Biodiversity in Drylands: Toward a Unified Framework (Long-Term Ecological Research Network Series)

Biodiversity in Drylands: Toward a Unified Framework Moshe Shachak, et al., Editors OXFORD UNIVERSITY PRESS BIODIVER...

Author: Moshe Shachak | James R. Gosz | Stewart T. A. Pickett | Avi Perevolotsky

49 downloads 487 Views 4MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

Biodiversity in Drylands: Toward a Unified Framework

Moshe Shachak, et al., Editors

OXFORD UNIVERSITY PRESS

BIODIVERSITY IN DRYLANDS

LONG-TERM ECOLOGICAL RESEARCH NETWORK SERIES LTER Publications Committee Climate Variability and Ecosystem Response at Long-Term Ecological Research Sites Edited by David Greenland, Douglas G. Goodin, and Raymond C. Smith Grassland Dynamics: Long-Term Ecological Research in Tallgrass Prairie Edited by Alan K. Knapp, John M. Briggs, David C. Hartnett, and Scott L. Collins Standard Soil Methods for Long-Term Ecological Research Edited by G. Philip Robertson, David C. Coleman, Caroline S. Bledsoe, and Phillip Sollins Structure and Function of an Alpine Ecosystem: Niwot Ridge, Colorado Edited by William D. Bowman and Timothy R. Seastedt Biodiversity in Drylands: Toward a Unified Framework Edited by Moshe Shachak, James R. Gosz, Avi Perevolotsky, and Steward T.A. Pickett

BIODIVERSITY IN DRYLANDS: TOWARD A UNIFIED FRAMEWORK Edited by

Moshe Shachak James R. Gosz Steward T.A. Pickett Avi Perevolotsky

1 2005

1 Oxford New York Auckland Bangkok Buenos Aires Cape Town Chennai Dar es Salaam Delhi Hong Kong Istanbul Karachi Kolkata Kuala Lumpur Madrid Melbourne Mexico City Mumbai Nairobi Saõ Paulo Shanghai Taipei Tokyo Toronto

Copyright # 2005 by Oxford University Press, Inc. Published by Oxford University Press, Inc., 198 Madison Avenue, New York, New York 10016 www.oup.com Oxford is a registered trademark of Oxford University Press 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, electronic, mechanical, photocopying, recording, or otherwise, without the prior permission of Oxford University Press. Library of Congress Cataloging-in-Publication Data Biodiversity in drylands: toward a unified framework/ edited by Moshe Shachak . . . [et al.]. p. cm—(Long-Term Ecological Research Network series) ISBN 0-19-513985-2 1. Arid regions ecology. 2. Biological diversity. I. Shachak, Moshe. II. Series. QH541.5.A74B56 2004 577.54—dc22 2003015271

9 8 7 6 5 4 3 2 1 Printed in the United States of America on acid-free paper

In Memory of Gary Allan Polis (28 August 1946–27 March 2000) Robert D. Holt Wendy B. Anderson

G

ary Polis loved deserts. Whenever he dealt with the topics of his major contributions to community ecology—the prevalence of intraguild predation and omnivory; the ubiquity of reticulate food web structures, with many weak and donor-controlled links; and the importance of allochthonous subsidies, detrital pathways, and temporal variability in food web dynamics—he would invariably lace the discussion with concrete examples from desert ecosystems. Among his many books is The Ecology of Desert Communities (1991), which in many ways can be viewed as a natural predecessor to the current volume. Gary’s boyhood fascination with deserts and scorpions broadened into a detailed understanding of food web interactions, providing an entreé to the ecological community, where he became a leader in food web ecology. It is in this arena that he made the greatest contributions to ecology, and this is one clear instance where an understanding of desert ecology has led to conceptual advances in ecology as a whole. Gary was a superior naturalist. He could develop a sense about a place because he spent huge amounts of time in the deserts he studied. He was accomplished at making acute observations about patterns, and relating them to broader ecological concepts. For example, while collecting data to describe patterns of scorpion diversity and distribution on desert islands in the Gulf of California, he noticed that patterns of spider and lizard abundances and distributions seemed to covary with scorpion distribution patterns. He began to understand the web of interactions that could exist among these three higher-level consumers. Furthermore, he became aware that all of these patterns were intricately tied to the influences of the

vi In Memory of Gary Allan Polis

surrounding ocean on the food webs of the desert islands, via the subsidy of their ecosystems by materials drawn from the marine environments. Gary’s emphasis on the ubiquity and importance of such subsidies was a major contribution made in his last few years. Again, Gary tapped his passion for desert ecology to inform at a deeper level for the discipline as a whole. The consummate educator, Gary taught everybody: undergraduate and graduate students, colleagues, his own children, dozens of volunteers who assisted him in the field, virtually anyone who would listen. Three generations of academic offspring have benefited from Gary’s wonderful insights and worldview directly or indirectly, and his ideas and perspectives continue to resonate through ecology today. Finally, Gary had an amazing talent for bringing together people who might not otherwise interact to create novel syntheses among previously disparate disciplines. His appreciation of people from diverse backgrounds (scientific and cultural) drew people to him and hence to each other. He ran his own research laboratory in an intellectually inclusive style—inviting students and post-docs with interests in plants, invertebrates, vertebrates, or soils, and ranging from taxonomy to physiology to landscape ecology–aiming toward a synthetic understanding of the desert systems he so loved. In particular, Gary believed that the interplay of temporal variability and spatial heterogeneity was fundamental to understanding desert communities. Valuing biological diversity, scientific diversity, and the diversity of human perspectives alike. Gary recognized that any healthy assembly of species—or people—must include numerous functional groups to thrive. In this volume, we acknowledge that we are poorer for his loss, but richer for his having been among us.

Reference Polis, G.A. (ed.). 1991. The Ecology of Desert Communities. University of Arizona Press, Tucson, Arizona.

Preface

Why Study Biodiversity in Drylands? Many international conventions have identified the primary environmental problems of the world as being urbanization, global change, desertification, and biodiversity (Vitousek 1994). These important environmental problems are interrelated and dramatically expressed in the world’s drylands. While all are important and actively studied topics, the role and significance of biodiversity in drylands is the least well understood scientifically topic. Biodiversity was coined as a term to promote public and political dialog about the state and future of the world’s biological richness. It is regarded as both a social–political construct and a scientific concept (Gaston 1996). While its history ensures social significance, it leaves the concept devoid of the theory and body of knowledge that typifies scientific disciplines. The need for the development of biodiversity as a scientific discipline is clear. First, although the concept is general, and appropriately applied to ecological realms ranging from the genetic to the ecosystem and landscape (Noss and Cooperider 1994), it has most often been tacitly restricted to the topic of species. Management, especially, requires the use of biodiversity to motivate and organize studies of ecosystems and landscapes. Second, while there is a rigorous general definition that identifies the core aspects of biodiversity as number and difference in ‘‘biological entities,’’ how this definition can and should be specified to the range of ‘‘ecological entities’’ is unclear. Third, the function of biodiversity must be determined in a wide variety of environments. Biodiversity is an important topic that requires a well-developed theory, and a clear strategy for application to management.

viii Preface

The aim of this book is to elucidate the scientific basis for biodiversity studies and management. We emphasize biodiversity as a powerful, integrative concept, but one that still requires careful articulation and application. Even though this book utilizes many case studies from drylands, it emphasizes the generality of the biodiversity concept. Drylands are experiencing an accelerating rate of change, mainly due to shifts in land use and climate change initiated by humans. These changes affect the distribution and abundance of species, habitats, and ecosystems, thereby creating new landscape mosaics. Biodiversity includes the diversity of organisms in complex assemblages of interacting communities and ecosystems, as changes in global systems accelerate, changes in dryland biodiversity are also accelerating. Understanding biodiversity changes in these systems requires the development of general guiding principles for the study and management of biodiversity of drylands. In order to address the problem we organized a workshop, ‘‘Biodiversity in Drylands: Toward a Unified Framework for Research and Management’’ (26 June–2 July 1999), at the Blaustein Institute for Desert Research of Ben Gurion University, Israel. The aim of the workshop was to confront three main problems that emerge as a consequence of the broad scope of the biodiversity concept. The first problem addressed how to incorporate processes (e.g., foraging, energy and nutrient flows, patch dynamics) into a concept that is based on entities (e.g., individual organisms, species, habitat types, patch types). The second involved how to integrate ecological subdisciplines (e.g., ecosystem, population, landscape ecology) that are involved in biodiversity studies. The third included how to use a theoretical framework that incorporates ecological processes and entities, and integrates across subdisciplines as a guideline for conservation, restoration, and management of biodiversity. We defined a number of objectives for the workshop participants: 1. Develop a conceptual framework that can integrate studies of biodiversity in a network of research and management sites in drylands. 2. Evaluate the state of knowledge of dryland biodiversity and pose questions to guide future research and management. 3. Generate new concepts and ideas needed for a theory and management of biodiversity in drylands. 4. Publish the findings in a book on biodiversity studies in drylands aimed at students, scientists, managers, and educators. This book represents the ideas, methodologies, and applications stimulated by workshop presentations, discussions, and group efforts. It addresses the question how diversity of entities in ecological systems, through their webs of interactions, affect the performance of the ecosystem. We believe that by addressing this question we have helped advance the aim of seeking a unified framework for biodiversity studies and management in many types of habitats. A central theme of the book is the relationship between the diversity of organisms and landscapes and the structure and function of the ecosystem.

Preface

ix

We have integrated processes and entities in biodiversity studies based on the relationship between primary production, food web interactions, community processes, resource distribution, habitat structures and species diversity, in relation to scale. We suggest general guidelines for integration across ecological subdisciplines in biodiversity studies. The integration is based on the sets of relationships between organisms as ecosystem engineers and species diversity, microbial and ecosystem processes, species and ecosystem processes, and landscape processes and species diversity. The applied aspects of this book suggest utility of the biodiversity concept for conservation, restoration, and management. The topics encompass how rangeland management and water harvesting support sustainability of biodiversity.

Acknowledgments We are grateful to a number of organizations and individuals whose efforts enabled us to hold our workshop and publish this book. An important organization that assists in studying biodiversity in drylands is the International Arid Lands Consortium (IALC). The IALC was very generous in its support for the workshop in Israel and for the agendum leading to the publication of this book. The IALC supports projects that are intended to lead to a better understanding of the management of fragile dryland ecosystems for sustainable human use. Their policy dictates that understanding biodiversity is critical for managers attempting to achieve the goals of maintenance of ecosystem function, optimizing yield of valuable plant and animal species, or maintaining niches for threatened or endangered species. The IALC experience in encouraging the linkage of biodiversity research with management needs illustrates the problems and potentials of addressing human needs more directly through research in biodiversity. The IALC is a partnership of organizations established in 1990 and dedicated to research, education, and training relative to development, management, restoration, and reclamation of arid and semiarid lands with a primary focus on the Middle East and the Southwestern United States. Fifty research and development projects have been funded from 1993 through 2000, of which 24 relate to a variety of biodiversity issues and concerns. A listing of those projects funded by the IALC since 1993 can be found in Hegwood (2000) and on the website http://ag.arizona.edu/OALS/IALC/Home.html. These projects dealt with the role of landscape and species diversities in the function of water-limited systems. In addition, the IALC supports demonstration projects that represent applications of available knowledge and technologies derived from research and development efforts for the management of sustainable ecological systems. The IALC provides unique opportunities to foster international collaboration for basic and applied research. Sponsors are particularly interested in the

x Preface

transfer of technology to citizens of countries most in need of this technology and in promoting cooperation and peaceful interaction among neighboring nations in the politically troubled Middle East. Please visit the IALC website for more information. We want to thank Ben Gurion University of the Negev for generous support and for providing campus facilities for the workshop; also the Blaustein Institute for Desert Research, and especially the Blaustein Center for Scientific Cooperation, for financial support and providing use of the Sede Boqer facilities for the workshop. We are indebted to the Jewis National Fund for their financial support. We are also grateful to Patty Sprott at the Long Term Ecological Research Network Office in Albuquerque, New Mexico for support with managing and editing the volume. We appreciate the fact that Oxford University Press agreed to publish our book and thank them for their patience. Many individuals assisted in helping this effort become a reality. We are grateful to them for that. They include Menachem Sachs for his encouragement and help in organizing support for the workshop and book; Andy Wilby and Yoram Ayal for helping in the organization and logistics of the workshop, which was significant for the success of the workshop; Bertrand Boeken, Yarden Oren, and David Ward for their help in planning and carrying out the field trips during the workshop; Yael Kaplan for her secretarial assistance; Bob Waide who agreed to be the link between Ben Gurion University and ILTER; and Sol Brand for helping in editing aspects of the book. In conclusion, we hope that this book will be beneficial to students, scientists, managers, and educators who are concerned with biodiversity issues. We believe that greater understanding of biodiversity of drylands, which occupy 40% of the world’s terrestrial systems and are extremely sensitive to desertification, will be the outcome of this book.

References Gaston, K.J. (ed.) 1996. Biodiversity. A biology of numbers and difference. Blackwell Science, Oxford, UK. Hegwood, D.A. (compiler). 2000. International Arid Lands Consortium: A compendium of funded projects. International Arid Lands Consortium, Tucson, Arizona. Noss, R.F., and A.Y. Cooperider. 1994. Saving Nature’s Legacy: Protecting and Restoring Biodiversity. Island Press, Washington, DC. Vitousek, P.M. 1994. Beyond global warming: ecology and global change. Ecology 75: 1861–1876.

Contents

Contributors 1

xv

Introduction A Framework for Biodiversity Studies

3

Moshe Shachak, James R. Gosz, Avi Perevolotsky, and Steward T.A. Pickett

I

Living Components of Biodiversity: Organisms

2

How Can High Animal Diversity Be Supported in Low-Productivity Deserts? The Role of Macrodetritivory and Habitat Physiognomy

15

Yoram Ayal, Gary A. Polis, Yael Lubin, and Deborah E. Goldberg

3

Biodiversity Along Core–Periphery Clines

30

Salit Kark, Sergei Volis, and Ariel Novoplansky

4

Species Diversity, Environmental Heterogeneity, and Species Interactions 57 William A. Mitchell, Burt P. Kotler, Joel S. Brown, Leon Blaustein, and Sasha R.X. Dall

xii Contents

5

SHALOM A Landscape Simulation Model for Understanding Animal Biodiversity 70 Yaron Ziv, Michael L. Rosenzweig, and Robert D. Holt

6

Spatial Scale and Species Diversity Building Species–Area Curves from Species Incidence

89

William Edward Kunin and Jack J. Lennon

7

Microbial Contributions to Biodiversity in Deserts

109

Peter M. Groffman, Eli Zaady, and Moshe Shachak

8

Unified Framework I Interspecific Interactions and Species Diversity in Drylands

122

Gary A. Polis, Yoram Ayal, Alona Bachi, Sasha R.X. Dall, Deborah E. Goldberg, Robert D. Holt, Salit Kark, Burt P. Kotler, and William A. Mitchell

II

Ecological Complexes of Biodiversity, Ecosystems, and Landscapes

9

Species Diversity and Ecosystem Processes in Water-Limited Systems 153 Moshe Shachak, Steward T.A. Pickett, and James R. Gosz

10

Linking Species Diversity and Landscape Diversity

167

Bertrand Boeken, Yarden Oren, Shlomo Brandwine, and Sol Brand

11

The Impact of Animals on Species Diversity in Arid-Land Plant Communities 189 Andrew Wilby, Bertrand Boeken, and Moshe Shachak

12

Resource Partitioning and Biodiversity in Fractal Environments with Applications to Dryland Communities 206 Mark E. Ritchie and Han Olff

13

Unified Framework II Ecosystem Processes: A Link Between Species and Landscape Diversity 220 Moshe Shachak, Robert Waide, and Peter M. Groffman

Contents

III

Biodiversity, Conservation, and Management

14

The Effects of Grazing on Plant Biodiversity in Arid Ecosystems David Ward

15

Sustainability in Arid Grasslands New Technology Applications for Management

250

Arian Pregenzer, Robert R. Parmenter, Howard Passell, John R. Vande Castle, Thomas K. Budge, and Gregory Michael Bonito

16

Reconciliation Ecology and the Future of Species Diversity Michael L. Rosenzweig

17

Management for Biodiversity Human and Landscape Effects on Dry Environments

286

Avi Perevolotsky, Moshe Shachak, and Steward T.A. Pickett

18

Unified Framework III Human Interactions with Biodiversity

305

Anna A. Sher, Bruce M. Kahn, and Christopher R. Dickman

19

Toward a Unified Framework in Biodiversity Studies Moshe Shachak, James R. Gosz, Avi Perevolotsky, and Steward T.A. Pickett

Index

337

320

266

xiii

233

This page intentionally left blank

Contributors

Yoram Ayal Blaustein Institute for Desert Research Ben Gurion University of the Negev Sede Boqer, Israel 84990

Sol Brand Blaustein Institute for Desert Research Ben Gurion University of the Negev Sede Boqer, Israel 84990

Alona Bachi Department of Ecology and Evolutionary Biology University of Arizona Tucson, AZ 85721-0088

Shlomo Brandwine Blaustein Institute for Desert Research Ben Gurion University of the Negev Sede Boqer, Israel 84990

Leon Blaustein Institute of Evolution University of Haifa, Israel 31905

Joel S. Brown Department of Biological Sciences University of Illinois Chicago, IL 60607

Bertrand Boeken Blaustein Institute for Desert Research Ben Gurion University of the Negev Sede Boqer, Israel 84990

Thomas K. Budge Earth Data Analysis Center University of New Mexico Albuquerque, NM 87131

Gregory Michael Bonito LTER Network Office University of New Mexico Albuquerque, NM 87106

Sasha R.X. Dall Blaustein Institute for Desert Research Ben Gurion University of the Negev Sede Boqer, Israel 84990 xv

xvi

Contributors

Christopher R. Dickman School of Biological Sciences and Institute of Wildlife Research University of Sydney NSW, Australia 2006 Deborah E. Goldberg Department of Ecology and Evolutionary Biology University of Michigan Ann Arbor, MI 48109-1048 James R. Gosz University of New Mexico Albuquerque NM 87131-1091 Peter M. Groffman Institute of Ecosystem Studies Milbrook, NY 12545 Robert D. Holt Department of Zoology University of Florida Gainesville, FL 32611 Bruce M. Kahn Department of Rural Sociology University of Wisconsin Madison, WI 53706 Salit Kark Institute of Life Sciences The Hebrew University of Jerusalem Jerusalem, Israel 91904 Burt P. Kotler Blaustein Institute for Desert Research Ben Gurion University of the Negev Sede Boqer, Israel 84990 William Edward Kunin School of Biology University of Leeds Leeds, United Kingdom LS2 9JT Jack J. Lennon Macaulay Institute Aberdeen, Scotland AB15 8QH

Yael Lubin Blaustein Institute for Desert Research Ben Gurion University of the Negev Sede Boqer, Israel 84990 William A. Mitchell Department of Life Sciences Indiana State University Terre Haute, IN 47809 Ariel Novoplansky Blaustein Institute for Desert Research Ben Gurion University of the Negev Sede Boqer, Israel 84990 Han Olff Department of Environmental Science Wageningen Agricultural University Bornsesteeg 69 6708 PD Wageningen The Netherlands Yarden Oren Sustainable Ecosystems Department Commonwealth Scientific & Industrial Research Organization (CSIRO) Canberra ACT Australia 2601 Robert R. Parmenter Department of Biology University of New Mexico Albuquerque, NM 87131 Howard Passell Cooperative Monitoring Center Sandia National Laboratory Albuquerque, NM 87185 Avi Perevolotsky Department of Natural Resources Agricultural Research Organization Bet Dagan, Israel 50250 Steward T.A. Pickett Institute of Ecosystem Studies Millbrook, NY 12545 Arian Pregenzer Cooperative Monitoring Center Sandia National Laboratory Albuquerque, NM 87185

Contributors Mark E. Ritchie Department of Biology Syracuse University Syracuse, NY 13244

Michael L. Rosenzweig Department of Ecology and Evolutionary Biology University of Arizona Tucson, AZ 85721-0088

Moshe Shachak Blaustein Institute for Desert Research Ben-Gurion University of the Negev Sede Boqer, Israel 84990

Anna A. Sher Weed Science Program University of California Davis, CA 95616

John R. Vande Castle LTER Network Office University of New Mexico Albuquerque, NM 87106

xvii

Sergei Volis Institutes for Applied Research Ben Gurion University Beer Sheva, Israel 84105 Robert Waide LTER Network Office University of New Mexico Albuquerque, NM 87106 David Ward Department of Conservation Ecology University of Stellenbosch Matieland, South Africa 7602 Andrew Wilby WERC Center for Population Biology Silwood Park, Ascot, Berks United Kingdom SL5 7PY Eli Zaady Blaustein Institute for Desert Research Ben Gurion University of the Negev Sede Boqer, Israel 84990 Yaron Ziv Department of Life Sciences Ben Gurion University Beer Sheva, Israel 84105


BIODIVERSITY IN DRYLANDS


1 Introduction A Framework for Biodiversity Studies Moshe Shachak James R. Gosz Avi Perevolotsky Steward T.A. Pickett

The Need for a Unified Framework of Biodiversity Biodiversity is regarded as a scientific concept, a measurable entity, as well as a social–political construct (Gaston 1996, Wilson 1993). The aim of this volume is to develop the scientific basis for biodiversity studies, and for the integration of the concept into management practice. We emphasize biodiversity as a powerful, integrative concept—one that requires careful articulation and further conceptualization before application. Diversity is a concept that refers to the range of variation or differences among a set of entities; biological diversity then refers to variety within the living world. An example of biological diversity is ‘‘species diversity,’’ which is commonly used to describe the number, variety, and variability of the assemblage of living organisms in a defined area or space. However, biodiversity as a concept has evolved. Current definitions expand the biological diversity concept to emphasize the multiple dimensions and ecological realms in which biodiversity can be observed. These definitions stress that biodiversity encompasses at least four kinds of diversities: genetic diversity, species or taxonomic diversity, ecosystem diversity, and landscape diversity (McAllister 1991; Solbrig 1993, Stuart and Adams 1991; Groombridge 1992; Heywood 1994, Wilson 1993). Two main problems emerge as a consequence of the broad scope that the biodiversity concept has taken at present. Cast as questions, the problems are: (1) How do we incorporate processes (e.g., foraging, energy and nutrient flows, patch dynamics) into a concept that is based on seemingly static entities (i.e., individual organisms, species, habitat types, patch types)? (2) How do we integrate across ecological subdisciplines (e.g., 3

4

Introduction

ecosystem, population, landscape ecology) and across scales that are involved in biodiversity studies? The two problems are not mutually exclusive. Indeed, they are inseparable and complementary. For example, to determine how species diversity and ecosystem processes interact requires incorporation of entities and processes, as well as integration of community and ecosystem ecology. The focus on both entities and processes reflects the long-recognized dichotomy of structure and function in biology and ecology. Clearly, both structure and function must be integrated in order to successfully solve ecological questions. Dealing with biodiversity brings this needed integration into focus. The history of science shows that integration across disciplines is a critical component of scientific progress (Cohen 1985, Pickett 1999). Integration forces us to ask new questions, fill gaps in understanding, facilitate information flow among disciplines, and bridge dichotomies that arise due to divergence in disciplinary paradigms (Pickett et al. 1994). In addition, integration creates new disciplines, such as applied ecology, and impels us to address issues of scale. Indeed the introduction of the concept of biodiversity has led to new questions that had not been raised within the research agendas of population, community, ecosystem, and landscape ecology until recently (Gaston 1996). The study of biodiversity as an interfacing process between populations and ecosystems is an excellent example of the role of the concept of biodiversity in generating new questions and facilitating information flows across subdisciplines (Levin 1997, Mooney et al. 1996). These interfacing studies address questions regarding the role of organisms (e.g., populations, species, functional groups) in system processes (e.g., nutrient retention, decomposition, production), and how system properties such as stability, resistance, invasibility, and predictability are affected by organismal diversity. The attempts to answer the above questions include experiments in the field and in artificial ecosystems (Naeem et al. 1995, Mooney et al. 1996, Lawton 1994, Naeem et al. 1994, Tilman et. al., 1996), field data collection, modeling, and the introduction of new concepts and theories. New empirical data provide evidence that biotic diversity at levels ranging from genetic diversity among populations to landscape diversity is critical for the maintenance of natural and agricultural ecosystems (Schulze and Mooney 1993, Allen-Wardell et al. 1998). In addition to empirically based new ideas, the interfacing studies generated the novel concept of ‘‘functional biodiversity’’ (Grassle et al. 1991, Solbrig 1994, Martinez 1995), ‘‘organisms as ecosystem engineers’’ (Jones et al. 1994, 1997), and ‘‘ecosystem predictability’’ (McGrady-Steed et al. 1997). These new concepts identify important arenas for integration. Functional biodiversity addresses the variety of the relationships between specific groups of living entities and certain ecological processes. The concept demonstrates the lack of a theoretical framework to guide and prioritize the broad scope of observations, experiments, and management so characteristic of the field of ecology (Lawton and Brown 1993). Essentially, the concept reverses the traditional relationship between biotic and abiotic factors. Whereas, classically, ecologists

Introduction

5

have used environmental parameters as predictors or independent variables, the concept of functional biodiversity suggests that the independent variables are biotic. The lexicon for such relationships (Lawton, 1994), the experimental methods, and the response dimensions to be measured in these variables, are not specified by existing ecological theory (McGrady-Steed et al. 1997). However, the concept of functional biodiversity advanced biodiversity studies by generating hypotheses that address the relationship among physiology, species diversity, and ecosystem function (Solbrig 1994, Martinez 1995, Schulze and Mooney 1993). An important conceptual aspect of functional biodiversity is the conceptual refinement and reinvigoration of the diversity/stability issue, which had been abandoned as unproductive. Formerly, the issue was dealt with descriptively, using indirect measures based on vague concepts. Currently, the issue is being addressed in a functional and experimental manner. The inclusion of functional relationships, process studies, and studies of dynamics of the entities involved are key aspects of the power of the functional approach. The concept of ecosystem engineering addresses the variety of the relationships among organism activities, landscape diversity, ecosystem processes, and species diversity. The concept of organisms as ecosystem engineers also relates to biodiversity questions at the interface between population/community, ecosystem, and landscape ecology. This concept also introduces new feedback pathways between organisms and their environment. Currently we lack a conceptual framework or methodology for understanding engineering– biodiversity relationships. This is because these relationships lie outside of the traditional domains of population and ecosystem models. However, we see initial effort for developing new models that combine population processes and engineering (Gurney and Lawton 1996). The concept of ‘‘ecosystem predictability’’ (McGrady-Steed et al. 1997) identifies variations in ecosystem processes that are subjected to control by the richness of species in those systems. There are other new and important concepts that have arisen from the study of the role of biodiversity in system processes. These include the ‘‘redundant,’’ ‘‘rivet,’’ and ‘‘idiosyncratic’’ hypotheses (Walker 1992, Lawton and Brown 1993, Vitousek and Hooper 1993, Lawton 1997, Ehrlich and Walker 1998). Biodiversity includes more than the interface of species and ecosystem concepts (Martinez 1995). We recognize at least two more interfaces: species and landscape, and ecosystem and landscape. Since the time of Watt (1947), ecologists have recognized the importance of spatial pattern and its relationship to processes. Therefore, it is important to know how spatial patterns are related to species richness. The landscape mosaic redistributes resources. This is the link between ecosystem processes and landscape. This link is vital for determination of species diversity. We assume that the essential role of biodiversity studies is to highlight new, interfacing questions, concepts, and theories. We believe that as was shown for species–ecosystem interfacing, adding new interfaces will contribute to our understanding of the relationship between biodiversity and ecological func-

6

Introduction

tion. Dryland biodiversity studies can be a good start in this direction. Dryland systems are especially amenable to experimentation; they have manageable faunal and floristic diversities and tractable physical structures. The openness of many dryland ecosystems allows a degree of visualization and understanding difficult to obtain in more complex environments. Although current expansions of the biodiversity concept are new, they have already demonstrated a remarkable ability to integrate formerly disparate areas, generate new data and related concepts, and serve as a foundation for a new theory. Still, the concept requires further elaboration and analysis. This book brings together chapters that focus on the wide range of interfaces between biodiversity and other ecological realms. To make the collection most useful, we present a framework to tie the variety of perspectives, concepts, and connections together. We introduce this framework below.

Toward a Framework for Biodiversity In this section we elaborate on the biodiversity concept and relate the book chapters to this comprehensive concept. We suggest that biodiversity refers to an assemblage of ecological entities, on various scales, appropriate to types and numbers of entities and the differences and interactions among them. In fig. 1.1, we clarify the concept of ecological entities, their relation to scale and interactions, and generate four focal questions as a foundation for integrative biodiversity. Types and Number of Ecological Entities (Fig. 1.1(II)) Ecologists recognize many types of tangible and abstract ecological entities. These include: genes, species, functional groups, trophic levels, compartments in ecosystem models, resources that organisms use, habitats, and patches. For our purposes, the diversity of entities can be grouped into three types: organism-, resource-, and landscape-related entities. For example, a small unit of soil (1 1 1 m) encompasses organisms such as bacteria and fungi that can be classified by entities such as genes, species, and functional groups. These entities are mixed with diverse nitrogen resources such as parent material, rainfall, nitrogen fixers and dry deposition. Another group of entities in this mixture is landscape entities, such as different soil horizons or soil patches varying in air, water, and solid particle content. A principal question pertaining to all ecological entities is: What are the processes that control the number of a specific entity and all types of entities in a biosphere unit? More specifically, we ask: What are the processes that control the number of species, resources, and patches in ecological systems at different spatial and temporal scales? (Fig. 1.1(II).) The numbers of ecological entities are usually scale dependent. An increase in spatial scale will increase the number of species and the number of habitats in which they live.

Introduction

7

Figure 1.1 A conceptual model capturing the essence and the focal questions of biodiversity studies. (I) Units of the biosphere are a mixture of ecological entities ( , *: ecological entities such as species and habitats). (II) A first set of questions related to the dynamics of the numbers of the entities (?1: first question). (III) The entities differ in their properties (*, *, , ; for example, species differ in body mass, while habitats differ in patch types). ?2: The second focal question refers to the relationship between the number and the differences. (IV) The entities are organized by interactions. ?3: The third focal question focuses on the interactions of depicts the different entity types and their effects on number and differences ( control of the dynamics of one entity type, i.e., species, on another, i.e., patches). (V) The organized entities affect ecosystem processes (primary production, decomposition, etc.). ?4: The fourth focal question deals with the number, differences, and interactions on system behavior.

8

Introduction

Several chapters in this book address the number of entities and scale relationships. Rosenzweig (chapter 15) explores what is known about the species–area relationship. Kark et al. (chapter 3) present a study of organismal diversity patterns across a distributional range of species, thus revealing areas with especially high genetic and morphological diversity. Ritchie and Olff (chapter 11) suggest a scale-dependent model of species coexistence in relation to body size. Boeken et al. (chapter 9) show how the scale, structure, and definition of patches are relevant for the distribution of the organisms. Shachak et al. (chapter 8) suggest that hypotheses are dependent on the assumptions of water accessibility and utilization by a diverse plant assemblage and scale of analysis. Mitchell et al. (chapter 4) show scale-dependent biodiversity by comparing small-scale interactions with large-scale patterns. Kunin and Lennon (chapter 6) refer to the relationship between scale and incidence. Differences Among Entities (Fig. 1.1(III)) Entities differ in their properties. Species differ in many properties such as body size, behavior, and abundance. Similarly, patches are distinguished by a suite of properties. Some examples are size, resources, and spatial distribution. In the biodiversity context, the question is: What is the relationship between the number of entities and the degree of differences among them? This last question is the second focal question for a unified biodiversity framework. The second focal question integrates types of entities, their number, and the differences among them. Specifically, we ask how differences in a trait or a set of traits of an assemblage of species affect the number of species in a given area. Or: How do differences in patch properties in a landscape mosaic affect the number of species? Ziv et al. (chapter 5) propose a model that simulates how the number and distribution of different-sized habitats affect the number of species, of different body mass. Shachak et al. (chapter 8) show how differences among species using water for biomass production along a gradient of soil moisture can affect the productivity–diversity relationship. Boeken et al. (chapter 9) present a conceptual framework for connecting species diversity and landscapes. They discuss how changes in species assemblages, which differ in abundance and frequency of occurrence in patches, coincide with changes in landscape structure. Ritchie and Olff (chapter 11) discuss how differences in habitat, food, and resources may contribute to higher species diversity. Wilby et al. (chapter 10) show the role of animals in controlling the differences in local species assemblage, in terms of the number of individuals and the frequency of species occurrence. Interactions Among Entities (Fig. 1.1(IV)) Each assemblage of ecological entities is characterized by a web of interactions. From a biodiversity perspective, the third focal question is: What is the

Introduction

9

outcome, in terms of numbers and differences, of the interactions between different types of entities? For example, we ask how changes in the number and properties of habitats in a landscape affect the number and properties of species, and vice versa. Several chapters address the question of number, difference, and interaction. Mitchell et al. (chapter 4) discuss how variation in substrate, slope, solar input, and productivity contribute to species interactions at a local scale and patterns of species diversity. Ayal et al. (chapter 2) present a relationship between landscape heterogeneity in plant cover, food web structure, predation, and species diversity. Boeken et al. (chapter 9) introduce a new method and Ziv et al. (chapter 5) propose a simulation model, for studying species, landscape diversity relationships. In addition to the reciprocal effect of the interactions among various entities, there is another outcome of interentity relationships; their effect on ecosystem processes (fig. 1.1(V)). This generates the fourth focal question: How does the organization of an assemblage of ecological entities control ecosystem processes and how do the ecosystem processes feedback to the number and differences of ecological entities? The existing studies in relation to entity organization, ecosystem processes and their feedbacks on organization focus on: (1) the relationship between species or functional group diversity and productivity, and (2) organisms as ecosystem engineers. In studies of the effect of diversity on productivity, the assumption is that rates of ecosystem processes are determined by complementarity in resources used by different species or functional groups. Under this assumption, the organization of the entities is controlled by the diversity of resources and their distribution in time and space. The study of organisms as ecosystem engineers refers to the modulation of the landscape by organisms that modify resource distribution. This activity reorganizes species and landscape diversity. Shachak et al. (chapter 8) present a conceptual model linking species properties and ecosystem processes for water-limited systems. They integrate species, water, and energy flows to demonstrate the relationships between species diversity and productivity. Wilby et al. (chapter 10) show how animals modify environment structure and reorganize the entities in the system, thereby affecting species number and differences. Boeken et al. (chapter 9) deal with the relationship between species and landscape diversity and how this relationship affects productivity.

