Aquatic Food Webs This page intentionally left blank Aquatic Food Webs An Ecosystem Approach
EDITED BY Andrea Belgrano National Center for Genome Resources (NCGR), Santa Fe, NM, USA Ursula M. Scharler University of Maryland Center for Environmental Science, Chesapeake Biological Laboratory (CBL), Solomons, MD, USA and Smithsonian Environmental Research Center, Edgewater, MD, USA Jennifer Dunne Pacific Ecoinformatics and Computational Ecology Lab, Berkeley, CA USA; Santa Fe Institute (SFI) Santa FE, NM, USA; Rocky Mountain Biological Laboratory, Crested Butte, CO USA
AND Robert E. Ulanowicz University of Maryland Center for Environmental Science, Chesapeake Biological Laboratory (CBL), Solomons, MD, USA
1 1 Great Clarendon Street, Oxford OX2 6DP Oxford University Press is a department of the University of Oxford. It furthers the University’s objective of excellence in research, scholarship, and education by publishing worldwide in Oxford New York Auckland Cape Town Dare es Salaam Hong Kong Karachi Kuala Lumpur Madrid Melbourne Mexico City Nairobi New Delhi Shanghai Taipei Toronto With offices in Argentina Austria Brazil Chile Czech Republic France Greece Guatemala Hungary Italy Japan South Korea Poland Portugal Singapore Switzerland Thailand Turkey Ukraine Vietnam Oxford is a registered trade mark of Oxford University Press in the UK and in certain other countries Published in the United States by Oxford University Press Inc., New York # Oxford University Press, 2005 The moral rights of the authors have been asserted Database right Oxford University Press (maker) First published 2005 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, or under terms agreed with the appropriate reprographics rights organization. Enquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above You must not circulate this book in any other binding or cover and you must impose this same condition on any acquirer British Library Cataloging in Publication Data (Data available) Library of Congress Cataloging-in-Publication Data Aquatic food webs : an ecosystem approach / edited by Andrea Belgrano ... [et al.]. p. cm. Includes bibliographical references and index. ISBN 0-19-856482-1 (alk. paper) – ISBN 0-19-856483-X (alk. paper) 1. Aquatic ecology. 2. Food chains (Ecology) I. Belgrano, Andrea. QH541.5.W3A68225 2004 577.6016–dc22 2004027135 ISBN 0 19 856482 1 (Hbk) 9780198564829 ISBN 0 19 856483 X (Pbk) 9780198564836 10987654321 Typeset by Newgen Imaging Systems (P) Ltd., Chennai, India Printed in Great Britain on acid-free paper by Antony Rowe, Chippenham FOREWORD CURRENT AND FUTURE PERSPECTIVES ON FOOD WEBS
Michel Loreau
Food webs have been approached from two basic on food webs that is complementary to the energetic perspectives in ecology. First is the energetic view approach (DeAngelis 1992). Material cycles are articulated by Lindeman (1942), and developed by among the most common of the positive feedback ecosystem ecology during the following decades. loops discussed by Ulanowicz in his concluding In this view, food webs are networks of pathways remarks, and may explain key properties of eco- for the flow of energy in ecosystems, from its systems (Loreau 1998). The stoichiometry of ecolo- capture by autotrophs in the process of photo- gical interactions may further strongly constrain synthesis to its ultimate dissipation by hetero- food-web structure (Sterner and Elser 2002; Elser trophic respiration. I would venture to say that the and Hessen’s chapter). There has also been con- ecological network analysis advocated by siderable interest in the relationship between bio- Ulanowicz and colleagues in this book is heir to diversity and ecosystem functioning during the last this tradition. A different approach, rooted in decade (Loreau et al. 2002). Merging the theories that community ecology, was initiated by May (1973) bear upon food webs and the maintenance of species and pursued by Pimm (1982) and others. This diversity is urgently needed today, and may provide approach focuses on the dynamical constraints new insights into food-webs structure and ecosys- that arise from species interactions, and empha- tem functioning (Hillebrand and Shurin’s chapter). sises the fact that too much interaction (whether in The structure and functioning of ecological sys- the form of a larger number of species, a greater tems is determined not only by local constraints connectance among these species, or a higher and interactions, but also by larger-scale processes. mean interaction strength) destabilises food webs The importance of regional and historical influ- and ecological systems. The predictions resulting ences has been increasingly recognised in com- from this theory regarding the diversity and con- munity ecology (Ricklefs and Schluter 1993). The nectance of ecological systems led to a wave of extent to which they shape food webs, however, comparative topological studies on the structure of has been relatively little explored. The recent food webs. Thus, the two traditions converge in development of metacommunity theory (Leibold the search for patterns in food-web structure et al. 2004) provides a framework to start exam- despite different starting points. This book results ining spatial constraints on the structure and from the confluence of these two perspectives, functioning of local food webs (Melian et al.‘s which are discussed in a number of chapters. chapter). At even larger time scales, food webs are Patterns, however, are generally insufficient to the result of evolutionary processes which deter- infer processes. Thus, the search for explanations of mine their current properties. Complex food webs these patterns in terms of processes is still very much may readily evolve based on simple ecological alive, and in this search the energetic and dynamical interactions (McKane 2004). The evolution of food- perspectives are not the only possible ones. Bio- web and ecosystem properties is a fascinating geochemical cycles provide a functional perspective topic for future research.
v vi FOREWORD
This book provides a good synthesis of recent Loreau, M. 1998. Ecosystem development explained by research into aquatic food webs. I hope this competition within and between material cycles. Pro- synthesis will stimulate the development of new ceedings of the Royal Society of London, Series B 265: 33–38. approaches that link communities and ecosystems. Loreau, M., S. Naeem, and P. Inchausti. Eds. 2002. Bio- diversity and ecosystem functioning: synthesis and per- spectives. Oxford University Press, Oxford. References May, R. M. 1973. Stability and complexity in model ecosys- tems. Princeton University Press, Princeton. DeAngelis, D. L. 1992. Dynamics of nutrient cycling and McKane, A. J. 2004. Evolving complex food webs. The food webs. Chapman & Hall, London. European Physical Journal B 38: 287–295. Leibold, M. A., M. Holyoak, N. Mouquet, P. Amarasekare, Pimm, S. L. 1982. Food webs. Chapman & Hall, London. J. M. Chase, M. F. Hoopes, R. D. Holt, J. B. Shurin, R. Law, Ricklefs, R. E., and D. Schluter. Eds 1993. Species diversity D. Tilman, M. Loreau, and A. Gonzalez. 2004. The in ecological communities: historical and geographical per- metacommunity concept: a framework for multi-scale spectives. University of Chicago Press, Chicago. community ecology. Ecology Letters 7: 601–613. Sterner, R. W., and J. J. Elser. 2002. Ecological stoichiometry: Lindeman, R. L. 1942. The trophic-dynamic aspect of the biology of elements from molecules to the biosphere. ecology. Ecology 23: 399–418. Princeton University Press, Princeton. Contents
Foreword v Michel Loreau
Contributors ix
Introduction 1 Andrea Belgrano
PART I Structure and function 5
1 Biosimplicity via stoichiometry: the evolution of food-web structure and processes 7 James J. Elser and Dag O. Hessen
2 Spatial structure and dynamics in a marine food web 19 Carlos J. Melia´n, Jordi Bascompte, and Pedro Jordano
3 Role of network analysis in comparative ecosystem ecology of estuaries 25 Robert R. Christian, Daniel Baird, Joseph Luczkovich, Jeffrey C. Johnson, Ursula M. Scharler, and Robert E. Ulanowicz
4 Food webs in lakes—seasonal dynamics and the impact of climate variability 41 Dietmar Straile
5 Pattern and process in food webs: evidence from running waters 51 Guy Woodward, Ross Thompson, Colin R. Townsend, and Alan G. Hildrew
PART II Examining food-web theories 67
6 Some random thoughts on the statistical analysis of food-web data 69 Andrew R. Solow
7 Analysis of size and complexity of randomly constructed food webs by information theoretic metrics 73 James T. Morris, Robert R. Christian, and Robert E. Ulanowicz
8 Size-based analyses of aquatic food webs 86 Simon Jennings
vii viii CONTENTS
9 Food-web theory in marine ecosystems 98 Jason S. Link, William T. Stockhausen, and Elizabeth T. Methratta
PART III Stability and diversity in food webs 115
10 Modeling food-web dynamics: complexity–stability implications 117 Jennifer A. Dunne, Ulrich Brose, Richard J. Williams, and Neo D. Martinez
11 Is biodiversity maintained by food-web complexity?—the adaptive food-web hypothesis 130 Michio Kondoh
12 Climate forcing, food web structure, and community dynamics in pelagic marine ecosystems 143 L. Ciannelli, D. Ø. Hjermann, P. Lehodey, G. Ottersen, J. T. Duffy-Anderson, and N. C. Stenseth
13 Food-web theory provides guidelines for marine conservation 170 Enric Sala and George Sugihara
14 Biodiversity and aquatic food webs 184 Helmut Hillebrand and Jonathan B. Shurin
PART IV Concluding remarks 199
15 Ecological network analysis: an escape from the machine 201 Robert E. Ulanowicz
Afterword 208 Mathew A. Leibold
References 211
Index 255 Contributors
Daniel Baird, Zoology Department, University of Port Simon Jennings, Centre for Environment, Fisheries Elizabeth, Port Elizabeth, South Africa. and Aquaculture Science, Lowestoft Laboratory NR33 Jordi Bascompte, Integrative Ecology Group, Estacio´n 0HT, UK. Email: [email protected] Biolo´gica de Don˜ana, CSIC, Apdo. 1056, E-41080, Jeffrey C. Johnson, Institute of Coastal and Marine Sevilla, Spain. Email: [email protected] Resources, East Carolina University, Greenville, NC Andrea Belgrano, National Center for Genome Resources 27858, USA. (NCGR), 2935 Rodeo Park Drive East, Santa Fe, NM Pedro Jordano, Integrative Ecology Group, Estacio´n 87505, USA. Email: [email protected] Biolo´gica de Don˜ana, CSIC, Apdo. 1056, E-41080, Ulrich Brose, Technical University of Darmstadt, Sevilla, Spain. Department of Biology, Schnittspahnstr. 3, 64287 Michio Kondoh, Center for Limnology, Netherlands Darmstadt, Germany. Institute of Ecology, Rijksstraatweg 6, Nieuwersluis, Robert R. Christian, Biology Department, East Carolina P.O. Box 1299, 3600 BG Maarssen, The Netherlands. University, Greenville, NC 27858, USA. Email: Email: [email protected] [email protected] P. Lehodey, Oceanic Fisheries Programme, Secretariat Lorenzo Ciannelli, Centre for Ecological and of the Pacific Community, BP D5, 98848 Noumea Evolutionary Synthesis (CEES), Department of Biology cedex, New Caledonia. University of Oslo, Post Office Box 1066, Blindern, Mathew Leibold, Section of Integrative Biology, The N-0316 Oslo, Norway. Email: lorenzo.ciannelli@ University of Texas at Austin, 1 University Station, bio.uio.no. C0930 Austin, TX 78712, USA. Email: mleibold@ Janet T. Duffy-Anderson, Alaska Fisheries Science mail.utexas.edu Center, NOAA, 7600 Sand Point Way NE, 98115 Jason S. Link, National Marine Fisheries Service, Seattle, WA, USA. Northeast Fisheries Science Center, 166 Water St., Jennifer A. Dunne, Pacific Ecoinformatics and Computa- Woods Hole, MA 02543, USA. Email: jlink@ tional Ecology Lab, P.O. Box 10106, Berkeley, CA 94709 whsunl.wh.whoi.edu USA; Santa Fe Institute, 1399 Hyde Park Road, Santa Fe, Michel Loreau, Laboratoire d’Ecologie, UMR 7625 NM 87501 USA; Rocky Mountain Biological Laboratory, Ecole Normale Superieure 46, rue d’ Ulm F-75230, P.O. Box 519, Crested Butte, CO 81224 USA Email: Paris Cedex 05, France. Email: [email protected] [email protected]. Joseph Luczkovich, Biology Department, East Carolina James J. Elser, School of Life Sciences, Arizona State University, Greenville, NC 27858, USA. University,Tempe,AZ85287,USA.Email:j.elser@ Neo D. Martinez, Pacific Ecoinformatics and asu.edu Computational Ecology Lab, P.O. Box 10106, Dag O. Hessen, Department of Biology, University of Berkeley, CA 94709 Rocky Mountain Biological Oslo, P.O. Box 1050, Blindern, N-0316 Oslo, Norway. Laboratory, P.O. Box 519, Crested Butte, CO 81224 Alan G. Hildrew, School of Biological Sciences, Queen USA. Mary, University of London, Mile End Road, London, Carlos J. Melia´n, Integrative Ecology Group, Estacio´n E1 4NS, UK. Email: [email protected] Biolo´gica de Don˜ana, CSIC, Apdo. 1056, E-41080, Helmut Hillebrand, Institute for Botany, University of Sevilla, Spain. Cologne, Gyrhofstrasse 15 D-50931 Ko¨ln, Germany. Elizabeth T. Methratta, National Marine Fisheries Email: [email protected] Service, Northeast Fisheries Science Center, D.Ø. Hjermann, Centre for Ecological and Evolutionary 166 Water St., Woods Hole, MA 02543, USA. Synthesis (CEES), Department of Biology University of James T. Morris, Department of Biological Sciences, Oslo, Post Office Box 1066 Blindern, N-0316 Oslo, University of South Carolina, Columbia, SC 29208, USA. Norway. Email: [email protected]
ix x CONTRIBUTORS
Geir Ottersen, Institute of Marine Research, P.O. Box William T. Stockhausen, National Marine Fisheries 1870 Nordnes, 5817 Bergen, NORWAY Current Service, Northeast Fisheries Science Center, Address: Centre for Ecological and Evolutionary 166 Water St., Woods Hole, MA 02543, USA. Synthesis, Department of Biology, P.O Box 1050 Andrew R. Solow, Woods Hole Oceanographic Blindern, N-0316 Oslo, NORWAY. Institution, Woods Hole, MA 02543, USA. Email: Enric Sala, Center for Marine Biodiversity and [email protected] Conservation, Scripps Institution of Oceanography, Geroge Sugihara, Center for Marine Biodiversity and La Jolla, CA 92093-0202, USA. Email: Conservation, Scripps Institution of Oceanography, [email protected] La Jolla, CA 92093-0202, USA. Ursula M. Scharler, University of Maryland, Ross Thompson, Biodiversity Research Centre, Center for Environmental Science, Chesapeake University of British Columbia, Vancouver, Canada. Biological Laboratory (CBL), Solomons, MD 20688, Colin R. Townsend, Department of Zoology, USA; Smithsonian Environmental Research University of Otago, New Zealand. Center, Edgewater, MD, USA. Email: scharler@ Robert E. Ulanowicz, University of Maryland, Center cbl.umces.edu for Environmental Science, Chesapeake Biological Jonathan B. Shurin, Department of Zoology, Laboratory (CBL), Solomons, MD 20688, USA. Email: University of British Columbia, 6270 University Blvd. [email protected] Vancouver, BC V6T 1Z4, Canada. Richard J. Williams, Pacific Ecoinformatics and Nils Chr. Stenseth, Centre for Ecological and Computational Ecology Lab, P.O. Box 10106, Berkeley, Evolutionary Synthesis (CEES), Department of Biology CA 94709 USA; Rocky Mountain Biological Laboratory, University of Oslo, Post Office Box 1066 Blindern, N- P.O. Box 519, Crested Butte, CO 81224 USA; San Fran- 0316 Oslo, Norway. cisco State University, Computer Science Department, Dietmar Straile, Dietmar StraileLimnological 1600 Holloway Avenue, San Francisco, CA 94132 USA. Institute, University of Konstanz, 78457 Konstanz, Guy Woodward, Department of Zoology, Ecology Germany. Email: dietmar.straile@ and Plant Science, University College, Cork, uni-konstanz.de Ireland. INTRODUCTION Aquatic food-webs’ ecology: old and new challenges
Andrea Belgrano
