Herbold, B., and P. B. Moyle. 1986. Introduced species and vacant niches. Amer. Nat. 128: 751–760. Hermes, N., D. Greenwood, D., and M. Hinchey. 1986. Eradication of rabbits from Phillip Island. Austr. Ranger Bull. 4(1): 34–37. Herre, E. A. 1996. An overview of studies on a community of Panamanian figs. J. Biogeogr. 23: 593–607. Herrera, C. M. 1987. Components of pollinator “quality”: comparative analysis of a diverse insect assemblage. Oikos 50: 79–90.

374

references

. 1988. Variation in mutualisms: the spatio-temporal mosaic of a pollinator assemblage. Biol. J. Linn. Soc. 35: 95–125. . 1989. Pollinator abundance, morphology, and flower visitation rate: analysis of the “quantity” component in a plant-pollinator system. Oecologia 80: 241–248. . 1996. Floral traits and plant adaptation to insect pollinators: a devil’s advocate approach. In D. G. Lloyd and S.C.H. Barrett, eds., Floral Biology, 65–87. New York, NY: Chapman & Hall. Herzog, D. C., and J. E. Funderburk. 1986. Ecological bases for habitat management and pest cultural control. In M. Kogan, ed., Ecological Theory and Integrated Pest Management Practice, 217–240. New York, NY: Wiley. Hill, W. R., M. G. Ryon, and E. M. Schilling. 1995. Light limitation in a stream ecosystem: responses by primary producers and consumers. Ecology 76: 1297–1309. Hixon, M. A. 1991. Predation as a process structuring coral reef fish communities. In P. F. Sale, ed., The Ecology of Fishes on Coral Reefs, 475–508. San Diego, CA: Academic Press. Hobbie, S. E. 1992. Effects of plant species on nutrient cycling. Trends Ecol. Evol. 7: 336–339. Hobbie, S. E., and F. S. Chapin III. 1998. The response of tundra plant biomass, aboveground production, nitrogen, and CO2 flux to experimental warming. Ecology 79: 1526–1544. Hobbs, R. J., and S. E. Humphries. 1995. An integrated approach to the ecology and management of plant invasions. Cons. Biol. 9: 761–770. Hodges, R. W. 1986. Moths of America north of Mexico. Fascicle 7.1. Gelechioidea: Gelechiidae (Part). In R. B. Dominick, ed., The Moths of North America North of Mexico. Washington, D.C.: Wedge Entomological Research Foundation. 1954 pages. Hodgson, J. G., K. Thompson, P. J. Wilson, and A. Bogaard 1998. Does biodiversity determine ecosystem function? The Ecotron experiment reconsidered. Funct. Ecol. 12: 843–848. Hofmann, G. E., and G. N. Somero. 1995. Evidence for protein damage at environmental temperatures: seasonal changes in levels of ubiquitin conjugates and hsp70 in the intertidal mussel Mytilus trossulus. J. Exp. Biol. 198: 1509–1518. . 1996a. Interspecific variation in thermal denaturation of proteins in the congeneric mussel Mytilus trossulus and M. galloprovincialis: evidence from the heat-shock response and protein ubiquitination. Mar. Biol. 126: 65–75. . 1996b. Protein ubiquitination and stress protein synthesis in Mytilus trossulus occurs during recovery from tidal emersion. Molec. Mar. Biol. Biotech. 5: 175–184. Holm, L. G. 1977. The World’s Worst Weeds: Distribution and Biology. Honolulu: University Press of Hawaii. Holmes, E. 2001. Estimating risks in declining populations with poor data. Proc. Nat. Acad. Sci. 98: 5072–5077.

references

375

Holmes, R. T., J. C. Schultz, and P. Nothnagle. 1979. Bird predation on forest insects: an enclosure experiment. Science 206: 462–463. Holt, R. D. 1977. Predation, apparent competition, and the structure of prey communities. Theor. Pop. Biol. 12: 197–229. Hooper, D. U., and P. M. Vitousek, 1997. The effects of plant composition and diversity on ecosystem processes. Science 277: 1302–1305. . 1998. Effects of plant composition and diversity on nutrient cycling. Ecol. Monogr. 68: 121–149. Horn, H. S. 1975. Markovian properties of forest succession. In M. L. Cody and J. M. Diamond, eds., Ecology and Evolution of Communities, 196– 211. Cambridge, MA: Belknap Press. Houde, A. E. 1997. Sex, Color, and Mate Choice in Guppies. Princeton, NJ: Princeton University Press. Howe, H. F. 1984. Constraints on the evolution of mutualism. Amer. Nat. 123: 764–777. Hubbes, T. 1999. The American elm and Dutch elm disease. For. Chron. 75: 265–273. Hughes, J. B., G. C. Daily, and P. R. Ehrlich. 1997. Population diversity: its extent and extinction. Science 278: 689–692. Hughes, J. B., and J. Roughgarden. 1998. Aggregate community properties and the strength of species interactions. Proc. Nat. Acad. Sci. 95: 6837– 6842. . 2000. Species diversity and biomass stability. Amer. Nat. 155: 618–627. Humphries, S. E., R. H. Groves, and D. S. Mitchell. 1992. Plant invasions of Australian ecosystems: a status review and management directions. In R. Longmore, ed., Plant Invasions: The Incidence of Environmental Weeds in Australia. Kowari 2, 1–127. Canberra, Australia: Australian National Parks Wildlife Service. Hunter, A. F. 1995. Ecology, life history, and phylogeny of outbreak and nonoutbreak species. In N. Cappuccino and P. W. Price, eds., Population Dynamics: New Approaches and Synthesis, 41–64. San Diego, CA: Academic Press. Huntley, N. 1991. Herbivores and the dynamics of communities and ecosystems. Ann. Rev. Ecol. Syst. 22: 477–503. Hurlbert, S. H. 1997. Functional importance vs. keystoneness: Reformulating some questions in theoretical biocenology. Austr. J. Ecol. 22: 369–382. Huston, M. A. 1979. A general hypothesis of species diversity. Amer. Nat. 113: 81–101. . 1997. Hidden treatments in ecological experiments: re-evaluating the ecosystem function of biodiversity. Oecologia 110: 449–460. Huston, M. A., L. W. Aarssen, M. P. Austin, B. S. Cade, J. D. Fridley, E. Garnier, J. P. Grime, J. Hodgson, W. K. Lauenroth, K. Thompson, J. H. Vandermeer, and D. A. Wardle. 2000. No consistent effect of plant diversity on productivity. Science 289: 1255a. Hutchinson, G. E. 1961. The paradox of plankton. Amer. Nat. 95: 137–147. . 1965. The Ecological Theater and the Evolutionary Play. New Haven, CT: Yale University Press.

376

references

Inouye, B. D. 1999. Integrating nested spatial scales: implications for the coexistence of competitors on a patchy resource. J. Anim. Ecol. 68: 150– 162. IUCN (International Union for the Conservation of Nature) 1994. IUCN Red List Categories. 21 pp. Gland, Switzerland: IUCN Council. Ives, A. R., K. Gross, and J. L. Klug. 1999. Stability and variability in competitive communities. Science 286: 542–544. Ives, A. R., J. L. Klug, and K. Gross. 2000. Stability and species richness in complex communities. Ecol. Lett. 3: 399–411. Jackson, J.B.C., M. X. Kirby, W. H. Berger, K. A. Bjorndal, L. W. Botsford, et al. 2001. Historical overfishing and the recent collapse of costal ecosystems. Science 293: 629–638. Jackson, K. S. 1998. Ecological factors affecting reproductive success of tall thistle (Cirsium altissimum): insect herbivory, nutrient availability, and competition. Master’s thesis, University of Nebraska. Jacobs, M. 1988. The Tropical Rain Forest. Berlin: Springer-Verlag. Janzen, D. H. 1970. Herbivores and the number of tree species in tropical forests. Amer. Nat. 104: 501–528. Janzen, D. H., and P. S. Martin. 1982. Neotropical anachronisms: the fruits the gomphotheres ate. Science 215: 19–27. Jarosz, A. M., and A. L. Davelos. 1995. Effects of disease in wild plant populations and the evolution of pathogen aggressiveness. New Phytol. 129: 371–387. Jobin, A., U. Schaffner, and W. Nentwig. 1996. The structure of the phytophagous insect fauna on the introduced weed Solidago altissima in Switzerland. Entomol. Exper. App. 79: 33–42. Joern, A. 1986. Experimental study of avian predation on coexisting grasshopper populations (Orthoptera: Acrididae) in a sandhills grassland. Oikos 46: 243–249. Johnson, K. G., A. F. Budd, and T. S. Stemann. 1995. Extinction, selectivity and ecology of Neogene Caribbean reef corals. Paleobiology 21: 52–73. Johnson, L. E., and J. T. Carlton. 1996. Post-establishment spread in large-scale invasions: dispersal mechanisms of the zebra mussel Dreissena polymorpha. Ecology 77: 1686–1690. Jones, C. G., and J. H. Lawton, eds. 1995. Linking Species and Ecosystems. New York, NY: Chapman & Hall. Jones, C. G., J. H. Lawton, and M. Shachak. 1994. Organisms as ecosystem engineers. Oikos 69: 373–386. . 1997. Positive and negative effects of organisms as physical ecosystem engineers. Ecology 78: 1946–1957. Julien, M. H., and M. W. Griffiths, eds. 1998. Biological Control of Weeds. A World Catalogue of Agents and Their Target Weeds, 4th ed. Wallingford, U.K.: CABI Publishing. Kaiser, J. 2000. Rift over biodiversity divides ecologists. Science 289: 1282– 1283. Kalko, E.K.V., E. A. Herre, and C. O. Handley Jr. 1996. Relation of fig fruit characteristics to fruit-eating bats in the New and Old World tropics. J. Biogeogr. 23: 565–576.

references

377

Kareiva, P. M. 1983. Influence of vegetation texture on herbivore populations: Resource concentration and herbivore movement. In R. Denno and M. McClure, eds., Variable Plants and Herbivores in Natural and Managed Systems, 259–289. San Diego, CA: Academic Press. . 1994. Higher-order interactions as a foil to reductionist ecology. Ecology 75: 1527–1528. Kareiva, P. M., and M. Andersen. 1988. Spatial aspects of species interactions: the wedding of models and experiments. In A. Hastings, ed., Community Ecology, 35–50. Berlin: Springer-Verlag. Kareiva, P. M., and U. Wennergren. 1995. Connecting landscape patterns to ecosystem and population processes. Nature 373: 299–302. Karki, J. B., Y. V. Jhala, and P. P. Khanna. 2000. Grazing lawns in Terai grasslands, Royal Bardia National Park, Nepal. Biotropica 32: 423–429. Karlin, S., and H. M. Taylor. 1975. A First Course in Stochastic Processes. New York, NY: Academic Press. Kartesz, J. T. 1998. Asteraceae. Biota of North America Program. College Station: Texas A&M University, Bioinformatics Working Group; http:// www.csdl.tamu.edu/FLORA/bonapfams/bonxxast.htm. Kearns, C. A., and D. W. Inouye. 1993. Techniques for Pollination Biologists. Niwot, CO: University Press of Colorado. . 1994. Fly pollination of Linum lewisii (Linaceae). Amer. J. Bot. 81: 1091– 1095. Kearns, C. A., D. W. Inouye, and N. M. Waser. 1998. Endangered mutualisms: the conservation of plant-pollinator interactions. Ann. Rev. Ecol. Syst. 29: 83–112. Kendall, D. A., and B. D. Smith. 1976. The pollinating efficiency of honeybee and bumblebee visits to flowers of the runner bean Phaseolus coccineus L. J. App. Ecol. 13: 749–752. Kessler, C. C. 2002. Eradication of feral goats and pigs from Sarigan Island, Commonwealth of the Northern Mariana Islands; methods and results. In C. R. Veitch and M. Clout, eds., Turning the Tide: The Eradication of Invasive Species. Auckland, N.Z.: World Conservation Union. Keys, R. N., S. L. Buchmann, and S. E. Smith, 1995. Pollination effectiveness and pollination efficiency of insects foraging Prosopis velutina in Southeastern Arizona. J. App. Ecol. 32: 519–527. King, C. 1984. Immigrant Killers. Auckland, N.Z.: Oxford University Press. Kirkpatrick, J. B. 1983. An iterative method for establishing priorities for selection of nature reserves: an example from Tasmania. Biol. Cons. 25: 127–134. Kirkpatrick, S., C. D. Gelatt Jr., and M. P. Vecchi. 1983. Optimization by simulated annealing. Science 220: 671–680. Kitajima, K. 1994. Relative importance of photosynthetic traits and allocation patterns as correlates of seedling shade tolerance of 13 tropical trees. Oecologia 98: 419–428. Kitchell, J. F., D. E. Schindler, R. Ogutu-Ohwayo, and P. N. Reinthal. 1997. The Nile Perch in Lake Victoria: interactions between predation and fisheries. Ecol. App. 7: 653–664. Knapp, A. K., J. M. Blair, J. M. Briggs, S. L. Collins, D. C. Hartnett, L. C.

378

references

Johnson, and G. Towne. 1999. The keystone role of bison in North American Tallgrass Prairie: Bison increase habitat heterogeneity and alter a broad array of plant, community, and ecosystem process. Bioscience 49: 39–50. Knowlton, N. 1992. Thresholds and multiple stable states in coral reef community dynamics. Amer. Zool. 32: 674–682. Knox, G. A. 1953. The intertidal ecology of Taylor’s Mistake, Banks Peninsula. Trans. Royal Soc. N.Z. 81: 189–220. . 2001. The Ecology of Seashores. Boca Raton, FL: CRC Press. Kohm, K. A., ed. 1991. Balancing on the brink of extinction: The Endangered Species Act and lessons for the future. Washington, D.C.: Island Press. Kohm, K. A., and J. F. Franklin, eds. 1997. Creating a Forestry for the 21st Century. Washington, D.C.: Island Press. Kozloff, E. N. 1973. Seashore Life of the Northern Pacific Coast. Seattle, WA: University of Washington Press. Krebs, C. J. 1999. Ecological Methodology. Menlo Park, CA: Benjamin/ Cummings. Kupferberg, S. J. 1997. Bullfrog (Rana catesbeiana) invasion of a California river: the role of larval competition. Ecology 78: 1736–1751. Lafferty, K. D., and A. M. Kuris. 1999. How environmental stress affects the impacts of parasites. Limnol. Oceanogr. 44: 925–931. Lamp, W. O., and M. K. McCarty. 1979. A preliminary study of seed predators of Platte thistle. Trans. Nebr. Acad. Sci. 7: 71–74. . 1981. Biology and ecology of Platte thistle (Cirsium canescens). Weed Sci. 29: 686–692. . 1982a. Biology of predispersal seed predators of the Platte thistle Cirsium canescens. J. Kans. Entomol. Soc. 55: 305–316. . 1982b. Predispersal seed predation of a native thistle, Cirsium canescens. Envir. Entomol. 11: 847–851. Laska, M. S., and J. T. Wootton. 1998. Theoretical concepts and empirical approaches to measuring interaction strength. Ecology 79: 461–476. Lawrence, J. M. 1990. The effect of stress and disturbance on echinoderms. Zool. Sci. 7: 17–28. Lawton, J. H. 1994. What do species do in ecosystems? Oikos 71: 367–374. Lawton, J. H., D. E. Bignell, G. F. Bloermers, P. Eggleton, and M. E. Hodda. 1996. Carbon flux and diversity of nematodes and termites in Cameroon forest soils. Biodivers. Conserv. 5: 261–273. Lawton, J. H., and V. K. Brown. 1993. Redundancy in ecosystems. In E. D. Schulze and H. A. Mooney, eds., Biodiversity and Ecosystem Function, 255–270. New York, NY: Springer-Verlag. Lawton, J. H., and R. M. May, eds. 1995. Extinction Rates. Oxford, U.K.: Oxford University Press. Lawton, J. H., S. Naeem, L. J. Thompson, A. Hector, and M. Crawley. 1998. Biodiversity and ecosystem function: getting the Ecotron experiment in its correct context. Funct. Ecol. 12: 848–852. Lehman, C. L., and D. Tilman. 2000. Biodiversity, stability, and productivity in competitive communities. Amer. Nat. 156: 534–552.

references

379

Leibold, M. A., J. M. Chase, J. B. Shurin, and A. L. Downing. 1997. Species turnover and the regulation of trophic structure. Ann. Rev. Ecol. Syst. 28: 467–494. Leider, S. A. 1989. Increasing straying by adult steelhead trout, Salmo gairdneri, following the 1980 eruption of Mount St. Helens. Envir. Biol. Fishes 24: 219–229. Leigh, E. G., Jr. 1971. Adaptation and Diversity. San Francisco, CA: Freeman Cooper. . 1975. Structure and climate in tropical rain forest. Ann. Rev. Ecol. Syst. 6: 67–86. . 1994. Do insect pests promote mutualism among tropical trees? J. Ecol. 82: 677–680. . 1999. Tropical Forest Ecology: A View from Barro Colorado Island. New York, NY: Oxford University Press. . 2001. Monteverde: ecology and conservation of a tropical cloud forest (book review). Envir. Pract. 3: 65–67. Leigh, E. G., Jr., and G. J. Vermeij. 2002. Does natural selection organize ecosystems for the maintenance of high productivity and diversity? Phil. Trans. Royal Soc. Lond. B. 357: 709–718. Leigh, E. G., Jr., A. S. Rand, and D. M. Windsor, eds. 1982. The Ecology of a Tropical Forest. Washington, D.C.: Smithsonian Institution Press. Leigh, E. G., Jr., and D. M. Windsor. 1982. Forest production and regulation of primary consumers on Barro Colorado Island. In E. G. Leigh Jr., A. S. Rand, and D. M. Windsor, eds., The Ecology of a Tropical Forest, 111– 122. Washington, D.C.: Smithsonian Institution Press. Leigh, E. G., Jr., S. J. Wright, F. E. Putz, and E. A. Herre. 1993. The decline of tree diversity on newly isolated tropical islands: a test of a null hypothesis and some implications. Evol. Ecol. 7: 76–102. Leonard, G. H., J. M. Levine, P. R. Schmidt, and M. D. Bertness. 1998. Flowdriven variation in intertidal community structure in a Maine estuary. Ecology 79: 1395–1411. Leslie, H., M. Ruckelshaus, I. Ball, H. Possingham, and S. Andelman. 2002. A habitat-based representative marine reserve system for the Florida Keys. Ecol. App. In press. Lessios, H. A. 1988. Mass mortality of Diadema antillarum in the Caribbean: what have we learned? Ann. Rev. Ecol. Syst. 19: 371–393. Letourneau, D. K., and L. A. Dyer. 1998. Experimental test in lowland tropical forest shows top-down effects through four trophic levels. Ecology 79: 1678–1687. Levin, S. A. 1992. The problem of pattern and scale in ecology. Ecology 73: 1943–1967. Levine, J. M., and C. M. D’Antonio. 1999. Elton revisited: a review of evidence linking diversity and invasibility. Oikos 87: 15–26. Likens, G. E. 1992. The Ecosystem Approach: Its Use and Abuse. Oldendorf/ Luhe, Germany: Ecology Institute. Likens, G. E., F. H. Bormann, N. M. Johnson, D. W. Fisher, and R. S. Pierce. 1970. Effects of forest cutting and herbicide treatment on nutrient