A Framework for Biodiversity Management Biodiversity management refers to human influences on the organization and interactions of an assemblage of ecological entities. The aim is to have a desired number and variety of ecological entities in a managed biosphere unit. Any biodiversity management is a manipulation of landscape diversity, succession, ecosystem processes, and species diversity (Pickett et al. 1997, 1999). This is the consequence, even when the objective of management is

10 Introduction

to manipulate only one of the components. This suggests that development of a biodiversity framework that integrates among ecological entities interactions with ecosystem processes should enable managers to follow the chain of interactions among landscape diversity, ecosystem processes, and species diversity. Therefore, biodiversity management is a part of ecosystem management. Ecosystem management is the manipulation of ecosystems to satisfy specified societal values. This definition is useful because biodiversity is managed to meet human values such as high species and landscape diversity and the goods and services that they provide. Several chapters address management issues. Perevolotzky et al. (chapter 16) present a conceptual model on human–biodiversity relationships in waterlimited systems. They demonstrate how humans have actively managed landscape diversity to direct ecosystem processes for enhancing the distribution and abundance of organisms that provide ecosystem services. Pregenzer et al. (chapter 14) outline four different kinds of large-scale data required by land managers for the development of sustainable land-use strategies that can be met with current or future technologies. Ward (chapter 13) contributes to the understanding of the effect of grazing by domestic animals on dryland biodiversity by comparing North American and African studies. Rosenzweig (chapter 15) discusses how to manage ecosystems to stop mass extinction of species. He suggests methods for sharing anthropogenic habitats with species. He also proposes techniques for providing conditions needed by both native species and human society.

Book Organization Most of the chapters in this book refer to the relationship among three components: organism diversity, ecosystem processes, and landscape diversity. However, in many of the chapters there are different emphases on these three constituents. Therefore, we grouped chapters within two main parts according to the gist of the particular chapter; whether they focus on the organism, ecosystem, or landscape perspective. We also included a third part dedicated to biodiversity, conservation, and management. In part I, ‘‘Living Components of Biodiversity: Organisms,’’ we determined the order of the chapters according to topics: chapters 2 and 3 are devoted to biodiversity studies on a relatively small scale; chapters 4 and 5 deal with concepts and models; chapter 6 deals with species diversity on a large scale. In part II, ‘‘Ecological Complexes of Biodiversity, Ecosystems, and Landscapes,’’ chapter 8 focuses on the role of ecosystem science in biodiversity, while chapters 9 to 11 emphasize the landscape aspect. In part III, ‘‘Biodiversity, Conservation, and Management,’’ chapters 13 and 14 deal with the relationship between biodiversity and grazing systems; chapter 15 concerns species conservation.

Introduction

11

At the end of each part, ideas for a unified framework are proposed. Chapter 7 provides a unified framework for biodiversity from the community and population ecology perspective. Chapter 12 suggests a unified framework for biodiversity from the ecosystem perspective, while chapter 17 indicates human interactions with biodiversity. The opening chapter suggests a conceptual framework for biodiversity studies and demonstrates how the ensuing chapters contribute to the idea. Chapter 18, the last chapter, suggests a unified framework for biodiversity studies.

References Allen-Wardell, G. et al. (21 authors). 1998. The potential consequences of pollinator declines on the conservation of biodiversity and stability of food crop yields. Conservation Biology 12: 8–17. Cohen, J.B. 1985. Revolution in Science. Belknap Press, Cambridge, MA. Ehrlich, P., and B. Walker. 1998. Rivets and redundancy. BioScience 48: 387. Gaston, K.J (ed.). 1996. Biodiversity – A Biology of Numbers and Differences. Blackwell Science, Oxford. Grassle, J.F., P. Lasserre, A.D. McIntyre, and G.C. Ray. 1991. Marine biodiversity and ecosystem function. Biology International Special Issue 23: 1–19. Groombridge, B. 1992. Global Biodiversity: Status of the Earth’s Living Resources. Chapman and Hall, London. Gurney, W.S.C., and J.H. Lawton. 1996. The population dynamics of ecosystem engineers. Oikos 76: 273–283. Heywood, V.H. 1994. The measurement of biodiversity and the politics of implementation. In: P.L. Forey, C.J. Humphries, and R.I. Vane-Wright (eds.), Systematics and Conservation Evaluation, pp. 15–22. Oxford University Press, Oxford. Jones, C.G., J.H. Lawton, and M. Shachak. 1994. Organisms as ecosystem engineers. Oikos 69: 373–386. Jones, C.G., J.H. Lawton, and M. Shachak. 1997. Positive and negative effects of organisms as physical ecosystem engineers. Ecology 78: 1946–1957. Lawton, J.H. 1994. What do species do in ecosystems? Oikos 71: 367–374. Lawton, J.H. 1997. The role of species in ecosystems: aspects of ecological complexity and biological diversity. In: A. Takuya, S.A. Levin, and M. Higashi (eds.), Biodiversity: An Ecological Perspective, pp. 215–228. Springer-Verlag, Berlin. Lawton, J.H., and V.K. Brown. 1993. Redundancy in ecosystems. In: E.D. Schulze and H.A. Mooney (eds.), Biodiversity and Ecosystem Function, pp. 255–270. Springer-Verlag, Berlin. Levin, S.A. 1997. Biodiversity: interfacing populations and ecosystems. In: A. Takuya, S.A. Levin, and M. Higashi (eds.), Biodiversity: An Ecological Perspective, pp. 277–287. Springer-Verlag, Berlin. Martinez, N.D. 1995. Unifying ecological subdisciplines with ecosystem food webs. In: C.G. Jones and J.H. Lawton (eds.), Linking Species and Ecosystems, pp. 166–175. Chapman and Hall, New York. McAllister, D.E. 1991. What is biodiversity? Canadian Biodiversity 1: 4–6. McGrady-Steed, J, P.M. Harris, and P.J. Morin. 1997. Biodiversity regulates ecosystem predictability. Nature 390: 162–165.

12 Introduction

Mooney, H.A., J.H. Cushman, E. Medin, O.E. Sala, and E.D. Schulze. 1996. What we have learned about the ecosystem functioning of biodiversity. In: H.A. Mooney, J.H. Cushman, E. Medlin, O.E. Sala, and E.D. Schulze (eds.), Functional Roles of Biodiversity: A Global Perpective, pp. 475–484. Scope 55, John Wiley and Sons, New York. Naeem, S., L.J. Thompson, S.P. Lawler, J.H. Lawton, and R.M. Woodfin. 1994. Declining biodiversity can alter the performance of ecosystems. Nature 368: 734–736. Naeem, S., L.J. Thompson, S.P. Lawler, J.H. Lawton, and R.M. Woodfin. 1995. Empirical evidence that declining species diversity may alter the performance of terrestrial ecosystems. Philosophical Transactions of the Royal Society of London B 347: 249–262. Pickett, S.T.A. 1999. The culture of synthesis: habits of mind in novel ecological integration. Oikos 87: 479–487. Pickett, S.T.A., J. Kolassa, and C.G. Jones. 1994. Ecological Understanding: The Nature of Theory and the Nature of Nature. Academic Press, Cambridge. Pickett, S.T.A., R.S. Ostfeld, M. Shachak, and G.E. Likens, (eds.) 1997. The Ecological Basis of Conservation: Heterogeneity, Ecosystems, and Biodiversity. Chapman and Hall, New York. Pickett, S.T.A., M. Shachak, B. Boeken, and J.J. Armesto. 1999. The management of ecological systems. In: T.W. Hoekstra and M. Shachak (eds.), Arid Lands Management: Toward Ecological Sustainability, pp. 8–17. University of Illinois Press, Urbana. Schulze, E.D., and H.A. Mooney (eds.) 1993. Ecosystem Function of Biodiversity. Springer-Verlag, Berlin. Solbrig, O.T. 1993. Plant traits and adaptive strategies: their role in ecosystem function. In: E.D. Schulze and H.A. Mooney (eds.), Biodiversity and Ecosystem Function, pp. 97–116. Springer-Verlag, Berlin. Stuart, S.N., and R.J. Adams. 1991. Biodiversity in Sun Saharan Africa and Its Islands. World Conservation Union, Gland, Switzerland. Tilman, D., D. Wedin, and J. Knops. 1996. Productivity and sustainability influenced by biodiversity in grassland ecosystems. Nature 379: 718–720. Vitousek, P.M., and D.U. Hooper. 1993. Biological diversity and terrestrial ecosystem biogeochemistry. In: E.D. Schulze and H.A. Mooney (eds.), Biodiversity and Ecosystem Function, pp. 3–14. Springer-Verlag, Berlin. Walker, B.H. 1992. Biodiversity and ecological redundancy. Biological Conservation 6: 18–23. Watt, A. S. 1947. Pattern and process in the plant community. Journal of Ecology 35: 1–22. Wilson, E.O. 1993. The Diversity of Life. Harvard University Press, Cambridge, MA.

Part I



2 How Can High Animal Diversity Be Supported in Low-Productivity Deserts? The Role of Macrodetritivory and Habitat Physiognomy Yoram Ayal Gary A. Polis* Yael Lubin Deborah E. Goldberg

O

n large spatial scales, species diversity is typically correlated positively with productivity or energy supply (Wright et al. 1993, Huston 1994, Waide et al. 1999). In line with this general pattern, deserts are assumed to have relatively few species for two main reasons. First, relatively few plants and animals have acquired the physiological capabilities to withstand the stresses exerted by the high temperatures and shortage of water found in deserts (reviewed by Noy-Meir 1974, Evenari 1985, Shmida et al. 1986). A second, more ecological mechanism is resource limitation. In deserts, the low and highly variable precipitation levels, high temperatures and high evapotranspiration ratios limit both plant abundance and productivity to very low levels (Noy-Meir 1973, 1985, Polis 1991d). This lack of material at the primary producer level should exacerbate the harsh abiotic conditions and reduce the abundance of animals at higher trophic levels by limiting the types of resources and their availability. Animal abundance should be even further reduced because primary productivity is not only low, but also tends to be sporadic in time and space (MacMahon 1981,

Gary Polis died tragically during the preparation of this chapter, before he could finish contributing his insights and examples based on his broad experience in many desert ecosystems. We dedicate our efforts to Gary, with the deepest sadness and regret for his loss. 15

16 Living Components of Biodiversity: Organisms

Crawford 1981, Ludwig 1986). Herbivores should have difficulties tracking these variations (e.g., Ayal 1994) and efficiently using the available food resources. Hence, herbivore populations in deserts have low densities relative to other biomes (Wisdom 1991) and most of the primary productivity remains unused (Crawford 1981, Noy-Meir 1985). This low abundance of herbivores should propagate through the food web and result as well in lower abundance of higher trophic levels. The number of individuals and the number of species are not always positively correlated; in particular, some examples of low diversity at high productivity with high densities are well documented (e.g., salt marshes, reviewed by Waide et al. 1999). However, several distinct mechanisms have led to the expectation that when productivity and the number of individuals are low, the number of species is also likely to be low. First, within trophic levels, the ‘‘statistical mechanics’’ model of Wright et al. (1993) may operate. In this model, the amount of energy present determines the probability distribution of population sizes for the members of the species pool in a region. A species with a larger population size has a higher probability of occurrence in a given patch and therefore a higher cumulative probability that it will be present in any patch in a region. As energy supply increases, the probability distribution of population sizes shifts upward and more species are likely to occur in the region. Sufficiently low food available at the base of the food chain means that fewer trophic levels (and fewer functional groups of species) can be supported because of the inefficiency of energy transfer along the food chain (Elton 1927, Lindemann 1942, Fretwell 1977, Oksanen et al. 1981). Fewer trophic levels are likely to lead to fewer species because of fewer possible roles in the community. A third mechanism by which low abundance leads to low diversity explicitly assumes that competition among consumers is an important process. Reduction of the number of species within any given trophic level should also lead to reduction in diversity at higher trophic level because of fewer opportunities for specialization and resource partitioning based on prey type. In addition, low prey density should support only generalist predators that can use most prey types they may encounter. Hence, desert predators are expected to overlap widely in their diets, which conventional competition theory tells us should limit the number of predators that can coexist (Polis 1991d). Despite the logic of these arguments for low species diversity in deserts, empirical studies demonstrate that diversity of at least above-ground desert animal communities is actually quite high (Polis 1991c and review in Polis 1991a). Representatives of almost all terrestrial animal taxa are found in deserts (Polis 1991c). Physiological and behavioral adaptations for desert conditions have appeared many times in evolutionary history, suggesting that these traits are acquired easily by animals. Thus, harsh environmental conditions have not directly limited the number of species in deserts. But, the above energy-based arguments still hold and pose the question how the high

High Animal Biodiversity in Low-Productivity Deserts

17

species diversity observed in deserts can be maintained, based on such a low productivity. In this chapter, we develop several hypotheses about the solutions to this apparent paradox. First, although primary productivity is low, deserts tend to have higher ecological efficiencies of transfer to higher trophic levels, and thus a given amount of food at the base can support higher abundances at higher trophic levels. Because several of the arguments described above for why low primary productivity should lead to low diversity are based on low densities of available prey for higher trophic levels, this means that diversity should also be higher than expected, based solely on energy available. The higher net ecological efficiencies in deserts are due to two mechanisms: (1) the dominance of macrodetritivores rather than herbivores as the central link between primary and secondary productivity, leading to a higher proportion of the energy in detritus being transferred to animals in above-ground food webs in deserts, and (2) the dominance of the above-ground food web by small arthropods and poikilotherms, with lower metabolic requirements and hence higher efficiencies (Turner 1970, McNeill and Lawton 1970). Second, the energy-based arguments for low diversity in deserts assume that resource limitation is ubiquitous and therefore that competition for resources is the dominant ecological interaction in deserts. In contrast, we argue that in many cases macrodetritivores are not limited by their resources and instead, that predator-mediated coexistence is a particularly important diversity-maintaining process in deserts. In particular, we argue that deserts have high spatial heterogeneity of plant cover and thus have high spatial heterogeneity of predation intensity due to use of cover as refugia. This in turn should lead to great potential for predator-mediated habitat partitioning (Holt 1977), which then allows higher diversity than expected, based solely on energy considerations. Because data on food web structure in deserts around the world is rather limited, our ideas are based primarily on our own studies in the Negev Desert (Israel), the North American deserts, and the Namib desert. We hope that our arguments will give new impetus to the study of desert communities to examine the generality of the hypotheses we develop to explain high diversity in deserts despite their low productivity.

Higher Ecological Efficiency in Deserts: The Desert Food Web In this section, we focus largely on why animal abundance in deserts is likely to be higher than expected, based solely on the amount of primary productivity available. As described above, the mechanisms that have led to the expectation of low diversity in deserts are direct consequences of this expected low abundance and, therefore, our arguments also directly lead to an explanation of higher than expected diversity.


Macrodetritivores as the Main Primary Consumers in Desert Food Webs In most terrestrial communities, 80–90% of primary productivity becomes plant litter (detritus) to be decomposed by microbes either in the soil, or on the soil surface. Most of this litter is mineralized or recycled within the soil food web and only a minor portion finds its way back to the above-ground food web (reviewed by Hairston and Hairston 1993), the one usually studied by community ecologists (Polis and Strong 1996). For deserts, we argue that even more of the primary productivity goes into litter but that less of it is mineralized or stays in the soil food web. Instead, we hypothesize that macrodetritivores transfer the energy and minerals from detritus directly to higher trophic levels in the above-ground food web, leading to greater overall efficient transfer from primary producers to higher trophic levels. As the growing season for plants is short in deserts, desert herbivores need to complete their development within this short period and survive the long period between consecutive growing seasons either by subsisting on a poor resource or by being inactive. Most herbivores in deserts are ephemeral insects (e.g., grasshoppers, moths, beetles) with only one reproductive cycle per year, therefore they cannot track the high between-year variability in productivity levels (Ayal 1994). In years of high productivity, most of the unconsumed plant production dries up and turns to plant litter. In addition, desert soils also should contain high amounts of plant detritus because of the high allocation to roots typical of desert plants. Plants in deserts are limited by soil resources, especially water, and by nitrogen in times of high precipitation (Noy-Meir 1973, 1985). Under such conditions, plants allocate a high proportion of their energy to develop their root system and less to their shoots (high root:shoot ratio, Tilman 1988). Indeed, Cody (1986) found that the root:shoot ratio in desert shrubs is around 2, while mesic habitats typically have root:shoot ratios less than 1 (Tilman 1988). Thus, compared with more mesic areas, deserts should have a higher proportion of primary production going into detritus than to herbivores. In addition to being more abundant, the detritus in deserts should be less likely to be broken down by free-living microbes. The low moisture in deserts during most of the year limits microbial decomposition of both above- and below-ground plant litter (Vishnevetsky and Steinberger 1997). This seasonal limitation of decomposition lent support to Huston’s (1994) argument on the importance of the length of the growing season to productivity (or, in this case, to decomposition). The low decomposition rate results in a rich and reliable resource (above- and below-ground litter) for those animals that use it, namely, macrodetritivorous arthropods such as isopods, termites, and tenebrionid beetles (larvae and adults) (Johnson and Whitford 1975, Crawford 1981, 1991, MacKay 1991). One can say, therefore, that in deserts most of the microbial decomposition takes place within the macrodetritivores’ gut rather than in the soil (Crawford and Taylor 1984). This explains


19

the high biomass of macrodetritivores in deserts (Crawford 1991, Mackay 1991). But more important to the current discussion of deserts, the diverse group of macrodetritivores forms an integral part of the above-ground food web (e.g., Polis 1991b, Ayal and Merkl 1994). This contrasts with more mesic communities in which the soil food web has relatively few links to the aboveground web (e.g., earthworms and larvae eaten by moles and birds). Accordingly, macrodetritivores in deserts take the place of herbivores as the major group of primary consumers linking primary producers to higher trophic levels. This replacement has several important implications for energy flow and species diversity in deserts. One critical distinction between herbivores and macrodetritivores is that macrodetritivores have no negative effect on plant productivity. This contrasts the common consumer-producer dynamics in which high consumption by the consumer has a negative effect on the producer dynamics, and the interaction drives both the producer and the consumer populations to low density levels. Thus in herbivore–plant interactions, high efficiency of the herbivore results in a reduction in plant biomass and consequently a reduction in plant productivity (e.g., Noy-Meir 1975). However, macrodetritivores consume the plant tissue after its death and thus do not affect plant dynamics or productivity directly. In fact in deserts, macrodetritivores are the main decomposers and thus contribute to nutrient recycling from plant litter back to the soil. In addition, many detritivores are burrowers (e.g., termites, isopods, and tenebrionid larvae) and thus contribute to soil turnover and increase water infiltration into the soil. Consequently, macrodetritivores may commonly have positive effects on plant productivity in deserts (e.g., Whitford 1986). The slow rate of microbial decomposition in deserts and the shift of plant litter decomposition to macrodetritivores are unique features of desert communities and are key factors in understanding their structure and function. This shift changes profoundly the proportion of primary productivity channeled into the above-ground food web relative to the proportion channeled in other terrestrial ecosystems. In other terrestrial ecosystems, herbivores use less than 20% of the primary productivity, and more than 80% is lost to microbial decomposition and does not find its way back to the above-ground food web (Hairston and Hairston 1993). In deserts, microbial decomposition accounts for less than 10% of plant primary productivity (Whitford 1986). Even if some of the plant litter is lost by physical degradation or transported by wind or water, the majority of the primary productivity is probably still consumed by macrodetritivores and channeled into the above-ground food web (reviewed by Whitford 1986). Thus, despite the low level of primary productivity in deserts, the energy base of the above-ground food web is likely to be much higher than is generally assumed based on patterns found in other terrestrial communities. This could lead to higher diversity either by permitting a greater degree of prey-specialization by consumers within a trophic level or by allowing the addition of more trophic levels.


Desert Communities Dominated by Small and Poikilothermic Consumers As early as 1927, Elton recognized the importance of the energy base of a community to the length of food chain found in it (his pyramid of numbers, and later, the pyramid of biomass; Bodenheimer 1938). This energy-centered approach to community ecology was elaborated by Lindemann (1942) and currently dominates the discussion on community structure and function (Fretwell 1977, Oksanen et al. 1981, Oksanen 1992, Hairston and Hairston 1993, Hairston 1997, Polis and Strong 1996). However, Elton (1927) also recognized that organism size plays an important role in food chains, although this issue has been relatively neglected in the discussion of food chain length (but see Hutchinson 1959, Yodzis 1984, Cousins 1987, Hairston and Hairston 1993). Carnivores are generally larger than their prey (the gape-limitation hypothesis, Zaret 1980). Thus, a food chain based on small primary consumers (i.e., insects, zooplankton) can include more links than one that starts with even a small mammalian herbivore. The desert food chain that is based on macrodetritivorous insects includes predatory insects, arachnids, and reptiles as primary predators (e.g., Morton and James 1988) and larger reptiles and predatory birds as secondary predators, with mammals (e.g., jackals or coyotes) becoming important only in relatively more productive arid grasslands (Brown 1986). The desert food web is also a poikilotherm-based web. The main primary and secondary consumers are poikilotherms. Food chains based on poikilotherms are energetically more efficient than food chains based on the homeothermic mammals or birds. This is because poikilotherms have lower basic metabolic rates and lower energy requirements for maintenance than homeotherms (Table 41-4 in Ricklefs 1973, Humphreys 1979, McNeill and Lawton 1970). Thus, a given amount of primary productivity will enable more trophic links in a poikilotherm- than in a homoeotherm-based food chain (Yodzis 1984). Reagan et al. (1996) evoked a similar argument when discussing an island tropical forest food web. They suggested that the absence of large herbivores and the dominance of poikilotherms in the food web in the community they studied resulted in longer food chains than in similar continental communities where mammal herbivores and predators are common. Being small and poikilothermic allows members of the desert food web to fulfill their energetic demands by utilizing a smaller area of habitat than their larger homeothermic counterparts in more productive habitats. Thus, higher densities of macrodetritivores can live in 100 m2 of desert than can the number of voles in a single hectare of productive meadow or ungulates in an African grassland. Similarly, at the next higher trophic level, more scorpions or lizards can live in the same area in deserts than the weasels or lions of similar trophic position in more mesic productive communities. Hence, all else being equal, food webs made up of largely small or poikilothermic organisms should be able to support higher population densities and thus more trophic levels and more species than larger, homeothermic organisms.


21

To summarize, we argue that deserts have greater ecological efficiency than many mesic environments. We also argue that, in contrast to the current expectation that deserts are one-link communities (Fretwell 1977, Oksanen et al. 1981, Oksanen and Oksanen 2000), desert communities have at least four links when reduced to their basic trophic levels (fig. 2.1; for a more detailed approach to desert food webs see Polis 1991b). Thus, both the within- and the between-trophic level mechanisms we described earlier to link increasing numbers of individuals with increasing numbers of species may operate, and at least partially account for the observed high diversity of desert animal communities.

Greater Importance of Predation in Deserts: The Role of Plant Cover in Mediating Predator–Prey Interactions and Promoting Diversity In this section, we develop the argument that predation is an important factor in promoting high diversity of animals in deserts. Many attributes of desert organisms and communities suggest that predation is a major force. For example, in desert communities there is a high proportion of predators and a high frequency of cryptic coloration, and activity is frequently limited to times of day when predators are inefficient or inactive (Louw and Seely 1980). In addition, both predatory arthropods (reviewed by Polis and Yamashita 1991) and predatory nonmammalian vertebrates (reviewed by Vitt 1991, Wiens 1991) are common in deserts (Polis 1991b). We first argue that the high degree of horizontal redistribution of water characteristic of deserts results in large spatial heterogeneity in productivity, and thus in plant cover. Plant cover is an important refuge for small animals, which are consumed by large predators. This results in high spatial heterogeneity in predation intensity and hence the potential for predator-mediated habitat segregation, yielding enhanced coexistence and diversity.

Figure 2.1 The basic above-ground trophic structure of desert communities. Main groups of macrodetritivores are termites, isopods, and tenebrionids; of primary predators are arachnids and reptiles; of secondary predators are birds. Note that macrodetritivores have no negative effect on plant productivity.


High Spatial Heterogeneity of Productivity and Plant Cover in Deserts Water is the main resource that limits plant productivity in arid habitats (Noy-Meir 1973, 1985). Because the amount of direct precipitation in arid environments is low, horizontal redistribution of water has a huge impact on local productivity at small and intermediate scales. Redistribution of water depends on the velocity of the rain event, substrate features at the contributing and recipient sites, plant cover, and relief (Yair and Danin 1980). Poor habitats are those with poor absorption capabilities and low water retention capacity (e.g., rocky surfaces or steep slopes). Rich habitats have high absorption and water retention capabilities and are surrounded by large contributing surfaces. Rich patches may be small depressions created by animals digging for food (Alkon 1999) or burrows (reviewed by Whitford and Kay 1999), or washes that collect water from small areas and ones that carry water from large catchments. This redistribution of water results in local gradients of productivity. In turn, the local variation in productivity results in especially strong variation in plant cover and microhabitat physiognomy in arid environments because small changes in the amount of available water determine whether a given site is barren or supports sporadic growth of annuals, a small perennial, or a large shrub. The spatial heterogeneity in productivity and cover within an area depends on the topography of the area. Regions with low relief (e.g., sand fields, large plains) will have relatively less spatial heterogeneity in productivity and plant cover than regions with high relief (e.g., high dunes, rocky ridges, and wadis). Therefore, areas with low relief should have overall productivity that is more easily predictable by rainfall than areas with high relief.

High Spatial Heterogeneity in Cover Leads to High Variation in Predation Efficiency by Large Secondary Predators Secondary predators of the desert food web consist primarily of large predatory birds, with reptiles and large omnivorous mammals becoming important in some areas. Birds can move easily to forage over large areas and exploit areas with episodic rich food, and migrate to other regions in seasons of low food availability (Wiens 1991). Most of these predators are visually oriented and hover, perch, or walk while foraging. Increasing the amount of plant cover in the habitat can therefore greatly reduce the efficiency of these secondary predators. Consequently, the intensity of predation by secondary predators in deserts is negatively correlated with the amount of plant cover in the habitat (e.g., Ayal and Merkl 1994, Seely 1985, Kotler 1997). It is interesting to note that in the Namib desert where plants are scarce and cover is minimal, a diverse fauna of tenebrionid beetles subsist on wind-driven detritus, and these macrodetritivores are active at the hottest period of the day. Although competition for resources was evoked to explain this unexpected


23

pattern (Hamilton 1971), an alternative and more plausible explanation is predation avoidance. The tenebrionids become active at times when their largely nocturnal avian and mammalian predators cannot be active.

High Heterogeneity in Secondary Predation Efficiency Leads to High Heterogeneity in Primary Predator Abundance and Predation Efficiency Primary predators in deserts are mainly arthropods (insects, spiders, scorpions, solifugids) and smaller vertebrates (lizards, snakes, shrews, and some insectivorous rodents). These predators hunt within the vegetation cover and utilize cover for their own protection against second-tier predators (e.g., Skutelsky 1996). The abundance and diversity of primary predators should be correlated positively with plant cover (fig. 2.2). This is true for spiders, for which plant cover, density, and structural diversity can strongly influence species diversity and abundance (Brandt and Lubin 1998, Robinson 1981). Primary predators are more abundant in areas of high plant cover both because the small (largely arthropod) detritivores are more abundant, and because they too will benefit from the structural diversity of plant cover for concealment and defense from their own predators. Finally, many secondary predators and parasitoids are small and also forage within the vegetation layer. Thus, for example, araneophagic and oophagic spiders (Whitehouse

Figure 2.2 The suggested effect of habitat physiognomy on trophic interactions in habitats of relatively low and high productivity within deserts. Arrows point to the affected trophic level and their width indicates the relative strength of the effect. In low-productivity habitats, plant cover is low and secondary predators are highly efficient and limit the density and activity of primary predators. As a result, macrodetritivores are released from predation and become food limited. In highproductivity habitats, plant cover is high and provides shelter to primary predators from secondary predators. As a result, primary predators are both abundant and highly active and limit the densities of macrodetritivores. Hence, in productive habitats much of the plant litter remains unconsumed.


and Lubin 1998), predatory hemipterans, and parasitoid wasps constitute a second tier of small predators whose abundance should correlate positively with plant cover. This diversity of small predators leads to efficient utilization of the macrodetritivore resource in those habitats rich in vegetation cover, but low predation intensity in nearby habitats with lower vegetation cover. Effects of Heterogeneity in Predation on Regulation and Abundance of Trophic Groups The above-described relationships between predation intensity and plant cover in deserts have a substantial effect on the abundance and regulation of different trophic groups in these habitats. In poor habitats (e.g., plains in the Negev Desert), plant cover is low and bird predation on primary predators is intense. Therefore, primary predators are found in low abundance and their activity is limited to areas around refuges such as shrubs (e.g., Skutelsky 1996) or burrows (Shachak and Brand 1983). As a result, macrodetritivores are released from predation, and are relatively abundant and should be limited only by their food resources. In contrast, plant cover is high in rich desert habitats (e.g., wadis) and predation intensity by birds is lower. In rich habitats, primary predators are abundant, leading to high predation on macrodetritivores. This high predation, combined with the high production of detritus, suggests that food is unlikely to be limiting to macrodetritores in rich desert habitats, so competition among the macrodetritivores is unlikely to be important. Effect of Heterogeneity in Predation Intensity on Species Diversity The regulation of abundance and diversity in deserts by means of predation, rather than competition, can influence species diversity through mechanisms operating either within or between habitats. High predation intensity within rich desert habitats means that food limitation for macrodetritivores and competition for resources among them is unlikely to be important within those habitats. Instead, we suggest that this release from competition directly increases the abundance and potential diversity of macrodetritivores. Hence, the food web based on macrodetritivores within rich desert habitats is also potentially more diverse. The spatial heterogeneity in predation intensity among desert habitats promotes habitat segregation and thus may also increase diversity. For example, Ayal and Merkl (1994) suggested that habitat and size-dependent predation on tenebrionids was the most probable explanation for the observed habitat segregation of tenebrionids in the Negev Desert. They found that in compact soil habitats, body size of tenebrionid adults and amount of plant cover are positively correlated: small species are common in plains with low plant cover, medium-size species are common in slope habitats with intermediate plant cover, and large species are common in wadis with high plant cover.


25

This habitat segregation is consistent with the preference of large predatory birds for large tenebrionids, which therefore only survive when plant cover is available as a refuge. In contrast, small tenebrionid species are eaten mainly by scorpions, which are most abundant in the plant-covered wadis but are less effective predators in the plain. Cage experiments with birds and artificial cover support this hypothesis (Groner and Ayal 2001). And birds were also reported as an important factor in determining prey communities in other deserts (Seely 1991, Wiens et al. 1991, Polis 1991a, Floyd 1996). In addition to spatial heterogeneity of predation, high diversity of detritivores is also likely promoted directly by other aspects of the high spatial heterogeneity of the environment described above, accompanied by habitat specialization of the detritivores (see chapter 7 this volume, Seely 1985). Even if plant detritus is not amenable to food partitioning (Crawford 1981, 1991), the habitat in which it is consumed may promote special adaptations. For example, tenebrionid beetles show diverse morphological adaptations that can be attributed to habitat-substrate specialization (e.g., Medvedev 1965, Louw and Seely 1982) and habitat segregation according to substrate type has been demonstrated in tenebrionids (Ayal and Merkl 1994, Krasnov and Ayal 1995, Seely 1985, 1991). However, despite the high spatial substrate diversity found in deserts, the number of coexisting species is an order of magnitude higher than the available types of substrate, so that direct effects of heterogeneity are unlikely to be the sole explanation of high diversity in macrodetritivores. The predation hypothesis (Ayal and Makl 1994, Groner and Ayal 2001) mentioned above is a possible solution to this apparent paradox of high diversity of macrodetritivores in deserts despite the fact that detritus is a highly accessible resource of low nutritional value and relatively uniform composition.