Looking up ‘‘aquatic food web’’ on Google provides that linked latitudinal gradients in aquatic species a dizzying array of eclectic sites and information diversity, food-web complexity, and community (and disinformation!) to choose from. However, stability. even within this morass it is clear that aquatic Following that early MacArthur hypothesis, we food-web research has expanded greatly over the find it timely to also ask, How complex are aquatic last couple of decades, and includes a wide array food webs? of studies from both theoretical and empirical The first book on theoretical food-web ecology perspectives. This book attempts to bring together was written by May (1973), followed by Cohen and synthesize some of the most recent perspec- (1978). Since then, Pimm (1982) and Polis and tives on aquatic food-web research, with a parti- Winemiller (1996) have revisited some of the ideas cular emphasis on integrating that knowledge proposed by May and Cohen and discussed them within an ecosystem framework. in different contexts, and trophic flow models have It is interesting to look back at the pioneering been proposed and used widely for aquatic and work of Sir Alister Hardy in the early 1920s at particularly marine ecosystems (e.g. Wulff et al. Lowestoft Fisheries Laboratory. Hardy studied the 1989; Christensen and Pauly 1993). However, feeding relationship of the North Sea herring with recent advances in ecosystem network analysis planktonic assemblages by looking at the species (e.g. Ulanowicz 1996, 1997; Ulanowicz and Abarca- distribution patterns in an attempt to provide Arenas 1997) and the network structure of food better insights for the stock assessment of the webs (e.g. Williams and Martinez 2000; Dunne North Sea fisheries. If we take a look in his food- et al. 2002a,b; Williams et al. 2002) in relation to web scheme (Figure 1), it is interesting to note that ecosystem dynamics, function, and stability clearly he considered species diversity in both phyto- set the path for a new, complementary research plankton and zooplankton, and also specified agenda in food-web analysis. These and many body-size data for the different organisms in the other studies suggest that a new synthesis of food web. Thus, it appears that already almost available information is necessary. This new 100 years ago the concept of constructing and synthesis is giving rise to novel basic research that drawing links among diverse species at multiple generalizes across habitats and scales, for example, trophic levels in a network-like fashion was in the the discovery of universal scaling relations in food- mind of many aquatic researchers. web structure (Garlaschelli et al. 2003), and is also In following decades, researchers began to underpinning new approaches and priorities for consider links between food-web complexity and whole-ecosystem conservation and management, ecological community stability. The classic, and still particularly in marine systems. contentious MacArthur hypothesis that ‘‘Stability Aquatic food-web research is also moving beyond increases as the number of link increase’’ (1955) an exclusive focus on taxa from phytoplankton gave rise to studies such as that by Paine (1966) to fish. A new look at the role that marine microbes
1 2 INTRODUCTION
Figure 1 The food web of herring Clupea harengus Hardy (1924). From Parables of Sea & Sky—The life, work and art of Sir Alister Hardy F. R. S. Courtesy of SAHFOS—The CPR Survey, Plymouth, UK.
Figure 2 The microbial loop: impressionist version. A bacteria-eye view of the ocean’s euphotic layer. Seawater is an organic matter continuum, a gel of tangled polymers with embedded strings, sheets, and bundles of fibrils and particles, including living organisms, as ‘‘hotspots.’’ Bacteria (red) acting on marine snow (black) or algae (green) can control sedimentation and primary productivity; diverse microniches (hotspots) can support high bacterial diversity. (Azam, F. 1998. Microbial control of oceanic carbon flux: the plot thickens. Science 280: 694–696.) (See Plate1) INTRODUCTION 3 play in the global ocean (Azam and Worden 2004) including humans. Aquatic food webs may pro- suggests that oceanic ecosystems can be character- vide a particularly useful empirical framework for ized as a complex dynamic molecular network. developing and testing an information theory of The role of microbial food webs (Figure 2—see ecology that will take into account the complex also, Plate 1—Azam 1998) needs to be considered network of interactions among biotic and abiotic to understand the nonlinearities underlying the components of ecosystems. relationship between the pelagic and benthic domains. Acknowledgments Emerging challenges in aquatic food-web research include integrating genomic, biogeochemical, This work was funded in part or in full by the US environmental, and economic data in a modeling Dept of Energy’s Genomes to Life program (www. effort that will elucidate the mechanisms govern- doegenomestolife.org) under the project ‘‘Carbon ing the ecosystem dynamics across temporal and Sequestration in Synechococcus sp.: From Mole- spatial scales at different levels of organization cular Machines to Hierarchical Modeling’’ (www. and across the whole variety of species diversity, genomes-to-life.org). This page intentionally left blank PART I Structure and function
Many scientists use food webs to portray ecolo- dimensions of trophic interactions. Along those gical communities as complex adaptive systems. lines, several of the chapters extend their scope However, as with other types of apparently com- beyond traditional food-web ‘‘snapshot’’ analyses plex systems, underlying mechanisms regulate to take into account space and time when assessing food-web function and can give rise to observed changes in food-web structure and species structure and dynamics. These mechanisms can composition. sometimes be summarized by relatively simple By comparing food webs from different envir- rules that generate the ecosystem properties that onments and by encompassing organisms from we observe. bacteria to vertebrates, we start to see some com- This section of the book presents and discusses mon, general constraints that act to shape and responses of food webs to trophic interactions, change food-web structure and function. These transfer efficiency, length of food chains, changes include biological stoichiometry, body-size, and in community composition, the relative importance the distribution of interaction strengths. Insights of grazing versus detrital pathways, climate from ecological network analysis also provide new change, and the effects of natural and anthro- tools for thinking about dynamical and energetic pogenic disturbances. In addition, research is properties of food webs, tools which complement beginning to incorporate spatial and temporal a wide array of more long-standing approaches. This page intentionally left blank CHAPTER 1 Biosimplicity via stoichiometry: the evolution of food-web structure and processes
James J. Elser and Dag O. Hessen
Introduction on purine metabolism would reveal yet more details, finally yielding individual molecules In these days of the vaunted genome and the and each individual chemical reaction pathway. rest of the proliferating ‘‘gnomes’’ (transcriptome, The fascinating but intimidating journey just proteome, metabolome, etc.) and the unveiling completed should be familiar to food-web eco- of astonishing complex pleiotropy and protein/ logists, for whom Figure 1.1(a) (Lavigne 1996) has genome interactions, it may seem headstrong to achieved near-iconic status as a symbol, sure to propose that there is something more complex stimulate uneasy laughter in the audience, of the than the genome currently under study in modern daunting complexity confronted by food-web biology. Nevertheless, we propose that the ecologists. If we were to follow the metabolic ‘‘entangled bank’’ of food webs, the trophic example and zoom in on the northwest Atlantic connections among interacting organisms in food web, we would, of course, encounter more ecosystems, is indeed as complex and bewildering and more detail. The node ‘‘cod,’’ for example, as the emerging genome and its products. The might resolve itself into larval, juvenile, and complexity increases further when considering mature cod, each connected, by its feeding, in the myriad pathways of matter and energy that different ways with other parts of the web. Further the species interactions build upon. Consider, for inspection might then reveal the individual cod example, a simplified map of central cellular themselves, each with a distinct genome and a metabolism (Figure 1.1). Here we can see, in basic unique physiological and behavioral repertoire. outline, the key pathways by which energy and How, then, can we deal with this layered com- key resources are metabolized in maintenance and plexity in food webs? And how could any con- growth of the organism. Note the complexity of necting thread of simplicity and unifying principles the diagram both in terms of the numbers of nodes be spotted in this overwhelming complexity? It is and the numbers and types of connections among our view that, just as the individual molecules in different components. As shown by the shading, metabolism are the critical level of resolution for different parts of the overall metabolism can be the molecular biologist confronting the genome classified into different functional roles, in this case and its products, the level of the individual into 11 categories. Note also that we used the word organism should be of central importance for the ‘‘simplified.’’ That is, if we were to zoom in on the food-web ecologist. This is because, just as particu- nucleotide synthesis area of the diagram, more lar individual molecules (not classes of molecules) details would emerge, with more nodes (chemical are the actual participants in metabolic networks, categories) and pathways appearing (you can do it is individual organisms (not species, popula- this yourself on the Internet at www.genome.ad. tions, functional groups) that do the actual eating jp/kegg/). Yet more magnification, for example,
7 8 AQUATIC FOOD WEBS
(and thus make the trophic connections) in food (a) webs. Thus, organism-focused reasoning based on sound physiological principles is likely to be of great assistance in unraveling food webs. Further- more, the constituents of food webs are not fixed entities; rather, they are products and agents of continuous evolutionary change. And since evolu- tion operates primarily at the level of individual reproductive success, it seems that evolutionary thinking should play a central role in under- standing how and why food webs are shaped the way they are. Molecular/cell biologists coming to grips with the daunting complexity of the genome and its products (Figure 1.1) have a powerful ally in the fact that each node of a metabolic network is the product (and a reactant) in an enzymatically driven chemical reaction. Thus, all parts of the network must obey strict rules of mass balance and stoichiometric combination in the formation and destruction of the constituent parts. Indeed, these simplifying principles form the basis of various emerging theories through which cell biologists hope to make progress in understanding func- (b) tional interconnections among genes and gene products in metabolism (e.g. metabolic control theory, Dykhuizen et al. 1987; Wildermuth 2000; stoichiometric network theory, Hobbie et al. 2003; metabolic flux balancing, Varma and Palsson 1994). But why should such powerful tools be left to molecular biologists? Luckily, food-web ecologists can also take advantage of the considerable traction afforded by the firm laws of chemistry because, just as every node in a biochemical network is a chemical entity, so is every node in a food web. That is, each individual organism forming a connection point in Figure 1.1(b) is an aggregation of biochemicals and chemical elements and is sustained by the net outcome of the coupled biochemical pathways shown in Figure 1.1(a). Thus, the interactions among food-web components in terms of consump- Figure 1.1 Two entangled banks demonstrating the intimidating tion (and the feedbacks imposed by nontrophic task that lies before biology. Ultimately, food webs relations of excretion and nutrient regeneration) (a) (Lavigne 1996) are the outcome of dynamic interactions are also constrained by the firm boundaries of among various organisms that acquire resources from the abiotic mass balance and stoichiometric combination. environment and each other in order to drive their metabolism (b) (www.genome.ad.jp/kegg/) and leave offspring. These principles and their applications are known as ‘‘ecological stoichiometry’’ (Sterner and BIOSIMPLICITY VIA STOICHIOMETRY 9
Elser 2002), while their more recent extension to the partnership of an intimate comedy; and if we would catch realms of evolutionary biology, behavior, physio- the spirit of the piece, our attention must not all be logy, and cellular/molecular biology are known as absorbed in the characters alone, but most also be ‘‘biological stoichiometry’’ (Elser et al. 2000b). The extended to the scene, of which they are born, on which they play their part, and with which, in a little while, approach of ecological stoichiometry simplifies the they merge again. bewildering ecological complexity in Figure 1.1(b) by focusing on key ecological players in food webs and characterizes them in terms of their relative Stoichiometric imbalance, ‘‘excess’’ carbon (C), nitrogen (N), and phosphorus (P) carbon, and the functioning of demands. In biological stoichiometry, metabolic food webs complexity in Figure 1.1(a) is simplified by focus- ing on major biochemical pools (e.g. rRNA, total Conventional food-web diagrams show binary protein, RUBISCO) that determine overall orga- feeding links among species. Flowchart analyses of nismal demands for C, N, and P and attempts food webs go beyond a binary depiction of feeding to connect those biochemical demands to major relations in using dry-weight, energy (Joules), or evolutionary forces operating on each organism’s carbon as a common currency to express the life history or metabolic strategy. magnitude of particular connections. The advant- In this chapter we will review some basic prin- age of using C-based flow charts is quite obvious ciples and highlight some of the most recent since, not only does C account for some 50% of dry findings from the realm of ecological stoichiometry mass in most species and taxa, it also allows for in food webs, to illustrate how a multivariate inclusion on import and export of inorganic carbon perspective on energy and chemical elements in the same scheme. However, with the realization improves our understanding of trophic relations. that the supply and availability of P is a key More details of these (and other) matters are avail- determinant of the binding, flux, and fate of C in able in Sterner and Elser (2002); in this chapter we freshwater food webs, in many cases more informa- seek to highlight some findings that have emerged tion may be gained from P-based flow charts. Such since publication of that work. We will then dis- a view will also provide a better representation of cuss recent movements to integrate stoichiometric the recycling of elements, and thus differentiate study of food webs with the fact that food-web between ‘‘old’’ and ‘‘new’’ production. In a P-lim- components are evolving entities and that major ited system, any extra atom of P will allow for evolutionary pressures impose functional trade- binding of more than 100 atoms of C in autotroph offs on organisms that may have profound impli- biomass. Due to different elemental ratios in dif- cations for the structure and dynamics of food ferent food-web compartments, conventional webs. Our overarching view is that stoichiometric flowchart diagrams will normally turn out quite theory can help in integrating food-web ecology different in terms of C or P (Figure 1.2). Neither is and evolution into a more comprehensive frame- more ‘‘true’’ or correct than the other; instead, each work capable of making a priori predictions about provides complementary information on pools and major food-web features from a relatively simple key processes. Since P is a conservative element set of fundamental assumptions. In advocating this that is lost from aquatic systems only by sedimenta- view we hope to continue to add substance to the tion or outflow, it will normally be frequently vision of food webs offered nearly 100 years ago recycled and may thus bind C in stoichiometric by Alfred Lotka (1925): proportions a number of times over a season. In addition to pictures of who is eating whom in For the drama of life is like a puppet show in which stage, scenery, actors, and all are made of the same stuff. food webs, we need to understand the outcome of The players indeed ‘have their exits and their entrances’, those feeding interactions for the consumer. but the exit is by way of a translation into the substance ‘‘Trophic efficiency’’ is key aspect of food webs of the stage; and each entrance is a transformation scene. that captures important aspects of this outcome So stage and players are bound together in the close (here we use ‘‘efficiency’’ to refer to the fraction of 10 AQUATIC FOOD WEBS