380

references

budgets in the Hubbard Brook watershed-ecosystem. Ecol. Monogr. 40: 23–47. Linthicum, K. J., A. Anyamba, C. J. Tucker, P. W. Kelley, M. F. Myers, and C. J. Peters. 1999. Climate and satellite indicators to forecast rift valley fever epidemics in Kenya. Science 285: 397–400. Lockwood, J. L. 1999. Using taxonomy to predict success among introduced avifauna: relative importance of transport and establishment. Cons. Biol. 13: 560–567. Lockwood, J. L., R. D. Powell, P. M. Nott, and S. L. Pimm. 1997. Assembling ecological communities in time and space. Oikos 80: 549–553. Lockwood, J., D. Simberloff, M. McKinney, and B. Von Holle. 2001. How many, and which, plants will invade natural areas? Biol. Inv. 3: 1–18. Lodge, D. M. 1993a. Biological invasions: lessons for ecology. Trends Ecol. Evol. 8: 133–137. . 1993b. Species invasions and deletions: community effects and responses to climate and habitat change. In P. M. Kareiva, J. G. Kingsolver, and R. B. Huey, eds., Biotic Interactions and Global Change, 367– 387. Sunderland, MA: Sinauer Associates Inc. Loeffler, C. C. 1993. Host plant and habitat effects on behavior, survival, and growth of early instar Dichomeris leuconotella (Lepidoptera: Gelechiidae), leaf-folders on goldenrods. J. Res. Lepidopt. 32: 53–74. . 1994. Natural history of leaf-folding caterpillars, Dichomeris spp. (Gelechiidae), on goldenrods and asters. J. N.Y. Entomol. Soc. 102: 405– 428. . 1996. Caterpillar leaf-folding as a defense against predation and dislodgment: staged encounters using Dichomeris (Gelechiidae) larvae on goldenrods. J. Lepidopt. Soc. 50: 245–260. Loiselle, B. A., and J. G. Blake. 1991. Temporal variation in birds and fruits along an elevational gradient in Costa Rica. Ecology 72: 180–193. Loope, L. L., and D. Mueller-Dombois. 1989. Characteristics of invaded islands, with special reference to Hawaii. In J. A. Drake, H. A. Mooney, R. diCastri, F. Kruger, R. Groves, M. Rejmanek, and M. Williamson, eds., Biological Invasions: A Global Perspective, 257–280. Chichester, U.K.: Wiley. Loreau, M. 1994. Material cycling and stability in ecosystems. Amer. Nat. 143: 508–513. . 1995. Consumers as maximizers of matter and energy flow in ecosystems. Amer. Nat. 145: 22–42. . 1998. Separating sampling and other effects in biodiversity experiments. Oikos 82: 600–602. . 2000. Biodiversity and ecosystem functioning: recent theoretical advances. Oikos 91: 3–17. Loreau, M., S. Naeem, P. Inchausti, J. Bengtsson, J. P. Grime, A. Hector, D. U. Hooper, M. A. Huston, D. Raffaelli, B. Schmid, D. Tilman, and D. A. Wardle. 2001. Biodiversity and ecosystem functioning: Current knowledge and future challenges. Science 294: 806–808. Louda, S. M. 1982. Distribution ecology: variation in plant recruitment over

references

381

a gradient in relation to insect seed predation. Ecol. Monogr. 52: 25– 41. . 1999a. Ecology of interactions needed in biological control practice and policy. Bull. Br. Ecol. Soc. 29: 8–11. . 1999b. Negative ecological effects of the musk thistle biocontrol agent, Rhinocyllus conicus Froh. In P. A. Follet and J. J. Duan, eds., Nontarget Effects of Biological Control, 215–243. Dordrecht: Kluwer Academic Publishers. Louda, S. M., and A. E. Arnett. 2000. Predicting non-target ecological effects of biological control agents: evidence from Rhinocyllus conicus. In N. Spencer, ed., Biological Control of Weeds, X International Symposium on Biological Control of Weeds, 551–567. Sydney: USDA, Agricultural Research Service. Louda, S. M., and S. K. Collinge. 1992. Plant resistance to insect herbivores: a field test of the environmental stress hypothesis. Ecology 73: 153–169. Louda, S. M., P. Dixon, and N. Huntly. 1987. Herbivory in sun versus shade at a natural meadow-woodland ecotone in the Rocky Mountains. Vegetatio 72: 141–149. Louda, S. M., D. Kendall, J. Connor, and D. Simberloff. 1997. Ecological effects of an insect introduced for the biological control of weeds. Science 277: 1088–1090. Louda, S. M., and A. K. McEachern. 1995. Insect damage to inflorescences of the threatened dune thistle Cirsium pitcheri. Bull. Ecol. Soc. Am. 76: 358. Louda, S. M., and C. W. O’Brien. 2002. Unexpected ecological effects of distributing the exotic weevil Larinus planus for biological control of Canada thistle. Cons. Biol. 16: 717–727. Louda, S. M., and M. A. Potvin. 1995. Effect of inflorescence-feeding insects in the demography and lifetime fitness of a native plant. Ecology 76: 229–245. Louda, S. M., M. A. Potvin, and S. K. Collinge. 1990. Predispersal seed predation, postdispersal seed predation and competition in the recruitment of seedlings of a native thistle in sandhills prairie. Am. Midl. Nat. 124: 105–113. . 1992. Predispersal seed predation in the limitation of native thistle. In S.B.J. Menken, J. H. Visser, and P. Harrewijn, eds., Insect-Plant Relationships, 30–32. Dordrecht: Kluwer Academic Publishers. Louda, S. M., D. Simberloff, G. Boettner, J. Connor, D. Kendall, and A. E. Arnett. 1998. Insights from data on the nontarget effects of the flowerhead weevil. Biocontr. News Inform. 19: 70N–71N. Lovelock, J. 1979. Gaia. Oxford, U.K.: Oxford University Press. Lowman, M. D., and J. D. Box. 1983. Variation in leaf toughness and phenolic content among five species of Australian rain forest trees. Austr. J. Ecol. 8: 17–25. Lubchenco, J. 1980. Algal zonation in the New England rocky intertidal community: an experimental analysis. Ecology 61: 333–344. . 1983. Littorina and Fucus: effects of herbivores, substratum heterogeneity, and plant escapes during succession. Ecology 64: 1116–1123.

382

references

Lubchenco, J., and B. A. Menge. 1978. Community development and persistence in a low rocky intertidal zone. Ecol. Monogr. 48: 67–94. Lubchenco, J., B. A. Menge, S. D. Garrity, P. J. Lubchenco, L. R. Ashkenas, S. D. Gaines, R. Emlet, J. Lucas, and S. Strauss. 1984. Structure, persistence, and role of consumers in a tropical rocky intertidal community (Taboguilla Island, Bay of Panama). J. Exp. Mar. Biol. Ecol. 78: 23–73. Lubchenco, J., A. M. Olson, L. B. Brubaker, S. R. Carpenter, M. M. Holland, S. Hubbell, S. A. Levin, J. A. MacMahon, P. A. Matson, J. M. Melillo, H. A. Mooney, C. H. Peterson, R. Pulliam, L. A. Real, P. J. Regal, and P. G. Risser. 1991. The Sustainable Biosphere Initiative: An ecological research agenda. Ecology 72: 371–412. Lubetkin, S. C. 1997. Multi-source mixing models: food web determination using stable isotope tracers. Master’s thesis, University of Washington School of Fisheries. Luckens, P. A. 1975. Predation and intertidal zonation of barnacles at Leigh, New Zealand. N. Z. J. Mar. Freshw. Res. 9: 355–378. Lynch, M. 1978. Complex interactions between natural coexploiters—Daphnia and Ceriodaphnia. Ecology 59: 552–564. MacArthur, R. H. 1961. Population effects of natural selection. Amer. Nat. 95: 195–199. . 1972. Geographical Ecology: Patterns in the Distribution of Species. New York, NY: Harper & Row. MacArthur, R. H., and E. O. Wilson. 1967. The Theory of Island Biogeography. Princeton, NJ: Princeton University Press. Macdonald, D. 1984. The Encyclopedia of Mammals. New York, NY: Facts on File Publications. Macdonald, I.A.W. 1984. Is the fynbos biome especially susceptible to invasion by exotics? A re-analysis of available data. South Afr. J. Sci. 80: 369–377. Macdonald, I.A.W., C. Th´ebaud, W. A. Strahm, and D. Strasberg. 1991. Effects of alien plant invasions on native vegetation remnants on La R´eunion (Mascarene Islands, Indian Ocean). Envir. Cons. 18: 51–61. Mace, G. M., and M. Kershaw. 1997. Extinction risk and rarity on an ecological time scale. In W. E. Kunin and K. J. Gaston, eds., The Biology of Rarity, 130–149. London: Chapman & Hall. Mack, R. N. 1996a. Biotic barriers to plant naturalization. In V. C. Moran and J. C. Hoffman, eds., Proceedings, IX International Symposium on Biological Control of Weeds, 39–46. Cape Town, South Africa: University of Cape Town. . 1996b. Predicting the identity and fate of plant invaders: emergent and emerging approaches. Biol. Cons. 78: 107–121. Magali, R., and G. Sorci. 1999. Rapid increase of host defense against brood parasites in a recently parasitized area: the case of village weavers in Hispaniola. Proc. Roy. Soc. B 266: 941–946. Maly, E. J. 1969. A laboratory study of the interaction between the predatory rotifer Asplanchna and Paramecium. Ecology 50: 59–73. . 1978. Stability of the interaction between Didinium and Paramecium: effects of dispersal and predator time lag. Ecology 59: 733–741.

references

383

Margules, C. R., A. O. Nicholls, and M. B. Usher. 1994. Apparent species turnover, probability of extinction and the selection of nature reserves: a case study of the Ingleborough limestone pavements. Cons. Biol. 8: 398–409. Marks, P. L. 1983. On the origin of the field plants of the Northeastern United States. Amer. Nat. 122: 210–228. Marquis, R. J., and C. J. Whelan. 1994. Insectivorous birds increase growth of white oak through consumption of leaf-chewing insects. Ecology 75: 2007–2014. Marshall, A., C. Smith, R. Brix, W. Dammers, J. Hymer, and L. Lavoy. 1995. Genetic diversity units and major ancestral lineages for chinook salmon in Washington. In C. Busack and J. Shaklee, eds., Genetic Diversity Units and Major Ancestral Lineages of Salmonid Fishes in Washington, 111–173. Washington Department of Fish and Wildlife Tech. Rep. RAD 95-02. Marshall, T. J., J. W. Holmes, and C. W. Rose. 1996. Soil Physics. Cambridge, U.K.: Cambridge University Press. Martinsen, G. D., E. M. Driebe, and T. G. Whitham. 1998. Indirect interactions mediated by changing plant chemistry: beaver browsing benefits beetles. Ecology 79:192–200. Maskell, L. C., A. F. Raybould, J. I. Cooper, M. L. Edwards, and A. J. Gray. 1999. Effects of turnip mosaic virus and turnip yellow mosaic virus on the survival, growth and reproduction of wild cabbage (Brassica oleracea). Ann. App. Biol. 135: 401–407. Mauzey, K. P. 1966. Feeding behavior and reproductive cycles in Pisaster ochraceus. Biol. Bull. 131: 127–144. May, R. M. 1971. Stability in multi-species community models. Math. Biosci. 12: 59–79. . 1973. Stability and Complexity in Model Ecosystems. Princeton, NJ: Princeton University Press. . 1974. Stability and Complexity in Model Ecosystems, 2nd ed. Princeton, NJ: Princeton University Press. . 1975. Patterns of species abundance and diversity. In M. L. Cody and J. M. Diamond, eds., Ecology and Evolution of Communities. Cambridge, MA: Belknap Press of Harvard University. McAlister, S. 1995. Species interactions and substrate specificity among loginhabiting bryophyte species. Ecology 76: 2184–2195. McAuliffe, J. R. 1988. Markovian dynamics of simple and complex desert plant communities. Amer. Nat. 131: 459–490. McCann, K. S. 2000. The diversity-stability debate. Nature 405: 228–233. McCann, K. S., A. Hastings, and G. R. Huxel. 1998. Weak trophic interactions and the balance of nature. Nature 395: 794–798. McCarty, M. K., C. J. Scifres, and L. R. Robison. 1967. A descriptive guide for major Nebraska thistles. University of Nebraska, Agricultural Experiment Station Bulletin No. 493: 1–24. McClanahan, T. R., N. A. Muthiga, A. T. Kamukuru, H. Machano, and R. W. Kiambo. 1999. The effects of marine parks and fishing on coral reefs of northern Tanzania. Biol. Cons. 89: 161–182.

384

references

McElhany, P., and J. Payne. Defining recovery: a Pacific salmon example. In prep. McElhany, P., M. Ruckelshaus, M. J. Ford, T. Wainwright, and E. Bjorkstedt. 2000. Viable salmonid populations and the recovery of evolutionarily significant units. U.S. Dept. of Commerce, NOAA Tech. Memo. National Marine Fisheries Service (NMFS)-Northwest Fisheries Service Center (NWFSC)-42, Northwest Fisheries Science Center, Seattle, WA. McGowan, J. A., and P. W. Walker. 1993. Pelagic diversity patterns. In R. E. Ricklefs and D. Schluter, eds., Species Diversity in Ecological Communities: Historical and Geographic Perspectives, 203–214. Chicago, IL: University of Chicago Press. McGrady-Steed, J., P. M. Harris, and P. J. Morin. 1997. Biodiversity regulates ecosystem predictability. Nature 390: 162–165. McIntosh, A. R., and C. R. Townsend. 1996. Interactions between fish, grazing invertebrates and algae in a New Zealand stream: a trophic cascade mediated by fish-induced changes in behavior. Oecologia 108: 174– 181. McIntosh, R. P. 1985. The Background of Ecology. Cambridge, U.K.: Cambridge University Press. McKinley, D. 1978. Clutch size, laying date, and incubation period in Carolina parakeet. Bird Banding 49: 223–233. McKnight, B. N., ed. 1993. Biological Pollution: The Control and Impact of Invasive Exotic Species. Indianapolis: Indiana Academy of Science. McNaughton, S. J. 1977. Diversity and stability of ecological communities: a comment on the role of empiricism in ecology. Amer. Nat. 111: 515–525. . 1978. Stability and diversity of ecological communities. Nature 274: 251–253. . 1979. Grazing as an optimization process: Grass ungulate relationships in the Serengeti. Amer. Nat. 113: 691–703. . 1983. Compensatory plant growth as a response to herbivory. Oikos 40: 329–336. . 1985. Ecology of a grazing ecosystem: the Serengeti. Ecol. Monogr. 55: 259–294. . 1994. Biodiversity and function of grazing ecosystems. In E.-D. Schulze and H. A. Mooney, eds., Biodiversity and Ecosystem Function, 361–383. New York: Springer-Verlag. Meffe, G. K., and C. R. Caroll. 1994. The species in conservation. In G. K. Meffe and C. R. Carroll, eds., Principles of Conservation Biology, 50–77. Sunderland, CT: Sinauer Associates, Inc. Meinesz, A. 1999. Killer Algae: The True Tale of a Biological Invasion. Chicago, IL: University of Chicago Press. Mellinger, M. V., and S. J. McNaughton. 1975. Structure and function of successional vascular plant communities in central New York. Ecol. Monogr. 45: 161–182. Menge, B. A. 1972. Competition for food between two intertidal starfish species and its effect on body size and feeding. Ecology 53: 635–644.

references

385

. 1976. Organization of the New England rocky intertidal community: role of predation, competition and environmental heterogeneity. Ecol. Monogr. 46: 355–393. . 1978a. Predation intensity in a rocky intertidal community: effect of an algal canopy, wave action and desiccation on predator feeding rates. Oecologia (Berlin) 34: 17–35. . 1978b. Predation intensity in a rocky intertidal community: relation between predator foraging activity and environmental harshness. Oecologia (Berlin) 34: 1–16. . 1983. Components of predation intensity in the low zone of the New England rocky intertidal region. Oecologia 58: 141–155. . 1991. Generalizing from experiments: Is predation strong or weak in the New England rocky intertidal? Oecologia 88: 1–8. . 1992. Community regulation: under what conditions are bottom-up factors important on rocky shores? Ecology 73: 755–765. . 1995. Indirect effects in marine rocky intertidal interaction webs: patterns and importance. Ecol. Monogr. 65: 21–74. . 1997. Detection of direct versus indirect effects: were experiments long enough? Amer. Nat. 149: 802–823. . 2000. Top-down and bottom-up community regulation in marine rocky intertidal habitats. J. Exp. Mar. Biol. Ecol. 250: 257–289. Menge, B. A., E. L. Berlow, C. A. Blanchette, S. A. Navarrete, and S. B. Yamada. 1994. The keystone species concept: variation in interaction strength in a rocky intertidal habitat. Ecol. Monogr. 64: 249–286. Menge, B. A., B. A. Daley, J. Lubchenco, E. Sanford, E. Dahlhoff, P. M. Halpin, G. Hudson, and J. L. Burnaford. 1999. Top-down and bottom-up regulation of New Zealand rocky intertidal communities. Ecol. Monogr. 69: 297–330. Menge, B. A., B. A. Daley, and P. A. Wheeler. 1996. Control of interaction strength in marine benthic communities. In G. A. Polis and K. O. Winemiller, eds. Food Webs: Integration of Pattern and Dynamics, 258– 274. New York, NY: Chapman & Hall. Menge, B. A., B. A. Daley, P. A. Wheeler, E. Dahlhoff, E. Sanford, and P. T. Strub. 1997a. Benthic-pelagic links and rocky intertidal communities: bottom-up effects on top-down control? Proc. Nat. Acad. Sci. USA 94: 14530–14535. Menge, B. A., B. A. Daley, P. A. Wheeler, and P. T. Strub. 1997b. Rocky intertidal oceanography: an association between community structure and nearshore phytoplankton concentration. Limnol. Oceanogr. 42: 57–66. Menge, B. A., and T. L. Freidenburg. 2001. Keystone species. In S. A. Levin, editor-in-chief, Encyclopedia of Biodiversity, 613–631. New York, NY: Academic Press. Menge, B. A., and J. Lubchenco. 1981. Community organization in temperate and tropical rocky intertidal habitats: prey refuges in relation to consumer pressure gradients. Ecol. Monogr. 51: 429–450. Menge, B. A., J. Lubchenco, L. R. Ashkenas, and F. Ramsey. 1986a. Experimental separation of effects of consumers on sessile prey in the low zone

386

references

of a rocky shore in the Bay of Panama: direct and indirect consequences of food web complexity. J. Exper. Mar. Biol. Ecol. 100: 225– 269. Menge, B. A., J. Lubchenco, S. D. Gaines, and L. R. Ashkenas. 1986b. A test of the Menge-Sutherland model of community organization in a tropical rocky intertidal food web. Oecologia (Berlin) 71: 75–89. Menge, B. A., and A. M. Olson. 1990. Role of scale and environmental factors in regulation of community structure. Trends Ecol. Evol. 5: 52–57. Menge, B. A., and J. P. Sutherland. 1976. Species diversity gradients: synthesis of the roles of predation, competition, and temporal heterogeneity. Amer. Nat. 110: 351–369. . 1987. Community regulation: variation in disturbance, competition, and predation in relation to environmental stress and recruitment. Amer. Nat. 130: 730–757. Menge, J. L., and B. A. Menge. 1974. Role of resource allocation, aggression and spatial heterogeneity in coexistence of two competing intertidal starfish. Ecol. Monogr. 44: 189–209. Merrill, G. G. 1995. An ecological framework for the management of the nonindigenous seagrass Zostera japonica in Washington state. Master’s thesis, University of Washington. Meserve, P. L., J. R. Gutierrez, J. A. Yunger, L. C. Contreras, and F. M. Jaksic. 1996. Role of biotic interactions in a small mammal assemblage in semiarid Chile. Ecology 77: 133–148. Messenger, S. L., I. J. Molineux, and J. J. Bull. 1999. Virulence evolution in a virus obeys a trade-off. Proc. Roy. Soc. London B. 266: 397–404. Meyer, O. 1993. Functional groups of microorganisms. In E.-D. Schulze and H. A. Mooney, eds., Biodiversity and Ecosystem Function, 67–96. Berlin: Springer-Verlag. Michalakis, Y., I. Olivieri, F. Renaud, and M. Raymond. 1992. Pleiotropic action of parasites: how to be good for the host. Trends Ecol. Evol. 7: 59– 62. Miller, R. J., and E. L. Brannon. 1981. The origin and development of life history patterns in Pacific salmonids. Proceedings of the Salmon and Trout Migratory Behavior Symposium, School of Fisheries, 296–330. Seattle: University of Washington. Mills, E. L., J. H. Leach, J. T. Carlton, and C. L. Secor. 1994. Exotic species and the integrity of the Great Lakes. BioScience 44: 666–676. Mills, K. E., and J. D. Bever. 1998. Maintenance of diversity within plant communities: soil pathogens as agents of negative feedback. Ecology 79: 1595–1601. Mills, L. S., M. E. Soul´e, and D. F. Doak. 1993. The keystone-species concept in ecology and conservation. BioScience 43: 219–224. Montalvo, A. M., and J. D. Ackerman. 1986. Relative pollinator effectiveness and evolution of floral traits in Spathiphyllum friedrichsthalii (Araceae). Amer. J. Bot. 73: 1665–1676. Mooney, H. A., and J. A. Drake, eds. 1986. Ecology of Biological Invasions of North America and Hawaii. New York, NY: Springer-Verlag.