Conclusions We challenge the idea that low productivity in deserts causes low diversity when the whole community is examined (rather than a specific guild, as is typically done). We also challenge the idea that desert communities are simple in structure and that biotic interactions are not important for understanding species distribution and abundance in them. We suggest that the desert aboveground food web is based on macrodetritivores as the primary consumers, and because both primary consumers and primary predators are ectotherms, the web is highly efficient energetically. Because of low microbial decomposition rates, macrodetritivores use a high proportion of the primary productivity, become abundant, and form a base to another two trophic levels. The heterogeneity in plant cover adds another dimension to the biotic interactions along the desert food chain. Plant cover mediates the strength of interactions between the secondary and primary predators. In this way plant cover also affects the strength of interactions between the primary predators and the macrodetritivores. Yet the low rate of microbial decomposition probably has


a negative effect on the diversity of soil organisms in deserts. This low belowground diversity may, in part, counteract the increase in the above-ground diversity that results from the shift of energy from the below- to the aboveground community typical of deserts. However, in contrast to the dominant paradigm about above-ground, desert community structure (Shmida et al. 1986), we suggest that at least the above-ground compartment of desert communities is shaped by biotic interactions, and desert communities as a whole have some unique features compared with other terrestrial communities.

References Alkon, P.U. 1999. Microhabitat to landscape impacts: crested porcupine digs in the Negev Desert highlands. Journal of Arid Environments 41: 183–202. Ayal, Y. 1994. Time-lags in insect response to plant productivity—significance for plant insect interactions in deserts. Ecological Entomology 19: 207–214. Ayal, Y., and O. Merkl. 1994. Spatial and temporal distribution of tenebrionid species (Coleoptera) in the Negev Highlands, Israel. Journal of Arid Environments 27: 347–361. Bodenheimer, F.S. 1938. Problems of Animal Ecology. Oxford University Press, Oxford. Brandt, J. and Y. Lubin. 1998. An experimental manipulation of vegetation structure: consequences for desert spiders. Israel Journal of Zoology 44: 201–216. Brown, J.H. 1986. The roles of vertebrates in desert ecosystems. Pp. 51–71. Whitford, W.G., ed., Pattern and Processes in Desert Ecosystems. University of New Mexico Press, Albuquerque. Cody, M. 1986. Structural niches in plant communities. Pp. 381–405 in J. Diamond and T. Case, eds., Community Ecology. Harper and Row, New York. Cousins, S. 1987. The decline of the trophic level concept. Trends in Ecology and Evolution 2: 312–316. Crawford, C. 1991. The community ecology of macroarthropod detritivores. Pp. 89–112 in G. Polis, ed., The Ecology of Desert Communities. The University of Arizona Press, Tucson. Crawford, C., and E. Taylor. 1984. Decomposition in arid environments: role of the detritivore gut. South African Journal of Science 80: 170–176. Crawford, C.S. 1981. Biology of Desert Invertebrates. Springer Verlag, New York. Elton, C. 1927. Animal Ecology. Sidgwick and Jackson, London. Evenari, M. 1985. Adaptations of plants and animals to the desert environment. Pp. 79–92 in M. Evenari, I. Noy-Meir, and D. Godall, eds., Hot Deserts and arid Shrublands. Elsevier, Amsterdam. Floyd, T. 1996. Top-down impacts on creosotebush herbivores in a spatially and temporally complex environment. Ecology 77: 1544–1555. Fretwell, S. 1977. The regulation of plant communities by the food chain exploiting them. Perspectives in Biology and Medicine 20: 169–185. Groner, E., and Y. Ayal. 2001. The interaction between bird predation and plant cover in determining habitat occupancy of darkling beetles. Oikos 93(1): 22–31. Hairston Jr., N., and N. Hairston Sr. 1993. Cause-effect relationships in energy flow, trophic structure, and interspecific interactions. American Naturalist 142: 379–411.


27

Hairston, N.G. 1997. Does food web complexity eliminate trophic-level dynamics? American Naturalist 149: 1001–1007. Hamilton, W.J. III. 1971. Competition and thermoregulatory behavior of the Namib Desert tenebrionid beetle genus Cardiosis. Ecology 52: 810–822. Holt, R. 1977. Predation, apparent competition, and the structure of prey communities. Theoretical Population Biology 12: 197–229. Humphreys, W. 1979. Production and respiration in animal populations. Journal of Animal Ecology 48: 427–453. Huston, M.A. 1994. Biological Diversity: the Coexistence of Species on Changing Landscapes. Cambridge University Press, Cambridge. Hutchinson, G. 1959. Homage to Santa Rosalia or why are there so many kind of animals. American Naturalist 93: 145–159. Johnson, K., and W. Whitford. 1975. Foraging ecology and relative importance of subterranean termites in Chihuahuan desert ecosystems. Environmental Entomology 4: 66–70. Kotler, B.P. 1997. Patch use by gerbils in a risky environment: Manipulating food and safety to test four models. Oikos 78: 274–282. Krasnov, B., and Y. Ayal. 1995. Seasonal changes in darkling beetle communities (Coleoptera, Tenebrionidae) in the Ramon erosion cirque, Negev Highlands, Israel. Journal of Arid Environments 31: 335–347. Lindemann, R. 1942. The trophic-dynamic aspect of ecology. Ecology 23: 399–418. Louw, G., and M.K. Seely 1982. Ecology of Desert Organisms. Longman, London. Ludwig, J.A. 1986. Primary production variability in desert ecosystems. Pp. 5–17 in W.G. Whitford, ed., Pattern and Processes in Desert Ecosystems. University of New Mexico Press, Albuquerque. MacKay, W. 1991. The role of ants and termites in desert communities. Pp. 113–150 in G. Polis, ed., The Ecology of Desert Communities. University of Arizona, Tucson. MacMahon, J.A. 1981. Introduction. Pp. 263–269 in D. Goodall and R. Perry, eds., Arid-Land Ecosystems: Structure, Functioning and Management, Vol. 2. Cambridge University Press, Cambridge. McNeill, S., and J. Lawton. 1970. Annual production and respiration in animal populations. Nature 225: 472–474. Medvedev, G.S. 1965. Adaptations of leg structure in desert darkling beetles. Entomological Review 44: 473–475. Morton, S., and C. James. 1988. The diversity and abundance of lizards in arid Australia: a new hypothesis. American Naturalist 132: 237–256. Noy-Meir, I. 1973. Desert ecosystems: environment and producers. Annual Review of Ecology and Systematics 4: 25–52. Noy-Meir, I. 1974. Desert ecosystems: higher trophic levels. Annual Review of Ecology and Systematics 5: 195–214. Noy-Meir, I. 1975. Stability of grazing systems: an application of predator-prey graphs. Journal of Ecology 63: 459–483. Noy-Meir, I. 1985. Desert ecosystem structure and function. Pp. 93–103 in M. Evenari, I. Noy-Meir, and D. Goodall, eds., Hot Deserts and Arid Shrublands, A. Elsevier Scientific Publications, Amsterdam. Oksanen, L. 1992. Evolution of exploitation ecosystems.1. Predation, foraging ecology and population-dynamics in herbivores. Evolutionary Ecology 6(1): 15–33. Oksanen, L., and T. Oksanen. 2000. The logic and realism of the hypothesis of exploitation ecosystems (EEH). American Naturalist 155: 703–723.


Oksanen, L., S.D. Fretwell, J. Arruda, and P. Niemela. 1981. Exploitation ecosystems in gradients of primary productivity. American Naturalist 118: 240–261. Polis, G.A., ed. 1991a. The Ecology of Desert Communities. University of Arizona Press, Tucson. Polis, G.A. 1991b. Food webs in desert communities: complexity via diversity and omnivory. Pp. 383–437 in G.A. Polis, ed., The Ecology of Desert Communities. University of Arizona Press, Tucson. Polis, G.A. 1991c. Complex trophic interactions in deserts: an empirical critique of food-web theory. American Naturalist 138: 123–155. Polis, G.A. 1991d. Desert communities: An overview of patterns and processes. Pp. 1–26 in G.A. Polis, ed., The Ecology of Desert Communities. University of Arizona Press, Tucson. Polis, G.A., and D.R. Strong. 1996. Food web complexity and community dynamics. American Naturalist 147: 813–846. Polis, G.A., and T. Yamashita. 1991. The ecology and importance of predaceous arthropods in desert communities. Pp. 180–222 in G.A. Polis, ed., The Ecology of Desert Communities. Arizona University Press, Tucson. Reagan, D.P., Camilo, R., and R.B. Waide. 1996. The community food web: major properties and patterns of organization. Pp. 463–510 in D.P. Reagan and R.B. Waide, eds., The Food-Web of a Tropical Rain Forest. University of Chicago Press, Chicago. Ricklefs, R. 1973. Ecology. Chiron, Newton, MA. Robinson, J.V. 1981. The effect of architectural variation in habitat on a spider community: An experimental field study. Ecology 62: 73–80. Seely, M.K. 1985. Predation and environment as selective forces in the Namib Desert. Pp. 161–165 in E. Vrba, ed., Species and Speciation. Transvaal Museum Monograph No. 4, Transvaal Museum, Pretoria. Seely, M.K. 1991. Sand dune communities. Pp. 348–382 in G.A. Polis, ed., The Ecology of Desert Communities. University of Arizona Press, Tuscon. Shachak, M. and S. Brand. 1983. The relationship between sit and wait foraging strategy and dispersal in the desert scorpion, Scorpio maurus palmtus. Oecologia 60: 371–377. Shmida, A., M. Evenari, and I. Noy-Meir. 1986. Hot desert ecosystems: an integrated view. Pp. 379–387 in M. Evenari, A. Shmida, and I. Noy-Meir, eds., Hot Deserts and Arid Shrublands. Elsevier Science Publications, Amsterdam. Skutelsky, O. 1996. Predation risk and state-dependent foraging in scorpions: Effects of moonlight on foraging in the scorpion Buthus occitanus. Animal Behaviour 52: 49–57. Tilman, D. 1988. Dynamics and Structure of Plant Communities. Princeton University Press, Princeton, NJ. Turner, F.B. 1970. The ecological efficiency of consumer populations. Ecology 51: 741–742. Vishnevetsky, S., and Y. Steinberger. 1997. Bacterial and fungal dynamics and their contribution to microbial biomass in desert soil. Journal of Arid Environments 37: 83–90. Vitt, L. 1991. Desert reptile communities. Pp. 249–277 in G. Polis, ed., The Ecology of Desert Communities. Arizona University Press, Tucson. Waide, R.B., M.R. Willig, C.F. Steiner, G. Mittelbach, L. Gough, S.I. Dodson, G.P. Juday, and R. Parmenter. 1999. The relationship between productivity and species richness. Annual Review of Ecology and Systematics 30: 257–300.


29

Whitehouse, M.E.A., and Y. Lubin. 1998. Relative seasonal abundance of five spider species in the Negev desert: Intraguild interactions and their implications. Israel Journal of Zoology 44: 187–200. Whitford, W.G. 1986. Decomposition and nutrient cycling in deserts. Pp. 93–117 in Whitford, W.G., ed., Pattern and Processes in Desert Ecosystems. University of New Mexico Press, Albuquerque. Whitford, W.G., and F.R. Kay. 1999. Bioperturbation by mammals in deserts: a review. Journal of Arid Environments 41: 203–230. Wiens, J. 1991. The ecology of desert birds. Pp. 278–310 in G.A. Polis, ed., The Ecology of Desert Communities. Arizona University Press, Tucson. Wiens, J.A., R.G. Cates, J.T. Rotenberry, N. Cobb, B. Vanhorne, and R.A. Redak. 1991. Arthropod dynamics on sagebrush (Artemisia tridentata)—Effects of plant chemistry and avian predation. Ecological Monographs 61: 299–321. Wisdom, C.S. 1991. Patterns of heterogeneity in desert herbivorous insect communities. Pp. 151–179 in G.A. Polis, ed., The Ecology of Desert Communities. The University of Arizona Press, Tucson. Wright, D.H., D.J. Currie, and B.A. Maurer. 1993. Energy supply and patterns of species richness on local and regional scales. Pp. 66–74 in R. Ricklefs and D. Schluter, eds., Species Diversity in Ecological Communities. University of Chicago Press, Chicago. Yair, A., and A. Danin. 1980. Spatial variation in vegetation as related to the soil moisture regime over an arid limestone hillside, northern Negev, Israel. Oecologia 47: 83–88. Yodzis, P. 1984. Energy flow and the vertical structure of real ecosystems. Oecologia 65: 86–88. Zaret, T. 1980. Predation and Fresh Water Communities. Yale University Press, New Haven, CT.

3 Biodiversity Along Core–Periphery Clines Salit Kark Sergei Volis Ariel Novoplansky

T

he study of biodiversity has received wide attention in recent decades. Biodiversity has been defined in various ways (Gaston and Spicer, 1998, Purvis and Hector 2000, and chapters in this volume). Discussion regarding its definitions is dynamic, with shifts between the more traditional emphasis on community structure to emphasis on the higher ecosystem level or the lower population levels (e.g., chapters in this volume, Poiani et al. 2000). One of the definitions, proposed in the United Nations Convention on Biological Diversity held in Rio de Janeiro (1992) is ‘‘the diversity within species, between species and of ecosystems.’’ The withinspecies component of diversity is further defined as ‘‘the frequency and diversity of different genes and/or genomes . . .’’ (IUCN 1993) as estimated by the genetic and morphological diversity within species. While research and conservation efforts in the past century have focused mainly on the community level, they have recently been extended to include the within-species (Hanski 1989) and the ecosystem levels. The component comprising within-species genetic and morphological diversity is increasingly emphasized as an important element of biodiversity (UN Convention 1992). Recent studies suggest that patterns of genetic diversity significantly influence the viability and persistence of local populations (Frankham 1996, Lacy 1997, Riddle 1996, Vrijenhoek et al. 1985). Revealing geographical patterns of genetic diversity is highly relevant to conservation biology and especially to explicit decision-making procedures allowing systematic rather than opportunistic selection of populations and areas for in situ protection (Pressey et al. 1993). Therefore, studying spatial patterns in within-species diversity may be vital in defining and prioritizing conservation efforts (Brooks et al. 1992). 30

Biodiversity Along Core–Periphery Clines 31

Local populations of a species often differ in the ecological conditions experienced by their members (Brown 1984, Gaston 1990, Lawton et al. 1994). These factors potentially affect population characteristics, structure, and within-population genetic and morphological diversity (Brussard 1984, Lawton 1995, Parsons 1991). The spatial location of a population within a species range may be related to its patterns of diversity (Lesica and Allendorf 1995). Thus, detecting within-species diversity patterns across distributional ranges is important for our understanding of ecological and evolutionary (e.g., speciation) processes (Smith et al. 1997), and for the determination of conservation priorities (Kark 1999). This is especially important in the face of recent climatic and environmental changes occurring at global, regional, and local spatial scales (Safriel et al. 1994). Much of the scientific focus at the community level is aimed at detecting areas especially rich in biological diversity, that is, ‘‘diversity hotspots’’ (Myers 1990, Myers et al. 2000). This approach focuses attention on revealing areas rich in species diversity, endemism, and rare and endangered species (Mittermeier et al. 1998). Similarly, at the within-species level, revealing areas with especially high genetic and morphological diversity, and rare or unique genetic structures may be important in setting research and conservation priorities (Kark 1999, Kark et al. 1999).

Core and Periphery Patterns of diversity within species may be studied along ‘‘clines’’ (Brussard 1984, Carson 1959, Da Cunha and Dobzhansky 1954, Lennon et al. 1997, Lesica and Allendorf 1995, Mayr 1963). Within the distribution range of a single species, the terms ‘‘periphery’’ and ‘‘core’’ are often used to refer to the physical location of a population within the range (Brown et al. 1996). Accordingly, ‘‘peripheral’’ populations are those located at the very edge of the distribution, while ‘‘core’’ populations are those found further away from the range boundaries (Brown et al. 1996, Channell and Lomolino 2000a). This geographical distinction may also have ecological relationships (Brussard 1984). ‘‘Marginal’’ areas are the ecologically least favorable (Brown 1984, Brussard 1984, Gaston 1990, Hengeveld and Haeck 1981, Wiens 1989), least predictable, and least suitable parts of the range (Hengeveld and Haeck 1981), in locations where extinction probabilities are relatively high (Lennon et al. 1997). In contrast, in ‘‘central’’ areas, extinction probabilities are lower and conditions are, over time, more favorable and predictable for the species (Brown 1984, Brussard 1984, Gaston 1990, Lesica and Allendorf 1995). There are specific cases in which the geographical and ecological areas are not congruent (Brussard 1984, Gaston 1990, Lesica and Allendorf 1995), but in many cases they do coincide (Brussard 1984, Hoffmann and Blows 1994, Wiens 1989). Core and peripheral populations are expected to experience different biotic and abiotic selection pressures (Andrewartha and Birch 1954,


Brown et al. 1995, Lawton 1995, Lesica and Allendorf 1995). In many cases, the optimal combination of ecological, environmental, and biotic factors for a species is found within the geographical core of its range (Lesica and Allendorf 1995, Wiens 1989). Population densities generally decrease and fluctuate more along such clines toward the periphery (Brown 1984, Brown et al. 1995, Brussard 1984, Caughley et al. 1988, Collins and Glenn 1991, Gaston 1990, Hengeveld and Haeck 1981, Hoffmann and Blows 1994, Lomolino and Channell 1995, Vrijenhoek et al. 1985, Wiens 1989), the range tends to become less continuous (Brown et al. 1996), and populations become more isolated, transient (Lomolino and Channell 1995), and patchily distributed (Boorman and Levitt 1973, Carter and Prince 1988), although exceptions to these trends exist (Blackburn et al. 1999, Lawton 1995, Svensson 1992). Fluctuations in population size and growth rate at the periphery may result in very small population size (Brussard 1984), and when dispersal from neighboring populations is limited, may lead to the extinction of local populations (Harrison 1994, Lennon et al. 1997, Thomas and Hanski 1997).

Diversity in Core Versus Periphery: Classical Hypotheses Three main hypotheses concerning trends in genetic diversity across core– periphery clines are found in the literature, each having different spatial implications (reviewed in Safriel et al. 1994).

Increasing Diversity from Periphery to Core— the ‘‘Carson Hypothesis’’ The hypothesis, developed by Carson (1959), argues that genetic diversity will increase from the range periphery toward the core. Carson suggested that core populations are more continuous and dense, undergo balancing selection, and are therefore expected to show higher levels of withinpopulation genetic diversity than peripheral populations that are relatively small, fragmented, and isolated (Carson 1959, Mayr 1963). This theory implicitly refers to a neutral model of gene diversity, large core populations maintain higher genetic diversity because they harbor more mutations and because genetic drift is less effective in them compared with small isolated peripheral populations (Lesica and Allendorf 1995). Yet this prediction may also be explained based on selective considerations. Accordingly, diversity of adaptive traits at the periphery is predicted to be lower if only a few genotypes can cope with its extreme conditions (Hoffmann and Parsons 1991). This hypothesis is supported by classical papers (e.g., Da Cunha and Dobzhansky 1954) and by more recent studies (Hoffmann and Parsons 1991, Parsons 1991, Vrijenhoek et al. 1985, reviewed in Lesica and Allendorf 1995, see table 3.1).

Table 3.1 A partial summary of studies testing trends in within-population diversity in core vs. peripheral populations. Among the papers published between 1978 and 1998 that deal with diversity in peripheral vs. core populations, some show evidence for higher diversity in core populations, supporting the Carson hypothesis, while others show evidence for the opposite trend, supporting the Fisher hypothesis. Several papers show no consistent trends or significant differences between core and peripheral populations. Only papers where authors refer to core (or central) and peripheral (or marginal) populations, rather than only one of these, were included. Incomplete data were not filled in

33

Study Area

Species Studied (Common Name)

Europe

Quercus petrea

Central and western Mediterranean basin; French Atlantic coast region

Quercus ilex (holm oak)

Central and northern Japan—Hokkaido and Honshu Islands

Pinus pumila (stone pine)

Western North America

Number of Populations Studied

Type of Diversity Measured

81

Genetic (allozyme)

Hypothesis Supported

Comments

Source Reference

Inconsistent

Higher heterozygosity in core than in peripheral populations, but higher number of alleles per locus in the periphery than in the core

Zanetto and Kremer (1994)

Genetic (allozyme)

Carson

Six main disjunct regions of the range

Michaud et al. (1995)

18

Genetic (allozyme)

Carson Fisher for some loci

15 enzyme systems, generally lower in peripheral populations (in some alleles higher in peripheral populations)

Tani et al. (1996)

Bromus tectorum (cheatgrass)

6

Genetic (quantitative)

Inconsistent mixed

Introduced species

Rice and Mack (1991)

California

Avena barbata

97

Genetic (allozyme)

Humpshaped

Genetic diversity in polymorphic populations was positively related to microhabitat heterogeneity (spatiotemporal) which had a bell shape. 35 loci

Allard et al. (1978)

Israel

Avena barbata

31

Genetic (allozyme)

Inconsistent

35 loci

Allard et al. (1978) continued

Table 3.1 continued

Study Area




Canada—USA

Carex lasiocarpa

39

Canada—USA

Carex pellita

20

Utah

34


Comments

Source Reference

Genetic (allozyme)

Fisher

12 loci

McClintock and Waterway (1993)

Genetic (allozyme)

Inconsistent

12 loci

McClintock and Waterway (1993)

Hordeum jubatum

Genetic (allozyme)

Carson

18 loci

Shumaker and Babble (1980)

Turkmenistan and Israel

Hordeum spontaneum (wild barley)

Morphological

Fisher

Higher in peripheral populations in most traits (14 of 18)

Volis et al. (1998)

Israel

Hordeum spontaneum (wild barley)

Genetic (allozyme)

Fisher

Utah

Pseudotsua menziesii

Genetic (allozyme)

Carson

20 loci

Schnabel et al. (1993)

New Zealand

Leptosepermum scoparium (Myrtaceae)

Morphological

Inconsistent mixed

Inconsistent for different Traits studied

Wilson et al. (1991)

Australian coast

Drosophila serrata

Genetic (quantitative)

Carson

Blows and Hoffmann (1993)

Kenya, Morroco, Italy, Reunion, Australia

Ceratitis capitata (medfly)

Genetic (allozyme and DNA)

Carson

Baruffi et al. (1995)

17

Nevo and Beiles (1988)

Study Area




Comments

Source Reference

23 loci; lower in periphery in allozyme analysis, nonregular yet ‘‘slightly impaired’’ in periphery in morphological traits

Descimon and Napolitano (1993)

35

Morphological and genetic (allozyme)

Carson for genetic and inconsistent for phenotypic

Theba pisana (Helicidae)

Genetic (allozyme)

Fisher


Israel

Hyla arborea (Hylidae)

Genetic (allozyme)

Fisher


Israel

Bufo viridis (Bufinidae)

Genetic (allozyme)

Fisher


Israel

Gryllotalpa gryllotalpa (Gryllotalpidae, mole cricket)

Genetic (allozyme)

Humpshaped

Israel

Acomys cahirinus (Muridae)

Genetic (allozyme)

Fisher


Israel

Agama stellio (Agamidae)

Genetic (allozyme)

Fisher


Arizona vs. Sonora, Mexico

Poeciliopsis occidentalis

Genetic (allozyme)

Carson

Southeastern France

Parnassius mnemosyne (Papilionidae)

Israel

24


21 (16 Sonoran)

Diversity first increases but declines toward the range extreme periphery

Heterozygosity


Vrijenhoek et al. (1985)


Decreasing Diversity from Periphery to Core— the ‘‘Fisher Hypothesis’’ This hypothesis, resulting from Fisher’s (1930) work, predicts that genetic diversity should decrease from the periphery toward the core of the species’ range. Accordingly, peripheral populations will sustain higher levels of genetic diversity due to fluctuating selection in spatially heterogeneous and unpredictable environments, while core populations will experience stabilizing selection, which reduces genetic diversity (Burger 1988, Fisher 1930, Hoffmann and Parsons 1991, Parsons 1989). This theory implicitly refers to adaptive considerations and mainly to the type and strength of the pressures of natural selection. This theory, too, is supported by empirical evidence from a wide array of species and study systems (Hoffmann and Parsons 1991, Nevo and Beiles 1988, Parsons 1991, table 3.1). Homogenous Diversity from Periphery to Core— the ‘‘Mayr Hypothesis’’ Mayr suggested that in some cases gene flow from the core may compensate for the effects of local selection and genetic drift at the periphery. In such cases genetic diversity may actually be homogenous throughout the species’ range (Mayr 1963, 1970, table 3.1).

Early Studies Early work done in the 1950s that compared patterns of genetic diversity across the distribution range yielded contradictory results. One of the earliest studies was performed by Da Cunha and Dobzhansky (1954), who compared chromosomal polymorphism in core and peripheral populations of Drosophila. Their hypothesis was that the amount of adaptive polymorphism carried in a population is correlated with the variety of the ecological niches its members exploit. They found that core Drosophila willistoni populations were highly polymorphic relative to those at the periphery, where the species was less common and less ubiquitous than its competitors. They interpreted this result in a ‘‘Carsonian’’ fashion, because their central populations were both richer and more diverse (Da Cunha and Dobzhansky 1954). However, White (1951) found no diminution of chromosomal variability toward the distribution periphery, and other studies found an increase in genetic diversity at the periphery of the range (summarized by Hoffmann and Parsons 1991). Although the topic continues to draw attention, no clear pattern has emerged (e.g., Brussard 1984, Parsons 1991). Work focusing on a wide range of animal and plant species has tested these hypotheses using various phenotypic and molecular genetic estimates. Overall, each one of the above-mentioned hypotheses has gained considerable support by empirical evidence (table 3.1). In addition, there are cases in which no obvious spatial trends appear


(Brussard 1984). Reviewing the case of protein electrophoretic diversity, Parsons (1989, p.43) notes that: ‘‘variability levels in central vs. marginal populations have revealed a rather confused situation. For an endangered fish, Poeciliopsis occidentalis, in Arizona, geographically peripheral populations show less electrophoretic variation than do central populations. In contrast, some Drosophila populations show higher electrophoretic variability at the margins . . . . Hence, comparisons of electrophoretic variability under differing ecological circumstances must be approached with extreme caution.’’

Discrepancies Between the Hypotheses Although some of the discrepancies found among the studies may reflect real differences between the study systems and different species, it appears that methodological and conceptual factors may have also contributed to the confusion: 1. Spatial Definitions of Core and Periphery: Studies of core versus periphery often compare two main distribution areas, one representing periphery and the other, core. Populations often are sampled from two extremes rather than along a continuum. The definition and logic behind the selection of these two areas very often differ between studies (Antonovics et al. 1994). In some cases, peripheral populations are sampled from the edge of the species continuous distribution range, where population density declines rapidly (Lennon et al. 1997). Yet this region may not represent the very edge of the range. Additional small population patches may occur beyond this region. Alternatively, peripheral populations may be sampled at the very extreme periphery of the range, representing small and isolated fragmented populations (Antonovics et al. 1994). In these cases, populations from intermediate areas of the range, located between the extreme periphery and the core, are not included in the study. These differences in sampling may easily lead to contradictory conclusions, because different sections of the distribution range are compared. While the sampling of the periphery is often inconsistent across studies, the definition of the term ‘‘core’’ is often even less clear. Thus, for example, some studies refer to the point that is geographically farthest away from all range peripheries, while others divide the range into two equal-sized areas, the one representing ‘‘core’’ and the other ‘‘periphery’’ (e.g., Lomolino and Channell 1995, Channell and Lomolino 2000b). The geometric shape of the range and the patterns of patchiness within it will largely determine which areas will be considered core versus peripheral. Other studies focus on the population density, where high-density populations are considered as central populations, yet these may not always be located at the geographical core. 2. Distance Between the Compared Populations: In some cases the two distribution areas representing core and periphery are compared in populations that are geographically very distant from each other (Brussard 1984),


sometimes from different continents (see, e.g., comparison of marginal vs. central populations of birds in Møller 1995). Populations from very distant areas may experience different evolutionary and recent histories, causing distinct patterns of genetic diversity within populations. 3. Anthropomorphic Influences: Differences may arise due to sampling of areas with differing levels of human-related disturbance. Because sampling of the extreme edge of the range may be very difficult, populations studied in these areas are often sampled in nonnatural-resource-rich (e.g., agricultural) and human-impacted areas, especially when the periphery occurs in the desert. Populations in these areas may actually have different patterns of genetic diversity relative to the more natural surrounding environment, where population density is lower and sampling becomes more difficult (see discussion in Kark 1999, Kark et al. 1999). 4. Diversity Estimates Used: Genetic diversity within a population may be affected by gene flow, population dynamics, and random processes, such as genetic drift (Slatkin 1994, Wade and Goodnight 1998, Wright 1932) and natural selection. The interaction between the neutral factors and the type and levels of selection pressure will largely determine the levels of genetic diversity in populations. While Carson’s theory implicitly refers to a neutral model of gene diversity, the Fisher hypothesis of hypervariable marginal populations due to fluctuating environments refers to selected markers and to more complicated genetic models that involve natural selection. Therefore, different trends in genetic diversity may be obtained by studying traits controlled by ‘‘neutral’’ versus ‘‘naturally selected’’ genes (Futuyma, 1997) or when comparing different loci. This may lead to contradictory results in different studies, because different genes and alleles may be subjected to varying pressures of natural selection (Randi et al. in review). Different studies have used diverse genetic and molecular methods to reveal genetic diversity (e.g., chromosome inversions vs. microsatellites) that may be affected to a different extent by random versus selective processes (table 3.1). However, even within a single class of genetic markers, such as allozymes, contradictory results are often revealed between studies testing trends across species ranges (e.g., Brussard 1984, Hoffmann and Parsons 1991). 5. Different Scales Are Studied: This could contribute at both the sampling and analysis steps to differences between studies focusing on unequal scales, or when species of different size and dispersal ability are compared, as discussed below.

A New Integrating Hypothesis Kark has recently proposed a new hypothesis, predicting a hump-shaped unimodal pattern of diversity across the range, with peak genetic diversity levels in intermediate populations, located between the range periphery and the core (Kark 1999). It suggests that part of the discrepancy between empiri-


cal findings appearing in the literature may be due to partial sampling of core–periphery gradients, representing the increase or decrease phase only. Maximum diversity is predicted to occur in the edge of the species continuous distribution, often congruent with areas of ecological transition (i.e., ecotones). As suggested by Kark (1999), in order to test the proposed hypotheses one would need to: (a) identify an area where steep environmental changes occur across short geographical distances and (b) select species that are distributed along the gradient that include more continuous populations, populations at the edge of their more continuous range, and small and isolated populations at the extreme periphery of their range. In this chapter, we aim to review new studies testing the above hypotheses across a steep climatic gradient in Israel, along which many species reach the edge of their distribution range (YomTov and Tchernov 1988). In the studies presented below, we define the periphery and core of the range based on population densities. Thus, the area beyond which the population density declines to zero will be considered the range periphery. The area where the species reaches the edge of its continuous distribution will be called the ‘‘turnover zone’’ (see below). The area where population densities are high and distribution is relatively continuous will be considered the core of the distribution range. We argue that the periphery can be more clearly and absolutely defined, and thus delimited on a spatial basis, compared with the core. Therefore, we recommend, when possible, to refer to the distance from the range periphery, rather than to an arbitrary core– periphery dichotomy.

Turnover Zones and Ecotones Areas of ecological transition, that is, ecotones and environmental gradients, have recently received scientific attention due to their potential importance in processes generating biodiversity (Schneider et al. 1999, Smith et al. 1997) and as potential biodiversity hotspots (Kark et al. 1999). Many species reach the limits of their continuous distributions in these areas of steep ecological transition between ecosystems (Danin 1998, Endler 1982). From here toward the most extreme periphery of the range, populations become very small and isolated, and eventually fade out, marking the edge of the species range (Kark 1999). Thus ‘‘turnover zones’’ of various species often are predicted to occur at areas of ecological transition, leading to congruence of the ecotone and the turnover zone (Kark 1999). In these areas, the environment fluctuates temporally and spatially between more favorable and more extreme (Kark et al. 1999). In this chapter, we will use the term ‘‘ecotone’’ for ecosystem-related transition zones (i.e., areas of transition between ecosystems) and the term ‘‘turnover zone’’ for density and distribution-related changes within a single species (i.e., transition between core and peripheral distribution).