explanations may be invoked to explain such 100 mg C m–3 100 mg C m–3 per day patterns but, as argued by Cebrian et al. (1998), surely the high transfer efficiency of C in plank- Z tonic systems may be attributed to both high cell A quotas of N and P relative to C (high-quality food B for reasons given below) coupled with decreased importance of low-quality structural matter like lignins and cellulose that are poorly assimilated. F A striking feature of the cross-ecosystem compar- ison compiled by Cebrian et al. (1998) is the close correlation paths among autotroph turnover rate, D specific nutrient (N, P) content, and the trophic efficiency. These associations make perfect sense SE D SE D from a stoichiometric point of view: while con- Egestion/exudation Grazing sumers are not perfectly homeostatic (cf. DeMott Death 2003), they have a far closer regulation of elemental ratios in somatic tissue than autotrophs (Andersen 1 mg P m–3 1mg P m–3 per day and Hessen 1991; Hessen and Lyche 1991; Sterner and Hessen 1994) and their input and output of elements must obey simple mass balance prin- A Z ciples. As a general rule, limiting elements are expected to be utilized for growth and transferred in food chains with high efficiency, while non- F limiting elements, by definition present in excess, B must be disposed of and may be recycled (Hessen 1992; Sterner and Elser 2002). Thus, when feeding on low C : N or low C : P food, a considerable share D of N and P may be recycled (Elser and Urabe Release by grazers Grazing 1999), while C-use efficiency is high (Sterner and Death Elser 2002). However, typically autotrophs have higher C : element ratios than consumers (Elser Figure 1.2 Pools and fluxes of carbon (a) and phosphorus (b) in a et al. 2000a). Thus, when consumers feed on diets pelagic food web of a eutrophic lake (data from Vadstein et al. that are high in C : N or C : P, nutrient elements are 1989). A: algae, B: bacteria, D: detritus and other kinds of nonliving reclaimed with higher efficiency by the animal dissolved and particulate matter, F: heterotrophic flagellates, and Z: metazoan zooplankton. Boxes denotes biomasses, arrows denote (Elser and Urabe 1999), while much of the C is fluxes. Note the entirely different size of pools and fluxes for C and P. unassimilated and must be egested, excreted, or respired (DeMott et al. 1998; Darchambeau et al. 2003). Since herbivore performance is strongly energy or C produced by a certain level that is impaired in these high C : nutrient systems, more C transferred to higher trophic levels). Efficient must enter detrital pathways, as is clearly shown trophic systems typically have steep slopes from by Cebrian’s studies. autotrophs at the base of the food web to top These factors point to fundamental differences predators, and subsequently should also support a between ecosystems not only with regard to the higher number of trophic levels than less efficient transfer and sequestration of carbon, but also systems. Typically, planktonic systems are among with regard to community composition and eco- those with high trophic efficiency, with forests on system function in more general terms. While the extreme low end (Hairston and Hairston 1993; realizing that pelagic food webs are among the Cebrian and Duarte 1995; Cebrian 1999). Several most ‘‘efficient’’ ecosystems in the world, there is BIOSIMPLICITY VIA STOICHIOMETRY 11 certainly a huge scatter in trophic efficiency also experiments (Faerovig et al. 2002; Urabe et al. among pelagic systems, that is, considerable 2002a). At a given (low) level of P, high light yields variation appears when phytoplankton biomass or more algal biomass than low light treatments, production is regressed against zooplankton but with lower food quality (higher C : P). The net (Hessen et al. 2003). Such scatter may be caused by outcome will be slow herbivore growth rates at time-lag effects, external forcing, algal species’ high light, with a higher asymptotic biomass composition, and associated biochemistry as well of adults. This is because high growth rate and as by top-down effects, but it is also clear that high reproduction require a diet that balances the alterations in trophic transfer efficiency due to grazer’s demands in terms of energy, elements, stoichiometric constraints could be a strong con- and macromolecules, while a standing stock of tributor. Thus, understanding food-web dynamics (nearly) nonreproducing adults can be sustained requires understanding the nature and impacts of on a low-quality diet since their basic metabolic nutrient limitation of primary production. requirements mostly rely on C (energy). This implies a shift from a high to low biomass: pro- Stoichiometry, nutrient limitation, duction ratio. Eventually, the nutrient constraint in and population dynamics in food webs low quality (high autotroph C : P) systems may be overcome by feedbacks from grazers. Such intra- Since nutrient limitation of autotroph production and interspecific facilitation (Sommer 1992; Urabe only occurs, by definition, when nonnutrient et al. 2002a) may induce a shift in population resources such as light are sufficient, a particularly dynamics under a scenario of increasing grazing intriguing outcome of stoichiometric analysis in since an increasing amount of P will be available freshwaters, and one that is rather counter- per unit of autotroph biomass due to the combined intuitive, is that high solar energy inputs in the effect of grazing and recycling (cf. Sterner 1986). form of photosynthetically active radiation may Thus, understanding the biological role of lim- reduce secondary (herbivore) production (Urabe iting nutrients in both autotrophs and consumers and Sterner 1996; Sterner et al. 1997; Hessen et al. provides a basis for better prediction of how 2002; Urabe et al. 2002b). The rationale is as population dynamics of herbivores should respond follows: when photosynthetic rates are high due to changing environmental conditions that alter to high light intensity but P availability is low nutrient supply, light intensity, or other environ- (a common situation in freshwaters), C is accumu- mental conditions. However, surprisingly little lated in biomass out of proportion with P. Thus, attention in mainstream textbooks on population C : P in the phytoplankton increases, meaning dynamics has been given to food quality aspects potential reduced C-use efficiency (P-limitation) (e.g. Turchin 2003). According to the stoichiometric in P-demanding grazers such as Daphnia. The growth rate hypothesis (described in more detail outcomes of such effects have been shown by later) and supported by an increasing body of Urabe et al. (2002b), who applied deep shading experimental data (Elser et al. 2003), taxa with that reduced light intensities nearly 10-fold to field high body P-contents commonly have high growth enclosures at the Experimental Lakes Area, where rates and can thus rapidly exploit available seston C : P ratios are generally high (Hassett et al. resources but are probably especially susceptible 1997) and Daphnia have been shown to be P-limited for stoichiometric food quality effects. What are (Elser et al. 2001). The outcome was a nearly five- the dynamic consequences of this under different fold increase in zooplankton biomass in unenriched conditions of nutrient limitation in the food web? enclosures after the five-week experiment. For reasons given above, the a priori assumption However, the negative effects of high algal would be that predator–prey interactions should C : P ratio on zooplankton can be a transient situa- be more dynamic when the system sets off with tion and high energy (light) input may eventually high quality (low C : P) autotroph biomass. Low sustain a high biomass of slow-growing zoo- autotroph C : P will stimulate fast growth of plankton, as demonstrated by long-term chemostat the consumer and relatively high recycling of P for 12 AQUATIC FOOD WEBS
autotroph reuse; the system should therefore (a) Qmin = 0.010, Qz = 0.018 rapidly reach an equilibrium where food abund- ance is limiting to the grazer. On the other hand, a system with high C : P in the autotrophs should have slow grazer response and low recycling of P, 1400 yielding sluggish and perhaps erratic dynamics as 1200 the system operates under the simultaneous effects 1000 of changing food abundance, quality, and nutrient 800 Bz recycling. Indeed, recent models (Andersen 1997; 600 Hessen and Bjerkeng 1997; Loladze et al. 2000; 400 Muller et al. 2001) taking grazer P-limitation 200 and recycling into account clearly demonstrate 0 1000 0.025 this kind of dynamic dependency on resource 2000 0.020 and consumer stoichiometry. As demonstrated by 3000 0.015 4000 Figure 1.3 (Hessen and Bjerkeng 1997), the ampli- Ba 0.010 5000 Qa tudes and periods of autotroph–grazer limit cycles 6000 0.005 depends both on food quantity and quality. When P : C in the autotroph becomes low, this constrains grazer performance and a high food biomass of low quality may accumulate before the grazer slowly builds up. With assumptions of a more (b) Qmin = 0.003, Qz = 0.018 efficient elemental regulation in the autotroph (i.e. lower minimum P : C), limit cycles or amplitudes will be smaller, but the periods will increase. One intriguing feature of stoichiometric model- 1400 ing is the potential extinction of the grazer, like a 1200 P-demanding Daphnia, under a scenario of high 1000 food biomass but low food quality (Andersen 1997; 800 B Hessen and Bjerkeng 1997). External enrichment z 600 of P to the system will also invoke strong shifts in 400 system dynamics due to stoichiometric mechan- 200 0 0.025 isms (Andersen 1997; Muller et al. 2001). The 1000 relevance for these theoretical exercises for natural 2000 0.020 Bz 0.015 systems remains to be tested, however. Clearly 3000 isocline 4000 0.010 Ba the assumption of two compartment dynamics 5000 0.005 Qa represent an oversimplification, since a consider- 6000 able share of recycled P and organic C will enter Figure 1.3 Three-dimensional limit cycles for two scenarios the bacteria or detritus pool, thus dampening the with Daphnia grazing on algae with different flexibility in their dynamics predicted from the simplified model P : C ratio (Qa). The solid line gives the trajectory, while projections of the three-dimensional trajectory are given on the Ba–Bz assumptions. 1 plane and the Qa–Ba plane. Ba: algal biomass (mgCl ), Bz: Thus, one central outcome of stoichiometric 1 grazer (Daphnia) biomass (mgCl ), Qa: algal P : C (mgP:mg C). theory in consumer–resource systems is deviation In the upper panel, the lower bound of Qa (Qmin) is set to 0.010(a), from the classical straight Lotka–Volterra isoclines while Qmin in the lower panel is 0.003(b). P : C in the grazer (Qz) (Andersen 1997; Murdoch et al. 2003). From is in both cases fixed at 0.018. By increasing Qz (higher P : C ratio, these analyses, it appears that a combination of lower C : P ratio) slightly in the lower scenario, the grazer will go extinct, and the system will stabilize at a high algal biomass inter- and intraspecific facilitation during periodic near Qmin. nutrient element limitation by consumers results in a deviation from straightforward negative density BIOSIMPLICITY VIA STOICHIOMETRY 13 dependence in consumer populations and a much and developmental stages according to their mode richer array of population dynamics appears. of feeding (carnivores, omnivores, herbivores, In this way, stoichiometry can provide a logical detritivores, filtrators, raptors, scavengers, etc). explanation for Allee effects (positive density- We suggest that it might also be useful to adopt dependence) and hump-shaped curves for density- a subtler categorization based on dietary, stoichio- dependent responses. metric requirements. In fact, in many cases the Much of the preceding discussion has had more specific dietary requirements of a taxon may herbivores and other primary consumers (e.g. be the ultimate cause for an organism being a detritivores) in mind. What about the role of carnivore or a detritivore and, as we discuss stoichiometry higher in food webs? Since meta- below, is probably also an important factor con- zoans do not vary too much in their biochemical tributing to widespread omnivory among taxa. makeup, predators are less likely to face food Hence one could speak about the ‘‘stoichiometric quality constraints compared with herbivores and niche’’ of a particular species, in the sense that especially detritivores (Sterner and Hessen 1994). species (or stages) with high P (or N) requirements Fish in general have high P requirements due to would succeed in situations that supply nutrient- investment in bone (Sterner and Elser 2002); this rich food compared with species with lower could be seen as another reason, in addition to nutrient requirements. For example, for freshwater their large body size, why P-rich Daphnia should food webs it has been suggested that when be preferred prey relative to P-poor copepods. planktonic algae are deprived of P and develop A more important issue is, however, how the high C : P ratios, P-demanding species like Daphnia predicted dynamics due to stoichiometric mechan- acutely suffer from ‘‘P-starvation’’ and, probably isms might be associated with the potential prey due to decreased C-use efficiency, become com- susceptibility to predators. A reasonable assump- petitively inferior to less P-demanding members of tion would be that grazers in low food quality the plankton community like Bosmina (DeMott and systems would be more at risk for predatory Gulati 1999; Schulz and Sterner 1999). Thus, the mortality simply because, all else being equal, stoichiometric niche space available to Bosmina slow growth would render the population more may extend to higher regions of food C : P than susceptible to the impacts of any given rate of in Daphnia. But