references

387

Mooney, H. A., J. Lubchenco, R. Dirzo, and O. E. Sala. 1995. Biodiversity and ecosystem functioning: basic principles. In V. H. Heywood, ed., Global Biodiversity Assessment, 275–325. Cambridge, U.K.: Cambridge University Press. Morin, P. J. 1995. Functional redundancy, non-additive interactions, and supply-side dynamics in experimental pond communities. Ecology 76: 133–149. Morin, P. J., S. P. Lawler, and E. A. Johnson. 1988. Competition between aquatic insects and vertebrates: experimental measures of interaction strength and higher order interactions. Ecology 69: 1401–1409. Morris, R. F. 1963. The dynamics of epidemic spruce budworm populations. Memoirs of the Entomological Society of Canada No. 31. 332 pp. Morris, W., D. Doak, M. Groom, P. Kareiva, J. Fieberg, L. Gerber, P. Murphy, and D. Thomson. 1999. A Practical Handbook for Population Viability Analysis. Arlington, VA: The Nature Conservancy. Moss, E. H. 1932. The vegetation of Alberta. IV. The poplar association and related vegetation of central Alberta. J. Ecol. 20: 380–415. Moulton, M. P., and J. G. Sanderson. 1997. Predicting the fates of passeriform introductions on oceanic islands. Cons. Biol. 11: 552–558. Moyle, P. B. 1986. Fish introductions into North America: patterns and ecological impact. In H. A. Mooney and J. A. Drake, eds. Ecology of Biological Invasions of North America and Hawaii, 27–43. New York, NY: Springer-Verlag. . 1999. Effects of invading species on freshwater and estuarine ecosystems. In O. T. Sandlund, P. J. Schei, and A. Viken, eds. Invasive Species and Biodiversity Management, 177–191. Leiden: Kluwer. Moyle, P. B., H. W. Li, and B. A. Barton. 1986. The Frankenstein effect: impact of introduced fishes on native fishes in North America. In R. H. Stroud, ed., Fish Culture in Fisheries Management, 415–426. Bethesda, MD: American Fisheries Society. Moyle, P. B., and T. Light. 1996. Fish invasions in California: do abiotic factors determine success? Ecology 77: 1666–1670. Mulder, C.P.H., J. Koricheva, K. Huss-Danell, P. Hogberg, ¨ and J. Joshi. 1999. Insects affect relationships between plant species richness and ecosystem processes. Ecol. Lett. 2: 237–246. Murtaugh, P. A. 1981. Selective predation by Neomysis mercedis in Lake Washington. Limnol. Oceanogr. 26: 445–453. Myers, J. M., R. G. Kope, G. J. Bryant, D. Teel, L. J. Lierheimer, T. C. Wainwright, W. S. Grant, F. W. Waknitz, K. Neely, S. T. Lindley, and R. S. Waples. 1998. Status review of Chinook salmon from Washington, Idaho, Oregon, and California. U.S. Dept. Commerce, National Oceanic & Atmospheric Administration (NOAA) Memo. National Marine Fisheries Service (NMFS)-Northwest Fisheries Service Center (NWFSC)-35, Northwest Fisheries Science Center, Seattle, WA. Nabhan, G. P., and S. L. Buchmann. 1997. Services provided by pollinators. In G. C. Daily, ed., Nature’s Services, 133–150. Washington, D.C.: Island Press.

388

references

Nadkarni, N. M., and N. T. Wheelwright. eds. 2000. Monteverde: Ecology and Conservation of a Tropical Cloud Forest. New York, NY: Oxford University Press. Naeem, S. 1998. Species redundancy and ecosystem reliability. Cons. Biol. 12: 39–45. . 2000. Reply to Wardle et al. Bull. Ecol. Soc. Amer. 81: 241–246. Naeem, S., T. Chapin, R. Costanza, P. Ehrlich, F. B. Golley, D. Hooper, J. H. Lawton, R. O’Neill, H. Mooney, O. Sala, A. Symstad, and D. Tilman. 1999. Biodiversity and Ecosystem Functioning: Maintaining Life Support Systems. Ecological Issues No. 4. Washington, D.C.: Ecological Society of America. Naeem, S., D. Hahn, and G. Schuurman. 2000. Producer-decomposer co-dependency influences biodiversity effects. Nature 403: 762–764. Naeem, S., J.M.H. Knops, D. Tilman, K. M. Howe, T. Kennedy, and S. Gale. 2000. Plant diversity increases resistance to invasion in the absence of covarying extrinsic factors. Oikos 91: 97–108. Naeem, S., and S. Li. 1997. Biodiversity enhances ecosystem reliability. Nature 390: 507–509. Naeem, S., L. J. Thompson, S. P. Lawler, J. H. Lawton, and R. M. Woodfin. 1994. Declining biodiversity can alter the performance of ecosystems. Nature 386: 734–737. . 1995. Empirical evidence that declining species diversity may alter the performance of terrestrial ecosystems. Phil. Trans. Roy. Soc. London B 347: 249–262. Naiman, R. J., J. M. Melillo, and J. E. Hobbie. 1986. Ecosystem alteration of boreal forest streams by beaver (Castor canadensis). Ecology 67: 1254– 1269. Nakano, S., H. Miyasake, and N. Kuhara. 1999. Terrestrial-aquatic linkages: riparian arthropod inputs alter trophic cascades in a stream food web. Ecology 80: 2435–2441. Nason, J. D., E. A. Herre, and J. L. Hamrick. 1996. Paternity analysis of the breeding structure of strangler fig populations: evidence for substantial long-distance wasp dispersal. J. Biogeogr. 23: 501–512. . 1998. The breeding structure of a tropical keystone plant resource. Nature 391: 685–687. Nasr, S. H. 1996. Religion and the Order of Nature. New York, NY: Oxford University Press. Nault, L. R. 1997. Arthropod transmission of plant viruses: a new synthesis. Ann. Entomol. Soc. Amer. 90: 521–541. Navarrete, S. A., and B. A. Menge. 1996. Keystone predation and interaction strength: interactive effects of predators on their main prey. Ecol. Monogr. 66: 409–429. Navarrete, S. A., and E. A. Wieters. 2000. Variation in barnacle recruitment over small scales: larval predation by adults and maintenance of community pattern. J. Exp. Mar. Biol. Ecol. 253: 131–148. Neale, D., and W. Nelson. 1998. Marine algae of the West Coast, South Island, New Zealand. Tuhinga 10: 87–118.

references

389

Nehlsen, W., J. E. Williams, and J. A. Lichatowich. 1991. Pacific salmon at the crossroads: stocks at risk from California, Oregon, Idaho and Washington. Fisheries 16: 4–21. Neill, W. 1974. The community matrix and interdependence of the competition coefficients. Amer. Nat. 108: 399–408. Nelson, J. S. 1994. Fishes of the World. New York, NY: Wiley. Netting, R. M. 1976. What alpine peasants have in common: observations on communal tenure in a Swiss village. Human Ecol. 4: 135–146. . 1993. Smallholders, Householders. Stanford, CA: Stanford University Press. Newman, R. M., D. C. Thompson, and D. B. Richman. 1998. Conservation strategies for the biological control of weeds. In P. Barbosa, ed., Conservation Biological Control, 371–396. San Diego, CA: Academic Press. Nico, L. G., and P. L. Fuller. 1999. Spatial and temporal patterns of nonindigenous fish introductions in the United States. Fisheries 24: 16–27. Nijs, I., and I. Impens. 2000. Biological diversity and probability of local extinction of ecosystems. Funct. Ecol. 14: 46–54. NMFS (National Marine Fisheries Service). 1991. Final Rule: Endangered and Threatened Species; Endangered Status for Snake River Sockeye Salmon. Federal Register 56, 58619. . 1992. Final Rule: Endangered and Threatened Species; Threatened Status for Snake River Spring/Summer Chinook Salmon, Threatened Status for Snake River Fall Chinook Salmon. Federal Register 57, 14653. . 1994. Final Rule: Endangered and Threatened Species; Status of Sacramento River Winter-run Chinook Salmon. Federal Register 59, 440. . 1995. Final Rule: Endangered and Threatened Species; Status of Snake River Spring/Summer Chinook Salmon and Snake River Fall Chinook Salmon. Federal Register 60, 19342. . 1996a. Final Rule: Endangered and Threatened Species; Endangered Status for Umpqua River Cutthroat Trout in Oregon. Federal Register 61, 41514. . 1996b. Final Rule: Endangered and Threatened Species: Threatened Status for Central California Coast Coho Salmon Evolutionarily Significant Unit (ESU). Federal Register 61, 56138. . 1997a. Final Rule: Endangered and Threatened Species; Listing of Several Evolutionarily Significant Units (ESUs) of West Coast Steelhead. Federal Register 62, 43937. . 1997b. Final Rule: Endangered and Threatened Wildlife and Plants; Threatened Status for the Southern Oregon/Northern California Coast Evolutionarily Significant Unit of Coho Salmon. Federal Register 62, 33038. . 1998a. Final Rule: Endangered and Threatened Species; Threatened Status for the Oregon Coast Evolutionarily Significant Unit of Coho Salmon. Federal Register 63, 42587. . 1998b. Final Rule: Endangered and Threatened Species; Threatened Status for Two ESUs of Steelhead in Washington, Oregon, and California. Federal Register 63, 13347.

390

references

. 1999a. Final Rule: Endangered and Threatened Species: Threatened Status for Two ESUs of Steelhead in Washington and Oregon. Federal Register 64, 14517. . 1999b. Final Rule: Endangered and Threatened Species; Threatened Status for Ozette Lake Sockeye Salmon in Washington. Federal Register 64, 14528. . 1999c. Final Rule: Endangered and Threatened Species; Threatened Status for Two Chinook Salmon Evolutionarily Significant Units (ESUs) in California. Federal Register 64, 50393. . 1999d. Final Rule: Endangered and Threatened Species; Threatened Status for Two ESUs of Chum Salmon in Washington and Oregon. Federal Register 64, 14508. . 1999e. Final Rule: Endangered and Threatened Species; Threatened Status for Three Chinook Salmon Evolutionarily Significant Units (ESUs) in Washington and Oregon, and Endangered Status for One Chinook Salmon ESU in Washington. Federal Register 64, 14308. Nomme, K. M., and P. G. Harrison. 1991a. A multivariate comparison of the seagrasses Zostera marina and Zostera japonica in monospecific versus mixed populations. Can. J. Bot. 69: 1984–1990. . 1991b. Evidence for interaction between the seagrasses Zostera marina and Zostera japonica on the Pacific coast of Canada. Can. J. Bot. 69: 2004–2010. Northcott, M. S. 1996. The Environment and Christian Ethics. Cambridge, U.K.: Cambridge University Press. Nowak, R. M. 1991. Walker’s Mammals of the World, 5th ed. Baltimore, MD: Johns Hopkins University Press. Nowlin, W. H., and R. W. Drenner. 2000. Context-dependent effects of bluegill in experimental mesocosm communities. Oecologia 122: 421–426. NRC (National Research Council). 1995. Science and the Endangered Species Act. Washington, D.C.: National Academy Press. O’Connor, T. G. 1995. Acacia karroo invasion of grassland: environmental and biotic effects influencing seedling emergence and establishment. Oecologia 103: 214–223. Odum, E. P. 1953. Fundamentals of Ecology. Philadelphia, PA: Saunders. Odum, H. T. 1994. Ecological and General Systems. Niwot, CO: University Press of Colorado. Oksanen, L., S. D. Fretwell, J. Arruda, and P. Niemela. 1981. Exploitation ecosystems in gradients of primary productivity. Amer. Nat. 118: 240– 261. Olckers, T., and M. P. Hill, eds. 1999. Biological Control of Weeds in South Africa (1990–1998). African Entomology Memoir No. 1. Hatfield, South Africa: Entomological Society of South Africa. Olsen, K. M. 1997. Pollination effectiveness and pollinator importance in a population of Heterotheca subaxillaris (Asteraceae). Oecologia 109: 114– 121. Olson, A. M. 1992. Evolutionary and ecological interactions affecting seaweeds. Ph.D. dissertation, Oregon State University.

references

391

Ooi, K., S. Ohshita, I. Ishii, and T. Yahara. 1997. Molecular phylogeny of geminivirus infecting wild plants in Japan. J. Plant Res. 110: 247–257. Ooi, K., and T. Yahara. 1999. Genetic variation of geminiviruses: comparison between sexual and asexual host plant populations. Molec. Ecol. 8: 89–97. Opler, P. A. 1979. Insects of American chestnut: possible importance and conservation concern. In W. McDonald, ed., The American Chestnut Symposium, 83–85. Morgantown, WV: University of West Virginia Press. Orians, G. H. 1986. Site characteristics favoring invasions. In H. A. Mooney and J. A. Drake, eds., Ecology of Biological Invasions of North America and Hawaii, 133–148. New York, NY: Springer-Verlag. OTA (Office of Technology Assessment). 1993. Harmful Non-Indigenous Species in the United States. Washington, D.C.: U.S. Government Printing Office. Owadally, A.W. 1979. The dodo and the tambalacoque tree. Science 203: 1363– 1364. Pacala, S. W., C. D. Canham, J. Saponara, J. A. Silander Jr., R. K. Kobe, and E. Ribbens. 1996. Forest models defined by field measurements: estimation, error analysis and dynamics. Ecol. Monogr. 66: 1–43. Pacala, S. W., and M. Rees. 1998. Models suggesting field experiments to test two hypotheses explaining successional diversity. Amer. Nat. 152: 729–737. Pacala, S., and J. Roughgarden. 1984. Control of arthropod abundance by Anolis lizards on St. Eustatius (Meth. Antilles). Oecologia 64: 160–162. Pace, M. L., and E. Funke. 1991. Regulation of planktonic microbial communities by nutrients and herbivores. Ecology 72: 904–914. Padis´ak, J. 1992. Seasonal succession of phytoplankton in a large shallow lake (Balaton, Hungary)—a dynamic approach to ecological memory, its possible role and mechanisms. Ecology 80: 217–230. Paine, R. T. 1966. Food web complexity and species diversity. Amer. Nat. 100: 65–75. . 1969a. A note on trophic complexity and community stability. Amer. Nat. 103: 91–93. . 1969b. The Pisaster-Tegula interaction: prey patches, predator food preference, and intertidal community structure. Ecology 50: 950–961. . 1971. A short-term experimental investigation of resource partitioning in a New Zealand rocky intertidal habitat. Ecology 52: 1096–1106. . 1974. Intertidal community structure: experimental studies on the relationship between a dominant competitor and its principal predator. Oecologia (Berlin) 15: 93–120. . 1976. Size-limited predation: an observational and experimental approach with the Mytilus-Pisaster association. Ecology 57: 858–873. . 1977. Controlled manipulations in the marine intertidal zone, and their contributions to ecological theory. In The Changing Scenes in Natural Sciences, 1776–1976, 245–270. Philadelphia: Academy of Natural Sciences at Philadelphia, Special Publication No. 12. . 1980. Food webs: linkage, interaction strength and community infrastructure. J. Anim. Ecol. 49: 667–685.

392

references

. 1984. Ecological determinism in the competition for space. Ecology 65: 1339–1348. . 1988. Food webs: road maps of interactions or grist for theoretical development? Ecology 69: 1648–1654. . 1992. Food-web analysis through field measurement of per capita interaction strength. Nature 355: 73–75. . 1994. Marine Rocky Shores and Community Ecology: An Experimentalist’s Perspective. Oldendorf/Luhe, Germany: Ecology Institute. . 1995. A conversation on refining the concept of keystone species. Cons. Biol. 9: 962–964. Paine, R. T., J. C. Castilla, and J. Cancino. 1985. Perturbation and recovery patterns of starfish-dominated intertidal assemblages in Chile, New Zealand, and Washington state. Amer. Nat. 125: 679–691. Paine, R. T., and S. A. Levin. 1981. Intertidal landscapes: disturbance and the dynamics of pattern. Ecol. Monogr. 51: 145–178. Paine, R. T., J. L. Ruesink, A. Sun, E. L. Soulanille, M. J. Wonham, C.D.G. Harley, D. R. Brumbaugh, and D. L. Secord. 1996. Trouble on oiled waters: Lessons from the Exxon Valdez Oil Spill. Annu. Rev. Ecol. Syst. 27: 197–235. Paine, R. T., and R. L. Vadas. 1969. The effects of grazing by sea urchins, Strongylocentrotus spp., on benthic algal populations. Limnol. Oceanogr. 14: 710–719. Paine, R. T., J. T. Wootton, and P. D. Boersma. 1990. Direct and indirect effects of peregrine falcon predation on seabird abundance. Auk 107: 1–9. Paine, R. T., and T. M. Zaret. 1975. Ecological gambling: the high risks and rewards of species introductions. JAMA 231: 471–473. Palmisano, J. F., and J. A. Estes. 1977. Ecological interactions involving the sea otter. In M. L. Merritt and R. G. Fuller, eds. The Environment of Amchitka Island, Alaska, 527–567. Springfield, VA: U.S. Energy Research & Development Administration. Palumbi, S. R. 1985. Spatial variation in an alga-sponge commensalism and the evolution of ecological interactions. Amer. Nat. 126: 267–274. . 2001. The Evolution Explosion: How Humans Cause Rapid Evolutionary Change. New York, NY: Norton. . 2002. Humans as the world’s greatest evolutionary force. Science 293: 1786–1790. Panzer, R., and M. W. Schwartz. 1998. Effectiveness of a vegetation-based approach to insect conservation. Cons. Biol. 12: 693–702. Park, T. 1948. Experimental studies of interspecies competition I. Competition between populations of the flour beetles, Tribolium confusum Duval and Tribolium castaneum Herbst. Ecol. Monogr. 18: 265–308. . 1954. Experimental studies of interspecies competition II. Temperature, humidity, and competition in two species of Tribolium. Physiol. Zool. 28: 177–238. Parker, F. D. 1981. How efficient are bees in pollinating sunflowers? J. Kans. Entomol. Soc. 54: 61–67.

references

393

Parker, I. M. 1997. Pollinator limitation of Cytisus scoparius (Scotch broom), and invasive exotic shrub. Ecology 78: 1457–1470. Parker, I. M., D. Simberloff, W. M. Lonsdale, K. Goodell, M. Wonham, P. M. Kareiva, M. H. Williamson, B. Von Holle, P. B. Moyle, J. E. Byers, and L. Goldwasser. 1999. Impact: toward a framework for understanding the ecological effects of invaders. Biol. Inv. 1: 3–19. Parker, R. C., and B. Aberle. 1979. A situation report on the Spartina infestation in northwest Washington. Unpublished report. Parmenter, R. R., and J. A. MacMahon. 1988. Factors influencing species composition and population sizes in a ground beetle community (Carabidae): predation by rodents. Oikos 52: 350–356. Parsons, T. R. 1996. The impact of industrial fisheries on the trophic structure of marine ecosystems. In G. A. Polis and K. O. Winemiller, eds., Food Webs. Integration of Patterns and Dynamics, 352–357. New York, NY: Chapman & Hall. Pawlak, B. T., and A. M. Olson. 1995. Policies and management practices of Washington state agencies as they pertain to the seagrasses Zostera marina and Zostera japonica: an ecological analysis. Puget Sound Res. 2: 516–521. PCE (Office of the Parliamentary Commissioner for the Environment). 2000. New Zealand Under Siege. A Review of the Management of Biosecurity Risks to the Environment. Parliamentary Commissioner for the Environment, Wellington, New Zealand. Pennings, S. C., and R. M. Callaway. 1996. Impact of a parasitic plant on the structure and dynamics of salt marsh vegetation. Ecology 77: 1410– 1419. Petchey, O. 2000. Species diversity, species extinction, and ecosystem function. Amer. Nat. 155: 696–702. Petchey, O. L., P. T. McPhearson, and P. J. Morin. 1999. Environmental warming alters food-web structure and ecosystem function. Nature 402: 69–72. Peterson, D. L., and K. T. Parker. 1998. Ecological Scale. New York, NY: Columbia University Press. Pettersson, M. W. 1991. Pollination by a guild of fluctuating moth populations: option for unspecialization in Silene vulgaris. J. Ecol. 79: 591– 604. Pianka, E. R. 1988. Evolutionary Ecology. New York, NY: Harper & Row. Pimentel, D., L. Lach, R. Zuniga, and D. Morrison. 2000. Environmental and economic costs of nonindigenous species in the United States. BioScience 50: 53–65. Pimm, S. L. 1979. The structure of food webs. Theor. Pop. Biol. 16: 144–158. . 1982. Food Webs. London: Chapman & Hall. . 1991. The Balance of Nature? Ecological Issues in the Conservation of Species and Communities. Chicago, IL: University of Chicago Press. Piperno, D. R., and D. M. Pearsall. 1998. The Origins of Agriculture in the Lowland Neotropics. San Diego, CA: Academic Press.