Core–Peripheral Clines in Israel The Ecogeographical Gradient in Israel Israel comprises a narrow land bridge between Europe, Asia, and Africa, with steep climatic and ecological clines across relatively short distances (Bitan and Rubin 1991, Kadmon and Danin 1997, Yom-Tov and Tchernov 1988). The sharp ecological gradient from Mediterranean to desert ecosystems is congruent with distributional margins of many Mediterranean, Irano-Turanian, and Saharo-Arabian organisms (Danin 1998, Yom-Tov and Tchernov 1988). While mean annual rainfall in the Mediterranean Galilee and Golan Heights in the north may exceed 900 mm per year, only 200–300 km to the south the mean annual rainfall in the southern Negev desert is less than 30 mm and is highly variable among years (Bitan and Rubin 1991, fig. 3.1). An especially steep climatic gradient occurs in the northern Negev ecotone, where mean annual rainfall decreases from over 450 mm to less than 150 mm within a narrow belt of less than 60 km (Danin 1998, Kark et al. 1999). It is here that many Mediterranean, steppe, and desert species reach the edge of their continuous distributions (Bitan and Rubin 1991, Kadmon and Danin 1997, Safriel et al. 1994, Yom-Tov and Tchernov 1988). Thus, the ecological cline in this region offers a unique opportunity to compare geographically proximate populations with very different population densities that are potentially connected by dispersal and gene flow. This gradient includes populations along a distribution gradient of increasing distance from the very edge of the distribution range. Furthermore, it allows testing of the presented hypotheses, comparing trends in genetic and phenotypic diversity along core–periphery clines. Earlier Studies Considerable work has been done over the years studying patterns of allozyme, DNA, and chromosmal diversity across the Mediterranean–desert gradient in Israel (Nevo and Beiles 1988, 1989). This work included many different groups, from insects to mammals and reptiles. The general pattern suggested by Nevo and Beiles is that diversity increases from the mesic toward the more arid environments. We suggest that an addition to this important work should be to relate the findings to the species’ distribution range patterns and processes. A certain environment that determines population sizes, gene flow, patterns of local selection, and the resulting phenotypic and genetic diversity (such as the Mediterranean environment), may be very different in the degree of stress it presents to a mesic versus a desert species. For example, patterns may differ greatly between desert species, for which the Mediterranean region is the periphery of the range, and for Mediterranean species, for which the desert provides the range periphery. The distribution range reflects the response of the species to the diverse environmental and biotic conditions. In many cases, the range periphery represents the area


Figure 3.1 Mean annual rainfall in Israel. The map was generated based on data in the GIS laboratory of the Hebrew University of Jerusalem. Note the sharp changes in rainfall across short geographical distances, and especially the Mediterraneandesert ecotone region where rainfall rapidly declines to the south and to the east.

beyond which the species cannot maintain viable populations at a certain point in time. Therefore, reference to the species’ distribution range, in addition to the environmental variables, may be highly important to our understanding of the ecological and evolutionary processes that determine diversity patterns and to conservation of the species. Here, we discuss some recent case studies from Israel that follow diversity patterns with reference to the species’ distribution range. Recent Work Following is a brief review of some of our recent studies from Israel, testing the above hypotheses on diversity across periphery-to-core (nonperiphery) clines within the distribution range. We look at a study of allozyme diversity


of a phasianid game bird, the chukar partridge (Alectoris chukar), a quantitative genetics study of an annual legume (Trifolium purpureum) and a perennial clonal grass (Dactylis glomerata), and a study of phenotypic and allozyme diversity of an annual grass, the wild progenitor of cultivated barley, Hordeum spontaneum. All these species have high densities in the Mediterranean region of Israel and their populations become smaller and more isolated toward the arid Negev Desert, which comprises their global distributional periphery.

The Chukar Partridge (Alectoris chukar) Chukar Distribution The chukar partridge (Alectoris chukar) generally inhabits the mesic and semiarid areas, and has relatively large and continuous populations in Mediterranean and steppe regions of Israel (Shirihai 1996). Deserts represent the margins of its range, where it occurs in isolated and sparse populations (Shirihai 1996, Liu Naifa pers. comm.). The chukar is a species indigenous to the region. The extreme desert regions of the southern Negev and Sinai comprise the global southwestern border of its distribution. The chukarcontinuous Mediterranean areas in the north and center of Israel are referred to in this work as part of the ‘‘core.’’ The Mediterranean desert ecotone of the northern Negev is the edge of the chukar-continuous distribution. This area comprises the ‘‘turnover zone’’ of the species’ range in Israel, where rapid thinning of chukar populations occurs across short geographic distances (Shirihai 1996). Marginal to the Negev Highlands in the south of Israel and toward the Sinai Desert, chukar density decreases, distribution becomes discontinuous, and local populations become patchy and isolated (Degen et al. 1984, Pinshow et al. 1982, Shirihai 1996). This area comprises the extreme periphery of the chukar range. An additional isolated population, most probably a relict from the late Pleistocene, is found in the mountains of the southern Sinai desert (see discussion in Kark et al. 1999). Chukars do not possess many physiological adaptations to heat stress (Carmi-Winkler et al. 1987, Frumkin 1983, Kam 1986), especially as compared with the partly sympatric sand partridge (Ammoperdix heyi), which is well adapted to the desert (Carmi-Winkler et al. 1987, Degen et al. 1984, Pinshow et al. 1982). A main limiting factor in the desert is the chukars’ ability to forage long enough to obtain their energy requirements without risking their heat balance. In arid hot environments, extremely high temperatures limit their foraging activity to approximately one hour a day, which cannot suffice for their energy and water demands (Carmi-Winkler et al. 1987). Therefore, in extremely arid regions the species usually occurs in small resource-rich patches (Carmi-Winkler et al. 1987, Pinshow et al. 1983, Shirihai 1996). These habitat patches must be rich enough to meet the birds’ energetic needs in the short available foraging time, limited to


the early morning hours (Carmi-Winkler et al. 1987, Degen et al. 1983, Pinshow et al. 1983). The patches must also provide sufficient water during the dry months when chukars need to drink water on a regular daily basis (Carmi-Winkler et al. 1987, Degen et al. 1983, Pinshow et al. 1983). As far as is known, chukars do not exhibit long-distance spatial or altitudinal migrations (Alkon 1974, Paz 1987), and available information from marked chukars in Israel suggests that movement of individual birds is usually limited to an area ranging several square kilometers in both northern and southern populations (Alkon 1974, P. Alkon unpublished data).

Trends in Diversity Across the Range Trends in within-population allozyme diversity were studied in the chukar partridge in Israel and were compared with three years of study (1990, 1993, 1995). Five chukar populations were sampled along the gradient in each of two years, 1990 and 1993. Three populations were sampled in both years in order to enable a comparison among years and to test robustness and variability of the trends along short-term time scales. Allozyme diversity in 32 allozyme loci was determined for birds collected in each population (see Kark 1999 for details). Trends revealed were very similar for the two years of study. Genetic diversity, as estimated by the percentage of polymorphic loci, mean number of alleles, and observed and expected heterozygosity increased from the core to the ecotone (Kark et al. 1999). Single and multiloci Hardy–Weinberg and linkage disequilibria increased significantly from populations in the Mediterranean region to those at the ecotone, despite the close geographical proximity between populations in these two regions. As predicted, peak diversity was found in the intermediate ecotone area, located between the extreme periphery of the range and the core (Kark et al. 1999). Local populations maintained distinct genetic characteristics even though population genetic data indicated the likelihood of substantial gene flow among populations, supporting the ‘‘divergence with gene flow’’ model of speciation (Rice and Hostert 1993, Smith et al. 1997, see Kark et al. 1999 for discussion). Study populations showed instances of isolation by distance effects in the face of both short distances among populations and the absence of significant geographical barriers (Kark et al. 1999). The highly diverse genetic structure of different chukar populations across the short geographical gradient given relatively high levels of gene flow is especially interesting. It could be maintained by a combination of stochastic population dynamics factors and natural selection acting on morphophysiological traits and on their linked allozyme loci (Kark et al. 1999). Increased intrapopulation genetic diversity in the ecotone region resulted from the addition of new alleles, not present in less variable populations, and increasing frequency of rare alleles at polymorphic loci. These results suggest a clear trend across the part of the range studied, yet it


should be emphasized that this sampling did not include the extreme periphery of the range. Following these findings, 13 chukar populations were sampled in 1995. These included the part of the range previously studied, but also added three populations from the extreme periphery of the range in the central and southern Negev, as described in detail in Kark (1999). Diversity across the more continuous distribution range shows a unimodal trend of genetic diversity from core to periphery. Peak diversity was found in all four populations sampled in the ecotone region of Israel; it decreased toward both the Mediterranean core, as in previous years, and toward the arid range periphery, showing a hump-shaped pattern across the range. Similar patterns were seen at the phenotypic level (Kark et al. 2002). These findings support the hypothesis presented by Kark (1999), suggesting that diversity will show a hump-shaped unimodal pattern across the distribution range, with peak levels in intermediate (subperipheral) populations located in the turnover zone region.

The Purple Clover (Trifolium purpureum) and the Common Dactyl (Dactylis glomerata) Distribution The purple clover, Trifolium purpureum, is distributed throughout the Mediterranean basin. The common dactyl, Dactylis glomerata, is Mediterranean, Irano-Turanian, and Euro-Siberian. Both species are found throughout the Mediterranean regions of northern and central Israel, and reach the edge of their global range in the ecotone region in the Northern Negev. Unlike the chukar partridge, these plant species do not extend toward the more extreme arid Negev desert. The two species exhibit a patchy distribution in all regions of Israel, including the more mesic northern regions with high local variation. In some cases local density is very high and they are the dominant species (Danino 2000). Due to the limitations of the study design, populations with very low local densities were not included in the sampling. Trends in Diversity Across the Range A greenhouse experiment was conducted with Trifolium purpureum (an annual legume) and Dactylis glomerata (a perennial grass). Details are given in Danino (2000). Parent plants were collected in the field at three core populations (Mediterranean, ca. 600 mm mean annual rainfall), and four peripheral (from the ecotone region with ca. 300 mm mean annual rainfall) populations in Israel. Plants (whole clones in Dactylis and individual plants in Trifolium) were collected from an area of circa 1 ha, using a random spatial design. Only individuals above a certain size that were suitable for the


experimental design were sampled in all locations. Therefore, the sampled plants did not represent the entire size range in the population. This limitation was especially important for Dactylis glomerata in which many clones were excluded due to their small size. Sampling aimed to include plants of homogeneous size. Plants (cuttings for Dactylis and seedlings for Trifolium) were grown in pots in a greenhouse at the Sede Boqer Campus, located in the central Negev desert of Israel. The offspring of each parental plant (i.e., ‘‘family’’) were grown under high or low water availability (Danino 2000) for one season (November 1994–April 1995). In all, 73 Dactylis and 90 Trifolium plant families were studied, with at least three replicates from each family in both low and high water treatments. Offspring were measured for the following traits: total shoot biomass; branch or tiller biomass; number of branches or tillers; mean individual branch or tiller biomass; inflorescence biomass; number of inflorescence; mean individual biomass; mean inflorescence length; proportion of aborted inflorescence; reproductive effort (inflorescence biomass/total shoot biomass). A factorial layout with the main factors being water treatment and population location across the range was used for region (core vs. periphery), population-within-region (three in core, four in periphery), or population alone (seven populations). A nested ANOVA was used on the pooled data to test the effects of the treatment on each of the dependent variables (trait measurements) depicting various growth and allocation characteristics (Danino 2000). A significant difference was found between regions (core and periphery) for inflorescence biomass and for number and biomass of tillers in Dactylis and for both total and shoot biomass in Trifolium, with a general tendency toward fewer and bigger inflorescence or shoots in the core and more but smaller inflorescence or shoots in the ecotone periphery (Danino 2000). The effect of the individual population was highly significant for the majority of traits in both species and under both water treatments when calculated across all the populations. In most cases, populations also differed significantly within region. Region-by-water interaction was highly significant for percent aborted inflorescence and for the reproductive effort and marginally significant for total biomass and for the number of branches in Trifolium, yet none of these was significant in Dactylis. When calculated from the pooled data, population-by-water interactions were significant for the vast majority of traits studied in both species. While significant differences were found between the estimated genetic variances of the individual populations, no generalization could be made for such differences between the core and the periphery. It may be that in species with highly patchy distributions the differences between core and periphery populations are less than those between different populations within the same geographical region. We conclude that the studied populations can respond to natural selective pressures imposed by a reduction in water supply comparable to the conditions created by the low water treatment to potted plants. Furthermore, the large differences found among the sampled populations imply that each and


every one of the populations studied may represent a unique collection of genetic backgrounds. This would be interesting to study further. A possible interpretation of these results is that both the Mediterranean and the semiarid populations of Trifulium purpureum and Dactylis glomerata in Israel represent the ‘‘peripheries’’ of their distributions. If this is so, the more favorable regions of these and other species with similar ranges would be found in the temperate regions of Europe or, in the case of Irano-Turanian species, the highlands of central Asia. As noted earlier, populations with very low local densities could not be included in the study due to the design, which required certain sample sizes. These may actually be representing the very extreme environmental patches and may show different patterns of phenotypic diversity.

Wild Barley (Hordeum spontaneum) Distribution Wild barley, Hordeum spontaneum, the ancestor of the cultivated barley, is widely distributed across the eastern Mediterranean basin, and western and central Asia to western China, Pakistan, and India. It is one of the main annual components of open park-forests of Quercus ithaburensis in central and northern Israel. It is also abundant in the Mediterranean grasslands of northern Israel, Jordan, southern Syria, and Lebanon (Harlan and Zohary, 1966). The species is seldom found in regions where mean annual rainfall is less than 150 mm, where mean winter temperatures go below 58C, or at altitudes over 1500 m (Harlan and Zohary 1966, Harlan 1968). The species’ range periphery is found in loess or sand deserts. In these environments, low and unpredictable rainfall are the major limiting factors for this species. Plants in these peripheral populations occupy only accumulating runoff wadis or ravines. Population size declines toward the geographical range periphery, but local population density within the wadis is often as high as at the species core (Gutterman 1992, Volis, pers. obs.). Therefore, the scale at which population density is studied and is estimated may largely determine the spatial diversity trends detected in this species. Distinction between wateraccumulating and non-water-accumulating habitats, therefore, is useful, and is the most reliable criterion for defining the species’ range periphery at the local scale (Volis et al. 2001). Barley densities increase and distribution becomes less patchy in the more mesic Mediterranean areas, toward the core of its range. In this region populations are large and are rarely isolated. In the Negev and Judean Deserts, wild barley density decreases, distribution becomes discontinuous and coincides with wadi distribution alone. Local populations become highly patchy and isolated from each other. These desert regions of Israel, similar to the chukar partridge, constitute the southwestern border of species global distribution range.


Trends in Diversity Across the Range Genetic diversity in wild barley was studied at two levels: quantitative (phenotypic) traits and allozymes.

Phenotypic Diversity Ten populations of wild barley were sampled in 1993 in Israel. Of these, five populations, representing the core, were sampled in central Israel in the mesic Shefela and Judea Hills (mean annual rainfall ranged between 400 and 600 mm). Five peripheral populations were sampled from the Negev and Judean Deserts at the species range periphery (mean annual rainfall between 70 and 160 mm). The ecotone between the Mediterranean and desert ecosystems, which is the turnover zone of this species distribution, was not sampled in this study. Seeds were sown in 1994 and three-week-old plantlets were transplanted into an experimental field of the Institutes for Applied Research in Beer Sheva. Located in the northern Negev ecotone region between the two sampled regions, Beer Sheva receives 250 mm of annual rainfall. The morphological traits included: culm length, flag and penultimate leaf length, spike length, awn length, number of nodes, internode length and total tiller height, number of spikelets per spike, and the average spikelet and seed weight. In addition, two phenological traits were measured: (i) the number of days to awn appearance, indicating the onset of the reproductive phase and (ii) the number of days to anthesis. Analysis of phenotypic variation included a nested ANOVA and MANOVA approach with regions as a fixed effect and populations nested within regions as a random effect (for details see Volis et al. 2002). The degree of variation showed opposite patterns for different traits in plants of Mediterranean and desert origin. Our explanation of the findings employs a combined effect of directional and diversifying selection, possibly as a result of temporal heterogeneity. Israeli peripheral populations inhabiting unpredictable desert environments with respect to water availability apparently represent a ‘‘spreading risk’’ strategy with alternative phenotypes present in a population (Ellner 1985, 1987, Kaplan and Cooper 1984). The traits associated with this strategy are those that enhance temporal variation in the germination of the seeds and in the maturation of the plants, including the onset of germination, seed dormancy, and start of reproduction. Days to awn appearance and anthesis (indicating transition to reproductive stage) and flag and penultimate leaf length (determining grain filling and thus duration of reproductive stage until seed maturation) were more variable in peripheral, than in core populations. Large differences were found between seed dormancy of Israeli peripheral and core populations (Volis et al. 2004). The core populations were more variable than peripheral populations in most of the other traits. This may be explained by the fact that variability in traits that are not directly related to fitness has a lower ‘‘cost’’ and therefore, variability may be maintained under more favorable conditions.


Genetic Diversity The same 10 populations analyzed for phenotypic diversity and two additional populations were used to study genetic variability, as estimated using starch-gel electrophoresis of nine water-soluble leaf proteins (Volis et al. 2002). At the first stage, populations were pooled into two groups, representing core and periphery. Following this approach, no significant differences were found between the core and the periphery in any of the three estimates of genetic diversity (mean number of alleles per locus, proportion of polymorphic loci, and expected heterozygosity). Following the hump-shaped pattern hypothesis presented by Kark (1999), trends were retested using a continuous approach to the range (rather than the previous more traditional dichotomous approach, in which populations were pooled into core and periphery and the means for each of the regions were compared). A quadratic regression between allozyme diversity and mean annual rainfall was performed. A noticeable trend, although nonsignificant, appeared (fig. 3.2). Diversity within populations showed a hump-shaped pattern across the rainfall gradient. As in the case of the chukar partridge, a polynomial quadratic regression explained the findings much better than a linear model. Diversity in all three diversity estimates increased toward the Mediterranean desert ecotone (which was not sampled), where the edge of the species’ continuous distribution edge occurs. These results are in accordance with the hypothesis predicting a hump-shaped curve of within-population diversity across core– periphery clines (Kark 1999).

Figure 3.2 Allozyme diversity in 12 wild barley populations in Israel, estimated by percentage of polymorphic loci (P) and by mean expected heterozygosity based on Hardy–Weinberg equilibrium (He), as a function of mean annual rainfall. Rainfall is strongly correlated with the distance of the population from the species’ range periphery.


Synthesis of Case Studies The three studies presented above tested patterns of within-population diversity across the distribution range of four species with rather similar ranges in Israel, ranging from a phasianid bird to annual and perennial plants. The studies examined different measures of genetic and morphological diversity and had different experimental and sampling designs. Unlike previous work on aridity gradients in Israel, our main goal in this review is not to compare the findings of the studies but rather to emphasize examples of the approach taken in these case studies and their implications to the findings. It is interesting that the two studies that focused on the chukar partridge and the wild barley showed rather similar patterns. When only part of the species’ range in Israel was studied, or when populations were pooled for analysis, rather than compared across a continuum, very different patterns appeared, compared with an analysis where patterns were tested across a more continuous distribution gradient extending from the very edge (periphery) toward the interior of the species distribution (core). In addition, as emphasized by the chukar work, sampling of the intermediate ecotone region, located between the core and the periphery of these species’ ranges revealed a hump-shaped pattern that would have otherwise been overlooked. In addition, when the extreme periphery of the chukar range was not sampled, a partial pattern appeared. The barley work at its first stage did not detect any differences between the core and the peripheral populations in their levels of diversity. Yet when analyzed in more detail across a rainfall gradient, representing a distributional cline for this species, a pattern seems to emerge, pointing toward a hump-shaped unimodal pattern of diversity across the range. This work suggests that the scale on which the study focuses may be crucial in determining the patterns detected. Thus even studies of a single species, which define and sample populations at different spatial scales, may lead to different conclusions. A study on patchily distributed plants from peripheral habitats which includes ten isolated subpopulations, for example, could lead to a different finding compared with work in which each subpopulation represents a study population. Because different organisms differ in the graininess in which they respond to the environment, differences between studies may also arise when species of different body sizes and life histories are studied at one spatial scale. Additionally, populations are often defined based on area size, and therefore the size of the area sampled for each species may largely determine the patterns found. For example, an area of 10 km2 for a medium-sized bird such as the chukar, and a plant species can represent very different patterns of patchiness at the local scale, especially closer to the range periphery, where distribution often becomes less continuous. For example, while chukar distribution becomes more patchy toward the range periphery, distribution of the purple clover is very patchy in both the Mediterranean and more arid periphery. Thus it is important to carefully determine the spatial sampling design, based on the study goals. It is also useful to give these details in the methodology


chapter of the scientific reports of the work, to enable comparisons between studies. As mentioned earlier, this chapter focuses on patterns in diversity across distribution clines, rather than on environmental clines alone. We believe that the distribution range and its spatial patterns summarize the way that individuals of different populations respond to the environments in which they are found. Climatic variables often used to correlate with diversity patterns, such as mean annual temperature, may not reflect the environment that an individual really perceives in its microhabitat. The distribution range and population density are the bottom-line summary of the way the environment (both biotic and abiotic) influences species, and therefore may be a more useful tool, especially in studies that aim to derive recommendations for conservation.

Conservation Implications As discussed above, spatial scale may largely determine the patterns of diversity found in populations across a species’ distribution range. Therefore, the goals of conservation programs should be clearly set, and these goals should be kept in mind when setting conservation priorities. We suggest that mapping patterns of biological diversity across species distribution ranges, and especially across ecological gradients, is desired for maintaining high levels of genetic diversity. Conserving the areas in which the processes that are generating this diversity are occurring may be an important first step. Ecological gradients and areas of ecological transition may be good candidates. We suggest that more attention must be paid by conservation biologists toward areas of transition between climatic regions and among ecosystems (such as Mediterranean and desert in the case of Israel). In these areas, many species attain the edge of their continuous distributions. Currently, very little conservation attention is directed toward these areas. For example, the Global 200 program (Olson and Dinerstein 1998) does not emphasize areas of transition or ecological gradients. If these areas are shown to harbor high levels of genetic and morphological diversity in additional species, investing more effort in conserving them may prove to be a cost-effective conservation strategy.

Emphasis on Diversity in Drylands This chapter has focused on the study of diversity across species distribution (core–periphery) clines. In many cases these gradients are congruent with climatic (e.g., increasing aridity) gradients. While various species change their distribution pattern across these climatic gradients, species differ in their distribution patterns. For example, many species reach the edge of their distribution range in the ecotones of Israel. For some species the


Mediterranean desert ecotone of the northern Negev comprises a southern limit to their Palearctic distribution, while for other species this same area is a northern edge to their Saharan distribution or the western margins of their Irano-Turanian distribution (Danin 1998). Peripheral, core and ‘‘turnover zone’’ populations may thus be located in different ecological regions. The case studies presented in this chapter were all species with a southern distribution margin in the Negev desert. It would be interesting to continue this work and compare trends in diversity across species with opposing distributions, for which the desert is the core of the range and the Mediterranean region is the periphery. We predict that diversity patterns will correspond to the location across the range (i.e., periphery–core cline) rather than the climatic changes alone (e.g., the degree of aridity). General patterns in diversity, if they exist, are expected to be found across distribution gradients rather than climatic gradients alone. For many Saharan and Arabian species the arid region may actually prove to be a more favorable region of the range where density, diversity, and persistence are higher.

Acknowledgments Our thanks go to Philip Alkon, Avigdor Cahaner, Ayelet Danino, Samuel Mendlinger, Imanuel Noy-Meir, Ettore Randi, and Uriel Safriel for their invaluable collaboration and discussion in various parts of this work, and the Israel Nature and Parks Authority scientists and rangers for their assistance in field work. Support for this research was granted to S.K. from the Pontremoli and the Rieger Research Funds through Keren Kayemet LeIsrael (JNF), The Ecology Fund founded by the JNF, The Blaustein International Center for Desert Studies of the Blaustein Institute for Desert Research, Ben-Gurion University of the Negev, and by grants to A.N. from the Israel Science Foundation founded by The Israel Academy of Sciences and Humanities and the International Arid Land Consortium. This is publication number 356 of the Mitrani Department of Desert Ecology.

References Alkon, P.U. 1974. Social behavior of a population of chukar partridge (Alectoris chukar) in Israel. Ph.D. thesis, Cornell University, Ithaca, New York. Allard, R.W., R.D. Miller, and A.L. Kahler. 1978. The relationship between degree of environmental heterogeneity and genetic popymorphism. Pp. 49–73 in A.H.J. Freysen and J.W. Woldendorp, eds., Structure and Functioning of Plant Populations. North-Holland, New York. Andrewartha, H.G., and L.C. Birch. 1954. The Distribution and Abundance of Animals. University of Chicago Press, Chicago. Antonovics, J., P. Thrall, A. Jarosz, and D. Stratton. 1994. Ecological genetics of metapopulations: The Silene-Ustilago plant-pathogen system. Pp. 146–170 in L. Real, ed., Ecological Genetics. Princeton University Press, Princeton. Baruffi, L., G. Damiani, C.R. Guglielmino, C. Bandi, A.R. Malacrida, and G. Gasperi. 1995. Polymorphism within and between populations of Ceratitis capitata: Comparison between RAPD and multilocus enzyme electrophoresis data. Heredity 74: 425–437.


Bitan, A., and S. Rubin. 1991. Climatic Atlas of Israel for Physical and Environmental Planning and Design. The Israeli Ministry of Transport, Tel Aviv. Blackburn, T.M., K.J. Gaston, R.M. Quinn, and R.D. Gregory. 1999 range. Do local abundances of British birds change with proximity to range edge? Journal of Biogeography 26: 493–505. Blows, M.W., and A.A. Hoffmann. 1993. The genetics of central and marginal populations of Drosophila serrata. I. Genetic variation for stress resistance and species borders. Evolution 47: 1255–1270. Boorman, S.A., and P.R. Levitt. 1973. Group selection on the boundary of a stable population. Theoretical Population Biology 4: 85–128. Brooks, D.R., R.L. Mayden, and D.A. McLennan. 1992. Phylogeny and biodiversity: Conserving our evolutionary legacy. Trends in Ecology and Evolution 7: 55–59. Brown, J.H. 1984. On the relationship between abundance and distribution of species. The American Naturalist 124: 255–279. Brown, J.H., D.W. Mehlman, and G.C. Stevens. 1995. Spatial variation in abundance. Ecology 76: 2028–2043. Brown, J.H., G.C. Stevens, and D.M. Kaufman. 1996. The geographic range: size, shape, boundaries, and internal structure. Annual Review of Ecology and Systematics 27: 597–623. Brussard, P.F. 1984. Geographic patterns and environmental gradients: the centralmarginal model in Drosophila revisited. Annual Review of Ecology and Systematics 15: 25–64. Burger, R. 1988. The maintenance of genetic variation: A functional analytic approach to quantitative genetic models. Pp. 63–72 in G. de Jong, ed., Population Genetics and Evolution. Springer-Verlag, Berlin. Carmi-Winkler, N., A.A. Degen, and B. Pinshow. 1987. Seasonal time-energy budgets of free-living chukars in the Negev Desert. The Condor 89: 594–601. Carson, H.L. 1959. Genetic conditions that promote or retard the formation of species. Cold Spring Harbor Symposium on Quantitative Biology 24: 87–103. Carter, R.N., and S.D. Prince. 1988. Distribution limits from a demographic viewpoint. Pp. 165–184 in A.J. Davy, M.J. Hutchings, and A.R. Watkinson, eds., Plant Population Ecology. Blackwell Scientific Publications, Oxford. Caughley, G., D. Grice, R. Barker, and B. Brown. 1988. The edge of the range. Journal of Animal Ecology 57: 771–785. Channell, R., and M. Lomolino. 2000a. Dynamic biogeography and conservation of endangered species. Nature 403: 84–86. Channell, R., and M. Lomolino. 2000b. Trajectories to extinction: spatial dynamics of the contraction of geographic ranges. Journal of Biogeography 27: 169–179. Collins, S. L., and S.M. Glenn. 1991. Importance of spatial and temporal dynamics in species regional abundance and distribution. Ecology 72: 654–664. Da Cunha, A.B., and T. Dobzhansky. 1954. A further study of chromosomal polymorphism in Drosophila willistoni in its relation to the environment. Evolution 8: 119–134. Danin, A. 1998. Man and the natural environment. Pp. 25–37 in T. E. Levy, ed., The Archaeology of Society in the Holy Land. Leicester University Press, Leicester. Danino, A. 2000. Biological responses to global change—differences between central and peripheral populations, M.Sc. thesis, Ben Gurion University of the Negev (in Hebrew with English summary).

Biodiversity Along Core–Periphery Clines 53 Degen, A.A., B. Pinshow, and P.U. Alkon. 1983. Summer water turnover weights in free-living chukars (Alectoris chukar) and sand partridges (Ammoperdix heyi) in the Negev desert. The Condor 85: 333–337. Degen, A.A., B. Pinshow, and P.J. Shaw. 1984. Must desert chukars (Alectoris chukar sinaica) drink water? Water influx and body mass changes in response to dietary water content. The Auk 101: 47–52. Descimon, H., and M. Napolitano. 1993. Enzyme polymorphism, wing pattern variability, and geographical isolation in an endangered butterfly species Biological Conservation 66: 117–123. Ellner, S. 1985. ESS germination strategies in randomly varying environments. Theoretical Population Biology 28: 50–116. Ellner, S. 1987. Alternative plant life history strategies and coexistence in randomly varying environments. Vegetatio 69: 199–208. Endler, J.A. 1982. Problems in distinguishing historical from ecological factors in biogeography. American Zoologist 22: 441–452. Fisher, R.A. 1930. The Genetical Theory of Natural Selection. Clarendon Press, Oxford. Frankham, R. 1996. Relationship of genetic variation to population size in wildlife. Conservation Biology 10: 1500–1508. Frumkin, R. 1983. Thermoregulation and acid base status in heat stressed desert phasianids: A comparison between the chukar partridge (Alectoris chukar) and the sand partridge (Ammoperdix heyi). Ben-Gurion University of the Negev (in Hebrew, abstract in English). Futuyma, D. 1997. Evolutionary Biology. Sinauer Associates, Sunderland, MA. Gaston, K. J. 1990. Patterns in the geographical ranges of species. Biological Reviews of the Cambridge Philosophical Society 65: 105–129. Gaston, K. J., and J. Spicer. 1998. Biodiversity: an Introduction. Blackwell Science, Oxford. Gutterman, Y. 1992. Ecophysiology of Negev upland annual grasses. Pp. 145–162 in G.P. Chapman, ed., Desertified Grasslands: Their Biology and Management. Linnean Society Symposium. Series 13, Academic Press, London. Hanski, I. 1989. Metapopulation dynamics: does it help to have more of the same? Trends in Ecology and Evolution 4: 113–114. Harlan, J.R. 1968. On the Origin of Barley. U.S.D.A. Agricultural Handbook 338: 9–13. Harlan, R.J., and D. Zohary. 1966. Distribution of wild wheats and barley. Science 153: 1074–1080. Harrison, S. 1994. Metapopulations and conservation. Pp. 111–128 in P. J. Edwards, R.M. May, and N.R. Webb, eds., Large Scale Ecology and Conservation Biology. Blackwell Scientific Publications, Oxford. Hengeveld, R., and J. Haeck. 1981. The distribution of abundance. II. Models and implications. Proc. Kon. Ned. Akad. Wetensch., C 84: 257–284. Hoffmann, A.A., and M.W. Blows. 1994. Species borders: ecological and evolutionary perspectives. Trends in Ecology and Evolution 9: 223–227. Hoffmann, A.A., and P.A. Parsons. 1991. Evolutionary Genetics and Environmental Stress. Oxford University Press, Oxford. IUCN. 1993. The Convention on Biological Diversity: An Explanatory Guide. The IUCN Environmental Law Centre (ELC), Bonn. Kadmon, R., and A. Danin. 1997. Floristic variation in Israel: a GIS analysis. Flora 192: 341–345.