what about the other end of the mortality loss. However, the effects might not C : P continuum? Interestingly, in a brand-new quite be so straightforward. For example, fast- stoichiometric wrinkle, recent evidence shows that growing individuals generally also require high extremely low C : P may cause decreased growth in rates of food intake; in turn, more active feeding Daphnia (Plath and Boersma 2001) and the cater- might increase predation risk (Lima and Dill pillar Manduca sexta (Perkins et al. 2003). While the 1990). Furthermore, there may be some inherent mechanistic bases of these responses remain and unappreciated physiological–developmental obscure, they appear to represent the other side of impacts associated with rapid growth such that the stoichiometric niche in the P-dimension. overall mortality is elevated in fast-growing indi- Another aspect of stoichiometric effects on bio- viduals, over and above potentially accentuated diversity and food-web structure relates to the predation risk (Munch and Conover 2003). number of trophic levels and the degree of omnivory. Hastings and Conrad (1979) argued that Stoichiometry, omnivory, and the the evolutionary stable length of food chains evolution of food-web structure would be three, and that the main determinant of the number of trophic levels is the quality of prim- The fact that different species or taxa have differ- ary production (and therefore, to at least some ent stoichiometric or dietary requirements has degree, its C : nutrient ratio) and not its quantity, as important bearings on the dietary preferences that is often implied in discussions of food-web length. weave food webs together. Ecologists have com- Omnivory may be seen as a compromise between monly generated a coarse classification of species exploiting large quantities of low quality resources 14 AQUATIC FOOD WEBS at low metabolic cost, or utilizing lower quantities that they often cannot even build their bodies of high-quality food at high metabolic costs. From from available food? Why do some species seem a stoichiometric point of view, omnivory may be to be more sensitive to the effects of stoichio- seen as a way of avoiding nutrient deficiency while metric food quality? In this section we follow at the same time having access to a large reservoir the advice of Holt (1995) by describing some of energy. This ‘‘best of two worlds’’ strategy is recent findings that illuminate some of these evolu- clearly expressed as life-cycle omnivory, like in tionary questions in the hopes that perhaps in the crayfish where fast-growing juveniles are carni- future we will encounter Darwin as well as Lin- vorous, while adults chiefly feed on detritus or deman in the reference sections of food-web plants (cf. Hessen and Skurdal 1987). papers. The fact that organisms can be potentially Beyond expanding the diet to include more limited not only by access to energy (carbon) but nutrient-rich prey items and thus inducing also by nutrients has obvious implications for omnivory as discussed earlier, another obvious coexistence of potential competitors. While this evolutionary response to stoichiometrically unbal- principle has been well explored for autotrophs anced food would be for a consumer to evolve a (e.g. Tilman 1982), the same principle may be lower body requirement for an element that is invoked for heterotrophs with different require- chronically deficient in its diet. Several recent ments for key elements (cf. Loladze et al. 2004). studies emphasizing terrestrial biota have pro- In fact, this will not only hold for interspecific vided evidence for just such a response. Fagan competition, but also for intraspecific competition, et al. (2002) examined the relative nitrogen content since most species undergo ontogenetic shifts in (%N of dry mass, N : C ratio) of folivorous insect nutrient requirements. Indeed, it now seems species and documented a significant phylogenetic that this coexistence principle can be extended signal in which the recently derived insect group to explain the evolution and maintenance of (the ‘‘Panorpida,’’ which includes Diptera and omnivory (Diehl 2003), since utilization of diff- Lepidoptera) have significantly lower body N erent food resources in species with different content than the more ancestral groups Coleoptera nutrient contents promotes and stabilizes feeding and Hemiptera which were themselves lower than diversification. the still older Lower Neoptera. Their analysis eliminated differences due to body size, gut con- Biological stoichiometry: the tents, or feeding mode as possible explanations for convergence of ecological the pattern. They noted that the divergence of and evolutionary time major insect groups appears to have coincided with major increases in atmospheric CO2 con- It should be clear from the preceding material that centrations (and thus high plant C : N ratio) stoichiometric imbalance between food items and and hypothesized that clades of insects that consumers has major effects on the dynamics emerged during these periods of ‘‘nitrogen crisis’’ and structure of key points in the food web and in plant biomass were those that had an efficient especially at the autotroph–herbivore interface. N economy. Signs of evolutionary response to Indeed, the effects of stoichiometric imbalance on stoichiometric imbalance in insects is also seen at a herbivores are often extreme and suggest that finer scale in studies by Jaenike and Markow there should be strong selective pressure to alle- (2002) and Markow et al. (1999), who examined the viate these impacts. The fact that such impacts body C : N : P stoichiometry of different species of nevertheless remain implies that there may be Drosophila in relation to the C : N : P stoichiometry fundamental trade-offs and constraints on evolu- of each species’ primary host resource. Host foods tionary response connected to organismal stoichio- involved different species of rotting cactus, fruit, metry. So, why is it that consumer organisms, mesquite exudates, and mushrooms and presented such as herbivorous zooplankton or insects, a considerable range in nutrient content. They maintain body nutrient contents that are so high showed a significant correlation between host BIOSIMPLICITY VIA STOICHIOMETRY 15 nutrient content and the nutrient content of the various comparisons involving physiological, associated fly species: taxa specializing on ontogenetic, and cross-species comparisons, nutrient-rich mushrooms had the highest nutrient P-content increases with growth rate (Main et al. contents (in terms of %N and %P) while those 1997; Carrillo et al. 2001; Elser et al. 2003; Makino specializing on the poorest quality resource and Cotner 2003), RNA content also increases with (mesquite flux) had the lowest body nutrient growth rate (Sutcliffe 1970; Vrede et al. 2002; Elser contents. Signs of evolutionary response to stoichio- et al. 2003; Makino and Cotner 2003; Makino et al. metric resource limitation can be found even in 2003), and, importantly, increased allocation to the amino acid structure of proteins themselves. RNA quantitatively accounts for the increased Using genomic data for the prokaryote Escherichia P content of rapidly growing organisms (Elser et al. coli and the eukaryote Saccharomyces cerevisiae, 2003; Figure 1.4(a)). The mechanistic connections Baudoin-Cornu et al. (2001) showed that protein among RNA, P, and growth appear to manifest enzymes involved in C, N, or S metabolism were quite directly in how nutrient supply affects herb- significantly biased in favor of amino acids having ivore dynamics, as illustrated in a recent study of a low content of C, N, and S, respectively. That is, plant–herbivore interaction in the Sonoran Desert enzymes involved in uptake and processing of N (Figure 1.4(b)). In this study (Schade et al. 2003), were disproportionately constituted of amino acids interannual variation in rainfall led to increased having relatively few N atoms, a response that soil P supply under wet conditions, which in turn makes adaptive sense. lowered foliar C : P in velvet mesquite trees. These studies suggest that heterotrophic con- Consequently, mesquite-feeding weevils had higher sumers can indeed respond in evolutionary time to RNA and correspondingly higher P contents, reduce the degree of stoichiometric imbalance consistent with a P-stimulated increase in growth between their biomass requirements and their rate, along with higher population densities on often-poor diets. And yet the overall nutrient mesquite branches with low foliar C : P. In sum, content (e.g. %N and %P) of herbivorous animals these findings of tight and ecologically significant (zooplankton, insects) remains at least 10-times associations of growth, RNA, and P contents and often 100- to 1,000-times higher than is found suggest that major aspects of an organism’s in autotroph biomass (Elser et al. 2000a). Some ecology and life history have an important fundamental benefit of body nutrient content must stoichiometric component. Given the strong effects exist. The nature of the benefits of higher body of stoichiometric imbalance in trophic relations nutrient content is perhaps becoming clearer now (Figure 1.3), it seems then that evolutionary based on work emerging from tests of the ‘‘growth adjustments bearing on growth rate will impact rate hypothesis’’ (GRH hereafter), which proposes on the types of organisms and species that come that variation in the C : P and N : P ratios of many to dominate key positions in food webs under organisms is associated with differences in growth different conditions. rate because rapid growth requires increased Thus, better understanding of the genetic basis allocation to P-rich RNA (Hessen and Lyche 1991; and evolutionary dynamics of the coupling among Sterner 1995; Elser et al. 1996; Elser et al. 2000b). growth, RNA, and P may help in understanding In this argument, a key life history parameter, how food webs self-organize. Elser et al. (2000b) growth or development rate, is seen to inherently suggested that the genetic basis of variation in require increased investment in P-intensive bio- growth and RNA (and thus C : P and N : P ratios) chemicals, implying a trade-off in that fast- lies in the genes encoding for RNA, the rDNA. growing organisms are likely to find themselves In particular, they reviewed evidence suggest- constrained by an inability to acquire sufficient P ing that rDNA copy number and length and con- from the environment or diet. Work with zoo- tent of the rDNA intergenic spacer (IGS, where plankton and other organisms such as insects and promoter–enhancer sequences are found) were key microbes now makes it clear that the fundamental variables underlying growth rate variation because core of the GRH is correct (Figure 1.4(a)): in of their effects on transcriptional capacity and ebvr iooe fet ouaindnmc.I td fadsr csse,itrnulvraini analldt neana difference interannual ( to trees led mesquite rainfall velvet in variation to interannual panel) ecosystem, left desert (upper a availability of study P a soil In in dynamics. population affects ribosomes herbivore -otn fdvrehtrtohcognss h aaicue nti ltivledt from data involve plot this in included data The organisms. heterotrophic diverse of P-content 1.4 Figure 16 ihfs-rwn ueie eedmntdb niiul arigol h ogIS iue rmGrkoae l (2002). al. et lines Gorokhova and fecundity from low long while Figures both genotype IGS. carrying mixed long individuals the the of by only dominated proportion were carrying relative juveniles individuals the slow-growing by in with generations dominated lines differed five fecundity were for lines high juveniles selection Selected IGS: Artificial fast-growing left). long (c) with (bottom the (2003). just contents al. or P et IGS and Schade short from RNA Figures parthenogenetic juvenile higher. in in was panel) and supply left (top P fecundity soil weight-specific and on rainfall when right) (bottom densities nrsos,msut-edn evl ( weevils mesquite-feeding response, In rsaenzolntnincluding zooplankton crustacean n hne u opyilgclPlmtto.I ahcs,RArc adtu -ih aawr rmognsswt eaieyfs rwhrate. growth an fast and relatively 0.97 with of organisms slope from a were has data combined P-rich) data thus the (and of RNA-rich all case, to each fit In relationship P-limitation. A physiological to due changes and (a)
% Total-P WEBS FOOD AQUATIC 0 1 2 3 012 rmtegnm hog eaoimt h odwb a aito nalcto oPrc iooa N xlisvraini the in variation explains RNA ribosomal P-rich to allocation in Variation (a) web. food the to metabolism through genome the From (c) % RNA-P Daphnia 0.2 0.4 0.6 0.8 1.2 1.4 10 12 0 2 4 6 8 0 1 iii setosa Sibinia niiulrltosisivleitrpcfi oprsn,cagsdrn noeei development, ontogenetic during changes comparisons, interspecific involve relationships Individual . y %RNA 0.97 = Hi R g %P 2 o Hi Low h 0.78 = X a ihrRAadPcnet bto et n hsahee ihrpopulation higher achieved thus and left) (bottom contents P and RNA higher had ) 0.62 + Fecundity %P ahi pulicaria Daphnia rspsvelutina Prosopis 3 (b) 100 R 0.1 0.2 0.3 0.4 0.5 20 40 60 80 2 0 0
f07.Fgr rmEsre l 20) b rnmsino olPinto P soil of Transmission (b) (2003). al. et Elser from Figure 0.78. of Sibinia %RNA or %P 0.5 1.5 2.5 0 1 2 3 Growth rate
May 2000 Soil P rdcdcreae hne njvnl rwhrt tpright) (top rate growth juvenile in changes correlated produced ,laigt eraei oirC:Prto(pe right). (upper ratio P : C foliar in decrease a to leading ), %RNA g May 2000 h %P Rainfall % Apr 2001 2001 Apr Low .cl,Doohl melanogaster Drosophila coli, E. %P