394

references

Poff, N. L., J. D. Allan, M. B. Bain, J. R. Karr, K. L. Prestegaard, B. D. Richter, R. E. Sparks, and J. C. Stromberg. 1997. The natural flow regime: a paradigm for river conservation and restoration. BioScience 47: 769– 784. Polis, G. A., R. D. Holt, B. A. Menge, and K. O. Winemiller. 1996. Time, space, and life history: influences on food webs. In G. A. Polis and K. Winemiller, eds., Food Webs: Integration of Pattern and Dynamics, 435–460. New York, NY: Chapman & Hall. Polis, G. A., S. D. Hurd, C. T. Jackson, and F. Sanchez Pinero. ˜ 1997. El Nino ˜ effects on the dynamics and control of an island ecosystem in the Gulf of California. Ecology 78: 1884–1897. Polis, G. A., and K. O. Winemiller. eds. 1996. Food Webs: Integration of Pattern and Dynamics. New York, NY: Chapman & Hall. Pollock, M. M., R. J. Naiman, H. E. Erickson, C. A. Johnston, J. Pastor, and G. Pinay. 1995. Beaver as engineers: influences on biotic and abiotic characteristics of drainage basins. In C. G. Jones and J. H. Lawton, eds., Linking Species and Ecosystems, 117–126. New York, NY: Chapman & Hall. Popay, A. I., and R. W. Medd. 1990. The biology of Australian weeds. 21. Carduus nutans L. Plant Prot. Q. 5: 3–13. Posey, M. H. 1988. Community changes associated with the spread of an introduced seagrass, Zostera japonica. Ecology 69: 974–983. Possingham, H. P., I. Ball, and S. Andelman. 2000. Mathematical methods for identifying representative reserve networks. In S. Ferson and M. A. Burgman, eds., Quantitative Methods in Conservation Biology, 291–306. New York, NY: Springer-Verlag. Power, A. G. 1992. Patterns of virulence and benevolence in insect-borne pathogens of plants. Crit. Rev. Plant Sci. 11: 351–372. . 2000. Insect transmission of plant viruses: a constraint on virus variability. Curr. Opin. Plant Biol. 3: 335–339. . 2002. Ecological risks of transgenic virus-resistant crops. In D. K. Letourneau and B. E. Burrows, eds., Genetically Engineered Organisms: Assessing Environmental and Human Health Effects. Boca Raton, FL: CRC Press. Power, A. G., and S. K. Remold. 1996. Incidence of barley yellow dwarf virus in wild grass populations: implications for biotechnology risk assessment. In M. Levin, C. Grim and S. Angle, eds., Proceedings of the Biotechnology Risk Assessment Symposium. Balmar, Gaithersburg, MD: University of Maryland. Power, M. E. 1992. Habitat heterogeneity and the functional significance of fish in river food webs. Ecology 73: 1675–1688. Power, M. E., M. S. Parker, and J. T. Wootton. 1996a. Disturbance and food chain length in rivers. In G. A. Polis and K. Winemiller, eds., Food Webs: Integration of Pattern and Dynamics, 286–297. New York, NY: Chapman & Hall. Power, M. E., D. Tilman, J. A. Estes, B. A. Menge, W. J. Bond, L. S. Mills, G. Daily, J. C. Castilla, J. Lubchenco, and R. T. Paine. 1996b. Challenges in the quest for keystones. BioScience 46: 609–620.

references

395

Preston, F. W. 1962. The canonical distribution of commonness and rarity. Ecology 43: 185–215, 410–432. Preszler, R. W., and W. J. Boecklen. 1996. The influence of elevation on tritrophic interactions: opposing gradients of top-down and bottom-up effects on a leaf-mining moth. Ecoscience 3: 75–80. Price, P. W., C. E. Bouton, P. Gross, B. A. McPheron, J. N. Thompson, and A. E. Weis. 1980. Interactions among three trophic levels: influence of plants on interactions between insect herbivores and natural enemies. Ann. Rev. Ecol. Syst. 11: 41–65. Price, P. W., M. Westoby, B. Rice, P. R. Atsatt, R. S. Fritz, J. N. Thompson, and K. Mobley. 1986. Parasite mediation in ecological interactions. Ann. Rev. Ecol. Syst. 17: 487–505. Price, R. D., D. H. Clayton, and R. J. Adams. 2000. Pigeon lice down under: taxonomy of Australian Campanulotes (Phthiraptera: Philopteridae), with a description of C. durdeni n. sp. J. Parasitol. 86: 948–950. Proulx, M., F. R. Pick, A. Mazumder, P. B. Hamilton, and D.R.S. Lean. 1996. Effects of nutrients and planktivorous fish on the phytoplankton of shallow and deep aquatic systems. Ecology 77: 1556–1572. PSTRT (Puget Sound Technical Recovery Team). 2001. Independent populations of chinook salmon in Puget Sound. Public review draft, April 2001; www.nwfsc.noaa.gov/cbd/trt/trt—puget.htm. Pullin, R.S.V., M. L. Palomares, C. V. Casal, M. M. Dey, and D. Pauly. 1997. Environmental impacts of tilapias. In K. Fitzsimmons, ed., Tilapia Aquaculture. Proceedings of the 4th International Symposium on Tilapia in Aquaculture, Vol. 2, 554–570. Ithaca, NY: Northeast Regional Agricultural Engineering Service Cooperative Extension. Quimby, P. C., Jr. 1982. Impact of diseases on plant populations. In R. Charudattan and H. L. Walker, eds., Biological Control of Weeds with Plant Pathogens, 47–60. New York, NY: Wiley. Quinn, J. F. 1979. Disturbance, predation and diversity in the rocky intertidal zone. Ph.D. dissertation, University of Washington. Quinn, J. F., and S. P. Harrison. 1988. Effects of habitat fragmentation and isolation on species richness: evidence from biogeographic patterns. Oecologia 75: 132–140. Rabinowitz, D. 1981. Seven forms of rarity. In H. Synge, ed., The Biological Aspects of Rare Plant Conservation, 205–217, London: Wiley. Raffaelli, D. G., and S. J. Hall. 1996. Assessing the relative importance of trophic links in food webs. In G. A. Polis and K. O. Winemiller, eds., Food Webs: Integration of Pattern and Dynamics, 185–191. New York, NY: Chapman & Hall. Ralls, K., D. P. Demaster, and J. A. Estes. 1996 Developing a criterion for delisting the southern sea otter under the U.S. Endangered Species Act. Cons. Biol. 10: 1528–1537. Randall, J. M. 1991. The ecology of an invasive biennial, Cirsium vulgare, in California. Ph.D. thesis. diss., University of California, Davis. Randall, R. B., M. C. Healey, and J. B. Dempson. 1987. Variability in length of freshwater residence of salmon, trout and char. Amer. Fish. Soc. Symp. 1: 27–41.

396

references

Rao, M. 2000. Variation in lead-cutter ant (Atta sp.) densities in forest isolates: the potential role of predation. J. Trop. Ecol. 16: 209–225. Rao, M., J. Terborgh, and P. Nunez. ˜ 2001. Increased herbivory in forest isolates: implications for plant community structure and composition. Cons. Biol. 15: 624–633. Rastetter, E. B., L. Gough, A. E. Hartley, D. A. Herbert, K. J. Nadelhoffer, and M. Williams. 1999. A revised assessment of species redundancy and ecosystem reliability. Cons. Biol. 13: 440–443. Ratchford, S. G. 1995. Changes in the density and size of newly settled clams in Willapa Bay, WA, due to the invasion of smooth cordgrass, Spartina alterniflora Loisel. Master’s thesis, University of Washington. Raunkier, C. 1934. The Life Form of Plants and Statistical Plant Geography. Oxford, U.K.: Oxford University Press. Raybould, A. F., L. C. Maskell, M. L. Edwards, J. I. Cooper, and A. J. Gray. 1999. The prevalence and spatial distribution of viruses in natural populations of Brassica oleracea. New Phytol. 141: 265–275. Real, L. A. 1996a. Disease ecology. Ecology 77: 989. . 1996b. Sustainability and the ecology of infectious disease—Diseases and their pathogenic agents must be viewed as important parts of any ecosystem management strategy. Bioscience 46: 88–97. Redfern, M. 1983. Insects and Thistles. Cambridge, U.K.: Cambridge University Press. Rees, N. E. 1977. Impact of Rhinocyllus conicus on thistles in southwestern Montana. Environ. Entomol. 6: 839–842. Reeves, B. 1998. Progress of the Spartina and purple loosestrife control programs. Washington State Department of Agriculture, Statewide Spartina Control Coordinator Report to the Legislature. Regal, P. J. 1977. Ecology and evolution of flowering plant dominance. Science 196: 622–629. Reich, P. B., B. D. Kloeppel, D. S. Ellsworth, and M. B. Walters. 1995. Different photosynthesis-nitrogen relations in deciduous hardwood and evergreen coniferous tree species. Oecologia 104: 24–30. Reichard, S. H., and C. W. Hamilton. 1997. Predicting invasions of woody plants introduced into North America. Cons. Biol. 11: 193–203. Rejm´anek, M. 1996. Species richness and resistance to invasions. In G. H. Orians, R. Dirzo, and J. H. Cushman, eds., Biodiversity and Ecosystem Processes in Tropical Forests, 153–172. Berlin: Springer-Verlag. Rejm´anek, M., and D. M. Richardson. 1996. What attributes make some plant species more invasive? Ecology 77: 1655–1661. Reynolds, C. S. 1984. The Ecology of Freshwater Phytoplankton. New York, NY: Cambridge University Press. Reynolds, C. S., J. Padis´ak, and I. Kobor. ´ 1993. A localized bloom of Dinobryon sociale in Lake Balaton: some implications for the perception of patchiness and the maintenance of species richness. Abstr. Bot. 17: 251–260. Reznick, D. N., H. Bryga, and J. A. Endler. 1990. Experimentally induced lifehistory evolution in an natural population. Nature 346: 357–359. Reznick, D. N., and J. A. Endler. 1982. The impact of predation on life history

references

397

evolution in Trinidadian guppies (Poecilia reticulata). Evolution 36: 160–177. Ricciardi, A., R. J. Neves, and J. B. Rasmussen. 1998. Impending extinctions of North American freshwater mussels (Unionida) following the zebra mussel (Dreissena polymorpha) invasion. J. Anim. Ecol. 67: 613–619. Ricciardi, A., and J. B. Rasmussen. 1998. Predicting the identity and impact of future biological invaders: a priority for aquatic resource management. Can. J. Fish. and Aquat. Sci. 55: 1759–1765. Ricciardi, A., W.W.M. Steiner, R. N. Mack, and D. Simberloff. 2000. Toward a global information system for invasive species. BioScience 50: 239– 244. Ricciardi, A., F. G. Whoriskey, and J. B. Rasmussen. 1997. The role of the zebra mussel (Dreissena polymorpha) in structuring macroinvertebrate communities on hard substrate. Can. J. Fish. Aquat. Sci. 54: 2596– 2608. Richens, R. H. 1983. Elm. Cambridge, U.K.: Cambridge University Press. Ricker, W. E. 1981. Changes in the average size and average age of Pacific salmon. Can. J. Fish. Aquat. Sci. 38: 1636–1656. Ricketts, E. F., J. Calvin, J. Hedgpeth, and D. W. Phillips. 1939. Between Pacific Tides. Stanford, CA: Stanford University Press. Ridley, H. N. 1930. The Dispersal of Plants Throughout the World. Ashford, Kent: Reeve & Company, Ltd. Riggs, S. R. 1992. Distribution of Spartina alterniflora in Padilla Bay, Washington, in 1991. Mount Vernon, WA: Padilla Bay National Estuarine Research Reserve Technical Report No. 3. Riggs, S. R., and D. A. Bulthuis. 1994. Estimated net aerial primary productivity and monitoring of selected characteristics of Spartina alterniflora in Padilla Bay, Washington, April 1992–May 1993. Mount Vernon, WA: Padilla Bay National Estuarine Research Reserve Technical Report No. 11. Rijnsdorp, A. D. 1993. Fisheries as a large-scale experiment in life history evolution: disentangling phenotypic and genetic effects in changes in maturation and reproduction of North Sea plaice, Pleuronectes platessa L. Oecologia 96: 391–401. Roberts, D. A., G. E. Hofmann, and G. N. Somero. 1997. Heat-shock protein expression in Mytilus californianus: acclimatization (seasonal and tidal-height comparisons) and acclimation effects. Biol. Bull. 192: 309– 320. Robertson, D. R. 1972. Social control of sex-reversal in a coral reef fish. Science 177: 1007–1009. Robles, C., and J. Robb. 1993. Varied carnivore effects and the prevalence of intertidal algal turfs. J. Exper. Mar. Biol. Ecol. 166: 65–91. Robles, C. D., R. Sherwood-Stephens, and M. Alvarado. 1995. Responses of a key intertidal predator to varying recruitment of its prey. Ecology 76: 565–579. Rogers, D. J., and S. E. Randolph, 2000. The global spread of malaria in a future, warmer world. Science 289: 1763–1766.

398

references

Rohlf, D. J. 1991. Six reasons why the Endangered Species Act doesn’t work— and what to do about it. Cons. Biol. 5: 273–282. Roosinck, M. J. 1997. Mechanisms of virus evolution. Ann. Rev. Phytopathol. 35: 191–209. Root, R. B. 1967. The niche exploitation pattern of the blue-gray gnatcatcher. Ecol. Monogr. 37: 317–350. . 1973. Organization of a plant-arthropod association in simple and diverse habitats: the fauna of collards (Brassica oleracea). Ecol. Monogr. 43: 95–124. . 1996. Herbivore pressure on goldenrods (Solidago altissima): its variation and cumulative effects. Ecology 77: 1074–1087. Root, R. B., and Cappuccino, N. 1992. Patterns in population change and the organization of the insect community associated with goldenrod. Ecol. Monogr. 62: 393–420. Root, R. B., C. C. Loeffler, and J. E. Rawlins. Trends in the traits, functions, and affinities of the caterpillars (Lepidoptera) associated with goldenrod (Solidago altissima). Unpublished manuscript. Rosenzweig, M. L., and Z. Abramsky. 1993. How are diversity and productivity related? In R. E. Ricklefs and D. Schluter, eds., Species Diversity in Ecological Communities: Historical and Geographical Perspectives, 52– 65. Chicago: University of Chicago Press. Rosenzweig, M. L., and R. H. MacArthur. 1963. Graphical representation and stability conditions of predator-prey interactions. Amer. Nat. 97: 209– 223. Ruesink, J. L. 1998. Variation in per capita interaction strength: thresholds due to nonlinear dynamics and nonequilibrium conditions. Proc. Nat. Acad. Sci. USA 95: 6843–6847. Ruesink, J. L., I. M. Parker, M. J. Groom, and P. M. Kareiva. 1995. Reducing the risks of nonindigenous species introductions: guilty until proven innocent. BioScience 45: 465–477. Ruiz, G. M., P. W. Fofonoff, J. T. Carlton, M. J. Wonham, and A. H. Hines. 2000. Invasion of coastal marine communities in North America: apparent patterns, processes, and biases. Ann. Rev. Ecol. Syst. 31: 481– 531. Runkle, J. R. 1981. Gap regeneration in some old-growth forests of the eastern United States. Ecology 62: 1041–1051. Sagers, C. L., and P. D. Coley. 1995. Benefits and costs of defense in a neotropical shrub. Ecology 76: 1835–1843. Sandgren, C. D., ed. 1988. Growth and Reproductive Strategies of Freshwater Phytoplankton. Cambridge, U.K.: Cambridge University Press. Sanford, E. 1999a. Oceanographic influences on rocky intertidal communities: coastal upwelling, invertebrate growth rates, and keystone predation. Ph.D. diss., Oregon State University. . 1999b. Regulation of keystone predation by small changes in ocean temperature. Science 283: 2095–2097. Sanford, E., and B. A. Menge. 2001. Spatial and temporal variation in barnacle

references

399

growth in a coastal upwelling system. Mar. Ecol. Prog. Ser. 209: 143– 157. Santos-Granero, F. 1991. The Power of Love. London: Athlone Press. Sanz, A. I., A. Fraile, J. M. Gallego, J. M. Malpica, and F. Garcia-Arenal. 1999. Genetic variability of natural populations of cotton leaf curl geminivirus, a single-stranded DNA virus. J. Mol. Evol. 49: 672–681. Sayce, J. R. 1991. Spartina in Willapa Bay: a case history. In T. F. Mumford, P. Peyton, J. R. Sayce, and S. Harbell, eds., Spartina Workshop Record; Seattle, WA; Nov. 14–15, 1990. Schardt, J. D. 1997. Maintenance control. In D. Simberloff, D. C. Schmitz, and T. C. Brown, eds., Strangers in Paradise. Impact and Management of Nonindigenous Species in Florida, 229–243. Washington, D.C.: Island Press. Scheffer, T. H. 1945. The introduction of Spartina alterniflora to Washington with oyster culture. Leafl. West. Bot. 4: 163–164. Schemske, D. W. 1983. Limits to specialization and coevolution in plant-animal mutualisms. In M. H. Nitecki, ed., Coevolution, 67–109. Chicago, IL: University of Chicago Press. Schemske, D. W., and C. C. Horvitz. 1984. Variation among floral visitors in pollination ability: a precondition for mutualism specialization. Science 225: 519–521. . 1989. Temporal variation in selection on a floral character. Evolution 43: 461–465. Schemske, D. W., B. C. Husband, M. H. Ruckelshaus, C. Goodwillie, I. M. Parker, and J. G. Bishop. 1994. Evaluating approaches to the conservation of rare and endangered plants. Ecology 75: 584–606. Schindler, D. E., S. R. Carpenter, J. J. Cole, J. F. Kitchell, and M. L. Pace. 1997. Influence of food web structure on carbon exchange between lakes and the atmosphere. Science 277: 248–251. Schindler, D. E., J. F. Kitchell, X. He, S. R. Carpenter, J. R. Hodgson, and K. L. Cottingham. 1993. Food web structure and phosphorus cycling in lakes. Trans. Amer. Fish. Soc. 122: 756–772. Schindler, D. W., K. H. Mills, D. F. Malley, D. L. Findlay, J. A. Shearer, I. J. Davies, M. A. Turner, G. A. Linsey, and D. R. Cruikshank. 1987. Long-term ecosystem stress: the effects of years of experimental acidification on a small lake. Science 228: 1395–1401. Schlesinger, W. H. 1997. Biogeochemistry: An Analysis of Global Change. New York, NY: Academic Press. Schluter, D. 1984. A variance test for detecting species associations, with some example applications. Ecology 65: 998–1005. Schmitz, D. C., D. Simberloff, R. H. Hofstetter, W. Haller, and D. Sutton. 1997. The ecological impact of nonindigenous plants. In D. Simberloff, D. C. Schmitz, and T. C. Brown, eds., Strangers in Paradise. Impact and Management of Nonindigenous Species in Florida, 39–61. Washington, D.C.: Island Press. Schmitz, O. J. 1998. Direct and indirect effects of predation and predation risk on old-field interaction webs. Amer. Nat. 151: 327–342.