Kam, M. 1986. Energy budget and seasonal water of free chukar (Alectoris chukar) and sand partridges (Ammoperdix heyi) in the Negev Highlands. Ben Gurion University (in Hebrew, abstract in English). Kaplan, R.H. and W.S. Cooper. 1984. The evolution of developmental plasticity in reproductive characteristics: an application of the adaptive coin-flip principle. The American Naturalist 123: 393–410. Kark, S. 1999. Within-population diversity in the distribution range: Partridges as a research model. Department of Evolution, Systematics and Ecology. The Hebrew University of Jerusalem, and Ecology Research Center. http://shum.huji.ac.il/ salit Kark, S., P.U. Alkon, U.N. Safriel, and E. Randi. 1999. Conservation priorities for the chukar partridge in Israel based on genetic diversity across an ecological gradient. Conservation Biology 13: 542–552. Kark, S., T. Mukerji, U.N. Safriel, R. Nissani, and A. Darvasi. 2002. Peak morphological diversity in an ecotone unveiled in the chukar partridge by a novel Estimator in a Dependent Sample (EDS). Journal of Animal Ecology 71: 1015–1029. Lacy, R.C. 1997. Importance of genetic variation to the viability of mammalian populations. Journal of Mammalogy 78: 320–335. Lawton, J.H. 1995. Population abundances, geographic ranges and conservation: 1994 Witherby lecture. Bird Study 43: 3–19. Lawton, J.H., S. Nee, A.J. Letcher, and P.H. Harvey. 1994. Animal distributions: patterns and processes. Pp. 41–58 in P.J. Edwards, R.M. May, and N.R. Webb, eds., Large Scale Ecology and Conservation Biology. Blackwell Scientific Publications, Oxford. Lennon, J.J., J.R.G. Turner, and D. Connell. 1997. A metapopulation model of species boundaries. Oikos 78: 486–502. Lesica, P., and F.W. Allendorf. 1995. When are peripheral populations valuable for conservation? Conservation Biology 9: 753–760. Lomolino, M.V., and R. Channell. 1995. Splendid isolation: patterns of geographic range collapse in endangered mammals. Journal of Mammalogy 76: 335–347. Mayr, E. 1963. Animal Species and Evolution. Belknap Press/Harvard University Press, Cambridge, MA. Mayr, E. 1970. Populations, Species, and Evolution. Belknap Press, Cambridge, MA. McClintock, K.A. and M.J. Waterway. 1993. Patterns of allozyme variation and clonal diversity in Carex lasiocarpa and C. pellita (Cyperaceae). American Journal of Botany, 80: 1251–1263. Michaud, H., L. Toumi, R. Lumaret, T.X. Li, F. Romane, and F. Di Giusto. 1995. Effect of geographical discontinuity on genetic variation in Quercus ilex (holm oak). Evidence from enzyme polymorphism. Heredity 74: 590–606. Mittermeier, R.A., N. Myers, J.B. Thomsen, and A.B. da Fonseca. 1998. Biodiversity hotspots and major tropical wilderness areas: approaches to setting conservation priorities. Conservation Biology, 12: 516–532. Møller, A.P. 1995. Patterns of fluctuating asymmetry in sexual ornaments of bird from marginal and central populations. The American Naturalist 145: 316–327. Myers, N. 1990. The biodiversity challenge: expanded hotspots analysis. Environmentalist 10: 243–256. Myers, N., R. Mittermeier, C. Mittermeier, G. da Fonseca, and J. Kent. 2000. Biodiversity hotspots for conservation priorities. Nature (London) 403: 853–858.

Biodiversity Along Core–Periphery Clines 55 Nevo, E., and A. Beiles. 1988. Genetic parallelism of protein polymorphism in nature: ecological test of the neutral theory of molecular evolution. Biological Journal of the Linnean Society 35: 229–245. Nevo, E., and A. Beiles. 1989. Genetic diversity in the desert: patterns and testable hypotheses. Journal of Arid Environments, 17: 241–244. Olson, D.M., and E. Dinerstein. 1998. The Global 200: A representation approach to conserving the Earth’s most biologically valuable ecoregions. Conservation Biology 12: 502–515. Parsons, P.A. 1989. Environmental stresses and conservation of natural populations. Annual Review of Ecology and Systematics 20: 29–49. Parsons, P.A. 1991. Evolutionary rates: stress and species boundaries. Annual Review of Ecology and Systematics 22: 1–18. Paz, U. 1987. The Birds of Israel. Steimatzky, Tel-Aviv. Pinshow, B., P.U. Alkon, and A.A. Degen. 1982. Ecophysiological studies on water, energy and thermoregulatory relationships of Negev desert chukars and sand partridges. Israel Journal of Zoology 31: 63. Pinshow, B., A.A. Degen, and P.U. Alkon. 1983. Water intake, existence energy and responses to dehydration in the sand partridge (Ammoperdix heyi) and the chukar (Alectoris chukar), two phasianids of the Negev desert. Physiological Zoology 56: 281–289. Poiani, K., B. Richter, M. Anderson, and H. Richter. 2000. Biodiversity conservation at multiple scales: Functional sites, landscapes and networks. BioScience 50: 133–146. Pressey, R.L., C.J. Humphries, C.R. Margules, R.I. Vane-Wright, and P.H. Williams. 1993. Beyond opportunism: key principles for systematic reserve selection. Trends in Ecology and Evolution 8: 124–128. Purvis, A., and A. Hector. 2000. Getting the measure of biodiversity. Nature 405: 412–419. Randi, E., S. Kark, and C. Tabarroni. In review. Population structure of chukars (Alectoris chukar) distributed across an ecotone: A nested clade analysis of mitochondrial DNA haplotypes. Journal of Molecular Ecology. Rice, K.J., and R.N. Mack. 1991. Ecological genetics of Bromus tectorum. 1. A hierarchical analysis of phenotypic variation. Oecologia 88: 77–83. Rice, W.R., and E.E. Hostert. 1993. Laboratory experiments on speciation: what have we learned in 40 years? Evolution 47: 1637–1653. Riddle, B.R. 1996. The molecular phylogeographic bridge between deep and shallow history in continental biotas. Trends in Ecology and Evolution 11: 207–211. Safriel, U.N., S. Volis, and S. Kark. 1994. Core and peripheral populations, and global climate change. Israel Journal of Plant Sciences 42: 331–345. Schnabel, A., J.L. Hamrick, and P.V. Wells. 1993. Influence of Quaternary history on the population genetic structure of Douglas-fir (Pseudotsuga menziestii) in the Great Basin. Canadian Journal of Forest Research 23: 1900–1906. Schneider, C., T. Smith, B. Larison, and C. Moritz. 1999. A test of alternative models of diversification in tropical rainforests: Ecological gradients vs. rainforest refujia. Proceedings to the National Academy of Sciences 96: 13869–13873. Shirihai, H. 1996. The Birds of Israel. Academic Press, London. Shumaker, K.M., and G.R. Babble. 1980. Patterns of allozymic similarity in ecologically central and marginal populations of Hordeum jubatum in Utah. Evolution 34: 110–116.


Slatkin, M. 1994. Gene flow and population structure. Pp. 146–170 in L. Real, ed., Ecological Genetics. Princeton University Press, Princeton. Smith, T.B., R.K. Wayne, D.J. Girman, and M.W. Bruford. 1997. A role for ecotones in generating rainforest biodiversity. Science 276: 1855–1857. Svensson, B.W. 1992. Changes in occupancy, niche breadth and abundance of three Gyrinus species as their respective range limits are approached. Oikos 63: 147–156. Tani, N., N. Tomaru, M. Araki, and K. Ohba. 1996. Genetic diversity and differentiation in populations of Japanese stone pine (Pinus pumila) in Japan. Canadian Journal of Forest Research, 26: 1454–1462. Thomas, C.D., and I. Hanski. 1997. Butterfly metapopulations. Pp. 359–386 in I. Hanski and M.E. Gilpin, eds., Metapopulation Biology: Ecology, Genetics, and Evolution. Academic Press, San Diego. United Nations Convention 1992. Earth Summit—convention on biological diversity, Rio de Janeiro. Volis, S., S. Mendlinger, L. Olsvig-Whittaker, U.N. Safriel, and N. Orlovsky. 1998. Phenotypic variation and stress resistance in core and peripheral populations of Hordeum spontaneum. Biodiversity and Conservation 7: 799–813. Volis, S., S. Mendlinger, and N. Orlovsky. 2001. Variability in phenotypic traits in core and peripheral populations of wild barley, Hordeum spontaneum Koch. Hereditas 133: 235–247. Volis, S., S. Mendlinger, Y. Turuspekov, and U. Esnazarov. 2002. Phenotypic and allozyme variation in Mediterranean and desert populations of wild barley, Hordeum spontaneum. Evolution 56: 1403–1415. Volis, S., S. Mendlinger, and D. Ward. 2004. Demography and role of the seed bank in Mediterranean and desert populations of wild barley, Hordeum spontaneum Koch. Basic and Applied Ecology 5: 53–64. Vrijenhoek, R.C., M.E. Douglas, and G.K. Meffe. 1985. Conservation genetics of endangered fish populations in Arizona. Science 229: 400–402. Wade, M., and C. Goodnight. 1998. Perspective: The theories of Fisher and Wright in the context of metapopulations: When nature does many small experiments. Evolution 52: 1537–1553. White, M.J.D. 1951. Structural heterozygosity in natural populations of the grasshopper Trimerotropis sparsa. Evolution 5: 376–394. Wiens, J.A. 1989. The Ecology of Bird Communities. Cambridge University Press, Cambridge. Wilson, J.B., Y. Ronghua, A.F. Mark, and A.D.Q. Agnew. 1991. A test of the low marginal variance (LMV) theory, in Leptospermum scoparium (Myrtaceae). Evolution 45: 780–784. Wright, S. 1932. The roles of mutation, inbreeding, crossbreeding, and selection in evolution. Proceedings of the Sixth International Congress of Genetics 1: 232–248. Yom-Tov, Y., and E. Tchernov. 1988. The Zoogeography of Israel. Dr W. Junk, Dordrecht. Zanetto, A., and Z. Kremer. 1994. Genetic variation in natural populations of oaks in Europe: Quercus petraea and Quercus robur. In Conservation of the Wild Relatives of European Cultivated Plants III Workshop. Gibilmanna–Palermo.

4 Species Diversity, Environmental Heterogeneity, and Species Interactions William A. Mitchell Burt P. Kotler Joel S. Brown Leon Blaustein Sasha R.X. Dall

D

espite their apparent simplicity, arid environments can be quite heterogeneous. From small-scale variation in substrate and slope to large-scale geographic variation in solar input and productivity, drylands and deserts provide organisms with a tremendous range of ecological challenges (Schmidt-Nielsen 1964, Huggett 1995). Any single species is unable to meet all of these challenges equally well. A species will do better in some environments than others because evolution in heterogeneous environments is constrained by fitness tradeoffs. Such tradeoffs prevent the evolution of a versatile species, competitively superior to all other species across the entire spectrum of heterogeneity (Rosenzweig 1987). Although fitness tradeoffs may hinder species’ evolution in heterogeneous environments, they are a blessing for biodiversity. The source of biodiversity that we address in this chapter is the interplay of heterogeneity, tradeoffs, and density dependence. While we focus on species interactions at the local scale, our presentation includes a model that predicts changes in local diversity as a function of climate. The model’s predictions are based on changes in the nature of competition wrought by changes in productivity levels and climatic regimes. Cast in terms of evolutionary stable strategies (ESSs), the predictions refer to evolutionary as well as ecological patterns.

Mechanisms of Coexistence A mechanism of coexistence consists of an axis of environmental heterogeneity together with an axis that indicates a tradeoff in the abilities of species 57


to exploit different parts of the axis. In the absence of some kind of heterogeneity, there is only one environmental type, and whatever species is best adapted to it will competitively exclude others. In the absence of a tradeoff, one species could evolve competitive superiority over the full range of heterogeneity, again resulting in a monomorphic community. Consider some examples of mechanisms of species’ coexistence from dryland communities (Kotler and Brown 1988, Brown et al. 1994). For many taxa, spatial heterogeneity in predation risk is a consequence of the pattern of bushy and open areas common in drylands. In certain rodent communities, some species are able to exploit the relatively riskier open microhabitats by virtue of antipredator morphologies (Kotler 1984). One of these morphologies is bipedalism, which confers agility and fast speed on its owner at the cost of maneuverability in shrubs. Another tradeoff is based on rodent body size. Larger rodents possess larger auditory bullae, especially in the family Heteromyidae (Kotler 1984). The size of the bullae reflects sensitivity to low field vibrations (Webster and Webster 1971, Lay 1974) and the ability to detect approaching predators (Webster and Strother 1972). Large rodents acquire these facilities, however, at the cost of an overall higher metabolic rate, whereas smaller rodents are able to forage profitably on lower seed densities. An example of these tradeoffs in action is a six-species rodent community at Tonopah Junction, Nevada, in the Great Basin Desert, which segregates species by body size along a bush-open microhabitat axis of heterogeneity (Kotler 1984). Other examples of mechanisms of species coexistence involve birds. Bushopen microhabitat selection among overwintering sparrows in semiarid grasslands apparently is based on differences in escape abilities, with some species being more vulnerable than others away from cover (Pulliam and Mills 1977). This can result in coexistence if the most vulnerable species is also the best resource competitor near protective cover. Each species then has a microhabitat in which it can profit more than its competitor. A rodent community in the creosote shrublands of the Sonoran Desert exemplifies a different sort of tradeoff, this time involving travel efficiency versus foraging efficiency (Brown 1989a). The salient feature of the environment is the tremendous spatial heterogeneity in seed densities (adjacent patches can vary by as much as 70-fold in seed availability; Price and Reichmann 1987). The tradeoff is mediated via body size. The round-tailed ground squirrel (Spermophilus tereticaudus) uses its larger size and speed to move quickly from patch to patch, visiting many patches in each foraging bout. Its high speed lowers its travel costs in both time and energy. This allows it to behave like a ‘‘cream skimmer’’ (Brown et al. 1994) finding and exploiting only the richest parts of the richest patches. The smaller Merriam’s kangaroo rat moves less frequently, but uses its lower metabolic rate and foraging costs to exploits patches more thoroughly and to lower seed densities like a ‘‘crumb picker’’ (Brown 1989b, Brown et al. 1994), thereby allowing it to coexist with the larger ground squirrel. A similar example involves the exploitation of a pulsed resource of nectar by solitary bees in the Sonoran Desert (Schaffer et al. 1979). Here, body size

Species Diversity, Environmental Heterogeneity, and Species Interactions 59

and wing disc loading appear to be involved in tradeoffs between harvest rate and foraging efficiency. Larger bees have higher wing disc loading, are faster fliers, and presumably can harvest nectar more quickly, but they have high foraging costs due to the high disc loading. In contrast, because smaller bees have lower wing disc loading, they have less expensive (albeit slower) flight, and presumably lower foraging costs. Bees show strong temporal partitioning, with larger bees foraging earlier in the day when nectar is most abundant, and smaller bees continuing to forage even as nectar is depleted. These examples illustrate the variety of circumstances under which heterogeneity and tradeoffs may combine to promote the ecological coexistence of competing species. In the model section of our chapter we employ the framework of heterogeneity and tradeoffs to study evolutionarily stable strategies in competitive communities. Using the model, we can predict how biodiversity changes as a function of productivity and climate (heat), parameters that show wide variation among dryland communities. We start with a brief overview of the two patterns of species diversity that motivated the modeling.

The Effects of Productivity and Climate (Heat) on Species Diversity One of the more puzzling patterns in ecology is the hump-shaped relationship between species richness and productivity (Rosenzweig and Abramsky 1993, Waide et al. 1999, Dodson 2000, Gross 2000, Mittelbach et al. 2001). When productivity is low, increased productivity is accompanied by increased diversity. But when productivity is high, increased in productivity corresponds to a decline in diversity. The first part of the pattern seems easier to explain: more productivity yields more individuals across species, and the more individuals there are in each species, the less likely a species is to go locally extinct. This has been referred to as the ‘‘more individuals’’ hypothesis (Srivastava and Lawton 1998). The declining part of the productivity–diversity curve is more problematic. Why should higher productivity yield fewer species? Rosenzweig (1995) reviewed nine hypotheses listed in the literature and found problems with most of them. Some lacked a mechanistic basis, others would not hold in evolutionary time, and still others appeared logically circular. There remains a need for a mechanistically based hypothesis that predicts the declining phase of the productivity–diversity curve in both ecological and evolutionary time. In the present chapter, we try to address this need (see next section). The presence of a hump-shaped productivity–diversity curve means that some taxa will reach their maximum diversity in relatively unproductive drylands. For example, North American rodent diversity peaks in the mixed desert grassland of the Sonoran desert, and declines as productivity increases eastward (Rosenzweig 1995). In other cases, the dryland system can include both the peak diversity and part of the declining phase. For example,


Rosenzweig (1995) finds the hump-shaped pattern in rodents of the Gobi desert, and Abramsky and Rosenzweig (1984) observe the pattern in Middle Eastern rodent assemblages. A different pattern of species diversity shows that diversity increases with environmental heat, such as solar input, temperature, and potential evapotranspiration (Wright et al. 1993). The correlation between diversity and environmental heat may appear at first glance to be an artifact of the wellknown correlation between diversity and latitude, but when latitude and heat compete for variance in diversity in regression analyses, heat usually wins (Currie 1991). Why should diversity increase with heat? Just as in the case of the increasing phase of the diversity–productivity curve, one explanation hinges upon population densities being higher where environments are warmer, due to lower metabolic costs of existence. If warmer climates support more total individuals, then species extinction rates will be lower, resulting in higher species diversity (Wright et al. 1993). This version of the ‘‘more individuals’’ hypothesis proposed to explain the increase phase of diversity with resource productivity (Srivastava and Lawton 1998). But there is a problem with this hypothesis. For example, in North America, the higher latitudes are both cooler and less speciose. Yet average population densities are actually higher in the north (Currie and Fritz 1993), rather than lower, as would be expected if the ‘‘more individuals’’ hypothesis accounted for the lower species diversity there. As in the case of the productivity–diversity pattern, the correlation of diversity with environmental heat calls for further work in developing and testing mechanistic models of species diversity. The need for predictive, mechanistic models is made imperative because purely correlational studies suffer from the inevitable correlations among the putative explanatory variables, including heat and productivity. Our model includes parameters for both productivity and maintenance cost, which in endotherms should be related to environmental heat. Consequently, we are able to use the model to predict how productivity and environmental heat can drive species diversity via mechanisms of coexistence. Furthermore, because it is an ESS model, the predictions remain valid for evolutionary time.

The Model Our model incorporates environmental heterogeneity and an evolutionary tradeoff. We model environmental heterogeneity as a continuous distribution of habitat types, that is, there exists an infinite variety of habitats. A continuous distribution is probably more realistic than a discrete and finite distribution. As ecologists, we may distinguish microhabitats by discrete categories such as ‘‘bush’’ and ‘‘open’’ for the sake of field experiments and observations, but bush size, percent bush cover, or distance from the bush may be more important, and these are continuous variables.


We also assume a continuous distribution of heritable phenotypes in the evolutionarily feasible set. These phenotypes are subject to an evolutionary tradeoff such that each particular phenotype is superior in one habitat over all other phenotypes, and different phenotypes are better suited to different habitats (fig. 4.1). By assuming a continuous distribution for habitat types and strategies, we avoid setting an a priori limit to the number of species that can coexist in either ecological or evolutionary time. Therefore, any limit to, or change in, diversity predicted by our model will flow from the integrated effects of environmental heterogeneity and fitness tradeoffs, together with climate and resource productivity (Mitchell 2000, Mitchell and Porter 2001). In overview, our model has the following characteristics. Individuals bias foraging to those habitats that are more profitable, i.e. where levels of food resource and habitat foraging costs define habitat profitability. Levels of food resource, and hence individual foraging effort, decrease with the density and foraging effort of intra- and interspecific competitors (Mitchell et al. 1990). Foraging cost depends on the combination of habitat and phenotype. For this model, we represent the cost function for a given strategy, u, as a quadratic with its minimum value at a habitat in which the strategy pays its lowest foraging cost. Different strategies possess different cost functions with their minima located in different habitats. Thus, for each of the infinite number of habitat types, there exists a strategy that is superior in that habitat to all other strategies (fig. 4.1). Finally, we assume that populations are limited by, and hence fitness dependent on, a combination of food and nonenergetic factors.

Figure 4.1 A representative set of foraging cost curves, Cost ðu; zÞ; used in our model. Different foraging cost curves correspond to different evolutionary strategies. A cost curve shows the rate of energy expended by an individual of a particular strategy while it forages as a function of habitat of type z. Each strategy pays its lowest foraging cost in one habitat type, and increasingly higher costs as the habitats become increasingly different from the lowest cost habitat. A fitness tradeoff results because different strategies pay their lowest cost in different habitats.


Details of the Model We assume that fitness is a function of both energetic and nonenergetic terms. The energetic term includes the energy available for reproduction, which is the difference between the net return from foraging and metabolic maintenance cost. The nonenergetic term represents the other effects of crowding or density dependence aside from energy intake (e.g., limited nest sites or burrow refuges). A general way to express these assumptions is with the function Fitness ¼ ½b ðDaily Netprofit MCÞ ndayseN

ð1Þ

where b = Conversion rate of energy into offspring (offspring/ KJ) Daily Netprofit = Daily foraging profit (KJ/(day individual) MC = Daily maintenance cost (KJ/(day individual) ndays = Number of days per generation (days/generation) b = Density-dependent term representing the effects of crowding, independent of resource competition (i.e., limited nest sites) N = Total population density in the community Here, we define maintenance cost (MC) to be the lowest rate of energy expenditure by a resting individual in its environment. Cooler environments impose higher MC due to thermoregulation. The activity cost attributable to foraging (shown in fig. 4.1), on the other hand, is accounted for in the Daily Netprofit, so it does not increase maintenance cost. We must expand the term for Daily Netprofit in equation (1) to account for environmental heterogeneity, evolutionary tradeoffs, and behavior. We calculate Daily Netprofit as the integral over all habitat types of an individual’s net profit in each habitat. The net profit in a habitat is the difference between energy intake and the energy expended due to foraging, which in turn depend on the combination of the strategy and the habitat type. Letting u represent strategy and z represent habitat, the term for net profit in a habitat becomes Netprofitðu; zÞ ¼ Energy Intakeðu; zÞ Energy Expenditureðu; zÞ

ð2Þ

We assume that individuals deplete resources in a habitat as they forage. Therefore, energy intake from a patch is simply the difference between the initial and quitting resource densities, multiplied by the per item value of the resource, v, Energy Intakeðu; zÞ ¼ ½Initial Resource DensityðzÞ Quitting Resource Densityðu; zÞv

ð3Þ

Energy expenditure in a habitat is equal to the rate of energy expenditure attributable to foraging multiplied by foraging time in the habitat. As stated previously, and represented in fig. 4.1, we indicate the rate of energy expenditure due to foraging in a habitat by the function Cost (u, z). Foraging time


in a habitat is the time it takes the forager to deplete resources from the initial resource density to the quitting resource density. For simplicity, we assume that the rate of harvest, and hence depletion, is proportional to resource density. Letting a be the coefficient relating resource density to the harvest rate, the foraging time becomes 1 Initial Resource DensityðzÞ Foraging timeðu; zÞ ¼ ln ð4Þ a Quitting Resource Densityðu; zÞ The quitting resource density represents the behavioral choice, or patchleaving rule, of the forager. The forager will achieve its greatest fitness by quitting a habitat patch when the rate of return from the patch equals the cost of foraging the patch (Charnov 1976, Brown 1988). In our model there is no missed opportunity cost or predation risk, so the best time for a forager to leave is when the rate of energy harvest equals Cost (u,v). The rate of energy harvest at any resource density is the density multiplied by av, so, the quitting resource density is defined by the equation Quitting Resource Densityðu; zÞ ¼ Costðu; zÞ=av

ð5Þ

By substituting equations (2–5) into equation (1), we can determine the behavior and fitness of any phenotype, u, once we know the initial resource levels for all the habitats. We solve for initial resource levels as a dynamic balance between resource exploitation and resource renewal. To account for exploitation of resources, we envision foragers that move randomly among habitats, so that the frequency of visits to a habitat is proportional to the density of searching foragers. Hence, the opportunity for a habitat to renew between forager visits decreases with forager density. To account for renewal, we model a rate of resource input to all habitats. This rate represents productivity in our model.

Determining the ESS and Community Invasibility Fitnesses in our model are frequency- and density-dependent. We can illustrate this fact graphically by mapping fitness on an adaptive landscape. An adaptive landscape illustrates fitness as a function of strategy, for a range of possible strategies. An evolutionarily stable strategy (ESS) is a strategy that cannot be invaded by a rare alternative strategy; that is, all feasible alternative strategies will possess lower fitness. Hence a resident strategy is an ESS if it resides on a ‘‘peak’’ in the adaptive landscape. Alternatively, if the resident resides on a slope in the adaptive landscape, that strategy can be invaded by an alternative strategy from the ‘‘upslope’’ direction. To generate an adaptive landscape, we first introduce one or more species as residents. Then we use equations (1–5) to solve numerically for resident equilibrium population densities, that is, the densities at which resident fitnesses equal one. By virtue of their foraging, the residents generate a resource distribution across the range of habitat types.


Any nonresident species that tries to invade will encounter the resource levels generated by the foraging of residents. We can use equations (1)–(5) to calculate fitnesses of nonresidents under the assumption that they are too rare to significantly impact resource distributions. If the nonresident’s fitness is less than one, it cannot successfully invade. But if its fitness is greater than one, the propagule increases in density until density-dependence reduces its fitness to one. In the process of growing to its equilibrium, the successful invader will reshape the habitat resource distribution. As a result, the adaptive landscape changes to reflect the new fitnesses of the remaining set of nonresidents. In some cases a resident will prevent all nonresidents from successfully invading. This resident (or set of residents) will occupy a global peak (or peaks) in the adaptive landscape. If, however, a resident occupies a hillside in the adaptive landscape, it is subject to invasion or directional selection. Other configurations of the adaptive landscape are possible, as we will show in the next section. All numerical analysis was performed in QuickBasic, version 4.5 (computer program available from the authors). Equilibrium densities were solved using a bisection routine, and the integration over habitats to calculate net foraging profit was performed using Romberg integration with an adaptive step size. Evolution of the Adaptive Landscape If a resident is not at an ESS, the system can be invaded, resulting in a modified adaptive landscape. Figure 4.2 illustrates this process in a system characterized by a low maintenance cost (high environmental heat) and low productivity. The endpoint of the invasions is a species that resides in a valley of the adaptive landscape. This valley is a stable minimum in that the species is not able to evolve out of it—any slight alteration in strategy by the species results in directional selection for the species to return to the stable minimum (Mitchell 2000). The stable minimum can be invaded by a second species or a reproductive isolate of the original resident. The two species are then subject to divergent selection, and evolve to occupy two distinct peaks in the adaptive landscape (fig. 4.2D). The Effects of Environmental Heat (Maintenance Cost) and Productivity on Diversity Holding productivity constant and increasing maintenance cost (MC) has a dramatic effect on the adaptive landscape and the invasibility of the system (fig. 4.3). At the low MC, a single species evolves to a stable minimum as described above. This adaptive landscape is susceptible to invasion, or competitive ecological speciation (Schluter 2001), followed by divergent selection to yield a two-species ESS. At a higher level of MC however, the ‘‘valley’’ of the landscape flattens out, producing at first a small local peak. This local ESS is not subject to directional selection, but it can be invaded by a species whose strategy differs significantly. At still higher MC, the ‘‘shoulders’’ of the


Figure 4.2 A sequence of adaptive landscapes in a system characterized by a low maintenance cost (MC ¼ 20 KJ/day) and low productivity (Prod ¼ 20). (A) A landscape generated by a single resident species that is not an evolutionarily stable strategy (ESS). The landscape rises sharply to the right of the strategy used by the species, indicating the strategies that could successfully invade the system. (B) The new adaptive landscape that results when the resident species evolves by directional selection or is simply replaced by the invasion of a new species positioned to the right on the adaptive landscape. (C) The endpoint of a single-species evolution is that the species occupies a stable minimum in the adaptive landscape. This adaptive landscape can be invaded by a different species, or disruptive selection may promote sympatric speciation. (D) A two-species ESS that results if the stable minimum is invaded.

landscape fall below a fitness value of one, and a single resident will inhabit a global peak, which is an ESS. By changing nothing more than MC, we have reduced the number of species that can coexist, even in evolutionary time. Increasing productivity produces results that are qualitatively similar to the results of increasing MC (fig. 4.3). The increase in productivity reduces the number of species in the ESS from two to one. Mechanistically, what determines whether two species can coexist? Coexistence is possible if the combination of MC and productivity allow different species to use the heterogeneity with sufficient difference. Maintenance cost and productivity determine the level of resources required by species to possess a fitness of one. When MC and productivity are low, species are able achieve a fitness of one even when competition holds the encountered resources in habitats to low levels. At low resource levels, habitat overlap among competitors is also low, because the rate of energy harvest


Figure 4.3 Diversity and community invasibility is reduced by increased maintenance cost and productivity. The effect of increased maintenance cost is shown in the three vertically stacked figures, where the adaptive landscape changes from stable minimum, to a local peak, to a global peak. The stable minimum is easily invadable by any strategy arbitrarily close to the resident. The global peak is not invadable, and constitutes a single species evolutionarily stable strategy (ESS). The effect of increased productivity is shown by the horizontally arrayed figures. As with maintenance cost, an increase in productivity changes the landscape from an invadable stable minimum to a single species ESS. The broad arrow indicates that the stable minimum, once invaded, can evolve to a two-species ESS.

from secondary habitats is too close to the energy cost of using those habitats. But when MC is high, species need higher ambient resource levels to achieve a fitness of one. Similarly, if productivity is high, higher resource levels are necessary to counteract the increased density-dependence due to the nonenergetic component of fitness (e.g., limited burrow sites). In either case, the increased resource levels result in secondary habitats increasing in relative profitability, and this encourages greater overlap in habitat use among com-


petitors. As overlap increases, competition begins to favor the strategy with the lowest average cost of foraging across all habitat types. For the tradeoff function we used (represented in fig. 4.1), this is the strategy u ¼ 5, which resides in the middle of the strategy space.

Discussion Theories that combine environmental heterogeneity and tradeoffs find their usefulness not only in helping us to understand coexistence of local competitors, but also in predicting how coexistence can depend on climate and productivity. For environmental heterogeneity to be relevant to diversity, it must induce fitness differences. Furthermore, it must be accompanied by features of the organisms that induce tradeoffs in the ability of individuals to exploit different parts of the heterogeneity (Kotler and Brown 1988). Finally, it helps if individuals can select from among the heterogeneity, avoiding areas and times where they are less efficient than competitors. We have shown that the conditions for heterogeneity and tradeoffs to promote diversity depend on features of the environment not normally associated with models of local coexistence. Furthermore, our model predicts patterns of diversity that have been observed to hold on a geographic scale. The prediction of increased diversity with environmental heat does not depend on a species-abundance curve. Instead, the prediction is derived from how maintenance cost changes frequency-dependent fitness in a competitive community. The model also avoids a problem that Rosenzweig and Abramsky (1993) had with some models that invoke environmental heterogeneity to explain the decrease in diversity with increased productivity. These authors note that explanations based on the assumption that productivity reduces heterogeneity (e.g., Tilman 1987) work fine in ecological time, but these same explanations ignore the fact that habitat and resource subdivision are evolved responses of organisms, and hence these models may not work on an evolutionary time scale. It is not obvious, therefore, why a reduced range of heterogeneity should result in reduced diversity, rather than evolution of narrower habitat and resource use. Mitchell (2000) showed that even when habitats are continuous, the range of heterogeneity can still set a limit to species diversity. And the model we present here demonstrates that, even when evolution is allowed to operate, increased productivity can still reduce diversity. Dryland species diversity undoubtedly depends on interactions among variables acting over a range of scales. We have focused on species coexistence at the local scale. Regional models, in contrast, simplify the mechanics of local interactions to study the roles of dispersal and local extinction. These models can predict diversity based on a tradeoff in dispersal and competitive ability (e.g., Tilman 1994), but in many such cases competitive abilities will still depend on local, within-patch heterogeneity. In these cases, the coexistence conditions will still be influenced by climate and productivity.


In summary, the model that we present here illustrates how the geographically scaled variables can alter the chance that local mechanisms permit coexistence. The results of our model suggest that large-scale correlations between diversity and heat or productivity may be the result of local coexistence mechanisms, which should be tested empirically.