No. of Sibinia per g leaf Mesquite leaf C : P 100 200 300 400 500 600 700 800 900 0 1 2 3 0 May 2000 0 800 400 efC:P : Leaf C Apr 2001 n various and , 1200 s BIOSIMPLICITY VIA STOICHIOMETRY 17 thus rates of RNA production. Thus, rDNA If the results of Gorokhova et al. are confirmed variation may also be linked to differences in and shown to hold for other biota that comprise stoichiometric requirements of biota. Emerging food webs, then we would suggest that there is data suggest that indeed it is the case that varia- reason to question the traditional distinction tions in the rDNA can have important stoichio- between ecological and evolutionary time. That is, metric ramifications (Gorokhova et al. 2002; ecologists are used to considering species shifts Weider et al. 2004). (e.g. during seasonal succession in the plankton) in Of these data, most striking is the study of terms of the sorting out of various ecological Gorokhova et al. (2002), who performed an artificial transactions such as competition and predation selection experiment on the progeny of a single among taxa that have fixed genetic structure on the female of Daphnia pulex, selecting on weight- timescale of the study. The results of Gorokhova specific fecundity (WSF) of the animals. The et al. suggest that, even with a clonal organism, treatments responded remarkably quickly to the genetic recombinations with important ecophysio- selection regimen—animals selected for low logical impacts can arise on time scales that would fecundity were significantly lower than random easily be encompassed, for example, by a single controls or high fecundity selected animals within growing season in a lake. This same point of rapid three rounds of selection. After five rounds of evolution in food webs is also demonstrated by selection, animals were assayed for juvenile studies of rapid evolution of digestibility-growth growth rate along with RNA and P contents. The trade-offs in rotifer-algae chemostats (Fussmann analyses showed that lineages showing low WSF et al. 2003, Yoshida et al. 2003). Not all evolutionary had experienced a correlated inverse response of change requires the slow propagation of small juvenile growth rate; that is, these females with point mutations through the vast protein library low WSF produced offsprings that grew faster, that comprises the genome. Indeed it is becoming offering an opportunity to test the growth rate increasingly obvious that the genome can reorgan- hypothesis. As predicted by the GRH, these fast- ize quickly via structural mutations that affect growing juveniles had elevated RNA and P regulatory regions and gene copy number and via contents compared to the slower-growing counter- gene silencing/unsilencing mechanisms such as parts in the control and high-selected lines (Figure those resulting from insertion/deletion of trans- 1.4(c)). To determine if these differences had a posable elements. It is interesting to note that genetic basis, animals were screened for variation transposable elements have a significant role in the IGS of their rDNA. Consistent with the idea in silencing copies of the rDNA in Drosophila above that high RNA (fast growth) phenotypes (Eickbush et al. 1997) and have been identified should be associated with long IGS, a dispropor- in the rDNA of various other taxa, including tionate number of the (fast-growing, high RNA, Daphnia (Sullender and Crease 2001). To the extent high P) animals in the low fecundity treatment that these reorganizations impact traits that are carried only long IGS variants while (slow- ecophysiologically relevant (such as those that growing, low RNA, low P) animals in the high affect RNA production and thus organismal WSF and control lines disproportionately carried C : N : P stoichiometry), the shifting genomes of both long and short IGS variants. Recall that this interacting species will need to be incorporated experiment involved selection on the offspring of a (somehow!) into food-web ecology. single parthenogenetic female and all offspring throughout were also produced by parthenogen- esis. Nevertheless, it appears that functionally Conclusions significant genetic variation can arise in a few generations even within a ‘‘clonal’’ organism, as is Lotka’s ecological play remains to a large degree a becoming increasingly recognized (Lushai and mysterious entertainment. There is a long road Loxdale 2002; Loxdale and Lushai 2003; Lushai ahead for food-web ecologists before we can expect et al. 2003). to really understand how food webs self-organize 18 AQUATIC FOOD WEBS and are sustained in their complex interplay Children’’ begins with what sounds like a familiar with the abiotic environment. So, in some problem: ways this chapter is a plea for a form of a ‘‘bio- The rising hills, the slopes, simplicity’’ research program in approaching of statistics food-web study. That is, Darwin unlocked the lie before us. key to the origin of the biosphere’s bewildering the steep climb biodiversity with his breathtakingly simple of everything, going up, algorithm of variation and selection. Perhaps up, as we all go down. the complexity we see in food webs is more apparent than real and will also come to be seen to The poem ends with the advice: be the product of a simple set of rules ramifying stay together through time and space. Bearing in mind the learn the flowers warning ‘‘if your only tool is a hammer then every go light problem looks like a nail,’’ for the moment we So, building on this counsel, perhaps we should propose that the minimal biosimplicity ‘‘rulebox’’ try to stay together with our colleagues who are should include the Darwinian paradigm along confronting the genome, adapting their tools, data, with the laws of thermodynamics, mass con- and ideas to better understand our own questions, servation, and stoichiometric combination. What and perhaps offering some insights of our own. kind of food-web structures can be built using We do need to learn the flowers, their names, and such tools given the raw materials circulating in the names of those who eat them, and of those who the biosphere? eat those too. But, especially these days when the Answering such a question will take some time genomics juggernaut threatens to bury us all in a and a large quantity of cleverness and a certain blizzard of data and detail, perhaps we should try amount of courage. We can take some solace in the to go light, approach the problems with a simpli- fact that we will be traveling this long road toward fied but powerful toolbox that includes stoichio- understanding of biological complexity with metric principles, and see what major parts of the our colleagues who study the genome and its puzzle will yield themselves to our best efforts. immediate metabolic products. While they might not themselves always realize that high- Acknowledgments throughput sequencers and microarray readers do not necessarily result in high throughput of the The authors are grateful for the support of the conceptual insights that might make sense of it all, Center for Advanced Study of the Norwegian a time will come (or has it already?) when all Academy of Letters and Sciences that made biologists, from the geneticists to the ecologists, this work possible. JJE also acknowledges NSF will lean on each other for insight and necessary grant DEB-9977047. We thank M. Boersma and data. For guidance at this point we turn to the K. Wiltshire for their helpful comments on an early American poet Gary Snyder whose poem ‘‘For the draft. CHAPTER 2 Spatial structure and dynamics in a marine food web
Carlos J. Melia´ n, Jordi Bascompte, and Pedro Jordano
Introduction studies in food-web structure and dynamics have dealt with either large (but see Hori and Noda 2001) The role of space in population and community or small (but see Caldarelli et al. 1998) number of dynamics has been recently emphasized (e.g. species, but make no explicit reference to space Hanski and Gilpin 1997; Tilman and Kareiva 1997; (Caswell and Cohen 1993; Holt 1996, 1997). Only a Bascompte and Sole´ 1998). Several models for the few studies have explored the role of space on a small coexistence of interacting species in heterogeneous subset of trophic interacting species (Holt 1997; environments have been formulated. These Melia´n and Bascompte 2002). include the energy and material transfer across The present study is an attempt to link structure ecosystem boundaries and its implication for and dynamics in a spatially structured large succession and diversity (Margalef 1963; Polis et al. marine food web. We use data on the diet of 5526 1997), the geographic mosaic of coevolution specimens belonging to 208 fish species (Randall (Thompson 1994), the regional coexistence of 1967) in a Caribbean community in five different competitors via a competition–colonization trade- habitats (Opitz 1996; Bascompte et al., submitted). off (Tilman 1994), the random assembly of com- First, we analyze structure by addressing how munities via recruitment limitation (Hubbell 2001), simple trophic modules (i.e. tri-trophic food chains and metacommunities (Wilson 1992). As a general (FCs) and chains with omnivory (OMN) with the conclusion of these approaches, succession, dis- same set of species are shared among the five persal, local interactions, and spatial heterogeneity habitats. Second, we extend a previous meta- have appeared strongly linked to the persistence community model (Mouquet and Loreau 2002) by of diversity. However, the underlying pattern of incorporating the dynamics of trophic modules ecological interactions in a spatially structured in a set of connected communities. Specifically, ecosystem and its implications for the persistence the following questions are addressed: of biodiversity remains elusive by the lack of synthetic data (Loreau et al. 2003). 1. How are simple trophic modules composed Introducing space and multiple species in a by the same set of species represented among single framework is a complicated task. As Caswell habitats? and Cohen (1993) argued, it is difficult to analyze 2. How does the interplay between dispersal and patch-occupancy models with a large number of food-web structure affect species dynamics at both species because the number of possible patch states local and regional scales? increases exponentially with species richness. Therefore, most spatial studies have dealt with a Data collection: peculiarities few number of species (Hanski 1983), predator– and limitations prey systems (Kareiva 1987), or n-competing species (Caswell and Cohen 1993; Tilman 1994; Mouquet The Caribbean fish community here studied covers and Loreau 2003). On the other hand, the bulk of the geographic area of Puerto Rico–Virgin Islands.
19 20 AQUATIC FOOD WEBS
Data were obtained in an area over more than number of chains (with identical species at all 1000 km2 covering the US Virgin Islands of trophic levels) shared by each pair of communities. St Thomas, St John, and St Croix (200 km2), the To assess whether this shared number is higher or British Virgin Islands (343 km2), and Puerto Rico lower than expected by chance we develop a null (554 km2). The fish species analyzed and asso- model. This algorithm randomizes the empirical ciated data were obtained mainly from the study data at each community, yet strictly preserves the by Randall (1967), synthesized by Opitz (1996). ingoing and outgoing links for each species. In this Spatially explicit presence/absence community algorithm, a pair of directed links A–B and C–D matrices were created by considering the presence are randomly selected. They are rewired in such a of each species in a specific habitat only when that way that A becomes connected to D, and C to B, particular species was recorded foraging or provided that none of these links already existed breeding in that area (Opitz 1996; Froese and Pauly in the network, in which case the rewiring stops, 2003). Community matrices include both the and a new pair of links is selected. trophic links and the spatial distribution of 208 fish We randomized each food web habitat 200 taxa identified to the species level. Randall’s list of times. For each pair of habitats we compare each shark species was completed by Opitz (1996), successive pair of replicates and count the shared which included more sharks with affinities to coral number of simple tri-trophic FCs and chains with reefs of the Puerto Rico–Virgin islands area, based OMN containing exactly the same set of species. on accounts in Fischer (1978). Note that our trophic Then we estimated the probability that a pair-wise modules are composed only by fishes, and that all comparison of a random replicate has a shared fish taxa is identified to the species level, which number of such modules equal or higher than the implies that results presented here are not affected observed value. Recent algorithm analysis suggest by trophic aggregation. that this null model represents a conservative The final spatially explicit community matrix test for presence–absence matrices (Miklo´s and includes 3,138 interactions, representing five Podani 2004). food webs in five habitat types. Specifically, the We calculated the number of tri-trophic FCs, and habitat types here studied are mangrove/estuaries OMN chains common to all pairs of communities, (m hereafter; 40 species and 94 interactions), coral and compared this number with that predicted by reefs (c hereafter; 170 species and 1,569 interactions), our null model (Figure 2.1(b) and (d)). The coral seagrass beds/algal mats (a hereafter; 98 species reef habitat shares with all other habitats a number and 651 interactions), sand (s hereafter; 89 species of FCs and OMN larger than expected by chance and 750 interactions), and offshore reefs (r hereafter; (P < 0.0001 in all pair-wise comparisons except 22 species and 74 interactions). To a single habitat for the mangrove comparison, where P < 0.002 85 species are restricted while 46, 63, 12, and 2 and P < 0.01 for FCs and OMN, respectively). species occupy 2, 3, 4, and 5 habitats, respectively. Similarly, seagrass beds/algal mats and sand (a/s Global connectivity values (C) within each habitat contrasts) share a significant number of FCs are similar to previously reported values for food and OMN (P < 0.0001). Globally, from the 10 webs (Dunne et al. 2002). Specifically, Cm ¼ 0.06, possible intercommunity comparisons, five share a
Cc ¼ 0.054, Ca ¼ 0.07, Cs ¼ 0.095, and Cr ¼ 0.15. number of modules higher than expected by chance (Figure 2.1(a) and (c) where arrows are Food-web structure and null model thick when the pair-wise comparison is statistically significant, and thin otherwise). This suggests that We consider tri-trophic FCs (Figure 2.1(a)) and FCs habitats sharing a significant proportion of trophic with OMN (Figure 2.1(c)). We count the number modules are mainly composed by a regional pool and species composition of such trophic modules of individuals. within the food web at each community. We then The average fraction of shared FCs and OMN make pair-wise comparisons among communities between habitat pairs is 38% 24.5% and 41% 25%, (n ¼ 10 pair-wise comparisons) and count the respectively, which still leaves more than 50% of SPATIAL STRUCTURE AND DYNAMICS 21
(b) 1,500 (a)
1,000 c m a Frequency (FC) Frequency 500
r s
0 c/s c/a a/s c/r m/c m/a m/s s/r a/r m/r Habitat pairs
(d) 1,500 (c)
1,000 c m a
Frequency (OMN) Frequency 500
r s
0 c/s c/a a/s c/r m/c m/a m/s s/r a/r m/r Habitat pairs
Figure 2.1 The food-web modules studied here are (a) tri-trophic FCs, and (c) OMN chains. Circles represent the five different habitat types. For each habitat pair, the link connecting the two habitats is thick if the number of shared trophic modules is significant, and thin otherwise; (b) and (d) represent the frequency of shared tri-trophic FCs and OMN chains, respectively in all pair-wise community comparisons. Black and white histograms represent the observed and the average expected value, respectively. Habitat types are mangrove/estuaries (m), coral reefs (c), seagrass beds/algal mats (a), sand (s), and offshore reefs (r). As noted, coral reefs (c), share with the rest of the habitats a number of FCs and OMN larger than expected by chance, which suggest a high degree of connectance promoted by dispersal. different species composition trophic modules trophic modules across space in larger structures, between habitats. However, it is interesting to note which suggest a cohesive spatial structure (Melia´n that 15 species (specifically, herbivorous species and Bascompte 2004). from Blenniidae and Scaridae families, and top species from Carcharhinidae and Sphyrnidae famil- Dynamic metacommunity model ies) are embedded in more than 75% of trophic modules, which suggests that a small number of In order to assess the local and regional dynamics species are playing an important role in connecting of the structure studied, we extend a previous through dispersal local community dynamics. metacommunity model (Mouquet and Loreau Note that these highly connected species link 2002, 2003) by incorporating trophic modules 22 AQUATIC FOOD WEBS