400

references

Schoener, T. W. 1974. Resource partitioning in ecological communities. Science 185: 27–39. . 1983. Field experiments on interspecific competition. Amer. Nat. 122: 240–285. . 1986. Patterns in terrestrial vertebrate versus arthropod communities: do systematic differences in regularity exist? In J. Diamond and T. J. Case, eds., Community Ecology, 556–586. New York, NY: Harper & Row. . 1988. Leaf damage in island buttonwood, Conocarpus erectus: correlations with pubescence, island area, isolation and the distribution of major carnivores. Oikos 53: 253–266. . 1989. Food webs from the small to the large. Ecology 70: 1559–1589. . 1993. On the relative importance of direct versus indirect effects in ecological communities. In H. Kawanabe, J. E. Cohen, and K. Iwasaki, eds., Mutualism and Community Organization: Behavioral, Theoretical and Food Web Approaches, 365–411. Oxford, U.K.: Oxford University Press. Schoener, T. W., and D. A. Spiller. 1987. Effect of lizards on spider populations: manipulative reconstruction of a natural experiment. Science 236: 949–952. . 1999a. Variation in the magnitude of a predator’s effect from small to large islands. In Monografies de la Societat d’Historia Natural de les Balears 6: 35–66. Ecologia de les illes. . 1999b. Indirect effects in an experimentally staged invasion by a major predator. Amer. Nat. 153: 347–358. . 1996. Devastation of prey diversity by experimentally introduced predators in the field. Nature 381: 691–694. Schulze, E.-D., and H. A. Mooney, eds. 1994. Biodiversity and Ecosystem Function. Berlin: Springer-Verlag. Schwartz, M. W., C. A. Brigham, J. D. Hoeksema, K. G. Lyons, M. H. Mills, and P. J. van Mantgem. 2000. Linking biodiversity to ecosystem function: implications for conservation ecology. Oecologia 122: 297–305. Scott, J. M., F. Davis, B. Csuti, R. Noss, B. Butterfield, C. Groves, H. Anderson, S. Caicco, F. Derchia, T. C. Edwards, J. Ulliman, and R. G. Wright. 1993. Gap analysis—a geographic approach to protection of biological diversity. Wildlife Monogr. 123: 1–41. Seeley, R. H. 1986. Intense natural selection caused a rapid morphological change in a living marine snail. Proc. Nat. Acad. Sci. USA 83: 6897– 6901. Set¨al¨a, H., J. Rissanen, and A. M. Markkola. 1997. Conditional outcomes in the relationship between pine and ectomycorrhizal fungi in relation to biotic and abiotic environment. Oikos 80: 112–122. Seymour, R. M., and R. H. Bradbury. 1999. Lengthening reef recovery times from crown-of-thorns outbreaks signal systemic degradation of the Great Barrier Reef. Mar. Ecol. Progr. Ser. 176: 1–10. Shafland, P. L. 1999. The introduced butterfly peacock (Cichla ocellaris) in Florida. I. Fish community analyses. Rev. Fish. Sci. 7: 71–94.

references

401

Shea, K., and P. Chesson. 2002. Community ecology theory as a framework for biological invasions. Trends Ecol. Evol. 17(4): 170–176. Sheldon, S. P., and R. P. Creed. 1995. Potential use of a native insect as a biological control for an introduced weed. Ecol. App. 5: 1122–1132. Shelford, V. E. 1913. Animal Communities in Temperate North America. Chicago, IL: University of Chicago Press. Shilling, F. 1997. Do habitat conservation plans protect endangered species? Science 276: 1662–1663. Shine, C., N. Williams, and L. Gundling. ¨ 2000. A guide to designing legal and institutional frameworks on alien invasive species. IUCN (World Conservation Union) Environmental Policy and Law Paper No. 40. Sih, A., P. Crowley, M. McPeek, J. Petranka, and K. Strohmeier. 1985. Predation, competition and prey communities: a review of field experiments. Ann. Rev. Ecol. Syst. 16: 269–311. Silver, W. L., S. Brown, and A. E. Lugo. 1996. Biodiversity and biogeochemical cycles. In G. H. Orians, R. Dirzo, and J. H. Cushman, eds., Biodiversity and Ecosystem Processes in Tropical Forests, 49–67. Berlin: SpringerVerlag. Simberloff, D. 1981. Community effects of introduced species. In H. Nitecki, ed., Biotic Crises in Ecological and Evolutionary Time, 53–81. New York, NY: Academic Press. . 1986. Introduced insects: a biogeographic and systematic perspective. In H. A. Mooney and J. A. Drake, eds., Ecology of Biological Invasions of North America and Hawaii, 3–26. New York, NY: Springer-Verlag. . 1991. Keystone species and community effects of biological introductions. In L. R. Ginzburg, ed., Assessing Ecological Risks of Biotechnology, 1–19. Boston: Butterworth-Heinemann. . 1995. Why do introduced species appear to devastate islands more than mainland areas? Pacif. Sci. 49: 87–97. . 1998. Flagships, umbrellas, and keystones: Is single-species management passe in the landscape era? Biol. Cons. 83: 247–257. . 1999. Nonindigenous species—a global threat to biodiversity and stability. In P. H. Raven, ed., Nature and Human Society. The Quest for a Sustainable World, 325–334. Washington, D.C.: National Academy Press. . 2001. Managing existing populations of introduced species. In R. Claudi, O. Hendrickson, and H. Ottens, eds., Alien Invasive Species: A Threat to Canadian Biodiversity. Ottawa: Natural Resources Canada, Canadian Forest Service. In press. Simberloff, D., D. C. Schmitz, and T. C. Brown, eds. 1997. Strangers in Paradise: Impact and Management of Nonindigenous Species in Florida. Washington, D.C.: Island Press. Simberloff, D., and B. Von Holle. 1999. Positive interactions of nonindigenous species: invasional meltdown? Biol. Inv. 1: 21–32. Simenstad, C. A. 1994. Faunal associations and ecological interactions in seagrass communities of the Pacific Northwest coast. In S. Wyllie-Echeverria, A. M. Olson, and M. J. Hershman, eds., Seagrass Science and

402

references

Policy in the Pacific Northwest: Proceedings of a Seminar Series (SMA 94-1), EPA 910/R-94-004, 11–18. Simenstad, C. A., J. R. Cordell, R. C. Wissmar, K. L. Fresh, S. L. Schroder, M. Carr, G. Sanborn, and M. Burg. 1988. Assemblage structure, microhabitat distribution, and food web linkages of epibenthic crustaceans in Padilla Bay National Estuarine Research Reserve, Washington. Mount Vernon, WA: Padilla Bay National Estuarine Research Reserve Reprint Series No. 9. Simenstad, C. A., and K. L. Fresh. 1995. Influence of intertidal aquaculture on benthic communities in Pacific Northwest estuaries: scales of disturbance. Estuaries 18(1a): 43–70. Simenstad, C. A., and R. M. Thom. 1995. Spartina alterniflora (smooth cordgrass) as an invasive haplophyte in Pacific Northwest estuaries. Hortus Northwest 6: 9–12, 38–40. Simms, E. L. 1996. The evolutionary genetics of plant-pathogen systems. BioScience 46: 36–145. Sinclair, A.R.E. 1979. Dynamics of the Serengeti ecosystem. In A.R.E. Sinclair and M. Norton-Griffiths, eds., Serengeti. Dynamics of an Ecosystem, 1– 30. Chicago, IL: University of Chicago Press. . 1995. Serengeti past and present. In A.R.E. Sinclair and P. Arcese, eds., Serengeti II, 3–30. Chicago, IL: University of Chicago Press. Skellam, J. G. 1951. Random dispersal in theoretical populations. Biometrika 38: 196–218. Smallwood, K. S., J. Beyea, and M. L. Morrison. 1999. Using the best scientific data for endangered species conservation. Envir. Manage. 24: 421– 435. Smith, N. G. 1968. The advantage of being parasitized. Nature 219: 690–694. Smith, T. B., L. A. Freed, J. Lepson, and J. Carothers. 1995. Evolutionary consequences of extinctions in populations of a Hawaiian honeycreeper. Cons. Biol. 9: 107–113. Smith, T. M., H. H. Shugart, and F. I. Woodward, eds. 1997. Plant Functional Types, International Geosphere-Biosphere Programme Book Series. Cambridge, U.K.: Cambridge University Press. Smythe, N. 1986. Competition and resource partitioning in the guild of neotropical terrestrial frugivorous mammals. Ann. Rev. Ecol. Syst. 17: 169–188. . 1989. Seed survival in the palm Astrocaryum standleyanum: evidence for dependence upon its seed dispersers. Biotropica 21: 50–56. Snow, A. A., and D. W. Roubik. 1987. Pollen deposition and removal by bees visiting two tree species in Panama. Biotropica 19: 57–63. Soler, J. J., J. G. Martinez, M. Soler, and A. P. Moller, 1999. Genetics and geographic variation in rejection behavior of cuckoo eggs by European magpie populations: an experimental test of rejecter-gene flow. Evolution 53: 947–956. Sorci, G., A. P. Moller, and J. Clobert. 1998. Plumage dichromatism of birds predicts introduction success in New Zealand. J. Anim. Ecol. 67: 263– 269.

references

403

Soul´e, M. E., and J. Terborgh, eds. 1999. Continental Conservation: Scientific Foundations of Regional Reserve Networks. Washington, D.C.: Island Press. Sousa, W. P. 1979. Experimental investigations of disturbance and ecological succession in a rocky intertidal community. Ecol. Monogr. 49: 227– 254. Spiller, D. A. 1984a. Competition between two spider species: experimental field study. Ecology 65: 909–919. . 1984b. Seasonal reversal of competitive advantage between two spider species. Oecologia 64: 322–331. Spiller, D. A., J. B. Losos, and T. W. Schoener. 1998. Impact of a catastrophic hurricane on island populations. Science 281: 695–697. Spiller, D. A., and T. W. Schoener. 1988. An experimental study of the effect of lizards on webspider communities. Ecol. Monogr. 58: 57–77. . 1990a. Lizards reduce food consumption by spiders: mechanisms and consequences. Oecologia 83: 150–161. . 1990b. A terrestrial field experiment showing the impact of eliminating top predators on foliage damage. Nature 347: 469–472. . 1994. Effects of top and intermediate predators in a terrestrial food web. Ecology 75: 182–196. . 1995. Long-term variation in effect of lizards on spider density is linked to rainfall. Oecologia 103: 133–139. . 1996. Food-web dynamics on some small subtropical islands: effects of top and intermediate predators. In G. A. Polis and K. Winemiller, eds, Food Webs: Integration of Pattern and Dynamics, 160–169. New York, NY: Chapman & Hall. . 1997. Folivory on islands with and without insectivorous lizards: an eight-year study. Oikos 78: 15–22. . 1998. Lizards reduced spider species richness by excluding rare species. Ecology 79: 503–516. . 2001. An experimental test for predator-mediated interactions among spider species. Ecology 82: 1560–1579. Stadler, B., and T. Muller. ¨ 1996. Aphid honeydew and its effect on the phyllosphere microflora of Picea abies (L.) Karst. Oecologia 108: 771–776. Stallard, R., T. Garc´ıa, and M. Mitre. 1999. Hidrologia y suelos. In S. Heckadon-Moreno, R. Ib´anez, ˜ and R. Condit, eds. La Cuenca del Canal: Deforestaci´on, Urbanizaci´on y Contaminaci´on, 57–83. Panama, R. P.: Smithsonian Tropical Research Institute. Stanforth, L. M., S. M. Louda, and R. L. Bevill. 1997. Insect herbivory on juveniles of a threatened plant, Cirsium pitcheri, in relation to plant size, density and distribution. EcoScience 4: 57–66. Stanley, S. 1979. Macroevolution: Pattern and Process. San Francisco, CA: Freeman. Steinberg, P. D., J. A. Estes, and F. C. Winter. 1995. Evolutionary consequences of food chain length in kelp forest communities. Proc. Nat. Acad. Sci. USA 92: 8145–8148. Steneck, R. S., and M. N. Dethier. 1994. A functional group approach to the structure of algal-dominated communities. Oikos 69: 476–498.

404

references

Sterner, R. W. 1986. Herbivores’ direct and indirect effects on algal populations. Science 231: 605–607. Stohlgren, T. J., D. Binkley, G. W. Chong, M. A. Kalkhan, L. D. Schell, K. A. Bull, Y. Otsuki, G. Newman, M. Bashkin, and Y. Son. 1999. Exotic plant species invade hot spots of native plant diversity. Ecol. Monogr. 69: 25–46. Stohlgren, T. J., K. A. Bull, and Y. Otsuki. 1998. Comparison of rangeland vegetation sampling techniques in the Central Grasslands. J. Range Manage. 51: 164–172. Stork, N. E., and C.H.C. Lyal, 1993. Extinction or co-extinction rates? Nature 366: 307. Strong, D. R., Jr. 1974. Rapid asymptotic species accumulation in phytophagous insect communities: the pests of cacao. Science 185: 1064– 1066. Strong, D. R., Jr., J. L. Maron, and P. G. Connors. 1996. Top down from underground? The underappreciated influence of subterranean food webs on above-ground ecology. In G. A. Polis and K. Winemiller, eds, Food Webs: Integration of Pattern and Dynamics, 170–175. New York, NY: Chapman & Hall. Strong, D. R., Jr., E. D. McCoy, and J. R. Rey. 1977. Time and number of herbivore species: the pests of sugarcane. Ecology 58: 167–175. Stubbendieck, J., G. Y. Friisoe, and M. R. Bolick. 1994. Weeds of Nebraska and the Great Plains. Lincoln: Nebraska Department of Agriculture. Sugihara, G. 1980. Minimal community structure: an explanation of species abundance patterns. Amer. Nat. 116: 770–787. Sullivan, K. M., and G. Bustamante. 1997. Geographic priorities of marine conservation: The Nature Conservancy’s ecoregional platform for the wider Caribbean. Presented at the 50th Annual Meeting of the Gulf and Caribbean Fisheries Institute, Merida, Yucatan, Mexico; November 9–14, 1997. Sutherland, J. P. 1974. Multiple stable points in natural communities. Amer. Nat. 108: 859–873. Swartz, M. B. 1998. Predation on an Atta cephalotes colony by an army ant, Nomamyrmex esenbeckii. Biotropica 30: 682–684. Symposium on spiders in agroecosystems. 1999. J. Arachnol. 27: 267–400. Tabashnik, B. E. 1983. Host range evolution: the shift from native legume hosts to alfalfa by the butterfly, Colias eriphyle. Evolution 37: 150–162. Tanner, J. E., T. P. Hughes, and J. H. Connell. 1994. Species coexistence, keystone species, and succession: a sensitivity analysis. Ecology 75: 2204– 2219. . 1996. The role of history in community dynamics: a modeling approach. Ecology 77: 108–117. Tansley, A. G. 1917. On competition between Galium saxatile L. (G. hercynicus Weig.) and Galium sylvestre Poll. (G. asperum Schreb.) on different types of soil. J. Ecol. 5: 173–179. . 1935. The use and abuse of vegetational concepts and terms. Ecology 16: 284–307.

references

405

Taylor, E. B. 1990a. Environmental correlates of life-history variation in juvenile chinook salmon, Oncorhynchus tshawytscha (Walbaun). J. Fish Biol. 37: 1–17. . 1990b. Phenotypic correlates of life-history variation in juvenile chinook salmon, Oncorhynchus tshawytscha. J. Anim. Ecol. 59: 455–468. . 1991. A review of local adaptation in Salmonidae, with particular reference to Pacific and Atlantic salmon. Aquaculture 98: 185–207. Taylor, E. B., and P. A. Larkin. 1986. Current response and agonistic behavior in newly emerged fry of chinook salmon, Oncorhynchus tshawytscha, from ocean- and stream-type populations. Can. J. Fish. Aquat. Sci. 43: 565–573. Taylor, P. W. 1986. Respect for Nature. Princeton, NJ: Princeton University Press. Tear, T. H., J. M. Scott, P. H. Hayward, and B. Griffith, 1993. Status and prospects for success of the Endangered Species Act: a look at recovery plans. Science 262: 976–977. . 1995. Recovery plans and the Endangered Species Act: are criticisms supported by data? Cons. Biol. 9: 182–195. Temple, S. A. 1977. Plant-animal mutualism: Coevolution with dodo leads to near extinction of plant. Science 197: 885–886. Templeton, A. R. 1989. The meaning of species and speciation: a genetic perspective. In D. Otte and J. A. Endler, eds., Speciation and its Consequences, 3–27. Sunderland, MA: Sinauer Associates. . 1994. The role of molecular genetics in speciation studies. In B. Schierwater, ed., Molecular Ecology and Evolution: Approaches and Applications, 455–477. Boston, MA: Birkhauser. Terborgh, J. 1986. Keystone plant resources in the tropical forest. In M. E. Soul´e, ed., Conservation Biology, 330–344. Sunderland, MA: Sinauer Associates. Terborgh, J., J. A. Estes, P. Paquet, K. Ralls, D. Boyd-Heger, B. J. Miller, and R. F. Noss. 1999. The role of top carnivores in regulating terrestrial ecosystems. In M. E. Soul´e and J. Terborgh, eds., Continental Conservation, 39–64. Washington, D.C.: Island Press. Terborgh, J., L. Lopez, P. Nunez, ˜ M. Rao, G. Shahabuddin, G. Orihuela, M. Riveros, R. Ascanio, G. H. Adler, T. D. Lambert, and L. Balbas. 2001. Ecological meltdown in predator-free forest fragments. Science 294: 1923–1926. Terborgh, J., L. Lopez, J. Tello, D. Yu, and A. R. Bruni. 1997. Transitory states in relaxing ecosystems of land bridge islands. In W. F. Laurance and R. O. Bierregaard, Jr., eds., Tropical Forest Remnants, 256–274. Chicago, IL: University of Chicago Press. Thom, R. M. 1988. Benthic primary production in the eelgrass meadow at the Padilla Bay National Estuarine Research Reserve, Washington. Report to the National Oceanic & Atmospheric Administration (NOAA), Ocean and Coastal Resource Management (OCRM), and Marine and Estuarine Management Division (MEMD) by University of Washington, Fisheries Research Institute (FRI-UW-8808). 33 pp.