References Abramsky, Z., and M.L. Rosenzweig. 1984. Tilman’s predicted productivity-diversity relationship shown by desert rodents. Nature 309: 150–151. Brown, J.S. 1988. Patch use as an indicator of habitat preference, predation risk, and competition. Behavioral Ecology and Sociobiology 22: 37–48. Brown, J.S. 1989a. Desert rodent community structure: a test of four mechanisms of coexistence. Ecological Monographs 20: 1–20. Brown, J.S. 1989b. Coexistence on a seasonal resource. American Naturalist 133: 168–182. Brown, J.S., B.P. Kotler, and W.A. Mitchell. 1994. Foraging theory, patch use and the structure of a Negev Desert granivore community. Ecology 75: 2286–2300 Charnov, E.L. 1976. Optimal foraging: the marginal value theorem. Theoretical Population Biology 9: 129–136. Currie, D.J. 1991. Energy and large-scale patterns of animal- and plant-species richness. American Naturalist 137: 27–49. Currie, D.J., and J.T. Fritz. 1993. Global patterns of animal abundance and species energy use. Oikos 67: 56–68. Dodson, S.I., S.E. Arnott, and K.L. Cottingham. 2000. The relationship in lake communities between primary productivity and species richness. Ecology 81: 2662–2679. Gross, K.L., M.R. Willig, L. Gough, R. Inouye, and S.B. Cox. 2000. Patterns of species density and productivity at different spatial scales in herbaceous plant communities. Oikos 89: 417–427. Huggett, R.J. 1995. Geoecology: an Evolutionary Approach. Routledge, New York. Kotler, B.P. 1984. Predation risk and the structure of desert rodent communities. Ecology 65: 91–96. Kotler, B.P., and J.S. Brown. 1988. Environmental heterogeneity and the coexistence of desert rodents. Annual Review of Ecology and Systematics 19: 281–307. Lay, D.M. 1974. Differential predation of gerbils (Meriones) by the little owl, Athene brahma. Journal of Mammalogy 55: 608–614. Mitchell, W.A. 2000. Limits to species richness in a continuum of habitat heterogeneity: an ESS approach. Evolutionary Ecology Research 2: 293–316. Mitchell, W.A., and W.P. Porter. 2001. Foraging games and species diversity. Annales Zoological Fennici 38: 89–98. Mitchell, W.A., Z. Abramsky, B.P. Kotler, B. Pinshow, and J.S. Brown. 1990. The effect of competition on foraging activity in desert rodents: Theory and experiments. Ecology 71: 844–854. Mittelbach, G.G, C.F. Steiner, S.M Scheiner, K.L Gross, H.L. Reynolds, R.B. Waide, M.R. Willig, S.I. Dodson, and L. Gough. 2001. What is the observed relationship between species richness and productivity? Ecology 82: 2381–2396.

Species Diversity, Environmental Heterogeneity, and Species Interactions 69 Price, M.V., and O.J. Reichmann. 1987. Spatial and temporal heterogeneity in Sonoran Desert soil seed pools, and implications for heteromyid rodent foraging. Ecology 68: 1797–1811. Pulliam, H.R., and G.S. Mills. 1977. The use of space by wintering sparrows. Ecology: 1393–1399. Rosenzweig, M.L. 1987. Habitat selection as a source of biological diversity. Evolutionary Ecology 1: 315–330. Rosenzweig, M.L., and Z. Abramsky. 1993. How are diversity and productivity related? In Species Diversity in Ecological Communities: Historical and Geographical Perspectives, R. Ricklefs and D. Schluter, eds., pp. 52–65. University of Chicago Press, Chicago. Rosenzweig, M. L. 1995. Species Diversity in Space and Time. Cambridge University Press, Cambridge. Schaffer, W.M., D.B. Jensen, D.E. Hobbs, J. Gurvetich, J.R. Todd, and M.V. Schaffer. 1979. Competition, foraging energetics, and the cost of sociality in three species of bees. Ecology 60: 976–987. Schluter, D. 2001. Ecology and the origin of species. Trends in Ecology and Evolution 16: 372–380. Schmidt-Nielsen, K. 1964. Desert Animals: Physiological Problems of Heat and Water. Oxford University Press, New York. Srivastava, D.S., and J.H. Lawton 1998. Why more productive sites have more species: an experimental test of theory using tree-hole communities. American Naturalist 152: 510–529. Tilman, D. 1987. Secondary succession and the pattern of plant dominance along experimental nitrogen gradients. Ecological Monographs 57: 189–214. Tilman, D. 1994. Competition and biodiversity in spatially structured habitats. Ecology 75: 2–16. Waide, R.B., M.R. Willig, C.F. Steiner, G.G. Mittelbach, L. Gough, S.I. Dodson, G.P. Juday, and R. Parmenter. 1999. The relationship between primary productivity and species richness. Annual Review of Ecology and Systematics 30: 257–300. Webster, D.B., and D.B Strother. 1972. Middle-ear morphology and auditory sensitivity of heteromyid rodents. American Zoologist 12: 727. Webster, D.B., and M. Webster. 1971. Adaptive value of hearing and vision in kangaroo rat predator avoidance. Brain, Behavior and Evolution 4: 310–322. Wright, D.H., D.J. Currie, and B. A. Maurer. 1993. Energy supply and patterns of species richness on local and regional scales. In Species Diversity in Ecological Communities, R.E. Ricklefs and D. Schluter, eds., pp. 66–74. University of Chicago Press, Chicago.

5 SHALOM A Landscape Simulation Model for Understanding Animal Biodiversity Yaron Ziv Michael L. Rosenzweig Robert D. Holt

T

he ecological complexity of landscape components of biodiversity may be understood by examining relatively simple landscapes such as those of arid and semiarid lands. It is believed that such lands provide easy recognition of their components and a relatively simple interaction between their different diversities (Safriel et al. 1989). In general, ecological complexity emerges from the existence of environmental heterogeneity and scaling effects. The effects of scaling include the differential changes in observed patterns produced by processes that operate and interact at different tempospatial scales. For example, interspecific competition may have a strong influence on species coexistence and, therefore, diversity, at a local scale, may be insignificant for determining species diversity compared with a regional scale, where colonization–extinction dynamics may be the major determinant for species diversity. Environmental heterogeneity mainly results from three components: habitat diversity (the number of different habitats), habitat size (the size of each habitat’s patch), and habitat patchiness (the distribution of the different habitats’ patches in the landscape). Each component may affect species diversity by providing specific processes for coexistence, colonization, extinction, and population-size dependent effects. Additionally, as emphasized by Kotliar and Wiens (1990), different scales (Wiens 1989) should introduce different levels of heterogeneity that may influence the way organisms respond to their environment. Morris (1987) suggested that an organism that does not respond to a particular heterogeneity presented at one scale may respond to the heterogeneity presented at another scale. This concept has led ecologists to accept the idea that ecological processes and patterns are not 70

SHALOM: A Landscape Simulation Model 71

fixed, but rather depend on the scale under study (e.g., Addicott et al. 1987, Kotliar and Wiens 1990, Dunning et al. 1992, Wiens et al. 1993). In this chapter we describe a spatially explicit, multispecies, process-based landscape simulation model, SHALOM (Species-Habitat ArrangementLandscape-Oriented-Model) that has been designed to explore ecological complexity of large scales. After describing the model, we will present several simulation results to demonstrate the strengths of using such models for understanding biodiversity processes and patterns. We believe that this model can serve an important tool for exploring biodiversity in arid and semiarid lands.

Model Design The model is coded in C++ (Stroustrup 1995) using object-oriented programming (Booch 1991, Martin 1995) for designing the different components of ecological structure (e.g., species, habitats) as classes of objects. A class is a general template of a particular component of a model, treated as an autonomic unit obtaining its own characteristics and functions (i.e., encapsulation). Object-oriented programming allows us to model natural systems realistically because different components of a model can be designed and coded as classes of objects. The model is based on ecological realism. First, it explicitly defines the processes affecting species, populations, and communities (hence, processbased model); in most cases it goes beyond the simple description of a process to characterize it by its mechanics. Second, it avoids arbitrary functions and arbitrary value assignments by relying on empirical ecological findings. Finally, many of the processes’ coefficients depend on body size via allometric equations where parameters for these equations come from the empirical literature (see Peters 1983, Schmidt-Nielsen 1984, Calder 1996). This, in turn, ensures that values for many processes of the model are realistic. Figure 5.1 describes the relationship between the different classes and the position of the processes between the classes according to the way they are modeled. Note the hierarchical structure of the model: the landscape-scale processes (described below) are invoked by the class landscape directly, while the local-scale processes are invoked at the patch-population level (described below). A detailed description of the model is found in Ziv (1998).

Model’s Classes and Their Characteristics The model defines seven biological (population, species, community) and physical (cell, patch, habitat, landscape) components that produce an ecological structure as the model’s classes. It uses the current terminology of landscape ecology (e.g., Forman and Godron 1986, Turner 1989) for the terms used here.


Figure 5.1 The class-relationship diagram of the model. Notice that, consistent with the multiscale design of the model, the landscape-scale processes are positioned between the landscape and the patch classes, while the local-scale processes are positioned at the population-community level.

A landscape (the coarse grain of the model) is the entire area under study composed by a row-by-column matrix of cells. The size of the landscape is determined by its number of cells and the area of each cell in the matrix. Two processes are directly controlled by the landscape: ‘‘catastrophic stochasticity’’ and ‘‘dispersal’’ (detailed description of these processes is given in the ‘‘Model’s Processes’’ section below). A cell is a square in the landscape matrix that serves technically to produce patches. Each cell has an ‘‘area’’ and contains a single habitat type. It is the fine grain of the model. A habitat has relatively homogeneous physical and biological attributes. The physical characteristics are temperature and precipitation, because at large scales the combination of temperature and precipitation distinguishes particular ecosystems and biomes (Holdridge 1947, Lieth and Whittaker 1975). Temperature and precipitation are characterized by their long-term annual mean and standard deviation. These statistics may be linked in a probabilistic manner (the higher the standard deviation, the less likely that the mean is met in a given year). The biological characteristics of a habitat are the list of ‘‘resources’’ it offers as well as the ‘‘resource-proportion distribution’’ of each of these resources. Resources are assumed to be discrete. ‘‘Resourceproportion distribution’’ represents the proportion of each resource


in the habitat (e.g., for two resources that occur equally in a habitat, each has a resource-proportion of 0.5). A patch is the area composed of all adjacent cells sharing a habitat type where the local-scale processes take place. Individuals of a species in one patch (population) interact among themselves independently of individuals in adjacent patches. Dispersal may connect patches. Variation in cells and habitat result in patch-specific characteristics, such as ‘‘energy supply’’ and ‘‘resource-proportion productivity.’’ ‘‘Energy supply’’ is given by multiplying productivity (energy per unit of time per unit of area) by the patch’s area, while productivity is calculated as a linear function of the product temperature times precipitation (Rosenzweig 1968, Lieth and Whittaker 1975). ‘‘Resourceproportion energy supply’’ is the amount of energy per unit time offered by each resource represented in the patch. The resources and their distribution are determined by the patch’s habitat. A species is the sum of all populations in the landscape, that is, a species is a metapopulation. Each species has ‘‘body size,’’ ‘‘niche position’’ (defined by habitat and resource utilization axes described below), and ‘‘dispersal coefficient.’’ Body size plays an important role in the model. ‘‘Birth rate,’’ ‘‘death rate,’’ and ‘‘metabolic rate’’ can be body-size dependent (Y ¼ aM b , where Y is a rate, M is body size, and a and b are coefficients; Calder 1996). Habitat utilization and resource utilization usually play important roles in a species’ niche position. These utilizations reflect the physical and biological characteristics of a habitat. Thus the model can compare what is offered by a patch with what is required by a species in it. (This comparison takes place in the class ‘‘population’’ and is called ‘‘species-habitat match.’’) Habitat utilization is defined by the ‘‘temperature’’ and ‘‘precipitation’’ requirements which, for simplicity, determine the species’ niche. The temperature and precipitation requirements of a species are set by each characteristic’s ‘‘mean’’ and ‘‘standard deviation.’’ We assume that the mean represents the value at which a species reproduces best, while the standard deviation represents the species’ tolerance to values that are different from the mean. We also assume a tradeoff between maximum performance and tolerance: the higher the standard deviation, the worse the species does at each point in its niche. This tradeoff allows for tolerance–intolerance community organization (see Colwell and Fuentes 1975, Rosenzweig 1991). Temperature and precipitation can be represented by a binormal distribution according to the ‘‘central limit theorem’’ (see Durrett 1991). Hence, a species’ niche is characterized by a binormal space, shaped by the temperature and precipitation’s mean and standard deviation. The lists of ‘‘resources’’ and ‘‘resource-proportion use’’ set the resource utilization of a species. As in class ‘‘habitat,’’ resources are distributed discretely.


Each species also has a ‘‘dispersal coefficient,’’ which determines the intensity of dispersal when and if it is invoked. The dispersal coefficient is a species-specific dimensionless value that allows the model to speed up or slow down the movement of populations relative to other populations or relative to the same populations in other simulations. A population is the group of individuals belonging to a given species in a particular patch. Many of the population’s characteristics are determined by the ‘‘species’’ it belongs to. Some of these characteristics do not change during a simulation (‘‘body size,’’ ‘‘birth rate,’’ ‘‘death rate,’’ ‘‘metabolic rate,’’ ‘‘habitat utilization,’’ and ‘‘dispersal coefficient’’). Other characteristics do change according to the requirements and pressures a particular population faces in each ‘‘patch.’’ The information from the patch sets such changes. The population’s ‘‘intrinsic rate of increase’’ (i.e., the maximal growth rate with no intra- and interspecific competitors) is calculated by subtracting the species’ death rate from its habitat-specific birth rate. The latter is obtained by multiplying the species–habitat match value (see ‘‘Model’s Processes’’ below) by the species birth rate. ‘‘Initial population size’’ is the number of individuals at the beginning of a run. The model allows initial population sizes to differ. Thus, one can explore how initial conditions may affect the community and landscape (e.g., priority effect; Quinn and Robinson 1987, Lawler and Morin 1993). The ‘‘carrying capacity’’ of a population is its population size at equilibrium in the absence of stochasticity. The list of ‘‘resources’’ used by a population results from the resources used by its species and the resources available in the patch. The population’s resource-proportion use is then rescaled accordingly (considering only the resources that are actually used), maintaining the ratios of all resources used in the patch. For example, if only two resources can be used by the population and they have fundamental proportions (according to the species’ ‘‘resource-proportion use’’) of 0.1 and 0.3 (i.e., 1:3 ratio), then they will be rescaled to have proportions of 0.25 and 0.75 in the population’s diet. A community is the set of nonzero populations in a patch.

Model’s Processes Ecological processes are simulated on two scales—local and landscape—similar to the general separation made by Whittaker and Levin (1977). Localscale processes occur within each patch, while the landscape-scale processes are those that occur across or between patches. This multiscale hierarchy allows most processes to work inside patches and to have a direct impact on population growth. Meanwhile, processes occurring between patches can affect population growth indirectly and at different temporal scales.


Local-Scale Processes Community-level saturation effect, fðsÞ The community-level saturation effect builds on the ratio between the energy offered by a patch (i.e., energy supply) and the overall energy consumed by all populations in a patch. The energy consumed by all populations in the patch is the sum of each population’s species-specific energy consumption, which is calculated by multiplying the metabolic rate of the species to which the population belongs by the number of individuals in that population. Because a patch’s energy supply and a species’ metabolic rate share units (energy/time), the division of these two gives a dimensionless variable (e.g., Vogel 1994) that ranges from zero (i.e., no individuals at all) to any positive value. A patch may offer more than one resource. A population may consume all of the patch’s resources or only a subset of them, depending on the population’s list of resources. Each resource’s energy in a patch is determined by its proportion (resource-proportion energy supply) out of the energy supply in that patch. An algorithm sets the relative use of each resource by those species that share it. The community-level saturation effect equation treats each resource one at a time and then sums all resources. The following equation describes the community-level saturation effect on population j, fðsÞj , given its species i, for K resources: fðsÞj ¼

K X S X RPUkl Nl EMi k¼1 l¼1

RPPk

ð1Þ

where l is a population selected from all S existing populations in a patch, RPUkl is the resource-proportion use of resource k by population l, Nl is the size of population l, EMi is the body-size dependent metabolic rate of species i, which population l belongs to, and RPPk is the resource-proportion energy supply of resource k in a patch. The community-level saturation effect is analogous to the carryingcapacity feedback function of the logistic equation (May 1981). However, the model does not assume an arbitrary value for carrying capacity. Instead, the value for carrying capacity comes from calculating the equilibrium of a population when saturation exists. It represents the density-dependent pressure a population experiences from all of a patch’s populations, including its own. Hence, it includes both intra- and interspecific density dependence. Species–habitat Match, fðmÞ The species–habitat match quantifies how well individuals of a particular population are suited to a particular patch, given the population’s species and the patch’s habitat. The function builds on the overlap between the temperature–precipitation binormal curve of the species and the temperature–precipitation biuniform curve of the habitat. Specifically, the population’s niche space is given by the following binormal distribution equation:


(

D1 ¼

"

1 exp 0:5 1 p2

!#) x XiT y XiP y XiP 2 þ 2p SDiT SDiP SDiP pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2SDiP SDiT ð1 p2 Þ ð2Þ

x XiT SDiT

2

where x and y are values of temperature and precipitation at the patch, XiT is the species’ temperature requirement’s mean, SDiT is the species’ temperature requirement’s standard deviation, XiP is the species’ precipitation requirement’s mean, SDiP is the species’ precipitation requirement’s standard deviation, and p is a covariance between the species temperature and precipitation. The patch’s habitat space is given by the following biuniform distribution equation: D2 ¼ 4SDhT 4SDhP D} 1

ð3Þ

where D} 1 is the highest distribution value of the population’s species niche space, SDhT is the habitat temperature characteristic’s standard deviation, and SDhP is the habitat precipitation characteristic’s standard deviation. The final species–habitat match value for a given population in a particular patch (fðmÞj Þ is given by dividing the population’s niche space nested within the patch’s habitat space by the patch’s entire habitat space. The species–habitat match value represents the fraction of the population’s species ability expressed in the particular patch given its habitat. A value of 1 represents a perfect match, while a value of 0 represents no match at all. The above form of calculating species–habitat match provides two major outcomes that we should expect to see in nature. First, the more tolerant a species, the more likely it will match a habitat far away from the species population’s temperature and precipitation mean values. Second, the lower the standard deviation of the habitat’s precipitation and temperature characteristics, the higher the species–habitat match. This should be true because a habitat’s standard deviations are negatively correlated with the probability of getting a particular value at a given time. Higher standard deviations represent a lower probability of any species finding a given value in a habitat. Ecologically, this should represent a measure of predictability: the lower the standard deviations of the habitat, the better it is for the populations occurring in that habitat. The population dynamics equation (eq. (5)) uses a single value for the species–habitat match. Other functions can be used to get the desired value. When possible, the species–habitat match should be generated with empirically derived functions that use the natural history of the species and more accurate measurements of how well the species thrives in the available habitats. Demographic stochasticity This refers to any change in population size caused by a chance event (resulting from sampling errors), independent of a


biological process. It tends to have critical effects when populations sizes are low. We used a simple descriptive equation to model stochastic deviations from the deterministic, body-size-dependent birth and death rates. The deviations are negatively correlated with population size: the larger the population, the lower the deviations are likely to be. Although the equation does not relate to any specific process (e.g., sex ratio or encounter rate), its behavior does follow the typical expectations of such stochasticity. The equation affects demographic parameters randomly and it is density-dependent (e.g., Diamond 1984, Shaffer and Samson 1985, Pimm et al. 1988, Lande 1993). The following equation defines the population’s stochasticity in birth or death rates, Zj , from a species’ deterministic birth or death rates, zi : ! "ð0:5zi Þ pffiffiffiffiffiffi Z j ¼ zi Nj

ð4Þ

where e is a random number sampled from a Gaussian probability distribution (with a mean of zero and a symmetrical truncation of two standard deviations, of one unit each), 0.5zi is a scaling term to make each distribution range between zero and twice the highest birth or death rate, g is a demographic stochasticity coefficient that allows for changing the ‘‘intensity’’ of the effect, and Nj is population size. We used a logistic-like continuous-time population growth for the localscale population dynamics. Birth rate and death rate are handled independently. This separation is realistic (Begon et al. 1986) because birth rate and death rate may be limited by different processes, such as a need for proteinrich resources for lactating females that are not required by the rest of the population. Overall, the equation by which a given population grows in a patch given the above processes is dNj ¼ Nj bi ffðmÞj ð1 fðsÞj Þþ g Nj di f1 þ fðsÞj g dt

ð5Þ

where fðmÞj and fðsÞj are the species–habitat match effect and the saturation effect, respectively, and (1 fðsÞj Þþ indicates that the latter term cannot take a value lower than zero (see Wiegert 1979). The community-level saturation effect ðfðsÞj ) enters the equation twice. First, we subtract the community-level saturation effect from one as in the carrying-capacity feedback function of the logistic equation (i.e., 1 N=K). This new term models the effect of the community saturation on birth. It is assumed (as in the logistic equation) that birth decreases linearly with an increase in community density. Oversaturation (i.e., 1 fðsÞj < 0 results in no birth. Second, we add one to the community-level saturation effect to model the effect of the community saturation on death. Here also, it is assumed that death rates decrease linearly with an increase in community density.


The local-scale population dynamics equation with its analytical solution and outcomes for body-size dependent habitat specificity are found in Ziv (2000). Landscape-Scale Processes Dispersal, fðdÞ This is the movement of individuals from one patch to another (e.g., Levin 1974, Andow et al. 1990, Johnson et al. 1992, Gustafson and Gardner 1996). In the model, individuals of a particular population in a given patch migrate to adjacent patches if they can gain a higher potential fitness there. The dispersal function builds on the optimization principles used for intraspecific density-dependent habitat selection suggested by Fretwell and Lucas (1969) and Fretwell (1972) (ideal free distribution). In the model, the dispersal process assumes that a population’s individuals can instantly assess the adjacent population’s per-capita growth rate. At each time step, the model calculates the per-capita growth rate of each population. Then it compares that rate with all adjacent populations’ percapita growth rate. Individuals move from patches with relative low percapita growth rate (i.e., low fitness potential) to patches with high per-capita growth rate (i.e., higher fitness potential). This results in equalizing the percapita growth rates of populations of the same species across patches (Fretwell 1972). Dispersal occurs on a continuous-time scale. Hence, dispersal from a given patch to patches that are unadjacent can happen fast in appropriate conditions (e.g., some patches of low potential fitness and a patch of a very high potential fitness). However, there is an implicit distance effect because individuals need to cross the adjacent patches first, and because each population in the different patches experiences population change due to other processes. This effect can be controlled by changing the species dispersal coefficient such that the rate at which individuals of populations of a given species move agrees with the user’s needs. Catastrophic stochasticity Also known as disturbance-induced extinction (Levin and Paine 1974, Pickett and White 1985, Turner et al. 1989), this is a density-independent loss of individuals due to some event (e.g., extreme cold weather or a drought) that has a random probability of occurrence. Some environments may have a higher probability of being affected by catastrophes than others. Catastrophes may cause the disappearance of entire populations of a given community or only partial disappearance. The same catastrophe may eliminate some species from a patch but only reduce others. A catastrophic event may be very local, such as within a single habitat (e.g., a falling tree in a forest), or may cover an extensive area and include many different types of habitats (see Turner et al. 1989). The catastrophic stochasticity of the model SHALOM relies on randomnumber-generating procedures (Press et al. 1995). These allow one to change the probability, intensity, and range of the density-independent loss of individuals and populations. The user sets the following options: the probability


function (either uniform or Gaussian) of the catastrophic stochasticity distribution, the threshold (a fraction between 0 and 1) below which catastrophic stochasticity is not invoked, the lower and the upper limits (a fraction between 0 and 1) for population loss once a catastrophic stochasticity is invoked, the probability function (either uniform or Gaussian) of the population loss, and the spatial distribution (either a random or a fixed distribution on a cell, or patch, or the entire landscape) of the catastrophic stochasticity. The two landscape-scale processes affect population growth on two different time scales. As mentioned above, dispersal is assumed to occur on a continuous-time scale similar to the continuous-time scale of the local population dynamics. In fact, dispersal at any time step of the model depends on the local-scale per-capita growth rate of each population. Defining the local growth of population j in equation (5) as FðlÞj , the overall population growth, including dispersal, becomes AP X dNj ¼ FðlÞj þ ðfðdÞjl NjðÞ =lðþÞ Þ dt l¼1

ð6Þ

where AP is the number of adjacent patches and NjðÞ=lðþÞ indicates that the per-capita migration is multiplied by the patch’s population size or by the adjacent patch’s population size, depending on the sign of the per-capita movement. A positive per-capita movement means that the particular patch’s per-capita growth rate is higher than the one adjacent. Hence, individuals from the adjacent patch disperse into it. In contrast, a negative per-capita movement means that individuals should disperse into the adjacent patch. Catastrophic stochasticity is simulated on a discrete time scale. Once a year (or on an interval that amounts to a year), the model invokes catastrophic stochasticity. Model Mechanics Before each run of the model, the user assigns the following: the species and their attributes, the habitats and their attributes, and the habitat arrangement in the landscape. Given this information, the model creates the patches as they would appear to organisms in the real world. Having modeled patches and species in the landscape, populations are then created. The species– habitat match of a population is then calculated. The option of invoking demographic stochasticity is set for each population. All populations of a particular patch create the patch’s community. The community monitors the overall saturation effect in a patch as well as the different species’ composition and diversity. Once the landscape is completely defined, the model asks for information about the large-scale processes. Dispersal may or may not be invoked by the user. Similarly, catastrophic stochasticity may or may not be invoked. If catastrophic stochasticity is invoked, the model asks for information about its different options.


Following the specification of the initial population size for each population and the run time (in years), the model runs a population-growth simulation of the different populations in the different patches. The Runge–Kutta method (Press et al. 1995) is used to integrate the small steps (dt ¼ 0:001 yr) on a continuous time axis. The model returns the value of population size for each population in the different patches every 100 time steps (i.e., 0.1 yr). The information is saved to an output file for further analysis. At the end of the run, the model calculates the ratio of each population’s size to its carrying capacity and returns values of the number of species and two species-diversity indices: Simpson’s diversity index (Simpson 1949) and Fisher’s alpha (Fisher et al. 1943).

Using the Model: An Example of the Effects of Ecological Processes on Community Structure in a Heterogeneous Landscape How do different processes—interspecific competition, demographic stochasticity, and dispersal—known to affect communities at a local scale, affect species composition and species-diversity patterns in a spatially heterogeneous landscape scale? Many studies have explored various processes that affect communities in heterogeneous landscapes. However, these studies treat each process discretely (e.g., Andow et al. 1990, Dunning et al. 1992, Holt 1992). How the interaction of multiple processes affects community structure is rarely explored, except in the context of metapopulation dynamics. In the following sections, we will describe a simulation design that allows the modeling of several species of different body sizes in a very simple heterogeneous landscape without losing track of the species diversity in each patch or in the entire landscape. As will be shown later, this simple simulation will provide enough information to make some sophisticated predictions. Simulation Design We simulated a landscape with 2 2 cells, each 100 m2, having its own unique habitat (total of four habitats). We chose this simple landscape design because the existence of the different processes in the current simulation added a tremendous amount of complexity to the model. Thus, the simple landscape design provides focus on the processes’ outcomes. We assigned realistic productivity values for the different habitats without a specific process in mind in order to keep the model as general as possible. Note that also in this simulation, patch and habitat are synonymous. We also assigned different species–habitat matches to the different habitats, such that habitat 1 was the best habitat and habitat 4 was the worst (species–habitat match ¼ 0.997, 0.987, 0.971, and 0.949 for habitats 1, 2, 3, and 4, respectively). To allow for competitive coexistence between the modeled species,


each habitat offered 28 different resources. To avoid a specific resourceproductivity distribution, we assigned an equal productivity for each resource out of the total productivity of the habitat. We simulated a total of 26 species. Species differed in only one characteristic—body size. Body size ranged between 5 g and 1585 g, corresponding to log values of body size ranging between 0.7 and 3.2. We assigned a unique preferred resource to each species and gave it a resource-proportion use of 0.5. Each species could consume two other resources, one on each side of the preferred resource; each of these had a resource-proportion use of 0.25 (e.g., species 1 is able to consume resources 1, 2, and 3 with a resource-proportion use of 0.25:0.5:0.25, species 2 is able to consume resources 2, 3, and 4 with a resource-proportion use of 0.25:0.5:0.25, and so on). Preliminary simulations have shown that this resource allocation was sufficient to produce a competitive relationship with resource partitioning without assuming any complex resource-use function. Throughout the simulations, we used the allometric power coefficients known for eutherian mammals for birth rate (0.33), death rate (0:56), and metabolic rate (0.75) (Calder 1996). When catastrophic stochasticity was invoked, we gave the system a 10% chance of suffering catastrophic stochasticity in a year (an average of one catastrophe every 10 years). In catastrophic years, stochasticity can affect up to 50% of the landscape with up to 50% loss of population size in those patches affected. These values were chosen after experimenting with many simulation designs. They are high enough to affect population and species distribution (Turner 1987), yet, low enough that no populations are driven to extinction. Other than these first-level assignments of values for cells, habitats, and species, no other assignments were made for second-level procedures such as habitat-specific population abundance, etc. Therefore, any bodysize-dependent patterns that emerge will result only from the basic rules described here. To understand the effects of the different ecological processes, we initially explored the patterns emerging from communities not affected by any of the above processes (i.e., in which competition is strictly intraspecific). We then introduced interspecific competition, and added, thereafter, demographic stochasticity to interspecific competition to explore how it changes the predicted patterns. Finally, we allowed dispersal to connect all patches. Results Carrying capacities All habitats were suitable for all species. That is, without any population-reducing processes—interspecific competition and demographic and catastrophic stochasticities—all populations in all habitats could maintain a persistent population size. Figure 5.2 shows the carrying capacities of the different populations in the different habitats as well as the species abundance in the entire landscape. Because all populations can persist in


Figure 5.2 Carrying capacities of species as a function of their body size in the different patches (habitats) and in the entire landscape.

all habitats, and because no process other than intraspecific competition affects population growth, the same population size pattern in all the habitats and the same species diversity pattern across the entire landscape (i.e., the sum of all population sizes of each species) emerge. The only difference between habitats is that carrying capacities of populations of the same species are lower in habitats with fewer species–habitat matches. The effect of interspecific competition Here we assumed that resource partitioning occurs such that the most preferred resource is different for each species. Because of the overlap in resource use, each resource is consumed by three species. This shared consumption can lead to competitive exclusion. When resources are equally shared by species of different body sizes, the larger species outcompetes the smaller species that use the same resources. This outcome results from the lower death rates of larger species. Regardless of the specific mechanism, this larger-species competitive advantage is consistent with competitive outcomes observed in many real systems (see Kotler and Brown 1988), confirming the effectiveness of the model SHALOM. Due to the modeling of resource partitioning as a deterministic process that does not change between habitats, the same species composition exists in all four habitats of the model as well as for the entire landscape (fig. 5.3). With interspecific competition, some populations are outcompeted, leaving a discontinuous distribution of body sizes. The absence of a particular species depends on an intratrophic level cascading effect: the largest species depresses the second largest species population size due to the largest species’ competitive advantage. Although the second largest species has the competitive


Figure 5.3 Sizes of all populations in the different patches (habitats) and in the entire landscape with interspecific competition. Interspecific competition deterministically affects all populations the same way. Identical pattern emerge for each patch and for the entire landscape.

advantage over the third largest species, the present but minimal effect of the third largest species on the second largest species is enough to depress the former further to local extinction. Because, in the present model, species share resources only with the species closest in body size, the third largest species, which does not share resources with the largest one, is saved from the potentially dominating effect by the extinction of the second largest species. The process repeats with the fourth, fifth, and sixth largest species, and so on. Because all interactions between all species are taking place simultaneously, the overall effect on the different species sometimes results in an absence of a species particular body size in between two coexisting species, each having close body sizes. The two species coexist because the larger species can consume its most preferred resource better, and has a competitive advantage, while the smaller species benefits from the other resource that is no longer used by the now extinct, smaller species. In the end, 12 species coexist in the landscape. Adding demographic stochasticity to interspecific competition Demographic stochasticity, or the sampling effects regarding sex ratio, litter size, etc., that may promote local extinctions of small populations, exists no matter what other processes affect population growth (Pimm et al. 1988, Lande 1993). With demographic stochasticity, different patterns appear in the different habitats (fig. 5.4). Populations of larger species are more likely to become extinct because they exist in fewer numbers. However, the particular population that ends up extinct is determined randomly. Once a particular population becomes extinct, its closest competitor in body size


Figure 5.4 Typical sizes of all populations in the different patches (habitats) and in the entire landscape with interspecific competition and demographic stochasticity. The community structure in each patch is determined by those large populations that escaped extinction. Once a random extinction of a large-body population takes place, the community is well structured according to competitive interactions. Demographic stochasticity increases species diversity at the landscape scale.

benefits from a competitive release and enjoys a higher population size, hence, showing a negative autocorrelation in population size. The rest of the community is now competitively determined by the particular large body-sized species that escaped extinction. Because demographic stochasticity reduces species diversity in each habitat, population size of the survivors, on average, is higher than with interspecific competition alone; that is, the same resources are now divided among fewer species comprising more individuals. At the landscape scale, more species exist because of the randomness of some extinction in the different habitats. Hence, demographic stochasticity increases species diversity at the landscape scale (see also Chesson and Case 1986). Overall, on average, 17:59 1:72 species exist in the landscape. The effect of dispersal with stochastic effects and interspecific competition Dispersal (e.g., Levin 1974, Johnson et al. 1992) has consistently been shown to have major effects on single-species distributions as well as on multispecies community structures. With dispersal (fig. 5.5), colonists can restore local populations of their species. When the species is competitively subordinate, a permanent recovery is unlikely. However, the recovery of a competitively dominant population has a significant effect on community composition. If dispersal is frequent enough, dominant species can establish in all patches and, on average, overcome the stochastic effects that tend to produce locally different patterns.