(tri-trophic FCs and FCs with OMN) in a set of where the sum stands for all the other commun- interacting communities. The model follows the ities l. Rik represents the amount of resources formalism of previous metapopulation models available to species i in community k (Levins 1969) applied to the scale of the individual XS : (Hastings 1980; Tilman 1994). At the local scale Rik ¼ aijkPjk,(24Þ (within communities), we consider a collection j¼1 of identical discrete sites given that no site is ever where aijk is the predation rate of species i on occupied by more than one individual. The regio- species j in community k, and the sum is for all nal dynamics is modeled as in mainland–island prey species. Similarly, Cik represents the amount models with immigration (Gotelli 1991), but with of consumption exerted on species i by all its an explicit origin of immigration that is a function predators in community k, and can be written as of emigration from other communities in the meta- follows: community (Mouquet and Loreau 2003). Therefore, XS : the model includes three hierarchical levels Cik ¼ ajikPjk,(25Þ (individual, community, and metacommunity). j¼1
The model reads as follows: where ajik is the predation rate of species j on dP species i in community k, and the sum is for all ik ¼ yI V þ (1 dÞc P V m P dt ik k ik ik k ik ik predator species. : : þ RikPik CikPik (2 1Þ We have numerically simulated a metacommu-
At the local scale, Pik is the proportion of sites nity consisting of six species in six communities. In occupied by species i in community k. Each com- each community, either two simple tri-trophic FCs, munity consists of S species that indirectly com- or two OMN chains are assembled with the six pete within each trophic level for a limited species. The two trophic modules within each proportion of vacant sites, Vk, defined as: community are linked only by indirect competition XS between species within the same trophic level. We : Vk ¼ 1 Pjk,(22Þ assumed a species was locally extinct when its j¼1 proportion of occupied sites was lower than 0.01. where Pjk represents the proportion of sites occu- Mortality rates (mik) are constant and equal for all pied by species j within the same trophic level in species. Dispersal success (y) was set to 1. community k. The metacommunity is constituted We considered potential reproductive rates to fit by N communities. d is the fraction of individuals the constraint of strict regional similarity, SRS dispersing to other habitats, and dispersal success, (Mouquet and Loreau 2003). That is, species within y, is the probability that a migrant will find a new each trophic level have the same regional basic community, cik is the local reproductive rate of reproductive rates, but these change locally among species i in community k, and mik is the mortality communities. Under SRS, each species within each rate of species i in community k. trophic level is the best competitor in one com- For each species in the community, we munity. Similarly, we introduce the constraint of considered an explicit immigration function Iik. strict regional trophic similarity (SRTS). That is, Emigrants were combined in a regional pool of each consumer has the same set of local energy dispersers that was equally redistributed to all requirements but distributed differently among other communities, except that no individual communities. Additionally, we assumed a direct returned to the community it came from (Mouquet relationship between the resource’s local repro- and Loreau 2003). After immigration, individuals ductive rate and the intensity it is predated with were associated to the parameters corresponding (Jennings and Mackinson 2003). to the community they immigrated to. Iik reads as: Under the SRS and SRTS scenarios, regional XN species abundance and intercommunity variance d I ¼ c P ,(2:3Þ are equal for each of the two species within the ik N 1 il il l6¼k same trophic level. Regional abundance in OMN is SPATIAL STRUCTURE AND DYNAMICS 23 higher, equal, and lower for top, intermediate, and (a) 0.4 basal species, respectively. Local abundances differ significantly between the two modules explored. 0.3 Specifically, when there is no dispersal (d ¼ 0) there is local exclusion by the competitively 0.2 superior species (Mouquet and Loreau 2002). This occurs for the basal and top species in the simple 0.1 trophic chain. The variance in the abundance of the basal and top species between local communities is 0 thus higher without dispersal for tri-trophic FCs 0 0.2 0.4 0.6 0.8 1 (Figure 2.2(a)). d However, the situation is completely different (b) 0.4 for OMN. Now, intercommunity variance is very low for both the basal and top species in the 0.3 absence of dispersal, and dramatically increases with d in the case of the top species. When the 0.2 communities are extremely interconnected, the top species disappears from the two communities < 0.1 (Pik 0.01), and is extremely abundant in the remaining communities. For intermediate species, varianceIntercommunity variance Intercommunity 0 increasing dispersal frequency decreases the 0 0.2 0.4 0.6 0.8 1 intercommunity variance, except when d ranges d between 0 and 0.1 in FCs (Figures 2.2(a) and (b)). Figure 2.2 Intercommunity variance in local species abundance Finally, we can see in Figure 2.2(b) (as compared for the basal (continuous line), intermediate (broken line), and top with Figure 2.2(a)) that intercommunity variance (dotted) species as a function of the proportion of dispersal between for high d-values is higher in a metacommunity communities (d ). (a) Represents tri-trophic FCs and (b) OMN with OMN. Thus, the interplay between dispersal chains. Parameter values are mik ¼ 0.2, cik for basal species is 3, among spatially structured communities and food- 2.8, 2.6, 2.4, 2.2, and 2 from the first to the sixth community, respectively. For intermediate species cik is 1.5, 1.4, 1.3, 1.2, 1.1, and web structure greatly affects local species abund- 1, respectively from the first to the sixth community. Top species ances. The results presented here were obtained reproductive values are 0.8, 0.75, 0.7, 0.65, 0.6, and 0.55, with a single set of species parameters. Under the respectively. Predation rates of intermediate and top species j on SRS and SRTS scenarios, results are qualitatively species i in community k are 0.6, 0.5, 0.4, 0.3, 0.2, and 0.1, robust to deviations from these parameter values. respectively. The initial proportion of sites occupied by species i in community k,(Pik) is set to 0.05. As noted, in closed metacommunities, tri-trophic FCs show an extreme variation in local < Summary and discussion abundances for both the basal and top species (Pik 0.01) in two and three communities, respectively. On the other hand, OMN shows It is well known that local communities can be the highest intercommunity variance for high dispersal rates (d ¼ 1). structured by both local and regional interactions The top species becomes unstable, and goes extinct in two local communities (P < 0.01). (Ricklefs 1987). However, it still remains unknown ik what trophic structures are shared by a set of interacting communities and its dynamical implications for the persistence of biodiversity. immigration (Karlson and Cornell 2002). It has The present study is an attempt to link local and been recently shown that mangroves in the regional food-web structure and dynamics in a Caribbean strongly influence the local community spatially structured marine food web. structure of fish on neighboring coral reefs Communities in five habitats of the Caribbean (Mumby et al. 2004). Additionally, empirical have shown significantly similar trophic structures studies have shown that dispersal among habitats which suggest that these communities are open to and local species interactions are key factors for 24 AQUATIC FOOD WEBS metacommunity structure (Shurin 2001; Cottenie In summary, the similarity in the trophic mod- et al. 2003; Kneitel and Miller 2003; Cottenie ules reported here suggests a strong link among and De Meester 2004), and the persistence of local the spatially structured communities. The level of and regional diversity (Mouquet and Loreau 2003). connectivity among these local communities and However, it still remains unclear how the interplay the type of trophic modules alter local abundance between dispersal and more complex trophic of species and promote local changes in diversity. structures affects species persistence in local com- It still remains unexplored how the results here munities (Carr et al. 2002; Kneitel and Miller 2003). presented change by the introduction of a larger In the present work, closed communities (d ¼ 0) number of interacting modules in a set of spatially with tri-trophic FCs showed an extreme variation structured communities. Our result predicts a in local abundances for both the basal and top relative stability in the composition of basal species (Figure 2.2(a)). On the other hand, OMN species, and a dramatic influence in the abundance shows the highest intercommunity variance for of top species depending on the connectivity high dispersal rates (d ¼ 1). The top species (i.e. dispersal) among distinct habitats. becomes unstable, and goes extinct in two local communities (Figure 2.2(b)). Recent empirical stu- Acknowledgments dies have shown that increasing dispersal fre- quency in intermediate species decreases the We thank the editors of this book for inviting us variance among local communities (Kneitel and to contribute this chapter. We thank Miguel A. Miller 2003), a pattern consistent with theoretical Fortuna and Mayte Valenciano for their useful results presented here (see dotted line in Figure comments on a previous draft. Funding was 2.2(a) and (b)). Further data synthesis and theore- provided by the Spanish Ministry of Science tical work is needed here to integrate the func- and Technology (Grants REN2003-04774 to JB and tional links between habitats and the local REN2003-00273 to PJ, and Ph.D. Fellowship dynamics of species embedded in food webs. FP2000-6137 to CJM). CHAPTER 3 Role of network analysis in comparative ecosystem ecology of estuaries
Robert R. Christian, Daniel Baird, Joseph Luczkovich, Jeffrey C. Johnson, Ursula M. Scharler, and Robert E. Ulanowicz
Introduction improve our inferences on trophic structure and dynamics. Assessments of trophic structure through eco- Estuaries are excellent ecosystems to test the logical network analysis (ENA) have been done in veracity of the inferences of ENA for three reasons. a wide variety of estuarine and coastal environ- First, more NAs have been conducted on estuaries ments. For example, some have used it to compare than on any other kind of ecosystem. Second, trophic structures within ecosystems focusing estuarine environments are often stressed by on temporal conditions (Baird and Ulanowicz natural and anthropogenic forcing functions. This 1989; Baird et al. 1998) and among ecosystems affords opportunities for evaluating controls on focusing on spatial conditions (e.g. Baird and trophic structure. Third, sampling of estuaries has Ulanowicz 1993; Christensen 1995). These compar- often been extensive, such that reasonable food isons have used carbon or energy as the currency webs can be constructed under different condi- with which to trace the interactions of the food tions of stress. Finally, other modeling approaches webs, although other key elements such as have been used in numerous estuarine ecosystems. nitrogen and phosphorus have also been used in Results of these alternate modeling approaches can ENA (Baird et al. 1995; Ulanowicz and Baird 1999; be compared to those of ENA to test the coherence Christian and Thomas 2003). One of the primary of inferences across perspectives of ecosystem features of ENA is that the interactions are structure and function. These conditions set the weighted. That is, they represent rates of flow of stage for an evaluation of the status of ENA as a energy or matter and not simply their existence. tool for comparative ecosystem ecology. Other kinds of comparisons have been attempted Comparative ecosystem ecology makes valuable less frequently. Effects of currency used to track contributions to both basic ecology and its applica- trophic dynamics has received little attention tion to environmental management. Given the (Christian et al. 1996; Ulanowicz and Baird 1999), critical position of estuaries as conduits for mater- and comparisons of ENA with other modeling ials to the oceans and often as sites of intense approaches are quite rare (Kremer 1989; Lin et al. human activities in close proximity to important 2001). There is a need to expand the applications natural resources, ENA has been used frequently of network analysis (NA) to address specific for the assessment of the effects of environmental questions in food-web ecology, and to use it conditions within estuaries related to management. more frequently to explain and resolve specific Early in the use of ENA in ecology, Finn and management issues. The NA approach must be Leschine (1980) examined the link between fertiliza- combined with other existing methods of identi- tion of saltmarsh grasses and shellfish production. fying ecosystem performance to validate and
25 26 AQUATIC FOOD WEBS
Baird and Ulanowicz (1989) expanded the detail Fath and Patten 1999), provides some of the same accessible in food webs and the consequences of analyses found in NETWRK4 but includes others this increased detail in their seminal paper of based on the theoretical considerations of how seasonal changes within the Chesapeake Bay. systems interact with their environment. Lastly, In 1992, Ulanowicz and Tuttle determined through social NA is beginning to be applied to ecological ENA and field data that the overharvesting of systems. A software package so used is UCINET oysters may have had significant effects on a (www.analytictech.com/ucinet.htm; Johnson et al. variety of aspects of the food web in Chesapeake 2001; Borgatti et al. 2002). Although several Bay. Baird and Heymans (1996) studied the methods and software packages exist for evaluat- reduction of freshwater inflow into an estuary ing weighted food webs, none has been developed in South Africa and noted changes in food-web and validated to an extent to give a good under- structure and trophic dynamics. More recently, standing of the full implications of the variety of Brando et al. (2004) and Baird et al. (2004) evalu- results. ated effects of eutrophication and its symptoms on We have organized this chapter to address the Orbetello Lagoon, Italy, and Neuse River Estuary, use of ENA associated with estuarine food webs in USA, respectively. All of these studies involved the context of comparative ecosystem ecology. comparisons of conditions linked to human Comparisons within and among estuaries are first impacts. considered. ENA provides numerous output The first comprehensive review of the meth- variables, but we focus largely on five ecosystem- odologies and use of ENA, an associated software level variables that index ecosystem activity NETWRK4, and application in marine ecology and organization. We address the ability of was published in 1989 (Wulff et al. 1989). Other recognizing ecosystem-level change and patterns approaches to ENA have been developed and of change through the use of these indices. applied to food webs. The software programs Then we compare several estuarine food webs to ECOPATH and ECOSIM have been used through- budgets of biogeochemical cycling to assess the out the world to address various aspects of aquatic correspondence of these two facets of ecosystems. resources management (see www.ecopath.org/for Again we use these same indices and relate them summary of activities; Christensen and Pauly to indices from the biogeochemical budgeting 1993). In parallel with NETWRK4, ECOPATH was approach of the Land–Ocean Interaction in developed by Christensen and Pauly (1992, 1995) the Coastal Zone (LOICZ) program. How do the and Christensen et al. (2000), based on the original two modeling approaches compare in assessing work of Polovina (1984). The dynamic simulation ecosystems? Finally, comparisons