406

references

Mount Vernon, WA: Padilla Bay National Estuarine Research Reserve Reprint Series No. 10. . 1990. Spatial and temporal patterns in plant standing stock and primary production in a temperate seagrass system. Bot. Mar. 33: 497– 510. Thom, R. M., J. Cordell, C. Simenstad, V. Luiting, and A. B. Borde. 1997. Autoecology of Spartina in Willapa Bay, Washington: benthic metabolism and below ground growth. Proceedings of the Second International Spartina Conference, Olympia, WA; March 20–21, 1997. Thom, R. M., B. Miller, and M. Kennedy. 1991. Temporal patterns of grazers and vegetation in a temperate seagrass system. Mount Vernon, WA: Padilla Bay National Estuarine Research Reserve Reprint Series No. 15. Thom, R. M., and S. Rumrill. 1999. Mapping and assessing estuarine habitat quality. In J. K. Parrish and S. Breslow, eds., Pacific Northwest Coastal Ecosystem Regional Study (PNCERS) 1998 Annual Report, Coastal Ocean Program, NOAA, 37–41. Thomas, C. D., D. Ng, M. C. Singer, L. B. Mallet, C. Parmesan, and H. L. Billington. 1987. Incorporation of a European weed into the diet of a North American herbivore. Evolution 41: 892–901. Thomas, S. C. 2002. Ecological correlates of tree species persistence in tropical forest fragments. In E. Losos, R. Condit, J. V. LaFrankie, and E. G. Leigh Jr., eds., Forest Diversity and Dynamism: Results from Large-Scale Demographic Plots. Chicago, IL: University of Chicago Press. In press. Thompson, J. N. 1988. Variation in interspecific interactions. Ann. Rev. Ecol. Syst. 19: 65–87. . 1996. Evolutionary ecology and the conservation of biodiversity. Trends Ecol. Evol. 11: 300–303. Thompson, J. N., and O. Pellmyr. 1992. Mutualism with pollinating seed parasites amid co-pollinators: constraints on specialization. Ecology 73: 1780–1791. Thomson, J. D., and B. A. Thomson. 1992. Pollen presentation and viability schedules and their consequences for reproductive success through animal pollination. In R. Wyatt, ed., Ecology and Evolution of Plant Reproduction: New Approaches, 1–24. New York, NY: Chapman & Hall. Thøstesen, A. M., and J. M. Olesen. 1996. Pollen removal and deposition by specialist and generalist bumblebees in Aconitum septentrionale. Oikos 77: 77–84. Tiffney, B. H. 1984. Seed size, dispersal syndromes, and the rise of the angiosperms: evidence and hypothesis. Ann. Missouri Bot. Gard. 71: 551–576. Tiffney, B. H., and S. J. Mazer. 1995. Angiosperm growth habit, dispersal and diversification reconsidered. Evol. Ecol. 9: 93–117. Tilman, D. 1977. Resource competition between planktonic algae: an experimental and theoretical approach. Ecology 58: 338–348.

references

407

. 1996. Biodiversity: population versus ecosystem stability. Ecology 77: 350–363. . 1997a. Community invasibility, recruitment limitation, and grassland biodiversity. Ecology 78: 81–92. . 1997b. Distinguishing between the effects of species diversity and species composition. Oikos 80: 185. . 1999. The ecological consequences of changes in biodiversity: a search for general principles. Ecology 80: 1455–1474. Tilman, D., and J. A. Downing, 1994. Biodiversity and stability in grasslands. Nature 367: 363–365. Tilman, D., and P. Kareiva, eds. 1997. Spatial Ecology. Princeton, NJ: Princeton University Press. Tilman, D., J. Knops, D. Wedin, P. Reich, M. Ritchie, and E. Siemann. 1997. The influence of functional diversity and composition on ecosystem processes. Science 277: 1300–1302. Tilman, D., C. L. Lehman, and C. E. Bristow. 1998. Diversity-stability relationships: statistical inevitability or ecological consequence? Amer. Nat. 151: 277–282. Tilman, D., C. L. Lehman, and K. T. Thomson. 1997. Plant diversity and ecosystem productivity: theoretical considerations. Proc. Nat. Acad. Sci. USA 94: 1857–1861. Tilman, D., M. Mattson, and S. Langer. 1981. Competition and nutrient kinetics along a temperature gradient: an experimental test of a mechanistic approach to niche theory. Limnol. Oceanogr. 26: 1020–1033. Tilman, D., R. M. May, C. L. Lehman, and M. A. Nowak. 1994. Habitat destruction and the extinction debt. Nature 371: 65–66. Tilman, D., D. Wedin, and J. Knops. 1996. Productivity and sustainability influenced by biodiversity in grassland ecosystems. Nature 379: 718– 720. Tonn, W. M. 1985. Density compensation in Umbra-perca fish assemblages of northern Wisconsin lakes. Ecology 66: 415–429. Towns, D. R. 2002. Unpredicted rebound responses following eradication of introduced rats and rabbits from islands in northern New Zealand. In C. R. Veitch and M. Clout, eds., Turning the Tide: The Eradication of Invasive Species. Auckland, N.Z.: World Conservation Union. Townsend, C. R., and M. J. Winterbourn. 1992. Assessment of the risk posed by an exotic fish: the proposed introduction of channel catfish (Ictalurus punctatus) to New Zealand. Cons. Biol. 6: 273–282. Travis, J. 1996. The significance of geographical variation in species interactions. Amer. Nat. 148: S1–S8. Trowbridge, C. D. 1998. Stenophagous, herbivorous sea slugs attack desiccation-prone, green algal hosts (Codium spp.): indirect evidence of prey-stress models (PSMs)? J. Exper. Mar. Biol. Ecol. 230: 31–53. Trussell, G. C. 1996. Phenotypic plasticity in an intertidal snail: the role of a common crab predator. Evolution 50: 448–454. Trussell, G. C., and L. D. Smith. 2000. Induced defenses in response to an

408

references

invading crab predator: An explanation of historical and geographic phenotypic change. Proc. Nat. Acad. Sci. USA 97: 2123–2127. U. S. Department of Agriculture. 1977. Biological control of alligatorweed, 1959–1972. A review and evaluation. Technical Bulletin No. 1547, Agricultural Research Service, U.S.D.A. USDA, N. R. C. S. (United States Department of Agriculture, Natural Resources Conservation Service). 1999. The PLANTS database (http:// plants.usda.gov/plants). Baton Rouge, FL: National Plant Data Center. Usher, M. B. 1979. Markovian approaches to ecological succession. J. Anim. Ecol. 48: 413–426. van den Heuvel, J., S. A. Hogenhout, and F. van der Wilk. 1999. Recognition and receptors in virus transmission by arthropods. Trends Microbiol. 7: 71–76. van der Heijden, M.G.A., J. N. Klironomos, M. Ursic, P. Moutoglis, R. Streitwolf-Engel, T. Boller, A. Wiemken, and I. R. Sanders. 1998. Mycorrhizal fungal diversity determines plant biodiversity, ecosystem variability and productivity. Nature 396: 69–72. Vanni, M. J., and D. L. Findlay. 1990. Trophic cascades and phytoplankton community structure. Ecology 71: 921–937. Vanni, M. J., and C. D. Layne. 1997. Nutrient recycling and herbivory as mechanisms in the “top-down” effect of fish on algae in lakes. Ecology 78: 21–40. Vanni, M. J., C. D. Layne, and S. E. Arnott. 1997. “Top-down” trophic interactions in lakes: effects of fish on nutrient dynamics. Ecology 78: 1–20. Veitch, C. R., and B. D. Bell. 1990. Eradication of introduced animals from the islands of New Zealand. In D. R. Towns, C. H. Daugherty, and I.A.E. Atkinson, eds., Ecological Restoration of New Zealand Islands, 137–146. Wellington, N.Z., Department of Conservation. Veltman, C. J., S. Nee, and M. J. Crawley. 1996. Correlates of introduction success in exotic New Zealand birds. Amer. Nat. 147: 542–557. Vermeij, G. 1982. Environmental change and the evolutionary history of the periwinkle (Littorina littorea) in North America. Evolution 36: 561– 580. Vincent, W. F., C. Howard-Williams, P. Tildesley, and E. Butler. 1991. Distribution and biological properties of oceanic water masses around the South Island, New Zealand. N. Z. J. Mar. Freshw. Res. 25: 21–42. Vinton, M. A., D. C. Hartnett, E. J. Finck, and J. M. Briggs. 1993. Interactive effects of fire, bison (Bison bison) grazing and plant community composition in tallgrass prairie. Amer. Midl. Nat. 129:10–18. Vitousek, P. M. 1986. Biological invasions and ecosystem properties: can species make a difference? In H. A. Mooney and J. A. Drake, eds., Ecology of Biological Invasions of North America and Hawaii, 163–176. New York, NY: Springer-Verlag. Vitousek, P. M., and D. U. Hooper. 1994. Biological diversity and terrestrial ecosystem biogeochemistry. In E.-D. Schulze and H. A. Mooney, eds., Biodiversity and Ecosystem Function, 3–14. Berlin: Springer-Verlag.

references

409

Vlok, J.H.J. 1988. Alpha diversity of lowland fynbos herbs at various levels of infestation by alien annuals. South Afr. J. Bot. 54: 623–627. Volterra, V. 1926. Fluctuations in the abundance of a species considered mathematically. Nature 118: 558–560. . 1931. Le¸cons Sur La Th´eorie Math´ematique de la Lutte pour la Vie. Paris: Gauthier-Villars. von Broembsen, S. L. 1989. Invasions of natural ecosystems by plant pathogens. In J. A. Drake, H. A. Mooney, F. di Castri, R. H. Groves, F. J. Kruger, M. Rejm´anek, and M. Williamson, eds., Biological Invasions. A Global Perspective, 77–83. Chichester, U.K.: Wiley. Waggoner, P. E., and G. R. Stephens. 1970. Transition probabilities for a forest. Nature 225: 1160–1161. Waide, R. B., M. R. Willig, C. F. Steiner, G. Mittelbach, L. Gough, S. I. Dodson, G. P. Juday, and R. Parmenter. 1999. The relationship between productivity and species richness. Ann. Rev. Ecol. Syst. 30: 257–300. Walker, B. H. 1991. Biodiversity and ecological redundancy. Cons. Biol. 6: 18– 23. . 1995. Conserving biological diversity through ecosystem resilience. Cons. Biol. 9: 747–752. Wall, D. H., and J. C. Moore. 1999. Interactions underground. BioScience 49: 109–117. Walton, W. C., and W. C. Walton. 2001. Problems, predators, and perception: management of quahog (hardclam), Mercenaria mercenaria, stock enhancement programs in Southern New England. J. Shellfish Res. 20: 127–134. Waples, R. 1995. Evolutionarily significant units and the conservation of biological diversity under the Endangered Species Act. Amer. Fish. Soc. Symp. 17: 8–27. Waples, R. S. 1991. Pacific salmon, Oncorhynchus spp., and the definition of “species” under the Endangered Species Act. Mar. Fish. Rev. 53: 11–22. Ward, T. J., M. A. Vanderklift, A. O. Nicholls, and R. A. Kenchington. 1999. Selecting marine reserves using habitats and species assemblages as surrogates for biological diversity. Ecol. App. 9: 691–698. Wardle, D. A. 1999. Is “sampling effect” a problem for experiments investigating biodiversity-ecosystem function relationships? Oikos 87: 403–407. Wardle, D. A., M. A. Huston, J. P. Grime, F. Berendse, E. Garnier, W. K. Lauenroth, H. Set¨al¨a, and S. D. Wilson. 2000. Biodiversity and ecosystem function: an issue in ecology. Bull. Ecol. Soc. Amer. 81: 235– 239. Wardle, D. A., O. Zackrisson, G. Hornberg, and C. Gallet. 1997. Biodiversity and ecosystem properties. Science 278: 1867–1869. Waring, R. H., and W. H. Schlesinger. 1985. Forest Ecosystems. Orlando, FL: Academic Press. Warner, S. C., J. Travis, and W. A. Dunson. 1993. Effect of pH variation on interspecific competition between two species of hylid tadpoles. Ecology 74: 183–194. Waser, N. M., L. Chittka, M. V. Price, N. M. Williams, and J. Ollerton. 1996.

410

references

Generalization in pollination systems, and why it matters. Ecology 77: 1043–1060. Waterhouse, P. M., N. A. Smith, and M. B. Wang. 1999. Virus resistance and gene silencing: killing the messenger. Trends Plant Sci. 4: 452–457. WDF (Washington Department of Fisheries), WDW (Washington Department of Wildlife), WWTIT (Western Washington Treaty Indian Tribes). 1993. 1992 Washington State Salmon and Steelhead Stock Inventory. Washington Department of Fisheries, Washington Department of Wildlife, Western Washington Treaty Indian Tribes, Olympia, WA. Weber, E. 1998. The dynamics of plant invasions: a case study of three exotic goldenrods species (Solidago L.) in Europe. J. Biogeogr. 25: 147–154. Weber, E., and B. Schmid. 1998. Latitudinal population differentiation in two species of Solidago (Asteraceae) introduced into Europe. Amer. J. Bot. 85: 1110–1121. Welcomme, R. L. 1988. International Introductions of Inland Aquatic Species. Food and Agriculture Organization (FAO) Fisheries Technical Paper No. 294. Wellnitz, T., and N. L. Poff. 2001. Functional redundancy in heterogeneous environments: implications for conservation. Ecol. Lett. 4: 177–179. Wells, M.J., R. J. Poynton, A. A. Balsinhas, K. J. Musil, H. Joffe, E. van Hoepen, and S. K. Abbott. 1986. The history of introduction of invasive alien plants to southern Africa. In I.A.W. Macdonald, F. J. Kruger, and A. A. Ferrar, eds., The Ecology and Management of Biological Invasions in Southern Africa, 21–35. Capetown, South Africa: Oxford University Press. Weltzin, J. F., S. Archer, and R. K. Heitschmidt. 1997. Small-mammal regulation of vegetation structure in a temperate savanna. Ecology 78: 751– 763. Werner, E. E. 1992. Individual behavior and higher-order species interactions. Amer. Nat. 140: S5–S32. Werner, E. E., and J. F. Gilliam. 1984. The ontogenetic niche and species interactions in size-structured populations. Ann. Rev. Ecol. Syst. 15: 393–425. Wheeler, A. G., Jr. 1974. Phytophagous arthropod fauna of crown vetch in Pennsylvania. Can. Entomol. 106: 897–908. White, G. G. 1981. Current status of prickly pear control by Cactoblastis cactorum in Queensland. In E. S. Del Fosse, ed., Proceedings of the Fifth International Symposium on the Biological Control of Weeds, Brisbane, 1980, 609–616. Melbourne: Commonwealth Scientific & Industrial Research Organisation (CSIRO). White, T.R.C. 1978. The importance of relative shortage of food in animal ecology. Oecologia 33: 233–242. Whitmore, T. C., and C. P. Burnham. 1969. The altitudinal sequence of forests and soils on granite near Kuala Lumpur. Malayan Nat. J. 22: 99–118. Wiens, J. A., R. G. Cates, J. T. Rotenberry, N. Cobb, B. Van Horne, and R. A. Redak. 1991. Arthropod dynamics on sagebrush (Artemisis tridentata): effects of plant chemistry and avian predation. Ecol. Monogr. 61: 299–321.

references

411

Wilbur, H. M. 1972. Competition, predation, and the structure of the Ambystoma-Rana sylvatica community. Ecology 57: 3–21. . 1987. Regulation of structure in complex experimental temporary pond communities. Ecology 68: 1437–1452. Wilbur, H. M., and J. E. Fauth. 1990. Experimental aquatic food webs: interactions between two predators and two prey. Amer. Nat. 135: 176–204. Wilcove, D. S., M. McMillan, and K. C. Winston. 1993. What exactly is an endangered species? An analysis of the U. S. Endangered Species List: 1985–1991. Cons. Biol. 7(1): 87–93. Wilcove, D. S., D. Rothstein, J. Dubow, A. Phillips, and E. Losos. 1998. Quantifying threats to imperiled species in the United States. BioScience 48(8): 607–615. Wilkinson, G. 1978. Epitaph for the Elm. London: Hutchinson. Williams, C. B. 1964. Patterns in the Balance of Nature. New York, NY: Academic Press. Williams, E. S., E. T. Thorne, M.J.G. Appel, and D. W. Belitsky. 1988. Canine distemper in black-footed ferrets (Mustela nigripes) from Wyoming. J. Wildlife Dis. 24: 385–398. Williamson, M. 1996. Biological Invasions. London: Chapman & Hall. Williamson, M. H., and A. Fitter. 1996a. The characters of successful invaders. Biol. Cons. 78: 163–170. . 1996b. The varying success of invaders. Ecology 77: 1661–1666. Wills, C., and R. Condit. 1999. Similar non-random processes maintain diversity in two tropical rainforests. Proc. Roy. Soc. London B 266: 1445–1452. Wills, C., R. Condit, R. B. Foster, and S. P. Hubbell. 1997. Strong density- and diversity-related effects help to maintain tree species diversity in a neotropical forest. Proc. Nat. Acad. Sci. USA 94: 1252–1257. Willson, M. F., S. M. Gende, and B. H. Marston. 1998. Fishes and the forest: expanding perspectives on fish-wildlife interactions. BioScience 48: 455–462. Wilson, E. O. 1985. The biological diversity crisis: a challenge to science. Issues Sci. Tech. 2: 20–29. . 1990. The Diversity of Life. New York, NY: Norton. Wilson, P., and J. D. Thomson. 1991. Heterogeneity among floral visitors leads to discordance between removal and deposition of pollen. Ecology 72: 1503–1507. Windsor, D. A. 1995. Equal-rights for parasites. Cons. Biol. 9: 1–2. Wing, S. L., L. J. Hickey, and C. C. Swisher. 1993. Implications of an exceptional fossil flora for Late Cretaceous vegetation. Nature 363: 342– 344. Wise, D. H. 1993. Spiders in Ecological Webs. Cambridge, U.K.: Cambridge University Press. Wise, D. H., and B. Chen. 1999. Vertebrate predation does not limit density of a common forestfloor wolf spider. Oikos 84: 209–215. Wissinger, S., and J. McGrady. 1993. Intraguild predation and competition between larval dragonflies: direct and indirect effects on shared prey. Ecology 74: 207–218.

412

references

Witkowski, E.T.F. 1991. Effects of alien invasive acacias on nutrient cycling in the coastal lowlands of the Cape fynbos. J. App. Ecol. 28: 1–15. Witman, J. D., and K. R. Grange. 1998. Links between rain, salinity, and predation in a rocky subtidal community. Ecology 79: 2429–2447. Wolda, H. 1983. Spatial and temporal variation in abundance in tropical animals. In S. L. Sutton, T. C. Whitmore, and A. C. Chadwick, eds. Tropical Rain Forest: Ecology and Management, 93–105. Oxford, U.K.: Blackwell Scientific. Wolf, C. M., B. Griffith, C. Reed, and S. A. Temple. 1996. Avian and mammalian translocations: update and reanalysis of 1987 survey data. Cons. Biol. 10: 1142–1154. Wonham, M. J., and J. T. Carlton. Patterns and processes in marine bioinvasions of the Pacific Northwest. Biol. Inv. in rev. Wonham, M. J., J. T. Carlton, G. M. Ruiz, and L. D. Smith. 2000. Fish and ships: relating dispersal frequency to success in biological invasions. Mar. Biol. 136: 1111–1121. Wood, C. C. 1995. Life history variation and population structure in sockeye salmon. Amer. Fish. Soc. Symp. 17: 195–216. Wootton, J. T. 1992. Indirect effects, prey susceptibility, and habitat selection: impacts of birds on limpets and algae. Ecology 73: 981–991. . 1993a. Indirect effects and habitat use in an intertidal community: interaction chains and interaction modifications. Amer. Nat. 141: 71– 89. . 1993b. Size-dependent competition: effects on the dynamics versus the endpoint of mussel bed succession. Ecology 74: 195–206. . 1994a. Putting the pieces together: testing the independence of interactions among organisms. Ecology 75: 1544–1551. . 1994b. The nature and consequences of indirect effects in ecological communities. Ann. Rev. Ecol. Syst. 25: 443–466. . 1997. Estimates and tests of per capita interaction strength: diet, abundance, and impact of intertidally foraging birds. Ecol. Monogr. 67: 45– 64. . 2001. Effects of extinction in complex ecosystems: predictions from Markov models. Ecology. 82: 580–5998. Wootton, J. T., M. S. Parker, and M. E. Power. 1996. Effects of disturbance on river food webs. Science 273: 1558–1561. Wootton, J. T., and M. E. Power. 1993. Productivity, consumers, and the structure of a river food chain. Proc. Nat. Acad. Sci. USA 90: 1384–1387. Wright, S. J. 1996. Plant species diversity and ecosystem functioning in tropical forests. In G. H. Orians, R. Dirzo, and J. H. Cushman, eds., Biodiversity and Ecosystem Processes in Tropical Forests, 11–31. Berlin: Springer-Verlag. Yachi, S., and M. Loreau. 1999. Biodiversity and ecosystem productivity in a fluctuating environment: the insurance hypothesis. Proc. Nat. Acad. Sci. 96: 1463–1468. Yodzis, P. 1989. An Introduction to Theoretical Ecology. New York, NY: Harper & Row.

references

413

Young, H. J. 1988. Differential importance of beetle species pollinating Dieffenbachia longispatha (Araceae). Ecology 69: 832–844. Young, H. J., and T. P. Young. 1992. Alternative outcomes of natural and experimental high pollen loads. Ecology 73: 639–647. Zaret, T. M., and R. T. Paine. 1973. Species introductions in a tropical lake. Science 182: 449–455. Zeh, D. W., and J. A. Zeh. 1992a. Dispersal-generated sexual selection in a beetle-riding pseudoscorpion. Behav. Ecol. Sociobiol. 30: 135–142. . 1992b. Emergence of a giant fly triggers phoretic dispersal in the neotropical pseudoscorpion, Seme¨ıochernes armiger (Balzan) (Pseudoscorpionida: Chernetidae). Bull. British Arachnol. Soc. 9: 43–46. Zeh, J. A., and D. W. Zeh. 1994. Tropical liaisons on a beetle’s back. Nat. Hist. 103: 36–43. ˚ Zheng, D. W., J. Bengtsson, and G. I. Agren. 1997. Soil food webs and ecosystem processes: decomposition in donor-control and Lotka-Volterra systems. Amer. Nat. 149: 125–148. Zimov, S. A., V. I. Chuprynin, A. P. Oreshko, F. S. Chapin III, J. F. Reynolds, and M. C. Chapin. 1995. Steppe-tundra transition: a herbivore-driven biome shift at the end of the Pleistocene. Amer. Nat. 146: 765–794. Zinn, J. L., K. A. Johnson, J. E. Sander, and J. L. Fryer. 1977. Susceptibility of salmonid species and hatchery stocks of chinook salmon (Oncorhynchus tshawytscha) to infections by Ceratomyxa shasta. J. Fish. Res. Board Can. 34: 933–936. Zipperer, V. T. 1996. Ecological effects of the introduced cordgrass, Spartina alterniflora, on the benthic community structure of Willapa Bay, Washington. Master’s thesis, University of Washington. Zotz, G., G. Harris, M. Koniger, ¨ and K. Winter. 1995. High rates of photosynthesis in the tropical pioneer tree, Ficus insipida Willd. Flora 190: 265–272. Zotz, G., and K. Winter. 1994. Predicting annual carbon balance from leaf nitrogen. Naturwissenschaften 81: 449. Zwolfer, ¨ H. 1965. Preliminary list of phytophagous insects attacking wild Cynareae (Compositae) in Europe. Commonw. Inst. Biol. Contr. Tech. Bull. 6: 81–154. . 1988. Evolutionary and ecological relationships of the insect fauna of thistles. Ann. Rev. Entomol. 33: 103–229. Zwolfer, ¨ H., and M. Romstock-V¨ ¨ olkl. 1991. Biotypes and the evolution of niches in phytophagous insects on Cardueae hosts. In P. W. Price, T. M. Lewinsohn, G. Wilson Fernandes, and W. W. Benson, eds., Plant-Animal Interactions: Evolutionary Ecology in Tropical and Temperate Regions, 487–507. New York, NY: Wiley.