Figure 5.5 Typical sizes of all populations in the different patches (habitats) and in the entire landscape with interspecific competition, demographic stochasticity, and dispersal. Dispersal allows dominant populations that became extinct from a particular patch due to a chance event (stochasticity) to recolonize that patch and increase in numbers. As a result, the dominant species in the landscape reestablish their populations in all patches and, on average, overcome the stochastic effects that might locally produce different patterns.

Knowing the outcomes (or fingerprints) of the different processes (i.e., competition and demographic stochasticity), we can now detect the fingerprints of the different processes. As before, demographic stochasticities are responsible for larger discontinuities of body sizes and for the disappearance of the largest species from the landscape (the local extinction of the largest species from all the patches deprives them of colonists that could otherwise restore extinct populations). Dispersal allows dominant species to recolonize habitats in which they have previously become extinct, resulting in homogeneity among habitats in a landscape. At the landscape scale, the existence of dispersal together with demographic stochasticity and interspecific competition produces the lowest species diversity (5.08 0.598). The main reason for this low species diversity is the ongoing disappearance of small populations that usually belong to species of large body size.

Discussion This model presents a new approach to the study of complex ecological systems. This new approach contributes to our understanding of large-scale ecological processes and patterns by providing us with nontrivial predictions on the combination of spatial heterogeneity and multiple-process interac-


tions. The example given in this chapter demonstrates this contribution by providing specific outcomes that could have not been predicted otherwise. Stochasticity depresses mean population sizes and allows different habitats to support different communities. These different communities are determined by which large species becomes locally extinct at random. The local extinction of a large species shifts the maximum body size of the competitively organized community. With both demographic and catastrophic stochasticities, species diversity is higher than with interspecific competition alone. The effect of demographic stochasticity on species composition differs from that of catastrophic stochasticity. With demographic stochasticity, discontinuities of body sizes are larger, and no species of very similar body size coexist. With catastrophic stochasticity, all of the largest species disappear. Combined, each of the two stochasticities affects species composition in the different habitats and in the landscape. Hence, such communities have large discontinuities of body size and none of the largest species. Dispersing individuals move between habitats and reestablish the local populations of their species. Thus, dispersal neutralizes the randomness of the assemblages produced by stochasticity. As a result, each habitat tends toward the same set of species. However, even with dispersal, stochasticity eliminates the largest species and produces large discontinuities in the body size distribution. Loss of randomness in the assemblages means that, at the landscape scale, dispersal reduces species diversity. The predictions presented here about species composition and species diversity demonstrate the usefulness of the current model. The ability to characterize specific fingerprints of different processes and then analyze the joint effect of multiple processes by tracking these fingerprints should help us to understand natural systems better. We recommend that ecologists adopt such an approach for understanding ecological complexity. Ecologists need also to set up studies that will allow them to test whether the predicted patterns produced by particular processes in the model are indeed observed in the field.

Note

Send all correspondence to Yaron Ziv.

References Addicott, J.F., Aho, J.M., Antolin, M.F., Padilla, D.K., Richardson, J.S., and Soluk, D.A. 1987. Ecological neighborhoods: scaling environmental patterns. Oikos 49: 340–346. Andow, D.A., Kareiva, P.M., Levin, S.A., and Okubo, A. 1990. Spread of invading organisms. Landscape Ecology 4: 177–188. Begon, M., Harper, J.L., and Townsend, C.R. 1986. Ecology: Individuals, Populations and Communities. Sinauer Associates, Sunderland. Booch, G. 1991. Object-Oriented Design with Applications. Benjamin/Cummings, Redwood City.

SHALOM: A Landscape Simulation Model 87 Calder, W.A. III. 1996. Size, Function, and Life History. Dover, New York. Chesson, P.L., and Case, T.J. 1986. Overview: nonequilibrium community theories: chance, variability, history, and coexistence. In Community Ecology, J. Diamond, and T.J. Case (eds.), pp. 229–239. Harper and Row, New York. Colwell, R.K. and Fuentes, E.R. 1975. Experimental studies of the niche. Annual Review of Ecology and Systematics 6: 281–310. Diamond, J.M. 1984. ‘‘Normal’’ extinctions of isolated populations. In Extinctions, M.H. Nitecki (ed.), pp. 191–246. University of Chicago Press, Chicago. Durrett, R. 1991. Probability: Theory and Examples. Wadsworth and Brooks/Cole Advanced Books and Software, Pacific Grove. Dunning, J.B., Danielson, B.J., and Pulliam, H.R. 1992. Ecological processes that affect populations in complex landscapes. Oikos 65: 169–174. Fisher, R.A., Corbet, A.S., and Williams, C.B. 1943. The relation between the number of species and the number of individuals in a random sample from an animal population. Journal of Animal Ecology 12: 42–58. Forman, R.T.T., and Godron, M. 1986. Landscape Ecology. John Wiley and Sons, New York. Fretwell, S.D. 1972. Populations in a Seasonal Environment. Princeton University Press, Princeton. Fretwell, S.D., and Lucas, H.L.J. 1969. On territorial behavior and other factors influencing habitat distribution in birds. Acta Biotheoretica 19: 16–36. Gustafson, E.J., and Gardner, R.H. 1996. The effect of landscape heterogeneity on the probability of patch colonization. Ecology 77: 94–107. Holdridge, L.R. 1947. Determination of world plant formations from simple climatic data. Science 105: 367–368. Holt, R.D. 1992. A neglected facet of island biogeography: the role of internal spatial dynamics in area effects. Theoretical Population Biology 41: 354–371. Johnson, A.R., Wiens, J.A., Milne, B.T., and Crist, T.O. 1992. Animal movements and population dynamics in heterogeneous landscapes. Landscape Ecology 7: 63–75. Kotler, B.P., and Brown, J.S. 1988. Environmental heterogeneity and the coexistence of desert rodents. Annual Review of Ecology and Systematics 19: 281–307. Kotliar, N.B., and Wiens, J.A. 1990. Multiple scales of patchiness and patch structure: a hierarchial framework for the study of heterogeneity. Oikos 59: 253–260. Lande, R.L. 1993. Risks of population extinction from demographic and environmental stochasticity and random catastrophes. American Naturalist 142: 911–927. Lawler, S.P., and Morin, P.J. 1993. Temporal overlap, competition, and priority effects in larval anurans. Ecology 74: 174–182. Levin, S.A., 1974. Dispersion and population interactions. American Naturalist 108: 207–228. Levin, S.A., and Paine, R.T. 1974. Disturbance, patch formation, and community structure. Proceedings of the National Academy of Sciences (USA) 71: 2744–2747. Lieth, H., and Whittaker, R.H. 1975. Primary Productivity of the Biosphere. SpringerVerlag, New York. Martin, R.C. 1995. Designing Object-Oriented C++ Applications Using the Booch Method. Prentice Hall, Englewood Cliffs. Morris, D.W. 1987. Ecological scale and habitat use. Ecology 68: 362–369. May, R.M. 1981. Theoretical Ecology. Blackwell Scientific Publications, Oxford. Peters, R.H. 1983. The Ecological Implications of Body Size. Cambridge University Press, Cambridge.


Pickett, S.T.A., and White, P.S. 1985. The Ecology of Natural Disturbance and Patch Dynamics. Academic Press, New York. Pimm, S.L., Jones, H.L., and Diamond, J. 1988. On the risk of extinction. American Naturalist 132: 757–785. Press, W.H., Teukolsky, S.A., Vetterling, W.T., and Flannery, B.P. 1995. Numerical Recipes in C (second edition). Cambridge University Press, Cambridge. Quinn, J.F., and Robinson, G.R. 1987. The effects of experimental subdivision of flowering plant diversity in a California annual grassland. Journal of Ecology 75: 837–856. Rosenzweig, M.L. 1968. Net primary productivity of terrestrial environments: predictions from climatological data. American Naturalist 102: 67–84. Rosenzweig, M.L. 1991. Habitat selection and population interactions: the search for mechanism. American Naturalist 137: S5–S28. Safriel, U.N., Ayal, Y., Kotler, B.P., Lubin, Y., Olsvig-Whittaker, L., and Pinshow, B. 1989. What’s special about desert ecology? Journal of Arid Environments 17: 125–130. Schmidt-Nielsen, K. 1984. Scaling: Why Is Animal Size So Important? Cambridge University Press, Cambridge. Shaffer, M.L. and Samson, F.B. 1985. Population size and extinction: a note on determining critical population sizes. American Naturalist 125: 144–152. Simpson, E.H. 1949. Measurement of diversity. Nature 163: 688. Stroustrup, B. 1995. The C++ Programming Language (second edition). Addison Wesley, Reading, MA. Turner, M.G. 1987. Landscape Heterogeneity and Disturbance. Springer-Verlag, New York. Turner, M.G. 1989. Landscape ecology: the effect of pattern on process. Annual Review of Ecology Systematics 20: 171–197. Turner, M.G., Gardner, R.H., Dale, V.H. and O’Neill, R.V. 1989. Predicting the spread of disturbance across heterogeneous landscapes. Oikos 55: 121–129. Vogel, S. 1994. Life in Moving Fluids: The Physical Biology of Flow (second edition). Princeton University Press, Princeton. Whittaker, R.H. and Levin, S.A. 1977. The role of mosaic phenomena in natural communities. Theoretical Population Biology 12: 117–139. Wiegert, R.G. 1979. Population models: experimental tools for analysis of ecosystems. In Analysis of Ecological Systems, D.J. Horn, G.R. Stairs, and R.D. Mitchell (eds.). Ohio State University Press, Columbus. Wiens, J.A. 1989. Spatial scales in ecology. Functional Ecology 3: 385–397. Wiens, J.A., Stenseth, N.C., Van Horne, B., and Ims, R.A. 1993. Ecological mechanisms and landscape ecology. Oikos 66: 369–380. Ziv, Y. 1998. The effect of habitat heterogeneity on species diversity patterns: a community-level approach using an object-oriented landscape simulation model (SHALOM). Ecological Modelling 111: 135–170. Ziv, Y. 2000. On the scaling of habitat specificity with body size. Ecology 81: 2932– 2938.

6 Spatial Scale and Species Diversity Building Species–Area Curves from Species Incidence William Edward Kunin Jack J. Lennon

Biodiversity and the Species–Area Relationship This chapter is largely focused on the species–area relationship (SAR), although it may not seem so for much of the time. Bear with us; we will get there in the end. Our aim is to provide insights into how the relationship works, and how it is built. This leads us to take a rather reductionist approach, and to break down the SAR into its component parts. We will spend a substantial section of this chapter examining these pieces and their properties. We will then explore the logic by which the parts are reassembled, and will explore how biological and biogeographical properties of a system may affect the SAR. Before attempting this feat, however, we should begin with a brief discussion of the SAR itself, to explain why it is worth making such a fuss over. The SAR is, after all, only a simple graph: a plot of the number of species found in a sample as a function of the area sampled. Ecologists being an argumentative lot, we cannot even all agree on what this plot should look like; Gleason (1922, see also Williams 1964) argued that the absolute number of species should be plotted as a function of the logarithm of area, whereas Arrhenius (1921, see also Preston 1960) suggested that both species and area should be plotted logarithmically. Connor and McCoy (1979) found cases that fit both models, and two others besides (log species by untransformed area, and neither variable transformed). However it’s plotted, the SAR is not even a particularly attractive or elegant graph—at its best (!) it is simply a straight diagonal line within a tight scatter of datapoints on a rectangular plot. Hardly something to set the pulse racing. 89


Yet the SAR is exciting stuff; that simple line encapsulates a great deal of information about the diversity of biological systems across a wide range of scales. As Rosenzweig (1995) points out, the SAR brings together local, landscape, and regional measures of ‘‘inventory’’ diversity (a, g, and e diversity, in Whittaker’s 1972 terminology)—extending even to global diversity in some cases; and then unites them with local and regional measures of ‘‘transitional’’ diversity (b and d diversity, respectively) into a single, unified framework. Hegel would be spinning in his proverbial resting place: here the ‘‘synthesis’’ (the SAR) was developed decades before any of the ‘‘theses’’ or ‘‘antitheses’’ (the various scale-specific measures of diversity) were devised. We tend to think of the issue of scale-dependence of ecological patterns as a subject of comparatively recent interest (e.g., Tilman and Kareiva 1997), yet here is one of the oldest of ecological tools, devoted specifically to this most modern of concerns. The problems with the SAR are related to its strengths: it encapsulates so much information in such a simple form that it’s difficult to know how to unpick all of the richness contained within it. How are we to understand the height of the curve, its slope, its curvature (when it is curved) or its straightness (when it’s straight)? How can we explain the tendency for the curve to follow Gleason’s logarithmic model in some cases, and to follow Arrhenius’ power law relationship in others? If we are to understand this tool (we would argue), we need to take it apart into its component pieces. If we can understand these parts and how they fit together, we should gain some insight into how this most fundamental measuring stick for biodiversity works.

Species Distributions and Incidence Curves If the SAR measures species diversity, then its component parts are individual species distributions. As the SAR is devoted to cross-scale patterns, so too these species distributions must be analyzed in a multiscale perspective. To be more precise, the species–area curve can be thought of as the sum of myriad species-specific ‘‘incidence area’’ curves, that is, the gradual increase in the probability of encountering each species as the area surveyed grows. Each of these curves must rise as we move from fine to coarser scales of analysis, because the probability of sampling any given species will rise as the size of the sampled ‘‘quadrat’’ increases. At the extreme of a vanishingly small (point) quadrat, the area occupied by the species will be its total cover, the fraction of the survey area covered by the species’ tissue. At the opposite extreme, a quadrat the size of the planet will be certain to contain all of the species on earth, and so the probability of each species’ incidence will be 1.0. Between these two extremes, the incidence function for any given species will rise monotonically. At each scale, the probability of including a particular species in a random quadrat will be determined by the species’ ubiquity at that scale, that is, the fraction of all quadrats that size that include

Spatial Scale and Species Diversity 91

the species in question. This fraction is simply the area occupied by the species when analyzed at this scale divided by the total area of the survey. The notion that the area ‘‘occupied’’ by a species differs at different scales of analysis may seem a bit peculiar, but it is key to what follows. An essential step toward understanding the scaling properties of diversity is to understand that the quantity of ‘‘stuff’’ measured almost always depends on the scale at which the stuff in question is measured. This can be best understood by means on a well-used analogy: measuring a coastline (see Richardson 1961). It may seem a straightforward matter to measure the length of, for instance, the Israeli coast, but the problem proves quite difficult on closer inspection. The obvious first attempt would involve lining up a tape measure on a map of the country, from the end of the Gaza strip to the Lebanese border, some 200 or so kilometers. But the coast is curved, and the tape would not follow it precisely. To get a better estimate, we might push the tape in at the center, making two straight segments, with a sum a bit greater than our first estimate. But neither of these segments would quite capture the coast’s shape, so we might wish to divide each of them in half, and then each of the halves yet again, and so on. At each step the apparent length of the coastline would increase. We can measure the ‘‘scale’’ of the analysis by the length of the individual tape segments used. As the scale of our measurement fell to, for instance, 1 km, we would begin to expose the larger bumps in the coastline, with a detour around the Carmel (for example) adding a bit of distance to the total. At a still finer scale (say, 1 m), we would begin to find all sorts of zigs and zags in our formerly straight coastline, with detours around rocks and jetties greatly lengthening the overall coast length. Moving to even finer (mm or mm) scales, we might eventually be forced to work our way around each pebble and sand grain on the edge of each beach, vastly expanding the measured distance. The measured length continues to grow as the scale of analysis shrinks. The logic of measuring species distributions across scales is very similar, but here the area grows progressively smaller (rather than larger, as above) as the scale of analysis grows finer. As alluded to above, at extremely fine scale, we would count only areas physically covered by the species in question as ‘‘occupied.’’ At somewhat coarser scales of analysis, we might count any square meter containing at least one individual as ‘‘occupied,’’ but in doing so we would include quite a bit of area that the species was not physically covering. At coarser scales still, all of the square kilometers containing the species might be toted up, again, bringing in a great deal of land that was deemed ‘‘unoccupied’’ at the finer scales of analysis. These different scales of analysis are not simply progressively coarser and less accurate ways of measuring distribution; they actually reveal different attributes of a species’ spatial distribution. Thus an extremely coarse-scale map (using, e.g., a 100 km grid) would reveal only the broadest outlines of a species’ geographical range, but progressively finer scale maps would begin to reveal geographically isolated subpopulations, habitat preferences, discrete populations, local patterns of aggregation, and so on (Kunin 1998). By plotting the area


occupied across a range of scales (a ‘‘scale–area’’ curve, Kunin 1998), a variety of aspects of abundance can be displayed simultaneously. Note that this scale–area curve is almost identical to the incidence function we require for building the SAR; we need only divide the area occupied by the total area surveyed. To understand the SAR, then, we need to learn more about these scale–area curves. There are reasons to believe that, for many species, these curves should be approximately straight when plotted on logarithmic axes (Kunin 1998). In the language of geometry, this is to say that these distributions are approximately fractal, at least over a reasonably wide range of scales. For the purposes of our argument, it would be sufficient to state this fact and to build off it, but it is fitting (given the mechanistic tone of this chapter) to propose a mechanism for these approximately fractal species distributions before proceeding further. Fractal Habitats and Species Distributions The first step toward understanding the distribution of a species is to map the area of potentially suitable habitat available to it. In principle, at least, one could map out all areas that fall within the fundamental niche (sensu Hutchinson 1957) of a given species. This ‘‘niche map’’ would include all sites at which a pioneering party of immigrants of the species in question would have a positive population growth rate ( > 1). A wide variety of variables might contribute to determining the suitability of a site to a given species, but even without knowing precisely what they are, we can guess something about their distribution in space. We might expect this map to be approximately fractal. Geometric analyses of various aspects of the physical environment, for instance, coastlines (Richardson 1961), rock substrates (Kulatilake et al. 1997), lava flows (Gaonach et al. 1992), topography (Klinkenburg and Goodchild 1992), drainage (Deschaux and Souriau 1996), soil moisture (Pelletier et al. 1997), and rainfall patterns (Onof and Wheater 1996) typically display some form of fractal scaling, at least over certain ranges of scales. Disturbance events such as earthquakes (Hastings and Sugihara 1993), landslides (Pelletier et al. 1997), erosion (Chase 1992), fires (McAlpine and Wotton 1993), and some human activities (Frohn et al. 1996) show evidence of fractal properties as well. As the suitability of a site for a given species depends at least in part on these and other physical factors, we might expect the distribution of habitat acceptable to a given species (i.e., all areas for which its > 1), to be approximately fractal as well. This prospect seems particularly appropriate in arid lands, where approximately fractal drainage and erosion patterns are often starkly evident (without the concealing mantle of vegetation), leading to similarly patchy and complex patterns of habitat availability for many species. The physical niche map of a species will generally be reshaped by biotic factors as well (chapter 11, this volume). Interactions with other species (dominant competitors, natural enemies, mutualists) could alter the map considerably, but so long as these other species also have roughly fractal distributions, the map of habitat


remaining available to our focal species is simply the intersection of two fractals, and thus likely to be approximately fractal as well (but see below). Of course, even if we could map all habitable land for a particular species, this would only be the first step—the playing board upon which the species’ career is acted out (to mix metaphors badly). The fraction of that map that is actually occupied by a species will be determined in large part by a process of colonization and local extinction within patches, akin to a metapopulation model (Levins 1970, Hanski and Gilpin 1997), but acting on multiple scales. Areas that are particularly isolated (with consequent low rates of colonization), or that are small or poor in quality (both leading to high extinction rates) may be probabilistically unoccupied, even if suitable. The same is true of the interacting species that influence our focal species’ available niche map—they may be unable to colonize some sites that would be suitable for them, thus perhaps allowing our focal species to persist in some sites from which it would otherwise be excluded. On the other hand, high dispersal of propagules into sites that are marginally unsuitable for our focal species may result in the maintenance of ‘‘sink’’ populations outside of the species original niche map (Pulliam 1988). These processes, taken together, will reshape the original map, removing potentially large areas of unrealized niche space, while perhaps adding a bit of populated area in ‘‘non-niche’’ sites (fig. 6.1). These deletions and additions may affect the species distribution differently at different scales of analysis. If the original niche map can be thought of as approximately fractal, it can be represented as a line on a scale–area graph. This is done by dividing the region into cells at different resolutions, and counting up the area occupied, a process which is conceptually akin to the computation of the so-called boxcounting fractal dimension (Db , see, e.g., Hastings and Sugihara 1993, Virkkala 1991, Kunin 1998), one of the simplest forms of fractal analysis. Some species may have highly aggregated clumps of habitat available to them (represented by a relatively flat scale–area curve for their niche maps and high values of Db ), while others occupy a sparsely scattered niche-scape (resulting in a steeply sloped scale–area relationship and a low Db value). The processes of colonization and local extinction will act most strongly at different characteristic ranges of scales. Large blocks of habitat may be removed from the map at very coarse scales due to biogeographic scale barriers to dispersal, oceans or mountain ranges cutting off entire continents of potential niche space. At finer scales of analysis, rather smaller isolation distances are involved, and patches below some critical size or quality threshold may begin dropping off the list. The tradeoff between isolation and size often found in metapopulation studies (e.g., Thomas and Jones 1993) means that progressively smaller chunks of habitat are eliminated as one goes to finer scales of analysis. Where ‘‘sink’’ populations extend the map into inhospitable terrain, however, they can only do so in sites very close to existing sources of propagules, and so the area they add to the map should generally act on quite fine spatial scales. Taken together, these processes are likely to result in a somewhat flatter (higher Db ) curve for actual species distributions across


Figure 6.1 Maps (a, b) and scale–area curves (c) for a hypothetical species’ fundamental niche space (a, solid line in c) and realized populations (b, off line in c). Note that colonization and extinction processes remove larger areas at coarse scales than at fine, while adding area (‘‘sink’’ populations, denoted by gray shading) at relatively fine scales only.

scales than for the corresponding niche map, although the degree of flattening should depend on dispersal ability (fig. 6.1). As noted above, there is an intimate connection between an individual species’ distribution, as represented by a scale–area curve, and the species– area curve commonly used in the ecological literature. If we examine a finite region, and the area within that region occupied by a particular species rises


in some fashion as we move from fine to coarse scales of analysis (as it must), then the fraction of all cells occupied by the species must also rise in a related manner. These fractions can be interpreted as probabilities; a species occupying half of the 10 km cells within a survey region will have a 50% chance of being present in any one arbitrarily chosen cell at that scale. The species–area curve is the sum of such probabilities across species, and so should reflect the shape of the scale area curves that make it up. If scale–area curves are approximately linear on logarithmic axes (that is, if distributions are fractal), then it seems reasonable to expect the species area curve to be linear when plotted on logarithmic axes as well (Harte et al. 1999). Thus we may have a mechanistic basis for the power law formulation of the SAR, originally proposed by Arrhenius (1921) and widely accepted in recent decades (e.g. Rosenzweig 1995). Three Opposing Twists Unfortunately, things are not quite that simple. At least three factors can affect the shape of species–area curves, even when all of the species involved are fractally distributed. The first concerns variation in the fractal properties of different species. So long as all of the species concerned have distributions with the same fractal dimension, the species–area curve they produce will indeed be linear when plotted on logarithmic axes. However, where species differ in their fractal scaling, as many species do, the species–area curve ceases to be linear, growing progressively steeper at increasingly coarse scales (fig. 6.2; see appendix to this chapter for a more mathematical approach). This is counterintuitive for those of us used to working in arithmetic (rather than logarithmic) space; we generally suppose that the sum of a group of straight lines ought to be a straight line itself. Not so in logarithmic space, however: the log of a sum is not equal to the sum of the logs. If two scale–area curves differ in slope, the multiplicative difference between them changes across scales, and so the degree to which the higher of the two dominates their sum also changes. This can be seen most clearly where scale–area curves of species cross (fig. 6.2). Whichever species has the highest incidence at a given scale has a disproportionate influence on this sum. Where scale–area curves approach one another and eventually cross, this lead role vanishes and is ultimately replaced by a different ‘‘leader,’’ one that, of necessity, will have a steeper scale–area curve than the species it supplanted. The resulting shift toward steeper slopes at coarse scales is certainly not what the extensive literature on species–area curves would lead us to believe occurs in nature. The long-standing dispute between Arrhenius’s power function SAR (1921) and the semilog version championed by Gleason (1922) suggests that the SAR should either be linear or decelerating when plotted on two logarithmic axes; we generally don’t expect to find the sort of accelerating function suggested by the argument above! A second effect may counteract the first one. One problem with box counting as a way to measure fractals is that they can become ‘‘saturated’’; at


Figure 6.2 Effects of variation between species in scale–area curves on the properties of the resulting species–area relationship. The scale–area curves illustrated in (a) differ in slope. The summed probability of the two species occurring in a sample (b) does not rise linearly with scale, but curves slightly upward. This can be seen more clearly by plotting the slope of the curve (c), here expressed as the ratio of probabilities between successive scales. Had parallel scale– area curves (i.e., equal fractal dimensions) been used, the summed probability of occurrence would have remained linear.

coarse scales of analysis, common species come to occupy all of the available cells (fig. 6.3). Once they reach this saturated state, such species necessarily cease to increase the area they occupy at still coarser scales, as they already occupy the entire area under consideration. This should result in an abrupt


Figure 6.3 Example of map saturation. Two fine-scale grids of similar fractal dimension but different abundance are shown on the left. The commoner case (lower row) quickly comes to occupy nearly the entire arena at coarser scales of analysis (center two panels), causing the scale–area curve (on the right) to decelerate.

shift downward in the slope of their scale–area curves. In practice, however, the effect is often much more gradual, with different parts of a species’ distribution saturating progressively across a range of scales. This curvature in scale–abundance curves should cause a similar curvature in the SAR as well— resulting in a decelerating species–area curve when plotted on logarithmic axes, consistent with Gleason’s viewpoint. Depending on its relative strength, this effect could partially or wholly counteract the effect described above, perhaps even reversing it. Before leaving the subject, there is a third process that may influence the shape of species–area curves away from perfect log-linearity: the single-cell anomaly. Real populations (unlike idealized fractals) have finite distributions. At a sufficiently coarse scale, the entire sampled distribution of the species will be enclosed in a single cell. At all scales coarser than that, the species will behave as if it has a fractal dimension of zero, causing its scale–area curve to rise very steeply (fig. 6.4). As this phenomenon only takes hold at scales coarser than a species’ regional distribution, we had not anticipated that it would be very important in most SAR analyses. However, it should be noted that rare species make up a large fraction of most community samples, and this process is likely to act most strongly on such species (see below). Thus we are left with a surprisingly complex story to explain a straight line. Even if we begin with approximately fractal species distributions—which are themselves supposed to be linear on such axes—we still must balance off at least three competing forces, all of which tend to bend these simple lines. To make matters worse, two of the processes (adding dissimilar fractals and the single-cell anomaly) tend to bend the SAR upward, instead of the downward


Figure 6.4 The single-cell anomaly. Two fine-scale distributions with similar fractal dimensions, but differing in abundance are shown on the left. The rarer site (lower row) has only a single occupied cell at the intermediate scale of analysis. This results in an abruptly steeper scale–area curve at coarser scales (shown on the right).

bend we might expect to find from the empirical literature. Where the SAR follows the simple log-linearity of Arrhenius’ power function, it could either be because none of these three forces acts with any significant force (leaving us with the simple addition of similar fractals posited by Harte et al. 1999), or else this apparent simplicity must hide a complex core, with the combined effect of upward and downward curvature somehow balancing one another out. Alternatively, if the SAR follows Gleason’s decelerating SAR, it either means that the underlying distributions are not fractal or, if they are, that the impact of map saturation significantly outweighs the combined effects of the other two forces.

Testing the Ideas: A Fractal Transect of the Northern Negev To test these ideas, we need a handy SAR to dissect into its component species distributions. As luck would have it, we have just such field data to hand. One of us (WEK) had earlier carried out a transect survey of the crucifers of the northern Negev (spanning the range of sites visited at the workshop from which this book originates, see preface for details), and the data are well suited to exploring some of the ideas discussed above. Before proceeding further, a bit of explanation about the transect itself is in order. The data were collected in an attempt to explore the scaling properties of species’ distributions, and in particular the declines in abundance as one approaches the margins of a species’ range. It is widely appreciated that species tend to grow rarer as they approach their range margins (e.g., Brown


1984), indeed in the absence of some abrupt barrier such as a coastline, it would be surprising if they did not (see Lennon et al. 1997). But the nature of that decline is not well known; does the local density of a species’ populations decline, or is there simply a decline in the fraction of the landscape that it can occupy? A similar question can be posed in the language of scale–area curves: Does the slope of the curve shift, or just its elevation? To rephrase this yet again into something a biologist might recognize: Does a species remain equally patchy as it becomes rarer, or does its distribution become clumpier or (alternatively) more evenly scattered as it approaches its range margins? To address issues of this sort, one needs distributional data collected across a wide geographical range, but with sufficient local intensity to pick up finescale aggregation patterns. Given a finite amount of time and labor, these two factors (the spatial extent of the transect and local intensity of sampling) are generally inversely correlated; a fixed total number of quadrats dispersed along a growing transect results in samples scattered ever more widely apart. Again, fractal geometry may come to our assistance, providing an efficient means of collecting samples of information across a wide range of spatial scales. If species distributions are approximately fractal, as we have suggested above, then a fractal transect can be an efficient sampling tool; its overlap with a species’ distribution is simply the intersection of two fractals, a quantity which is well understood in theory: Dint ¼ Dspecies þ Dsample 2

ð1Þ

where Dspecies is the fractal dimension of the species’ distribution, Dsample is the fractal dimension of the transect used to study it, and Dint is the fractal dimension of the intersection between the two—the population captured within the sample (Mandelbrot 1982, J. Halley pers. comm.). Thus, for the simplest case of a linear transect (Dsample ¼ 1) across a species’ range, the resulting data should reveal a fractal dimension of Dint ¼ Dspecies þ 1 2 ¼ Dspecies 1

ð2Þ

Here one need only add 1 to the fractal dimension revealed by the transect to estimate the true fractal dimension of the species’ population. Note that this constrains the range of fractals that can be studied; fractal dimensions cannot be negative, so species exhibiting fractal dimensions less than 1 will not be amenable to study in this case. The same principle should apply to any form of fractal sampling; the application of equation (1) should allow the true fractal dimension to be calculated from that revealed in the sample. The Survey To study the crucifers of the Northern Negev, a variant of the well-known Cantor set (see, e.g., Hastings and Sugihara 1993) was used as a transect (fig. 6.5). The basic unit was a 0:5 0:5 m quadrat. Four of these quadrats were arranged in a line at 1 m spacings to form a sample 5 m in total length. Four sets of such samples were then arranged at 10 m spacings to form a local

100


Figure 6.5 Schematic representation of the sampling design used in the Northern Negev transect. Details are given in the text.

survey 50 m in length. Similar surveys were performed at 100 m intervals, covering a total distance of 500 m. In a similar manner, four such sections spaced 1 km apart were surveyed within a 5 km region. A total of five such regions were examined at 10 km spacings, creating a total transect 65 km in length, stretching from the edge of the Zin canyon (just south of Sede Boqer) northwards to Devira, some 16 km north of Beer Sheva. For most of its length, the transect used a gas pipeline maintenance road as an access route, which resulted in a slight west-to-east shift between survey areas; transect segments, however were oriented directly north–south. Note that the coarsest scale (with five rather than four sample areas) is a slight variation from the Cantor set pattern; at all other scales each tenfold increase in transect length is accompanied by only a fourfold increase in the area surveyed, giving a fractal dimension of logð4Þ= logð10Þ ¼ 0:602

ð3Þ

Consequently, a sample of this sort can only be used on populations with fairly high fractal dimensions; applying equation (1) suggests that only distributions with D > 1:4 should provide meaningful positive dimension estimates. The transect was designed to study population shifts across range margins, and so it was placed across a rather steep environmental gradient to improve the chances of encountering such boundaries. Over the length of the transect,

Spatial Scale and Species Diversity

101

there is a dramatic shift in water availability, with mean annual rainfall shifting from approximately 100 mm/yr in the south to about 300 mm/yr in the north. This results in a pronounced shift in vegetation from desert to nearly Mediterranean, with a corresponding turnover in species composition. Relatively uniform substrates were sampled, however, with the gaps in the transect so positioned as to skip over the bands of rock outcrops north of Sede Boqer, so that the survey included almost exclusively loess soils. The clusters of samples at different scales can be used to devise species– area curve for the transect, as well as to construct incidence curves for each of the species sampled. This is done by considering individual quadrats (0.25 m2 in area), sets of four quadrats (1 m2 total area), sets of four such groups (4 m2), and so on. Note that this does not quite fit Rosenzweig’s (1995) criterion, that SARs ideally should be composed of contiguous samples, but at least the groupings are composed of closest neighbors within the sample. While the resulting compression of space may exaggerate the rate of species turnover, the bias should act uniformly across scales, allowing a rather unnaturally steep but otherwise realistic SAR to be estimated. It should certainly serve adequately for demonstration purposes.