of food-web module, ECOSIM, was developed to facilitate diagrams are problematic if the food webs are the simulation of fishing effects on ecosystems at all complex. Recently, visualization tools (Walters et al. 1997). NETWRK4 and ECOPATH from biochemistry and social networks have include, to various extents, similar analytical tech- been used to portray food webs. We explore this niques, such as input–output analysis, Lindeman new approach in the context of intrasystem trophic analysis, a biogeochemical cycle analysis, comparisons. and the calculation of information-theoretical indices to characterize organization and develop- Estuarine food-web comparisons ment. However, some analyses are unique to each. There are several differences in the input meth- We highlight how food webs are perceived to odology between the NETWRK4 and ECOPATH change or remain stable across a variety of condi- software, which lead to differences in their tions. First, we compare systems temporally from outputs. Heymans and Baird (2000) assessed intra and interseasonal to longer-term changes. these differences in a case study of the northern Within a relatively unimpacted ecosystem, food Benguela upwelling system. Environs analysis, webs may tend to be relatively stable with developed by Patten and colleagues (reviewed by differences among times related to altered, ROLE OF NETWORK ANALYSIS 27 weather-related metabolism and differential of each level represents the ingestion of the growth, migrations and ontogenetic changes in next level as a percentage of the ingestion of populations (Baird and Ulanowicz 1989). Human the focal level. The geometric mean of individual impacts may alter these drivers of change and add level efficiencies is the system’s TE. Ulanowicz new ones. Multiple food-web networks for an has characterized the degree of organization and ecosystem tend to be constructed under common maturity of an ecosystem through a group of sets of rules, facilitating temporal comparisons. information-based indices (Ulanowicz 1986). Then we compare food webs among ecosystems Ascendency/developmental capacity (A/C) is a where major differences may exist in the very ratio of how organized, or mature, systems are, nature of the food webs. Interpreting such differ- where ecosystems with higher values reflect ences is more difficult than intrasystem com- relatively higher levels of organization. Thus, these parisons and must be viewed with more caution. five indices can be used to describe both extensive We have used studies of intersystem comparisons and intensive aspects of food webs. While our where effort was made by the authors to minimize focus is on these indices, we incorporate others as differences in rulemaking and network structure. appropriate to interpret comparisons. Should networks be constructed under different constraints, such as inconsistent rules for aggre- Temporal comparisons gation, the interpretation of differences in the NA results is difficult and should be viewed with more There are surprisingly few estuarine ecosystems caution. for which food-web networks have been examined Ecological network analysis provides a myriad during different times. Most networks represent of output variables and indices. Each has its annual mean food webs. We provide a brief review own sensitivity to differences in network of some for which we have direct experience structure. Generally, indices of population (i.e. at and can readily assess the focal ecosystem-level compartment-level) and cycling structure are more indices. These are ecosystems for which NETWRK4 sensitive than ecosystem-level indices in terms of was applied rather than ECOPATH, because of responsiveness to flow structure and magnitude of some differences in model construction and flows (Baird et al. 1998; Christian et al. 2004). Also, analysis (e.g. general use of gross primary pro- because currency and timescale may differ among duction in NETWRK4 and net primary production networks, direct comparisons using different in ECOPATH). The shortest timescale examined flow currencies are difficult. We focus on five has been for a winter’s Halodule wrightii ecosystem ecosystem-level output variables of ENA, four of in Florida, USA (Baird et al. 1998) where two which are ratios. These are described in greater sequential months were sampled and networks detail elsewhere (Kay et al. 1989; Christian and analyzed. Seasonal differences between food webs Ulanowicz 2001; Baird et al. 2004). The first adds were a central part of the Baird and Ulanowicz all flows within a network, total system through- (1989) analysis of the food web in Chesapeake put (TST), and reflects the size, through activity, Bay. Almunia et al. (1999) analyzed seasonal of the food web. Combinations of flows may be differences in Maspalomas Lagoon, Gran Canaria, interpreted as occurring in cycles, and the per- following the cycle of domination by benthic centage of TST involved in cycling is called the versus pelagic primary producers. Florida Bay, Finn Cycling Index (FCI; Finn 1976). The turnover which constitutes the most detailed quantified rate of biomass of the entire ecosystem can be network to date, has been analyzed for seasonal calculated as the sum of compartment production differences (Ulanowicz et al. 1999). Finally, inter- values divided by the sum of biomass (P/B). decadal changes, associated with hydrological Networks can be collapsed, mathematically into a modifications, were assessed for the Kromme food chain, or Lindeman Spine, with the proces- Estuary, South Africa (Baird and Heymans 1996). sing of energy or matter by each trophic level iden- Table 3.1 shows the five indices for each temporal tified (Ulanowicz 1995). The trophic efficiency (TE) condition for these ecosystems. 28 AQUATIC FOOD WEBS
Table 3.1 Temporal changes in ecosystem-level attributes for different estuarine ecosystems
Time period TST (mg C m 2 per day) FCI (%) P/B (day 1) TE (%) A/C (%)
St Marks, intraseasonal January 1994 1,900 16 0.037 4.9 36 February 1994 2,300 20 0.041 3.3 32 Neuse, intraseasonal Early summer 1997 18,200 14 0.15 5.0 47 Late summer 1997 17,700 16 0.30 4.7 47 Early summer 1998 18,600 16 0.24 3.3 47 Late summer 1998 20,700 16 0.33 4.9 46 Chesapeake, interseasonal Spring 1,300,000 24 n.a. 9.6 45 Summer 1,700,000 23 n.a. 8.1 44 Fall 800,000 22 n.a. 10.9 48 Winter 600,000 23 n.a. 8.6 49 Maspalomas Lagoon, interseasonal Benthic-producer-dominated system 13,600 18 n.a. 11.4 40 Transitional 12,300 23 n.a. 12.8 38 Pelagic-producer-dominated system 51,500 42 n.a. 8.7 45 Florida Bay, interseasonal Wet 3,460 26 n.a. n.a. 38 Dry 2,330 n.a. n.a. n.a. 38 Kromme, interdecadal 1981–84 42,830 12 0.012 4.5 48 1992–94 45,784 10 0.011 2.8 46
Note: Flow currency of networks is carbon; n.a. means not available.
Ecological network analysis was applied to a and a 12% increase in P/B. However, organization winter’s H. wrightii ecosystem, St Marks National of the food web (A/C) decreased, and dissipation Wildlife Refuge, Florida, USA (Baird et al. 1998; of energy increased lowering the TE. Although Christian and Luczkovich 1999; Luczkovich et al. statistical analysis of these changes was not done, it 2003). Unlike most applications of ENA, the field would appear that the indices do reflect perceived sampling design was specific for network con- effects of increased metabolism. struction. From these data and from literature The food web of the Neuse River Estuary, NC, values, the authors constructed and analyzed one was assessed during summer conditions over two of the most complex, highly articulated, time-and years (Baird et al. 2004; Christian et al. 2004). The site-specific food-web networks to date. Two Neuse River Estuary is a highly eutrophic estuary sequential months within the winter of 1994 were with high primary production and long residence sampled with the temperature increase of 5 C times of water. Temperature was not considered to from January to February. Metabolic rates, calcu- differ as dramatically from early to late summer, lated for the different temperatures and migrations but two major differences distinguished early and of fish and waterfowl affected numerous attributes late summer food webs. First was the immigration of the food webs (Baird et al. 1998). The changes in and growth of animals to the estuary during the focal indices are shown in Table 3.1. Activity summer, which greatly increased the biomass of estimated by the three indices was higher during several nekton compartments. Second, hypoxia the warmer period with >20% more TST, and FCI, commonly occurs during summer, stressing both ROLE OF NETWORK ANALYSIS 29 nekton and benthos. Hypoxia was more dramatic increased during the pelagic phase (Table 3.1). in 1997 (Baird et al. 2004). Benthic biomass The proportional increase in TST could be inter- decreased during both summers, but the decrease preted as eutrophication, but the system has no big was far more dramatic during the year of more sources of material input from outside the system. severe hypoxia. Changes in the ecosystem-level Almunia et al. (1999) explained the increase in A/C indices were mostly either small or failed to show as a shift in resources from one subsystem the same pattern for both years (Table 3.1). A/C (benthic) to another (pelagic). The FCI was lowest changed little over summers or across years. TST, during the benthic-dominated stage and highest FCI, and TE had different trends from early to late during the pelagic-dominated stage, and matter summer for the two years. Only P/B showed was cycled mainly over short fast loops. The relatively large increases from early to late pelagic-dominated stage was interpreted as being summer. Thus, inferences regarding both activity in an immature state, but this interpretation is and organization across the summer are not counter to the highest A/C during the pelagic readily discerned. We have interpreted the results stage. The average TE dropped from the benthic- to indicate that the severe hypoxia of 1997 reduced dominated stage to the pelagic-dominated stage, the overall activity (TST) by reducing benthos and and the ratio of detritivory to herbivory increased their ability to serve as a food resource for nekton. accordingly. Highest values of detritivory coin- But these ecosystem-level indices do not demon- cided with lowest values of TE. strate a stress response as effectively as others Florida Bay showed remarkably little change considered by Baird et al. (2004). in whole-system indices between wet and dry The food web in Chesapeake Bay was analyzed seasons (Ulanowicz et al. 1999; www.cbl.umces. for four seasons (Baird and Ulanowicz 1989). Many edu/ bonda/FBay701.html). Although system-level of the changes linked to temperature noted for the indices during the wet season were about 37% within-season changes of the food web in St Marks greater than the same indices during the dry hold here (Table 3.1). TST and P/B are highest in season, it became apparent that this difference was summer and lowest in winter, although the other almost exclusively caused by the change in system measure of activity, FCI, does not follow this activity (measured as TST), which was used to pattern. However, FCI is a percentage of TST. scale the system-level indices to the size of the The actual amount of cycled flow (TST FCI) does system. The fractions of A/C and the distribution follow the temperature-linked pattern. TE failed to of the different components of the overhead were show a pattern of increased dissipation with almost identical during both seasons. Ulanowicz higher temperatures, although it was lowest et al. (1999) concluded that the Florida Bay eco- during summer. Organization, as indexed by A/C, system structure is remarkably stable between the showed the greatest organization in winter and two seasons. (FCI was high during the wet season least in summer. Hence, in both of the aforemen- (>26%) but could not be calculated for the dry tioned examples, times of higher temperature and season since the computer capacity was exceeded therefore, higher rates of activity and dissipation by the amount of cycles (>10 billion).) of energy were linked to transient conditions of Lastly, we consider a larger timescale of a decreased organization. These findings are corrob- decade for the Kromme Estuary, South Africa. orated for spring—summer comparisons of these Freshwater discharge to this estuary was greatly food webs are discussed later in the chapter. reduced by 1983 due to water diversion and Maspalomas Lagoon, Gran Canaria, shows, over damming projects, greatly lessening nutrient the year, three successive stages of predominance additions, salinity gradients, and pulsing (i.e. flood- of primary producers (Almunia et al. 1999). The ing; Baird and Heymans 1996). Can ecosystem- system moves from a benthic-producer-dominated level indices identify resultant changes to the food system via an intermediate stage to a pelagic- web? Although there was a slight increase in TST, producer-dominated system. The analysis of the trend was for a decrease in all other measures system-level indices revealed that TST and A/C (Table 3.1). However, all of these were decreases of 30 AQUATIC FOOD WEBS less than 20%, with the exception of TE. This and processes among systems of different geo- general, albeit slight, decline has been attributed to graphic locations, ranging from studies on estuaries the stress of the reduced flow regime (Baird and in relatively close proximity (Monaco and Heymans 1996). The TE decreased during the Ulanowicz, 1997; Scharler and Baird, in press) decade to less than half the original amount. to those of estuarine/marine systems spanning Thus, much less of the primary production was three continents (Baird et al. 1991). Perhaps inferred to pass to higher, commercially important, the most extensive comparison has been done trophic levels. Further, TE of the Kromme under a by Christensen (1995) on ecosystems using reduced flow regime was among the smallest for ECOPATH to evaluate indices of maturity. These ecosystems reviewed here. comparisons are limited, as discussed previously, In summary, most temporal comparisons because of differences in rules for constructing and considered were intra-annual, either within or analyzing networks. We review here some of the among seasons. Seasons did not have comparable estuarine and coastal marine comparisons that meaning among ecosystems. The Chesapeake have taken into account these issues, beginning Bay networks were based on solar seasons, but with our focal indices. Maspalomas Lagoon and Florida Bay networks The geographically close Kromme, Swartkops, were not. All indices demonstrated intra-annual and Sundays Estuaries, differ in the amount of change, although the least was associated with A/C. freshwater they receive, and consequently in the This is to be expected as both A and C are logar- amount of nutrients and their habitat structure ithmically based indices. Summer or warmer (Scharler and Baird 2003). Input–output analysis seasons tend to have higher activity (TST and FCI highlighted the differences in the dependencies or FCI TST), as expected. In some cases this was (or extended diets) of exploited fish and inverteb- linked to lower organization, but this was not rate bait species. Microalgae were found to play an consistent across systems. We only include one important role in the Sundays Estuary (high interannual, actually interdecadal, comparison, but freshwater and nutrient input) as a food source to differences within a year for several systems were exploited fish and invertebrate bait