Index

m

aerial arthropods removal experiment, 70–73; magnitudes of effect and, 72t; mean leaf damage and, 75fig; results of, 73; results of, as representative of effect ratios in system, 75–79; significance/interpretation/implications of, 73–75; species importance concept and, 82–84; weaker effects of spiders vs. lizards in, 79–82 African walking catfish, 169 alewife (Alosa pseudoharengus), 228 alligatorweed (Alternanthera philoxeroides), 230 Allison, G., 43 Alouatta seniculus (howler monkeys), 249 Ameiurus melas (black bullhead), 168 American elm (Ulmus americana), 226–27 Andelman, S., 329 Anolis sagrei. See lizard (Anolis sagrei) removal experiment Antarctic blue whales (Balaenoptera musculus), 228 aphids (Aphidiae), 7 apple maggot (Rhagoletis pomonella), 288 Aristotle, 244 Ascophyllum nodosum, 19 Asian chestnut blight fungus (Cryphonectria parasitica), 226 Asian eelgrass. See Zostera japonica (Asian eelgrass) “the assertion of lesser effect” proposition, 82–83 Atlantic cordgrass. See Spartina alterniflora (Atlantic cordgrass) autotrophs: community with coexisting decomposers and, 127fig-32; functional role in ecosystem by, 117, 118, 120–21 autotrophy, 117

Balaenoptera musculus (Antarctic blue whales), 228 Balanus glandula, 22, 24, 48, 49 Ball, I., 329 Baris subsimilis Casey (Curculionidae), 9 barnacle (Balanus glandula), 46 barnacle (Chthamalus dalli), 46 barnacles: further observations in New Zealand on, 40–41; hydrodynamic forces/thermal stress on, 19, 20fig, 21– 23; productivity/prey recruitment effects studies on, 24–27, 29–30; studies on interactions of in high intertidal communities, 46–60 Bay of Panama biodiversity manipulation study, 95–97 Berlow, E., 43 biodiversity manipulation studies: compared to targeted species removals, 86–87; emergence of, 86–87; field experiments revealing impact of depletion, 1–4; goals of, 90–93; hybrid approaches using alternatives to, 100– 104; questions answered by targeted species reductions for, 93–98; results of Bay of Panama, 95–97; results of frog production/amphibian diversity, 94fig-95; strengths/shortcomings of, 99–100; summary of expected patterns of species contributions during, 92fig biological insurance/ecosystem reliability relationship, 119–20 biotic context of species impact: described, 31; effects of species characteristics, 38; effects of species composition, 33–38; potential for compensation, 31–33 Bison bison, 227

416

Black Bass Act of 1926, 185 Blossey, B., 291 brachyuran crabs (Carcinus maenas, Cancer irroratus, C. borealis), 19 Bronstein, J., 280 Brown, J., 291 “buffering” of functioning, 120 Buhle, E., 329 bull thistle (Cirsium vulgare), 8–10fig, 11– 12 Bulthuis, D. A., 66 butterflies (Nymphalidae, herperiidae), 7 Cactoblastis cactorum (moth), 231 Campbell, T., 233 Campephilus principalis (ivory-billed woodpecker), 224 Cancer borealis (shore crab), 296 Carassius auratus (goldfish), 168 Carcinus maenas (European green crab), 198 Carlton, J., 205 Carolina parakeet (Conuropsis carolinensis), 224 Carr, T., 291 Carter, J., 186 Castor canadensis (North American beaver), 227 catfish (Clarius gariepinus), 169, 172 Cecropia peltata, 252 Ceratomyxa shasta (freshwater myxosporean parasite), 317 chestnut blight (Cryphonectria parasitica), 330 chinook salmon (Oncorhynchus tshawytscha): evolutionary significant units /listing status of, 320t; geographic boundaries of ESUs of, 309fig, 323fig; population features/regional conservation planning for viable ESUs of, 317–24; population number/persistence theory applied to, 311; ranking populations for protection of, 324– 26fig, 327fig, 328fig; susceptibility to parasite by, 317 Chondrus crispus, 19 Chthamalus dalli, 24, 48, 49, 55 Cirsium spp. (thistles): described, 5; discussion on field experiments on insects and density of, 10–14; expendability of, 12; insect herbivores limiting densi-

index

ties of indigenous, 7–8; insects adapted to, 7; insects harbored by native thistles/attacking exotic, 8–10fig, 11fig; natural history background of, 6–7; vital ecosystem resistance role of, 14–15 Clarius gariepinus (catfish), 169, 172 Clinton, B., 186 Coccoloba uvifera (sea grape), 70 Coelosphaerium kutzingianum, 214 cold-water rock shrimp (Pandalus borealis), 300–301 Collier, B., 15 colonization: ecosystem reliability used to estimate, 133; ecosystem reliability used to explore, 118–19, 123–25 Colorado potato beetle (Leptinotarsa decemlineata), 288–89 community: disproportionately exert effects of keystone species on, 158; impact of disease-causing microbes on, 330–46; impact of single-species extinctions on, 221–33; implications of plant viruses specificity for, 344–46; rarity/ functional importance in phytoplankton, 206–20; relationship between species removals on biomass of, 156fig; role of disturbance in natural, 348. See also keystone species concept community perspective: on autotroph/ decomposer coexistence, 127fig-32; comparing dynamics of ecosystemfunction to, 115–18; ecosystemfunction perspective vs., 112–15, 114fig; implications for ecosystems of, 88–89. See also keystone species concept community stability: addition of weakly interacting species and, 142–43; averaging effects (portfolio effects) and, 142; disparity of competition impact on, 143–44; distributions of species abundance for two community rules, 150fig; histograms of percentage change in, 152fig; importance of stability-diversity question and, 140– 41; mean species abundance and, 156t; niche complementarity/compensatory competition and, 141, 142; predicting importance of rare species for, 155t; recent studies on stability-diversity patterns, 141–45; relationship between mean abundance of species/survival

index

rate/space vacated by other species, 157fig; relationship between species richness and, 151fig; relationships between dominance of removed species and, 153fig; simple sampling effect on, 141, 142 competition: addition of weakly interacting species and, 142–43; community stability and disparity of, 143–44; community stability and niche complementarity/compensatory, 141, 142; LotkaVolterra models on, 143; population regulation and role of, 247–48 Connell, J. H., 15, 97 conservation: challenges of maintenance of ecosystem services/process, 349–51; consequences of legislation for, 187; decisions regarding extinctions and goals of, 243–57; implications of plant viruses specificity for, 344–46; implications of pollinator studies for, 279–80; keeping status quo motivation of, 3; recovering species of concern to, 305– 29. See also ecology science Conservation Biology, 349 consumers, 117, 118, 121 context dependency, 2 Conuropsis carolinensis (Carolina parakeet), 224 Cordell, J., 205 crab (Carcinus maenas), 33 Crooks, K. R., 133 Crowder, L., 329 Crozier, E., 66 Cryphonectria parasitica (Asian chestnut blight fungus), 226 Cryphonectria parasitica (chestnut blight), 330 Ctenopharyngodon idella (grass carp), 168 Cyprinus carpio (common carp), 168 Dahlhoff, E., 43 Damiani, C., 280 D. antarctica, 27 D’Antonio, C. M., 162 “Daphnia era” (Lake Washington), 210, 215 Daphnia spp., 210 Dayton, P. K., 15, 223, 233 decomposers: community with coexisting autotrophs and, 127fig-32; functional role in ecosystem by, 117, 118, 120

417

deforestation, 244–45 Diatoma elongatum, 214 Dichomeris Leuconotella (leaf-rolling caterpillar): case for concluding expendability of, 283–86fig, 289–90; frequently encountered parasitoids that attack, 287t; host ranges of, 287t; phenology of, 286fig; selected traits of, 285t diffuse predation, defining, 18 disease-causing microbes. See plant viruses Doak, D. F., 106, 142, 143, 160 dodo (Raphus cucullatus), 225 dominant species: community stability changes and removed, 153fig; defining, 17–18 Downing, A. L., 3, 100 Dreissena polymorpha (zebra mussel), 183 Duggins, D. O., 98 Durvillea willana, 27 Dutch elm disease (Ophiostoma ulmi/O. novo-ulmi), 330 Dwyer, G., 160 Ebert, T., 15 Ecological Applications, 349 ecologically expendable, 64–65 Ecological Monographs (1995–1999), 60 Ecology (1995–1997), 60 Ecology Institute Prize, 349 ecology science: decisions regarding extinctions and goals of, 243–57; two emerging realities of, 44–46. See also conservation ecosystem functional groups: in phytoplankton community, 206–20; reliability block diagram to depict, 122fig; reliability of ecosystem with sequential/dependent, 126t; three main, 117–18, 120–21 ecosystem-function approach: community perspective vs., 112–15, 114fig; comparing dynamics of community to, 115–18; described, 110–11; ecosystem reliability using, 119–38; implications/ limitations of, 135–38; three main functional groups of interest in, 117–18 ecosystem functioning: determinants of, 120–21; lessons of holistic ecology on, 244–46; reliability block diagram to depict groups in, 122fig, 138; role of

418

ecosystem functioning (cont.) Earth’s biota in, 110–11; with sequentially/serially dependent groups, 126t; three main groups of interest in, 117– 18, 120–21 ecosystem reliability: described, 115, 138; estimating parameters and applying, 132–35; extinction probability and, 118–19, 121, 123; four different scenarios for, 131fig; fundamental, 120–23; implications/limitations of, 136–38; including colonization in, 118–19, 123– 25; including nonindependent extinctions in, 125, 126–32; load-sharing models of, 127–32; problematic timescale of, 136; redundancy concept and, 137–38; relationship between biological insurance and, 119–20; with three functional sequential/dependent groups, 126t. See also expendability ecosystem reliability applications: determining independence of extinction probabilities, 134–35; estimating extinction probabilities, 133–34; estimating extinctions/colonizations, 133 ecosystem resistance: competition from native plants element of, 8–10; provided by indigenous insect herbivores, 13–14; role of thistles in, 14–15 ecosystems: deciding what features to preserve in world, 240–43; decisions regarding extinctions and conservation goals for, 243–57; fundamental structure of, 117fig; identifying indicators of processes of, 349; impact of diseasecausing microbes on, 330–32; impact of single-species extinctions on, 221–33; implications of community/ecosystem variables of, 88–89; invasion/extinction and evolution of, 302–3; lessons from holistic treatment of, 244–46; pest pressure hypothesis and, 250–53; predicting extinctions based on understanding of, 246–47; role of Earth’s biota in functioning of, 110–11; sixteen possible states in near-minimal, 128t; species expendability and reliability of, 118–35; studies on invasion of, 161–78; Tansley’s definition of, 88; three main functional groups of interest in, 117–18

index

ecosystem services/processes: challenges of maintenance of, 349–51; indicators of, 349 Ectopistes migratorius (passenger pigeon), 224 Edmondson, W. T., 106–7, 211 effects of species characteristics studies, 38 effects of species composition studies, 33–38 “egg-inspection” behaviors, 303 Ehrl´en, J., 268 Eichhornia crassipes (water hyacinth), 230 endangered species: Fish and Wildlife Service (FWS) list of, 182–83; legislation regarding, 183–85, 187; West Coast salmon status as, 349 Endangered Species Conservation Act of 1969, 185 Enhydra lutris (sea otter), 227–28 environment: concepts of species impact on, 17–18; decisions regarding extinctions and conservation goals for, 243– 57; ethical/moral responsibilities to, 240–43; invasion and interaction of traits of species and, 16; speciesspecific basis of interaction with humans with, 88–89 environmental conditions: oceanographic productivity/recruitment, 23–31, 28fig; studies on hydrodynamic forces/thermal stress, 19, 20fig, 21–23 Eriksson, O., 268 ESA (U.S. Endangered Species Act) [1973], 180, 183, 185–86, 187, 306–8 ESUs (Evolutionarily Significant Units): conservation status of listed Pacific salmonids, 307t; geographic boundaries of chinook slamon, 309fig; MARXAN used to assess populations of, 325t, 326t, 327t; population characteristics/persistence in salmonids, 316–17; population features/regional conservation planning for viable, 317– 26fig; population number/persistence theory applied to salmon, 310–15fig; population number/persistence under conservation frameworks, 308–10; probability of extinction of, 313fig; probability of simultaneous extinction of all populations in, 315fig; which

index

population combinations constitute viable, 315–17 evolution: continuous process of new species, 292–94; examples of rapid ecological, 294–301; invasion/extinction and ecology, 302–3 exotics. See non-native species expendability: case for concluding Dichomeris Leuconotella, 283–86fig; consequences logic of, 138; defining, 45; ecosystem-function perspective vs. community perspective on, 110–18; ecosystem reliability and species, 118– 35; implications of species removal experiments for, 82–84; importance of understanding effects of, 85–86; species ranking and, 110; studies answering question of, 305–8; test case on caterpillars on goldenrods, 281–90; thistle, 12; two issues to consider in, 240; utility of, 63–65; value context of, 3. See also ecosystem reliability; stability-diversity patterns extinctions: biology and management of invasions vs., 181–82fig; common features of invasions and, 180; community/ecosystem impacts of singlespecies, 221–33; decisions regarding preservation goals and, 243–57; disease-causing microbes and, 330–31; ecological/economic scale of invasions vs., 182–83; ecosystem reliability and estimating, 133; ecosystem reliability and nonindependent, 125, 127–32; ecosystem reliability and probability of, 118–19, 121, 123, 133–35; ecosystemresponse variables affected by, 97–98; ethical/scientific grounds for decisions leading to, 281–82; evolution of ecology and invasions and, 302–3; expendable vs. acceptable, 179–204; global, 223–25; human responses to, 240; implications of plant viruses specificity for, 344–46; legislation regarding, 183– 85, 187; Paine’s seminal contribution approach to, 109–10, 347–48; probability of ESU extinction after 100 years, 313fig; probability of simultaneous extinction of all ESU populations, 315fig; removal experiment and percentages of spider, 78fig; Schluter’s variance test

419

determining probabilities of, 134–35t; species coexistence and predicting effects of, 250–53; synergistic effects of multiple species, 98; targeted species reduction/biodiversity manipulation for exploring, 88t; understanding ecosystems to predict effects of, 246–47 Exxon Valdez spill, 349 Fauth, J. E., 94, 95 Federal Noxious Weed Act of 1974, 186 Federal Plant Pest Act of 1957, 186 Federal Seed Act of 1939, 186 Feeley, K., 280 Feldman, T., 280 Felsenstein, J., 167 fennel (Foeniculum vulgare), 229 Findlay, D. L., 98 FishBase database: analyses of patterns of impact reported in, 164; binary ecological effects of invaders reported in, 164; cases of established freshwater fishes in, 168; described, 162–64; on high-impact species invasions, 168–69, 172; on life history for high-impact fish invaders, 170t–71t Fisher, R. A., 244 fish species: determinant of sex change in protogynous, 301; ecological effects/ cumulative number of established freshwater, 176fig; endemism/effects of invaders across countries by freshwater, 175fig; evolution of size in salmon, 296–99, 298fig; FishBase on life history of high-impact, 170t–71t; impacts of established freshwater fishes by, 169fig, 173fig; life history evolution of guppies, 299–300; multiple logistic regression of invasion impact of, 173t–75; phylogenetic relationships among, 167–68; traits predicting invasion impact by, 172–77. See also invasion studies Fish and Wildlife Service (FWS), 182–83 Flecker, A. S., 237 Foeniculum vulgare (fennel), 229 food webs: impact of high-impact species on, 169; of Lake Washington planktonic, 211; stability-diversity patterns and, 144 Ford, M. J., 236

420

index

forest fragmentation: mutualism and tropical forest diversity/productivity and, 255; to understand forest population regulation, 248–49 “forgotten” pollinators, 274 freshwater myxosporean parasite (Ceratomyxa shasta), 317 Fucus evanescens, 19 “functionally extinct” species, 225–28 functions. See ecosystem functioning

responses to extinction by, 240; speciesspecific basis of interaction with environment by, 88–89; values placed on specific species by, 206–7; as world’s greatest marine predators, 297 Hydrodamalis gigas (Steller’s sea cow), 225 hydrodynamic forces/thermal stress studies: New England shores, 19, 20fig, 21–23; Oregon Coast, 21–23; other examples of, 23

Gaines, Steve, 43 Gambusia affinis (mosquitofish), 168 goldenrod (Solidago altissima), 288 goldenrod system, 288, 290 Graham, M., 66 Grantham, B., 43 green crab (Carcinus maenas), 19 Greene, H., 291 green sea urchin (Strongylocentrotus droebachiensis), 228 guppy life history evolution, 299–300

i’iwi bills, 302 indigenous thistles: harboring insects attacking exotic thistles, 8–10; insect herbivores limiting densities of, 7–8 insect herbivores: adapted to Cirsium spp. (thistles), 7; difficulties of predicting long-term status of, 286–89; ecosystem resistance provided by indigenous, 13–14; expendability study on Dichomeris Leuconotella, 283–90; field studies on densities of exotic/native thistles and, 7–14, 10fig; limited densities of Cirsium spp. (thistles), 7– 10fig; “taxonomic impediment” problem and, 345; traits contributing to functional insignificance in, 282–83. See also plant viruses interaction strength: comparing absolute difference index/species importance/ Euclidean index and, 56fig; contextdependency scatter plots for, 62fig; interaction in context of species importance and, 60–63; multiple contexts scatter plot of, 61fig; quantification of, 50–51 introduced species: legislation regarding, 184fig, 186–87; proposals for developing federal legislation on, 199–204 Introduced Species Act (proposed): ecology vs. economics of, 202–3; management/management plans of, 201–2; objective criteria of, 200; species listed under, 199–200; species vs. ecosystems under, 203–4; speed and timing of, 201; taxonomic equality of, 200–201 invasions: biology and management of extinctions vs., 181–82fig; common features of extinctions and, 180; current theory on, 162; ecological/economic scale of extinctions vs., 182–83; evolution of ecology and extinction and,

Halpin, P., 43 Hard, J., 329 Harley, C.D.G., 2, 106 Harrison, S. P., 263 Hawaiian honeycreeper (Vestiaria coccinea), 302 Helfman, G. S., 167 herbivores. See insect herbivores hermit crab (Pagurus longicarpus), 294–95 Herrera, C. M., 276, 277 heterotroph, 117 high-impact species: defining, 164; FishBase database on invasions by, 168–69, 172; FishBase database on life history of fish invasions by, 170t–71t; impacts of established freshwater, 169fig Hixon, M., 43 Hofmann, G., 43 Holsinger, L., 329 Hooper, D. U., 245 horizontal transmission, 332 howler monkeys (Alouatta seniculus), 249 Hudgens, B., 280 Hufbauer, R., 291 Hughes, J. B., 143 humans: dependence on pollinator service by, 260–62; ethical/moral responsibilities to environment by, 240–43;

index

302–3; FishBase database on, 162–64; Pacific Northwest Tideflats case study on, 187–99; revelations about natural ecosystems through study of, 255–57. See also non-native species invasion studies: conclusions/findings of, 177–78; on cumulative number of established freshwater fishes/those with ecological effects, 176fig; data analysis in, 165–68; factors predicting invasion impact, 172–77; on freshwater fish endemisim/effects, 175fig; on highimpact species, 164, 168–69, 172; importance of species examined through, 161–62; methods used in, 163–68; on phylogenetic relationships, 167–68; results/discussion of, 168–77 inverse load sharing, 129–30 ISCO (Partnership for Interndisciplinary Studies of Coastal Oceans), 43 Italian thistle (Carduus pycnocephalus L.), 13 IUCN (International Union for the Conservation of Nature), 310 Ives, A. R., 143, 144 ivory-billed woodpecker (Campephilus principalis), 224

421

Kareiva, P., 42, 66, 84, 139, 178, 205, 220, 233, 291, 329 keystone species: definitions of, 17–18, 221–22, 348; disproportionately exert effects of, 158; forest fragmentation and identification of, 248–49; mutualists among, 253–55; relative scarcity of some, 224; single species in community as equivalent, 109–10; single-species extinctions and, 221–23; substitutability/redundancy criteria of, 222– 23; understanding population regulation to identify, 247–48. See also community; community perspective; Pisaster ochraceus (sea star) keystone species concept: criticisms of, 18; ecosystem-function vs. community perspective and, 112

Lake Washington phytoplankton community study: average annual contributions to community biomass, 215fig; discussion/findings on functioning of, 217–20; history of Lake Washington and, 210–11; relationship between numerical and temporal rarity in, 214fig; similarities in community composition between 1960–2000, 216fig-17; temporal rarity in, 211–17 legislation: conservation consequences of, 187; endangered species, 183–85, 187; governing Pacific Northwest Tideflats invading species, 197–99; introduced species, 186–87; Lacey Act of 1900, 183, 186, 198; proposals for an Introduced Species Act, 199–204 Lehman, C. L., 143 Leigh, E. G., Jr., 235 Leopold, A., 243 Leptinotarsa decemlineata (Colorado potato beetle), 288–89 Leslie, H., 329 Levin, S. A., 42, 220, 291 Levine, J. M., 162 Likens, G. E., 98 Littorina littorea, 294, 295 Littorina obtusata (snail), 295–96 lizard (Anolis sagrei) removal experiment: effect magnitudes of, 72t-73; experimental design for, 71fig; implications of, 82–84; numbers/densities during introduction experiment, 77fig; previous removal experiences with lizards and, 76–77; results of, 73; results as representative of relative effect ratios in system, 75–79; significance/interpretation/implications of, 73–75fig; spider extinction percentages during, 78fig; weaker effects of spiders over, 79–82 load-sharing models, 127–32 local/near-extinctions, 228–31 Loeffler, C., 291 Loreau, M., 14 Lotka-Volterra competition models, 143 Louda, S. M., 2, 15 Lubchenco, J., 43 Luck, R., 15

lacebugs (Tingidae), 7 Lacey Act of 1900, 183, 186, 198

Mack, R. N., 13 magpie behavior, 303

Jadera haematoloma (rhopalid bug), 289

422

Makuna, the (Amazonian Columbia), 241 Markov model of community replacement dynamics, 145 Marks, P., 291 Marvier, M., 106 MARXAN v2.1, 318–19, 325t, 326t, 327t Mastocarpus stellatus, 19 May, R., 144 M. californianus, 24 McCann, K. S., 139, 144 McElhany, P., 236, 311 Menge, B., 2, 15, 106 Meteperia datona. See spider (Meteperia datona) removal experiment Meyer, G., 291 Micropterus salmoides (largemouth bass), 168 Migratory Bird Conservation Act of 1929, 185 Migratory Bird Treaty Act of 1918, 185 milk thistle (Silybum marianum Gaertn.), 13 Morris, W., 159, 235–36 moth (Cactoblastis cactorum), 231 moths (Pyralidae, Pterophoridae), 7 M. trossulus, 24, 33 multiple regression of invasion impact model, 173t–77 Murdoch, W., 15 mussel (Mytilus trossulus), 46, 48, 49 mussels: further observations in New Zealand on, 40–41; hydrodynamic forces/thermal stress on, 19, 20fig, 21– 23; potential for compensation studies and, 31–33; productivity/prey recruitment effects studies on, 24–27, 29–30; studies on interactions in high intertidal communities, 46–60 mussels (Mytilus californianus), 22, 247 mutualism interactions: between keystone species, 253–55; forest fragmentation/tropical forest diversity and productivity and, 255; identifying most essential pollinators in, 260–80 Mytilus edulis, 19 Naeem, S., 91, 105, 106, 142 National Aquatic Nuisance Prevention and Control Act of 1990 (1996), 186 National Aquatic Nuisance Species Task Force, 186

index

National Invasive Species Act, 180 National Invasive Species Management Plan, 180, 187 natural history, 257–58 Navarrete, S., 43 near-extinctions, 228–31 Nelson, J. S., 167 Nepal’s Royal Bardia National Park, 257 nest parasites, 303 New England shores environmental conditions, 19, 20fig, 21 New Zealand: prey productivity/recruitment effects in, 27, 29–30; species impact observations in, 40–41 niche complementarity/compensatory competition, 141, 142 Nielsen, K., 43 Nile tilapia (Oreochromis niloticus), 168, 169, 172 NMFS (National Marine Fisheries Service), 306, 308 non-native species: insect herbivores/ ecosystem resistance to plant, 7–14, 10fig, 13–14; revelations about natural ecosystems through invasions by, 255– 57. See also invasions North American beaver (Castor canadensis), 227 Nuttallia obscurata (Asian varnish clam), 198 oceanographic context, and productivity/recruitment studies: described, 23–24; in New Zealand, 27, 29–30; in Oregon, 24–27; in South Africa, 30–31 Oecologia (1995–1997), 60 Oedoparena glauca, 47 Olson, A., 43 Olson, M. “Cap’n,” 66 Oncorhynchus mykiss (rainbow trout), 168 Oncorhynchus tshawytscha (chinook salmon), 309fig, 311, 317, 320t Oncorynchus spp. (Pacific salmon), 306–8, 309fig Operculina ventricosa, 229 Ophiostoma ulmi/O. novo-ulmi (Dutch elm disease), 330 Opuntia stricta (prickly pear), 231 Oregon: hydrodynamic forces/thermal stress studies in, 21–23; productivity/ prey recruitment effects in, 24–27

index

Oreochromis mossambicus (tilapia), 168 Oreochromis niloticus (Nile tilapia), 168, 169, 172 Oryctilagus cuniculus (rabbits), 230 Oscillatoria aghardii, 214 Pacific Northwest Tideflats invasion case study: background information on, 187–89; impacts of the invaders in, 190–95; interaction webs during, 195– 97; legislation governing invading species in, 197–99; major estuarine systems in area, 188fig; mudflat areas in, 189fig; natural history of invaders in, 189–90 Pacific salmon (Oncorynchus spp.), 306–8, 309fig Padilla Bay National Estuarine Research Reserve, 46–47 Pagurus longicarpus (hermit crab), 294–95 Paine, R. T., 1, 2, 12, 15, 16, 18, 24, 32, 43, 46, 66, 86, 97, 109, 139, 159, 160, 204, 205, 220, 237, 246, 247, 263, 291, 347–51 Palumbi, S. R., 236 Pandalus borealis (cold-water rock shrimp), 300–301 Panzer, R., 345 Papaipema mitela Guen. (Noctuidae), 9 Paracantha culta Wiedeman (Tephritidae), 9 Pasoh Forest (Malaysia), 255 passenger pigeon (Ectopistes migratorius), 224 Pauly, D., 178 “performance enhancing” effects, 120 pest pressure hypothesis, 250–53 Petromyzon marinus (sea lamprey), 228 phylogenetic relationships, 167–68 physical context of species impact: hydrodynamic forces/thermal stress environmental conditions, 19–23; oceanographic context of productivity/ recruitment, 23–31 picture-winged flies (Tephritidae), 7 Pietsch, T., 178 pink salmon: evolution of size in, 296– 99; size of 2-year-old (1951–1974), 298fig. See also salmonids Pisaster ochraceus (sea star): differing importance under different conditions of, 98; hydrodynamic forces/thermal

423

stress studies on, 21–23; as keystone species, 21; population regulation role by, 247; productivity/recruitment studies and, 22–33; significance of Paine’s study on, 86; species interactions in context of interaction strength/species importance studies on, 60–63; species interactions/exclusion studies on, 46–60 Plant Protection Act of 2000, 186 Plant Quarantine Act of 1912, 186 plants: Cirsium spp. (thistles), 5–15; consequences of viruses for, 331 plant viruses: percentage with specialized/narrow/broad host ranges, 343fig; predicted proportion falling into host/vector ranges, 333fig; proportion falling into host/vector ranges, 336fig; relationship between host/vector species and, 338fig; relationship between number of hosts/vectors for, 337fig; relationship between number of host/ vector species for, 340fig; seedtransmitted, 341–43fig. See also insect herbivores plant viruses specificity: horizontal/vertical transmission and, 332; plausible scenarios of, 332, 334; VIDE database on, 334–35 plant viruses specificity studies: on implications of specificity for species losses/community structure/conservation, 344–46; questions addressed by, 334; on seed transmission and host specificity, 341–43fig; sources of data for, 334–35; on vector vs. host specificity, 335–41 Platyptilia carduidactyla Riley (Pterophoridae), 9 pollinators: alarm over “forgotten,” 274; dependence of humans/other species on services by, 260–62; tool for identifying most essential, 263, 266–67 pollinator studies: assessing most essential of reliable visitors, 267–69; buffering of pollination against loss of different types of visitors, 274–77; caveats to findings of, 278–79; conservation implications of, 279–80; description of data in, 262–63; graphical tool for identifying most essential

424

pollinator studies (cont.) pollinators, 263, 266–67; list of empirical, 264t–65t; results of, 269–74. See also total pollination service pollock (Theragra chalcogramma), 228 pond biodiversity-ecosystem function study, 100–103, 101t population density calculations, 258–59 population number/persistence: existing conservation frameworks for, 308–10; regional conservation planning for viable salmonid, 317–26fig; in salmonids, 316–17; theory and applications to salmon, 310–15fig population regulation: forest fragmentation as tool for forest, 248–49; to identify keystone species, 247–48; role of competition in, 247–48 Populus tremuloides (quaking aspen), 227 Possingham, Hugh, 329 potential for compensation studies, 31– 33 Power, M. E., 237 predators: effects of species characteristics studies on, 38; effects of species composition studies on, 33–38; humans as world’s greatest marine, 297; population regulation of, 247–248; potential for compensation studies on, 31–33; species interactions in context of interaction strength/species importance, 60–63; species interactions/exclusion studies on Pisaster ochraceus, 46–60; studies on productivity/recruitment effects of, 24–27 Presidential Executive Order (11987, 5/1977), 186 Presidential Executive Order (13112, 2/1999), 180, 186 prickly pear (Opuntia stricta), 231 productivity/recruitment studies, 24–31, 28fig Pseudorasbora parva (stone moroko), 168 Psychotria horizontalis, 252 Quaker society, 242 quaking aspen (Populus tremuloides), 227 Quinn, J. F., 263 rabbits (Oryctilagus cuniculus), 230 Rabinowtiz, D., 207

index

Rand, T. A., 2 Rapana venosa (Asian veined whelk), 198 Raphus cucullatus (dodo), 225 rare species: described, 207–9; numerical/temporal components of, 208–9fig; as predisposed to extinction, 223–24; temporal rarity in phytoplankton, 211– 17 Rastetter, E. B., 123, 125 reduced species, 225–28 redundancy concept, 137–38 reliability block diagram, 122fig, 138 removal experiments. See speciesdeletion experiments reserve sitting algorithms (for salmonid ESUs), 317–26fig Rhagoletis pomonella (apple maggot), 288 Rhinocyllus conicus Fr¨ol, 7 Rhinocyllus conicus (weevil), 302 rhopalid bug ( Jadera haematoloma), 289 Ridley, H. N., 251 Riggs, S. R., 66, 205 Rodman, J., 15 Root, R. B., 12, 218, 236 Roughgarden, J., 143 Ruckelshaus, M., 236, 237 Ruesink, J. L., 66, 106 Saccharum spontaneum (southeast Asian grass), 256 salmonids: application of population number/persistence theory to, 310– 15fig; conservation status of ESUs (Evolutionarily Significant Units), 307t; endangered status of West Coast salmon, 349; Pacific salmon (Oncorynchus spp.), 306–8, 309fig; population characteristics/persistence in, 316– 17; regional conservation planning for viable, 317–26fig; size evolution of, 296–99, 298fig. See also pink salmon Sander lucioperca (zander), 168 Sanders, N., 233 Sanford, E., 43 Scheuerell, M. D., 66 Schindler, D. E., 66, 98, 107 Schluter’s variance test, 134–35t Schoener, T. W., 3 Schwartz, M. W., 345 sea grape (Coccoloba uvifera), 70

index

sea lamprey (Petromyzon marinus), 228 sea otter (Enhydra lutris), 227–28 sea star. See Pisaster ochraceus (sea star) sea star Stichaster australis, 29 sea stars (Asterias forbesi, A. vulgaris), 19 sea urchins (Strongylocentrotus spp.), 247–48 seed-transmitted viruses, 341–43fig. See also plant viruses specificity Seeley, R. H., 296 Semibalanus balanoides, 19 sexual evolution of Pandalus borealis, 300–301 shore crab (Cancer borealis), 296 Simberloff, D., 107, 178 Simenstad, S., 205 simple sampling effect, 141, 142 single-species extinctions: findings on, 231–33; impact of global, 223–25; impact of great reductions or “functional,” 225–28; impact of local or near-extinctions, 228–31; keystone species concept and, 221–23 Slatyer, R. O., 97 snail (Littorina obtusata), 295–96 Solidago altissima (goldenrod), 288 Somero, G., 43 Spartina alterniflora (Atlantic cordgrass): background information on, 187, 189; ecosystem impacts of, 190–91, 192t, 194–95; interaction webs and, 195–97; legislation governing, 197–99; natural history of, 189–90 species: aggregate measures of importance of, 54b–55b; calculating population densities of, 258–59; continuous evolution of new, 292–94; difficulties in predicting long-term status of, 286–89; ecosystem-function perspective of, 110–15; experimental removals of, 1–4; high-impact, 164, 168–69, 170t-71t, 172; historic average life span of, 292–93; human value of specific, 206–7; identification of ecologically important/nonexpendable, 65; interactions in high intertidal communities, 46–60; recovering those of concern to conservation, 305–29; stability-diversity modeling on loss of, 140–59; traits contributing to functional insignificance by insect, 282–83

425

species-deletion experiments: considering environmental context of, 16–17; on insect herbivores and density of native/ exotic thistles, 7–14, 10fig; revealing impact of biodiversity depletion, 1–4; simultaneous removal of lizard and spider, 69–84 species impact: biotic context of, 31–38; concepts of, 17–18; discussion of implications of, 38–40; future studies/research on, 41–42; invasions and highimpact, 164, 168–69, 172; New Zealand observations on, 40–41; physical context of, 19–31 species importance: ability to predict removals and, 154t; aggregate measures of, 54b–55b; comparing absolute difference index/interaction strength/ Euclidean index and, 56fig; contextdependency scatter plots for, 62fig; differing under different conditions, 98; evolutionary perspective on, 292–304; implications of removal experiments for, 82–84; interaction in context of interaction strength and, 60–63; invasion of species and, 161–62; Paine’s contributions to assessment of, 347–51 species interactions: aggregate measures of species importance and, 54b-55b; between keystone mutualists, 253–55; community stability and introduction of weakly, 142–43; comparing absolute difference index/interaction strength/ Euclidean index/species importance, 56fig; in context of interaction strength/species importance, 60–63; current theory on invasion and, 162; identifying most essential pollinators in mutualism, 260–80; negative and positive, 47fig; Pisaster ochraceus exclusion studies and, 46–60; predicting effects of extinction and understanding, 250–53; quantification of interaction strength, 50–51; studies on high intertidal communities, 46; utility of expendability concept and, 63–65; variation in space, 56–60; variation in time, 46– 56 species ranking: of chinook salmon populations for protection, 324–26fig; expendability issue and, 110

426

species traits: contributing to functional insignificance, 282–83; of Dichomeris Leuconotella, 285fig; environment and interaction/invasion and, 16; predicting fish invasion impact by, 172–77 spider (Eustala cazieri), 70 spider (Meteperia datona) removal experiment: effect magnitudes during, 72t–73; experimental design for, 71fig; extinction percentages during, 78fig; implications of, 82–84; numbers/densities during introduction experiment, 77fig; results of, 73; results as representative of relative effect ratios in system, 75– 79; significance/interpretation/implications of results for, 73–75fig; weaker effects of, 79–82 Spiller, D. A., 3 stability-diversity model: basic structure of, 145–47; discussion of, 157–59; results of, 149–56; stochastic simulations, 147–49 stability-diversity patterns: ability to predict importance of species removals, 154t; distributions of species abundance for two community rules, 150fig; four categories of, 141–42; histograms of percentage change in instability following removals, 152fig; importance of predicting community, 140–41; mean species abundance and, 156t; past modeling/empirical work on, 142–45; predicting importance of rare species for, 155t; recent studies on, 141–45; relationship between mean abundance of species/survival rate/space vacated by other species, 157fig; relationship between species removals/community and, 156fig; relationship between stability/species richness, 151fig; relationships between community stability/ dominance of removed species, 153fig. See also expendability Steller’s sea cow (Hydrodamalis gigas), 225 Sterner, R. W., 98 Stinchcombe, J., 280 Strawberry Hill experiments: on hydrodynamic forces/thermal stress, 22–23; on productivity/prey recruitment effects, 24–27, 28fig

index

Strongylocentrotus droebachiensis (green sea urchin), 228 Strongylocentrotus spp. (sea urchins), 247– 48 Strub, T., 43 sucking bugs (Hemiptera), 7 Sustainable Biosphere Initiative, 348–49 Sutherland, J. P., 19 tall thistle (Cirsium altissimum), 8–10t, 11fig-13 Tansley, A. G., 88 targeted species removal approach: compared to biodiversity manipulation studies, 87–89; comparing biodiversity study goals to, 90–93; described, 86; hybrid approaches using alternatives to, 100–104; key questions answered through, 93–98; strengths/shortcomings of, 99–100 “taxonomic impediment” problem, 345 TBT (tri-butyl tin), 31–32 Tebo, M., 233 temporal rarity: described, 208–9fig; in Lake Washington phytoplankton, 211– 17, 212fig Terborgh, J., 249 Theragra chalcogramma (pollock), 228 thistles. See Cirsium spp. (thistles) Thomson, J., 280 Tilman, D., 143 Tjosssem, S. F., 139 total pollination service: calculating, 263, 266–67; caveats to findings on, 278–79; decline as taxa are deleted from visitor pool, 270fig–72fig; effects of deleting taxa in order of increasing mean visitation frequency on, 275fig; rank in visitation frequency and, 276fig; testing for buffering and, 267, 273t, 274t; three basic patterns for effect of visitor deletions on, 266fig. See also pollinator studies tropical forest ecosystems, 255 Ulmus americana (American elm), 226–27 Uriarte, M., 291 U.S. Department of Agriculture, thistlefeeding insects released by, 7 U.S. Endangered Species Act (ESA) [1973], 180, 183, 185–86, 187, 306–8

index

USGS (United States Geological Survey), 198 Vanni, M. J., 98 Venerupis philippinarum (Asian littleneck clam), 198 vertical transmission, 332 Vestiaria coccinea (Hawaiian honeycreeper), 302 VIDE database, 334–35 Vitousek, P. M., 245 Volterra, V., 258 Waples, R., 329 water hyacinth (Eichhornia crassipes), 230 weevil (Rhinocyllus conicus), 302 weevils (Curculionidae), 7 weevils (Larinus planus), 14 weevils (Rhinocyllus conicus Fr¨ol), 7, 14 Welcomme, R. L., 163 West Coast salmon, 349 Wheeler, P., 43 whelk (Concholepas concholepas), 32

427

whelk (Nucella lapillus), 19, 21, 33 whelks: effects of species composition studies and, 33–38; hydrodynamic forces/thermal stress studies and, 19– 23; impact of TBT on, 31–32; potential for compensation studies and, 31–33 Wilbur, H. M., 94, 95 Wonham, M. J., 66, 106, 233 Wootton, J. T., 3 Wright, S. J., 245 Yachi, S., 143 Zaret, T. M., 204 zebra mussel (Dreissena polymorpha), 183 Zedler, P., 15 Zinn, J. L., 317 zonation, 19 Zostera japonica (Asian eelgrass): background information on, 187, 189; ecosystem impacts of, 190–91, 193t–95; interaction webs and, 195–97; legislation governing, 197–99; natural history of, 190