species, as great as those between decades for the Kromme whereas detritus and detritus producers were of Estuary. Interannual differences in these indices comparatively greater importance in the Kromme for other coastal ecosystems have been calculated (low nutrients) and Swartkops (pristine freshwater but with different currencies and software (Brando inflow, high nutrients) Estuaries (Scharler and et al. 2004; others). Elmgren (1989) has successfully Baird, in press). used trophic relationships and production esti- When comparing some indicators of system mates to assess how eutrophication of the Baltic performance such as TST, FCI, A/C, and TE of Sea over decades of enhanced nutrient loading has the Kromme, Swartkops, and Sundays Estuaries, modified production at higher trophic levels. Even it revealed an interplay between the various though the sample size remains small, it appears degrees of physical and chemical forcings. The that intra-annual changes in food-web structure Kromme Estuary is severely freshwater starved and trophic dynamics can equal or exceed those and so lacks a frequent renewal of the nutrient across years and across different management pool. Freshets have largely disappeared as a phy- regimes. Obviously, more examples and more sical disturbance. The Sundays system features thorough exploration of different indices are nee- increased freshwater input due to an interbasin ded to establish the sensitivities of ecosystem-level transfer, and the Swartkops Estuary has a relat- indices to uncertainties in ecosystem condition. ively pristine state of the amount of freshwater inflow but some degree of anthropogenic pollution Interecosystem comparisons (Scharler and Baird, in press). NA results showed that the Swartkops was more impacted Ecological network analysis has been used in inter- due to a low TST and a high average residence system comparisons to investigate the structure time (ART, as total system biomass divided by ROLE OF NETWORK ANALYSIS 31 total outputs) of material and least efficient to pass estuarine and marine upwelling systems. They on material to higher trophic levels. The Kromme used a slightly different approach, in that only the was more self-reliant (higher FCI) than the Sundays residual flow matrices (i.e the straight through (lowest FCI). The Sundays was also the most active flows) were used for this calculation, since the featuring a comparatively high TST and low ART. cycled flows were believed to inflate the inputs However, a comparatively high ascendency in the to the various end compartments. In this study, Sundays Estuary was not only a result of the high the most productive systems in terms of producing TST, which could have implicated the con- planktivorous fish from a unit of primary pro- sequences of eutrophication (Ulanowicz 1995a), duction were the upwelling systems (Benguela and but it also featured the highest AMI (the informa- Peruvian) and the Swartkops Estuary, compared to tion-based component of ascendency)(Scharler and the Baltic, Ems, and Chesapeake (Baird et al. 1991). Baird, in press). The Kromme Estuary had the In terms of carnivorous fish, the Benguela upwel- comparatively lowest A/C, and lowest AMI, ling system was the most efficient, followed by the and Baird and Heymans (1996) showed that since Peruvian and Baltic (Baird et al. 1991). the severe freshwater inflow restrictions, a decline Trophic efficiencies have also been used to make of the internal organization and maturity was assumptions about the productivity of a system. apparent. In perhaps the first intersystem comparison using Intercomparisons of estuaries and coastal aquatic NA, Ulanowicz (1984) considered the efficiencies ecosystems have often focused on other issues with which primary production reached the top in addition to the focal indices of this chapter. predators in two marsh gut ecosystems in Crystal One important issue has been the secondary River, Florida. Monaco and Ulanowicz (1997) production of ecosystems, which is of special identified that fish and macroinvertebrate catches interest in terms of commercially exploited species. in the Chesapeake Bay were higher compared to As Monaco and Ulanowicz (1997) stated, there can the Narragansett and Delaware Bay, despite be differences in the efficiency of the transforma- its lower system biomass, because the transfer tion of energy or carbon from primary production efficiencies between trophic levels were higher. to the commercial species of interest. By relating Similarly, transfer efficiencies calculated from the output of planktivorous and carnivorous fish, material flow networks were used to estimate the and that of suspension feeders to primary produc- primary production required to sustain global tion, it became apparent that in Narragansett Bay fisheries (Pauly and Christensen 1995). Based on twice as many planktivorous fish and 4.6–7.4 as a mean energy transfer efficiency between trophic many carnivorous fish were produced per unit levels of 48 ecosystems of 10%, the primary primary production than in Delaware or Chesapeake production required to sustain reported catches Bay, respectively. The latter, on the other hand, and bycatch was adjusted to 8% from a previous produced 1.3 and 3.5 as much suspension feeding estimate of 2.2%. biomass than Narragansett and Delaware Bay In the context of the direct and indirect diet of from one unit of phytoplankton production exploited and other species, it can be of interest (Monaco and Ulanowicz 1997). This analysis was to investigate the role of benthic and pelagic performed on the diet matrix to quantify a con- compartments. The importance of benthic pro- tribution from a compartment (in this case the cesses in the indirect diet of various age groups primary producers) in the network to any other, of harvestable fish was determined with input– over all direct and indirect feeding pathways, and output analysis by Monaco and Ulanowicz (1997). is described as part of an input–output analysis in The indirect diet is the quantified total consump- Szyrmer and Ulanowicz (1987). tion by species j that has passed through species i This approach of tracing the fate of a unit along its way to j (Kay et al. 1989). They found of primary production through the system was that benthic processes in the Chesapeake Bay also applied by Baird et al. (1991) who calculated was highly important to particular populations of the fish yield per unit of primary production in juvenile and adult piscivores. Indirect material 32 AQUATIC FOOD WEBS transfer effects revealed that the Chesapeake Bay pointed out that the system P/B ratio is not relied more heavily on its benthic compartments necessarily a reflection of the maturity of the sys- compared to the Narragansett and Delaware Bays tem, but due to NA results reinterpreted maturity and that disturbances to benthic compartments in the context of physical forcing (e.g. the upwel- may have a comparatively greater impact on the ling systems (Peruvian, Benguela) are considered system (Monaco and Ulanowicz 1997). The pattern to be mature under their relatively extreme phy- changes somewhat with season, as discussed in sical forcings, although they have a higher system the section, ‘‘visualization of network dyna´mics’’ P/B ratio than the estuarine systems (Chesapeake, of this chapter. Ems, Baltic, Swartkops)). The shallow Kromme, Swartkops, and Sundays Comparison of whole-system indices between Estuaries were found to rely more on their benthic the Chesapeake and Baltic ecosystems provided biota in terms of compartmental throughput and managers with a surprise (Wulff and Ulanowicz the total contribution coefficients in terms of 1989). The conventional wisdom was that the compartmental input (Scharler and Baird, in Baltic, being more oligohaline than the Chesapeake, press). In terms of carbon requirements, the would be less resilient to stress. The organ- Kromme and Swartkops Estuaries depended two- izational status of the Baltic, as reflected in the third on the benthic components and one-third on relative ascendency (A/C) was greater (55.6%) the pelagic components, whereas the Sundays than that of the Chesapeake (49.5%) by a significant Estuary depended to just over half on its benthic amount. The relative redundancy (R/C) of the components. The Sundays Estuary was always Chesapeake (28.1%) was correspondingly greater perceived to be ‘‘pelagic driven,’’ probably due to than that of the Baltic (22.0%), indicating that the high phyto and zooplankton standing stocks, the Chesapeake might be more stressed than the which are a result of the regular freshwater and Baltic. The FCI in the Chesapeake was higher nutrient input. By considering not only direct (30%) than in the Baltic (23%). As greater cycling is effects, but also all indirect effects between the indicative of more mature ecosystems (Odum compartments, the regular freshwater input sup- 1969), this result seemed at first to be a counter- pressed somewhat the dependence on benthic indication that the Chesapeake was more stressed, compartments, but has not switched the system to but Ulanowicz (1984) had earlier remarked that a predominantly pelagic dependence (Scharler and a high FCI could actually be a sign of stress, Baird, in press). especially if most of the cycling occurs over short Indicators of stress, as derived from ENA, have cycles near the base of the trophic ladder. This been discussed in several comparative studies. was also the case in this comparison, as a decom- Baird et al. (1991) proposed a distinction between position of cycled flow according to cycle length physical stress and chemical stress. The former has revealed that indeed most of the cycling in the in general been influencing ecosystems, such as Chesapeake occurred over very short cycles (one upwelling systems, for a time long enough so that or two components in length), whereas recycle the systems themselves could evolve under the over loops that were three or four units long was influence of this type of physical forcing. Fresh- significantly greater in the Baltic. The overall water inflow into estuaries similarly determines picture indicated that managerial wisdom had the frequency of physical disturbance, due to been mistaken in this comparison, as the saltier frequent flooding in pristine systems and restric- Chesapeake was definitely more disrupted than tions thereof in impounded systems. On the other the Baltic. hand, chemical influences are in general more recent through anthropogenic pollution, and the Intermodel and technique comparisons systems are in the process of changing from one response type (unpolluted) to another (polluted) Another modeling protocol was developed that adjusts to the chemical type of forcing (Baird under the auspices of the International Geosphere– et al. 1991). With this perspective, Baird et al. (1991) Biosphere Program (IGBP), an outcome of the 1992 ROLE OF NETWORK ANALYSIS 33
Rio Earth Summit and established in 1993. The inorganic P), one can derive whether (1) an estuary aims of the IGBP are ‘‘to describe and understand is a sink or a source of C, N, and P, that is D the physical, chemical and biological processes Y ¼ fluxout fluxin, where Y ¼ C, N, or P; (2) that regulate the earth system, the environment the system’s metabolism is predominantly auto- provided for life, the changes occurring in the trophic or heterotrophic, that is, (p r) ¼ D system, and the influence of human actions.’’ DIP(C : P)part, where (p r) is photosynthesis In this context, the Land Ocean Interactions in the minus respiration; and (3) nitrogen fixation (nfix) Coastal Zone (LOICZ) core project of the IGBP was or denitrification (denit) predominates in the sys- D D established. LOICZ focuses specifically on the tem, where (nfix denit) ¼ DIN DIP(N : P)part functioning of coastal zone ecosystems and their (Gordon et al. 1996). A summary of attributes for role in the fluxes of materials among land, sea, and this modeling approach is shown in Table 3.2. atmosphere; the capacity of the coastal ecosystems This section explores the possibility of linkages to transform and store particulate and dissolved between the two different methodologies of ENA matter; and the effects of changes in external and LOICZ biogeochemical budgeting protocol. forcing conditions on the structure and functioning The rationale for this hypothesis is: of coastal ecosystems (Holligan and de Boois 1993; Pernetta and Milliman 1995). The magnitude and frequency of N and P loadings and The LOICZ biogeochemical budgeting proce- the transformation of these elements within the system, ultimately affect the system’s function. Since we postulate dure was subsequently developed that essentially that system function is reflected in network analysis consists of three parts: budgets for water and salt outputs, we infer that there should exist correspondence movement through coastal systems, calculation of in the biogeochemical processing, as indexed by the rates of material delivery (or inputs) to and LOICZ approach, and trophic dynamics, as indexed by removal from the system, and calculations of rate network analysis outputs. of change of material mass within the system (particularly C, N, and P). Water and salt are To do this, we used ENA and LOICZ variables considered to behave conservatively, as opposed and output results from six estuarine or brackish to the nonconservative behavior of C, N, and P. ecosystems based on input data with a high level Assuming a constant stoichiometric relationship of confidence (Table 3.3). We first performed (e.g. the Redfield ratio) among the nonconservative Spearman’s and Kendall’s correlation analyses nutrient budgets, deviations of the fluxes from the between the ENA and LOICZ output results of a expected C : N : P composition ratios can thus be number of system indices of the six ecosystems. assigned to other processes in a quantitative From the correlation matrices we selected those fashion. Using the flux of P (particularly dissolved variables which showed correlation values of 80%
Table 3.2 System properties and variables derived from NA and the LOICZ biogeochemical budgeting protocol used in factor analysis
LOICZ variables Description of variable/system property
Nutrient loading From land to ocean, two macronutrients and their possible origins Dissolved inorganic nitrogen (DIN) (mol m 2 per year) Products of landscape biogeochemical reactions Dissolved inorganic phosphorus (DIP) (mol m 2 per year) Materials responding to human production, that is, domestic (animal, human) and industrial waste, and sewage, fertilizer, atmospheric fallout from vehicular and industrial emissions 2 DIN (mol m per year) Fluxout Fluxin 2 DIP (mol m per year) Fluxout Fluxin Net ecosystem metabolism (NEM) (mol m 2 per year) Assumed that the nonconservative flux of DIP is an approximation of net metabolism: (p r) ¼ DIP(C : P) NFIXDNIT (mol m 2 per year) Assumed that the nonconservative flux of DIN approximates N fixation minus denitrification: (nfix denit) ¼ DIN DIP(N : P) 34 AQUATIC FOOD WEBS
Table 3.3 Ecosystem-level attributes used for comparison of NA results of estuarine food webs with biogeochemical budgeting models
System Morphology ENA
Volume (m3) Area (m2) A/C (%) TST (mgC m 2 per day) FCI (%) P/B (day 1) TE (%)
Kromme 9.00Eþ06 3.00Eþ06 33.7 13,641 26 0.73 6.2 Swartkops 1.20Eþ07 4.00Eþ06 28 11,809 44 3.65 4 Sundays 1.40Eþ07 3.00Eþ06 43 16,385 20 10.95 2.6 Baltic Sea 1.74Eþ13 3.70Eþ11 55.6 2,577 23 29.2 16.2 Cheasapeake 3.63Eþ10 5.90Eþ09 49.5 11,224 23 51.1 9 Neuse 1.60Eþ09 4.60Eþ08 46.8 11,222 15.4 94.9 4.5
LOICZ models (mol m 2 per year)