Transect Results Despite the rather unconventional nature of the sampling regime used here, the results appear reassuringly familiar: taken as a whole, the dataset displays a classic power law (Arrhenius) SAR (fig. 6.6), although there appears to be a bit of sigmoid curvature around the line of best fit. A total of 19 species were encountered in the survey, although more than half (10 species) were encountered in fewer than 10 of the 1280 quadrats sampled.

Figure 6.6 The species–area relationship found in the Northern Negev transect samples. The relationship is well represented by a line in log–log space, implying a power law relationship. Note, however, that there is a noticeable curvature to the relationship, with generally steeper slopes at the finest and coarsest scales, and somewhat shallower slopes at intermediate scales.

102


Examining individual species–area distributions, it quickly becomes clear that all three of the problems discussed above act within the sampled biota. First of all, it is apparent that the range of fractal properties exhibited by these species varies considerably, with estimated slopes of scale–area curves ranging from 0.135 to 0.422. This is easily sufficient to trigger the first of our biases: the accelerating curve produced when dissimilar fractals are added together. Our second bias is also apparent; many of the graphs show noticeable deceleration, presumably due to saturation. This tendency is most notable among the commonest species (which approach saturation most quickly), but the trend is visible even in species that are not particularly common. Finally, several of the rarer species show a distinct upturn at coarse scales, which is often clearly attributable to the single-cell anomaly. However, the very rarest of these species were encountered only in a single quadrat each, providing a strictly linear, but unusually steep, scale–area curve. This is simply the single cell anomaly taken to its extreme, creating an artifactual estimate of Db ¼ 0. For some of these species, the problem may be exacerbated by the sampling technique; these rare scattered species are precisely the sort for which meaningful fractal estimates may be difficult with a lowdimension sample. Until now, we have treated the transect as a single sample, 65 km long, crossing a wide diversity of conditions. However, for our purposes it may be better to consider it as five shorter (5 km long) transects, each in a different environment. We can then plot the SAR for each (fig. 6.7). Now, rather than a single SAR, we have five different curves, which conveniently exhibit a range of behaviors, stretching from pure power law dynamics (Sites I and V) through intermediate levels of curvature (Site II) to Gleasonian logarithmic curves (Sites III and IV). This puts us in a position to try to dissect what aspects of the sites and their individual species distributions contribute to these differing patterns.

Figure 6.7 Species–area relationships for the five component segments of the northern Negev transect (a) plotted on log–log axes (after Arrhenius 1921); (b) plotted on semilog axes (after Gleason 1922).


103

We can now examine the various sources of curvature, to see which is most likely to be responsible for the differences observed. All five sites contain species with markedly different fractal properties, suggesting that the dissimilar fractals effect should apply to all of them. However, the variance in slopes is fairly similar in all five sites, and is in any case uncorrelated with SAR properties, with the two power law sites, I and V, having the highest and lowest variances of the group, respectively. Consequently, we cannot rely on this factor to explain the observed differences in curve shapes between sites. There is also relatively little difference between the sites in the degree of ‘‘fractalness’’ of their component species distributions. In the absence of a widely agreed measure for such a property, we will use a rough but convenient measure: the departure of each species’ scale–area curve from the bestfit line in log–log space, as measured by the R2 value (table 6.1). While there are some difference between species and sites in this respect, the overall difference between sites is surprisingly small. Certainly we cannot distinguish the power law sites (I and V, with a mean R2 value of 0.972) from the logarithmic sites (III and IV, with a mean R2 value of 0.971) on this basis. There is a much more striking difference between sites, however, in the nature of this departure from pure fractal linearity. In the power law sites, there is a mixture of upward and downward (and sigmoid) curvature of scale–area curves, whereas the more Gleasonian sites display almost exclusively downward curvature. This, in turn, reflects a difference in the species–abundance mixtures in the various sites; the straight power law sites had a much greater share of rare species sampled, while the curved (logarithmic function) sites were dominated by a few relatively common species.

General Conclusions It has become apparent, to us at least, that the species–area relationship is a rather more complex beast than we have usually given it credit for. In its most widely accepted (power law) formulation, it is built upon a tacitly fractal perspective on species distributions, a perspective that has gained popularity long after the original model was proposed. Yet, in a way, the departures from pure fractal scaling are as interesting as the fractal pattern itself, and these may hold the key to understanding the diversity of SARs documented in the literature. We have identified at least three such distorting processes, and have demonstrated through a case study how they may combine to determine the overall shape of the curve. What appears to be simple linearity is more akin to a ceasefire line: a balance of opposing forces that are attempting to bend the curve in different characteristic ways. In the example we examined above, the observed differences in SAR shapes seem to be caused by a shifting balance between what we termed ‘‘map saturation’’ and ‘‘single cell anomalies.’’ These, in turn, may reflect shifts in the mix of species, with the former process most affecting common and widely scattered populations, while the latter is characteristic of rare and tightly clustered distributions (fig. 6.8).

104


Table 6.1 Summary of scale–area curve properties for species in the Northern Negev transect data. Values are given as R 2 values of linear relationships between log species and log area occupied R2 of correlations between log scale and log area occupied Site I Power law Torularia

0.986 (

Reboudia

0.980 0.973 (

Carrichtera

Site II Intermediate

Site III Exponential

Site IV Exponential

0.962 ( 0.886 ( 0.921 (

0.988 ( 0.870 (

0.980 ( 0.914 (

1*

0.997

0.991 (

0.944 (

0.999 (

0.992

0.990 )

Matthiola

0.946 ( 0.973 (

Enarthrocarpus Biscutella

0.971 ) 0.937 (

0.947 (

Leptoleum

1* 0.992

Erucaria Nasturtiopsis Diplotaxis harra

Site V Power law

0.985 0.966 ( 0.999 (

0.933 ( 0.980 (

D. erucoides

0.963

Hirschfeldia

0.954

Others:

1* 0.939 )

0.997 ( 1* 1*

0.980 0.985

Mean

0.965

0.958

0.964

0.978

0.979

Mean ignoring singletons

0.962

0.953

0.964

0.974

0.972

Summed departure from linearity Decelerating (( ) Mixed () Accelerating ( ) )

P ð1 R2 )

0.272 (71%)

0.284 (86%)

0.143 (98%)

0.115 (88%)

0.094 (49%)

0.020 (5%)

0.035 (11%)

0.003 (2%)

0.016 (12%)

0.098 (51%)

0.090 (24%)

0.010 (3%)

0

0

0

): accelerating curve; (: decelerating curve; : curve with different directions of curvature over different ranges of scales. *Site with only one occupied quadrat of the species indicated. Such species necessarily had perfectly linear scale–area plots, with R2 therefore equal to 1.

There are, of course, limits to how much we can prove on the basis of one transect sample, whether viewed as a whole or decomposed into its five component sites. However, if the patterns uncovered above generalize to other SARs, there seem to be clear indications that the shape of such curves can reveal interesting and rather subtle bits of information about the communities of organisms they represent. Areas dominated by a few ubiquitous species, showing saturating distributional dynamics, should display tendencies toward Gleasonian SARs, whereas sites with large numbers of rare and restricted species might be expected to produce curves more akin to Arrhenius’ original power law model. Clearly, this is only a first step, but it is encouraging to find


105

Figure 6.8 Effect of distributional properties on two processes that affect SAR shape. Rare and/or highly aggregated species are particularly prone to single-cell anomalies, as their populations are readily enclosed within a single occupied cell. This creates a steep scale–area curve at coarse scales, biasing estimates of Db downward (as indicated by the arrow). Very common or widely scattered populations, on the other hand, tend to result in map saturation, causing a flattening of scale–area curves and an upward bias in Db estimates (again, indicated by an arrow).

that SARs can be successfully dissected, and that potentially useful information can be found from among their entrails. The SAR may be among the oldest tools in the study of biodiversity, but it looks set to begin a second career.

Acknowledgments The authors would like to thank the Israeli hosts of the 1999 workshop (see preface for details) for their hospitality and encouragement, both at the workshop and subsequently. WEK would also like to thank the British Royal Society for supporting the fieldwork described in the latter half of the chapter (twice!). JJL is supported by a Co-operative Joint Venture Agreement with the U.S.D.A. Forest Service.

Appendix: On the Sum of Scale–Area Curves The following sections provide the mathematical foundations for a few points made verbally in the text, and underline the foundations of a point not now included in the text but which was included in the original oral presentation of this chapter. The SAR for Fractal Species Distributions Let the number of occupied quadrats n at a particular scale s be given by n ¼ n0 ð1=sÞD

106


where n0 is a constant (which may be interpreted as the probability that a unit quadrat, of s = 1, is occupied), and D is the box-counting fractal dimension. The probability that a randomly selected quadrat at a particular scale is occupied is the ratio of the number of squares occupied at this scale divided by the total number of squares at this scale. This latter is simply (1/s)2, so the probability is p ¼ nð1=sÞ2 ¼ n0 ð1=sÞD2 ¼ n0 s2D

If species are distributed independently, then the expected number of species in a quadrat of scale s is the sum of the probabilities of the individual species: X RðsÞ ¼ pi Which on expansion of p gives RðsÞ ¼

X

n0i s2Di

This is the fractal SAR. It is not linear in log–log axes unless all Di s are equal, since different species make different contributions at different scales. To demonstrate this, we explicitly solve for the slope of this SAR.

The Slope of the SAR Instantaneous Scale Change Partial differentiation with respect to s gives the slope of the curve at each s: X R=s ¼ ð2 Di Þn0i s1Di It can be seen from this that the contribution of a particular species to the overall rate of accumulation of species with area depends on two main factors: (i) the abundance of the species at a reference scale (n0) and (ii) its sensitivity to scale, 1 D. A species with a relatively large n0 but with a small D 1 may make a relatively large contribution at coarse scales but a relatively small contribution at fine scales.

Discrete Scale Change If we consider a discrete change in scale, then the change in richness with scale is given by X 2Di R ¼ ai s ; ai ¼ n0i ð1 1=kÞ2Di Where k is the change in linear scale (e.g., k ¼ 10 means a 100-fold change in quadrat area). This equation is very similar to the continuous scale change version.


107

The SAR and Correlated Species Incidence One final issue we considered is the degree of spatial correlation between species incidence. In an earlier draft of this chapter (the one presented orally at the workshop), WEK asserted that the scaling patterns of species correlation might affect the SAR in interesting ways, especially in arid lands, where complex patterns of positive and negative correlations at different scales might be expected. We tested these ideas by simulating species distributions, with different degrees of spatial correlation (from completely independent distributions to strong positive and negative spatial correlations). Spectral synthesis simulation of intercorrelated species distribution patterns (with fixed D ¼ 1:5) gave a fixed SAR, independent of the level of correlation! It appears not to matter whether or not species tend to occur together—if we accumulated species from a randomly chosen point in the landscape, then the same average SAR curve occurs even if species are perfectly correlated, that is, distributed identically.

References Arrhenius, O. 1921. Species and area. Journal of Ecology 9: 95–99. Brown, J.H. 1984. On the relationship between the abundance and distribution of species. The American Naturalist 124: 255–279. Chase, C.G. 1992. Fluvial landsculpting and the fractal dimension of topography. Geomorphology 5: 39–57. Connor, E.F., and E.D. McCoy. 1979. The statistics and biology of the species-area relationship. The American Naturalist 113: 791–833 Deschaux, V., and M. Souriau. 1996. Topography of large-scale watersheds—fractal texture and global drift—application to the Mississippi basin. Earth and Planetary Science Letters 143: 257–267. Frohn, R.C., K.C. McGwire, V.H. Dale, and J.E. Estes. 1996. Using satellite remotesensing analysis to evaluate a socioeconomic and ecological model of deforestation in Rondonia, Brazil. International Journal of Remote Sensing 17: 3233–3255. Gaonach, H., S. Lovejoy, and J. Stix. 1992. Scale invariance of basaltic lava flows and their fractal dimensions. Geophysical Research Letters 19: 785–788. Gleason, H.A. 1922. On the relation between species and area. Ecology 3: 158–162 Hanski, I.A., and M.E. Gilpin. 1997. Metapopulation Biology. Academic Press, San Diego. Harte, J., Kinzig, A., and Green, J. 1999. Self-similarity and the distribution and abundance of species. Science 284: 334–336. Hastings, H.M., and G Sugihara. 1993. Fractals: a User’s Guide for the Natural Sciences. Oxford University Press, Oxford. Hutchinson, G.E. 1957. Concluding remarks. Cold Spring Harbor Symposium on Quantitative Biology 22: 415–427. Klinkenburg, B., and M.F. Goodchild. 1992. The fractal properties of topography—a comparison of methods. Earth Surface Processes and Landforms 17: 217–234. Kulatilake, P.H.S.W., R Fiedler, and B.B. Panda. 1997. Box fractal dimension as a measure of statistical homogeneity of jointed rock masses. Engineering Geology 48: 217–229.

108


Kunin, W.E. 1998. Extrapolating species abundance across spatial scales. Science 281: 1513–1515. Lennon, J.J., J.R.G. Turner, and D. Connell. 1997 A metapopulation model of species boundaries. Oikos 78: 486–502. Levins, R. 1970. Extinction. In M. Gerstenharber (ed.), Some Mathematical Questions in Biology. American Mathematical Society 2: 77–107. Mandelbrot, B.B. 1982. The Fractal Geometry of Nature. W.H. Freeman, New York. McAlpine, R.S., and B.M. Wotton. 1993. The use of fractal dimension to improve wildland fire perimeter predictions. Canadian Journal of Forest Research (Journal Canadien de la Recherche Forestiere) 23:1073–1077. Onof, C., and H.S. Wheater. 1996. Analysis of the spatial coverage of British rainfall fields. Journal of Hydrology 176: 97–113. Pelletier, J.D., B.D. Malamud, T. Blodgett, and D.L. Turcotte. 1997. Scale-invariance of soil moisture variability and its implications for the frequency-size distribution of landslides. Engineering Geology 48: 255–268. Preston, F.W. 1960. Time and space and the variation of species. Ecology 41: 611–627. Pulliam, H.R. 1988. Sources, sinks, and population regulation. The American Naturalist 132: 652–661. Richardson, L.F. 1961. The problem of contiguity: an appendix of statistics of deadly quarrels. General Systems Yearbook 6: 139–187. Rosenzweig, M.L. 1995. Species Diversity in Space and Time. Cambridge University Press, Cambridge. Thomas, C.D., and T.M. Jones. 1993. Partial recovery of a skipper butterfly (Hesperia comma) from population refuges: lessons for conservation in a fragmented landscape. Journal of Animal Ecology 62: 472–481. Tilman, D., and P. Karieva (eds.) 1997. Spatial Ecology: The Role of Space in Population Dynamics and Interspecific Interactions. Princeton University Press, Princeton. Virkkala, R. 1991. Spatial and temporal variation in bird communities and populations in north-boreal coniferous forests: a multi-scale approach. Oikos 62: 59–66. Whittakker, R.H. 1972. Evolution and measurement of species diversity. Taxon 21: 213–251. Williams, C.B. 1964. Patterns and the Balance of Nature. Academic Press, New York.

7 Microbial Contributions to Biodiversity in Deserts Peter M. Groffman Eli Zaady Moshe Shachak

T

he number of species living in the soil may well represent the largest reservoir of biodiversity on earth (Giller 1996, Wardle and Giller 1996, Service 1997). Five thousand microbial species have been described and identified (Amann and Kuhl 1998), but the actual number of species may be greater than 1 million (American Society for Microbiology 1994), larger even than the number of insect species (Service 1997). Over the last 10 to 15 years, interest in soil biodiversity has soared, driven by advances in molecular techniques that allow for identification and analysis of soil microbes, many of which are difficult to extract and culture (Kennedy and Gewin 1997). However, the factors that control soil microbial biodiversity and the links between soil biodiversity and ecosystem function are still unclear (Beare et al. 1995, Schimel 1995, Freckman et al. 1997, Brussard et al. 1997, Wall and Moore 1999). Soil may represent an excellent venue for exploring links between biodiversity and ecosystem function. The vast numbers of species in soil and methodological problems have long necessitated a functional approach in soil studies. As a result, soil functions important to organic matter degradation, nutrient cycling, water quality, and air chemistry are well studied (Groffman and Bohlen 1999). As our knowledge of soil biodiversity increases, this information may provide a strong basis for evaluating links between biodiversity and these functions. Evaluating functional diversity of soil communities requires considering how microbes interact with plants and soil fauna to produce patterns of ecosystem processes (Wall and Moore 1999). These interactions vary within and between ecosystems (i.e., across landscapes). Throughout this book, we 109

110


suggest that the science of biodiversity must consider links to ecosystem processes and interactions with landscape diversity (Shachak et al. this volume). The need for these links is particularly clear when considering soil biodiversity. There have been relatively few studies of microbial processes in desert soils, and very little analysis of desert soil biodiversity (Parker et al. 1984, Schlesinger et al. 1987, Peterjohn 1991, Fließbach et al. 1994, Zaady et al. 1996a,b, Steinberger et al. 1999). Given this lack of data, we cannot provide a comprehensive review of desert soil biodiversity, with comparison to more mesic ecosytems. Rather, we can review some of the unique features of desert microbial ecology and pose questions about desert soil biodiversity that should be addressed in future research. Our analysis, which is be framed in the context of ‘‘ecosystem processes and landscape biodiversity,’’ which is the theme of this book (Shachak et al. this volume), addresses the importance of (1) wetting events, (2) ecosystem diversity, and (3) landscape diversity as controllers of microbial biomass, activity, and biodiversity in desert soils. We address these features and questions primarily with data that we have collected in the Negev desert in Israel, but focus on results that are relevant to other deserts around the world.

The Importance of Wetting Events as Controllers of Microbial Biomass, Activity and Biodiversity in Desert Soils Many studies have found that much of the microbial activity that occurs in desert soils is concentrated during brief periods of high activity following wetting events (Parker et al. 1984, Schlesinger et al. 1987, Peterjohn 1991, Fließbach et al. 1994, Zaady et al. 1996a,b). During these periods, soils are warm and wet, that is, nearly optimal for microbial activity. Given the short generation times, dispersal possibilities, and survival capacities of microbes, we hypothesize that these brief periods of optimal conditions should allow for the development of diverse microbial communities. In one of the few studies of desert soils using recently developed molecular techniques (phospholipids fatty acid analysis, or PLFA), Steinberger et al. (1999) evaluated microbial biomass and diversity in Judean desert (Israel) soils along a precipitation gradient, during wet and dry periods. They found that biomass and diversity were highest in the wettest sites, during the wettest periods. Our own work in Negev desert (Israel) soils has found that the intensity of soil respiration during wetting events is high enough to deplete soil O2 concentration, providing the anaerobic conditions required for denitrification (Zaady et al. 1996b). We hypothesize that anaerobic respiration processes like denitrification are actually more important in arid ecosystems than in more temperate areas (Peterjohn and Schlesinger 1990, Groffman et al. 1993, Frank and Groffman 1998) because they are so prevalent during the pulses of activity following wetting. The presence of anaerobic respirers like denitrifiers


111

may represent an important functional diversity, maximizing the extent of nutrient cycling activity that can occur during the brief periods of water availability following rain events. On the other hand, the presence of denitrifiers fosters gaseous N loss from these systems, which may be a critical constraint to primary production in these ecosystems (Peterjohn and Schlesinger 1990).

Ecosystem Diversity and Microbial Biomass, Activity, and Biodiversity in Deserts Ecosystem diversity is related to the number of (1) compartments that partake in ecosystem processes, (2) trajectories in ecosystem processes, and (3) functional groups involved in ecosystem processes. In its simplest form, nutrient cycling involves the movement of nutrients between plants, animals and inorganic forms in the soil (fig. 7.1). However, as shown in figure 7.1, ecosystem biodiversity can considerably complicate nutrient cycles. The presence of different plants, which produce different amounts and types of organic matter (Pastor et al. 1984, Scott and Binkley 1997, Finzi et al. 1998), and the importance of fauna (Beare et al. 1992) can vary greatly in different ecosystems. The importance of ecosystem diversity to soil biodiversity and its role in sustaining ecosystem functions are not clear. Our work in the Negev suggests that ecosystem diversity may be important to nutrient cycling functions in aridlands. As shown in figure 7.1, Negev desert ecosystems have a large number of ecosystem components and complex nutrient cycling patterns. This ecosystem diversity may lead to high soil microbial diversity. For example, snails have long been recognized as an important component of Negev desert ecosystems (fig. 7.2, Shachak et al. 1987, Jones and Shachak 1990, 1994). Due to their high density and high feeding rates they produce large amounts of feces. These feces may greatly facilitate the cycling of nutrients from plants back to inorganic soil pools by functioning as a pool of nutrients that is readily mobilized when increases in soil moisture permit (fig. 7.3, Zaady et al. 1996a). Alternatively, snail feces may detract from soil nutrient cycling functions in that this mobilizable pool of nutrients is subject to gaseous and leaching loss. Thus snails have both positive and negative effects on ecosystem nutrient cycling and retention. Ecosystems with snails clearly have higher N availability during wetting events (fig. 7.3), but they also likely have higher N losses during these periods as well. An important question for future research is to determine if ecosystems with snails have higher soil microbial biodiversity and if this biodiversity influences nutrient cycling and loss functions during wetting events. More fundamentally, it will be important to examine relationships between ecosystem diversity and nutrient cycling processes in other ecosystems to determine if high ecosystem diversity facilitates nutrient cycling processes in a wide range of systems (e.g., it would be interesting to evaluate deserts with no snails).

112


Figure 7.1 Conceptual diagram of a nutrient cycle in a Negev desert ecosystem highlighting the importance of ecosystem diversity (i.e., multiple ecosystem compartments, trajectories, and functional groups) as a regulator of the transformation of organic matter and the importance of microbes (MO) as decomposers.

It is important to note that there are several types of snails in the Negev. There are rock-eating snails that feed on endolithic lichens, soil-eating snails that feed on cyanobacteria in soil crusts, and plant-eating snails that feed on detritus (Yom-Tov 1971, Shachak et al. 1987, Jones and Shachak 1990). Snail consumption of these diverse food sources adds a high diversity of organic materials to the microbial community. It remains to be seen if this high diversity of food sources increases soil biodiversity. In contrast to snail feces, plant litter appears to act as an important sink for nutrients released during wetting events (fig. 7.4, Zaady et al. 1996b). In simulated wetting events with litter, microbial N pools in soil decreased over time, suggesting that N was being sequestered in litter rather than remaining


113

Figure 7.2 Negev desert landscape in Sayeret Shaked Park near beer Sheva in the northern Negev Desert, Israel (31 170N, 34 370E), showing high density of snails.

Figure 7.3 Inorganic N during a simulated wetting event (a 12-day laboratory incubation at 25 C) in Negev desert soils with and without snail feces. Note that a large pulse of inorganic N was released from snail feces but that this pulse disappeared over the course of the incubation. Feces were collected from sites in the rocky hills of the northern Negev Desert, Israel near Sede Boqer (30 520N 34 470E) and were derived from the snail Trochoidea seetzenii feeding on litter the shrub Zygophyllum dumosum. Data from Zaady et al. (1996a).

114


Figure 7.4 Chloroform-labile (microbial) N during a simulated wetting event (a 30day laboratory incubation at 25 C) in Negev desert soils with and without plant litter. Note that soil microbial N decreases over time in the litter-amended incubations, suggesting that N released from soil by wetting was sequestered in the litter. Litter was collected from 20 macrophytic patch mounds in Sayeret Shaked Park near Beer Sheva in the northern Negev Desert, Israel, (31 170N, 34 370E), mixed, and then sieved to five different size classes; 250, 500, 1000, 2000 mm. Material less than 250 mm was considered to be soil. Values are mean standard error of all litter size classes combined versus the soil-alone treatment. Data from Zaady et al. (1996b).

in the actively cycling (and more easily lost) soil microbial pool. Litter may counteract some of the stimulatory effects of snail feces, which are often mixed in with litter. The snail and litter examples suggest that arid conditions do not inherently inhibit the development of high ecosystem diversity and the emergence of complex nutrient cycling activity. Indeed, complex nutrient cycles may greatly facilitate nutrient availability and plant growth under arid conditions. However, these complex nutrient cycles may also stimulate nutrient loss. In all ecosystems, there is an inherent tension between processes that facilitate nutrient cycling and availability and those that act to prevent nutrient loss (Bormann and Likens 1979, Vitousek et al. 1998). Determining if and how ecosystem diversity affects this balance is not clear, and represents an important challenge for ecology over the next decade or so. Determining how ecosystem diversity influences soil biodiversity may be critical to resolving this functional question.


115

Landscape Diversity and Microbial Biomass, Activity, and Biodiversity in Deserts Many desert areas are characterized by high landscape diversity, for example marked patchiness in the distribution of vegetation (Barth and Klemmedson 1978, West 1989, Schlesinger et al. 1990, Allen 1991, Agular and Sala 1999). This landscape-scale diversity is thought to have important effects on desert ecosystem functions related to nutrient cycling and productivity (Shachak et al. 1998). Our data from the Negev confirm that landscape diversity is important to ecosystem function, but suggest that this diversity may have both positive and negative effects on nutrient cycling and retention. In the Negev there are two main types of patches (fig. 7.5, Friedmann and Galun 1974, West 1990, Shachak et al. 1993, Shachak and Boeken 1994, Zaady and Shachak 1994): (1) macrophytic patches consisting of herbs and shrubs on slightly raised mounds and (2) microphytic patches consisting of cyanobacteria, algae, mosses, and lichens that are characterized by the presence of a relatively impermeable soil crust. This landscape diversity likely influences microbial biodiversity through its effects on key factors that control microbial community composition and function, such as soil moisture and organic matter. The macrophytic patches, which are common to many deserts, are considered to be ‘‘islands of fertility’’ with high levels of water and nutrients (West 1989, Smith et al.1994) relative to the microphytic patches. The extreme differences in the nature and extent of water and nutrient availability in the different patch types likely increases overall microbial biodiversity in aridlands.

Figure 7.5 Conceptual diagram of the Negev desert landscape showing macrophytic patches dominated by shrubs and microphytic patches dominated by cyanobacteria, algae, mosses, lichens, and soil crust. From Zaady et al. (1998).

116


Landscape diversity effects on microbial processes have important implications for ecosystem productivity and nutrient retention in aridlands (Shachak et al. 1998, Agular and Sala 1999). For example in the Negev, there is a strong assumption that both major patch types are important components of the landscape nutrient cycle. Plant growth in the macrophytic patches is thought to be strongly limited by N (Evenari 1985), while N-fixing cyanobacteria are thought to contribute significant amounts of N to the microphytic patches (West 1990, Evans and Ehleringer 1993). Runoff following rainfall events may transfer N from the crust-covered microphytic patches to the macrophytic patches (Zaady and Shachak 1994). A major area of research is to determine the optimal juxtaposition and abundance of different patch types for specific ecosystem management objectives (Shachak et al. 1998). Some of our research in the Negev raises questions about the role of different patch types in landscape nutrient cycling and loss. First, the role of the biological crusts as sources of N to the landscape may be overstated. While there is a general assumption that the crust areas fix N at relatively high rates, and that this N contributes to productivity in the macrophytic mounds, we have found equal or higher rates of N fixation in the mounds than in the crusts (fig. 7.6, Zaady et al. 1998). Second, while the macrophytic patches are clearly ‘‘hotspots’’ of higher plant productivity in the landscape, they support

Figure 7.6 N fixation during a simulated wetting event (a 72-hour laboratory incubation at 25 C) in soils from macro-and microphytic patches in the Negev desert. Soils were collected from nine sites located along a rainfall gradient in the Negev Deset, Israel, ranging from from 50 to 325 mm per annum. At each site, samples were collected from soils underlying shrubs, from cyanobacteria-dominated microphytic crusts (less than 125 mm rainfall) and moss-dominated microphytic crusts (more than 125 mm rainfall). N fixation was measured with an acetylene reduction method. Values are mean standard error.


117

Figure 7.7 Denitrification during a simulated wetting event (a 72-hour laboratory incubation at 25 C) in soils from macro-and microphytic patches in the Negev desert. Soils were collected from nine sites located along a rainfall gradient in the Negev Deset, Israel, ranging from from 50 to 325 mm per annum. At each site, samples were collected from soils underlying shrubs, from cyanobacteria-dominated microphytic crusts (less than 125 mm rainfall) and moss-dominated microphytic crusts (more than 125 mm rainfall). Denitrification was measured with an acetylene inhibition method. Values are mean standard error.

high rates of denitrification, and are therefore also ‘‘hotspots’’ of N loss (fig. 7.7). There is a clear need for more specific evaluation of the functional role of different patch types to determine just how landscape diversity is important to water and nutrient cycling functions in deserts. Future research should also determine if soil microbial biodiversity plays a key role in the functional differences between different patch types.

Conclusions Soil microbial ecology has been dominated by functional studies of processes, especially those related to nutrient cycling. While we know a lot about how ecosystem conditions influence microbial activity, we know little about the role that microbial diversity and community structure play in this activity. Studies of microbial response to variation in ecosystem conditions should be a productive platform for posing questions about links between biodiversity and ecosystem function. Deserts, where conditions vary greatly in time and space, may be especially valuable venues for such studies.

118


Microbial activity in deserts is concentrated in brief periods of optimal conditions during wetting events. A full range of processes (e.g., aerobic and anaerobic) occurs during these events. This functional diversity is important to patterns of nutrient cycling and nutrient loss in desert ecosystems. High ecosystem diversity could facilitate microbial diversity and does facilitate microbial activity in desert ecosystems. A high diversity of ecosystem compartments, trajectories in ecosystem processes, and functional groups facilitates high rates of microbial activity during wetting events, allowing for large amounts of nutrient cycling to occur during these events, at least in the Negev. The links between ecosystem diversity, soil microbial biodiversity, and these nutrient cycling functions should be explored in future research in a wide range of ecosystem types. While aridity may not constrain microbial diversity and activity, it may constrain biotic control over nutrient losses from desert ecosystems. The potential for gaseous and hydrologic loss during wetting events is high. In all ecosystems, there is an inherent tension between processes that facilitate nutrient cycling and availability and those that act to prevent nutrient loss. A favorable balance of these processes may be less common in deserts than in more mesic ecosystems. The role of soil microbial biodiversity in this balance may be important, for example if there are more functional types of microbes in the soil, there may be a better balance between cycling, availability, and loss. The balance between processes that facilitate both nutrient cycling and loss needs to be considered in assessments of patch dynamics and management in desert landscapes. Decisions about managing the abundance and juxtaposition of patch types should consider effects on nutrient losses as well as the nutrient input and plant diversity considerations that usually drive these assessments. Evaluating the effects of soil microbial biodiversity on the nutrient cycling and loss functions in different patch types should be a priority for future research.

References Agular M.R., and O.E. Sala. 1999. Patch structure, dynamics and implications for the functioning of arid ecosystems. TREE 14: 273–277. Allen, E.B. 1991. Temporal and spatial organization of desert plant communities. Pp. 295–332 in J. Skujins (ed.). Semiarid Lands and Desert: Soil Resource and Reclamation. Marcel Decker, New York. American Society for Microbiology. 1994. Microbial diversity research priorities. American Society for Microbiology, Washington, D.C. Amman, R., and M. Kuhl. 1998. In situ methods for assessment of microorganisms and their activities. Current Opinion in Microbiology 1: 352–358.


119

Barth, R.C., and J.O. Klemmedson. 1978. Shrub-induced spatial patterns of dry matter, nitrogen, and organic carbon. Soil Science Society of America Journal 42: 804–809. Beare, M.H., R.W. Parmelee, P.F. Hendrix, W. Cheng, D.C. Coleman, and D.A. Crossley, Jr. 1992. Microbial and faunal interactions and effects on litter nitrogen and decomposition in agroecosystems. Ecological Monographs 62: 569–591. Beare, M.H., D.C. Coleman, D.A. Crossley, Jr., P.F. Hendrix, and E.P. Odum. 1995. A hierarchical approach to evaluating the significance of soil biodiversity to biogeochemical cycling. Pp. 5–22 in H.P. Collins, G.P. Robertson and M.J. Klug (eds.). The Significance and Regulation of Soil Biodiversity, Kluwer Academic Publishers, The Netherlands. Bormann, F.H., and G.E. Likens. 1979. Pattern and Process in a Forested Ecosystem. Springer-Verlag, New York. Brussaard, L., V.M. Behan-Pelletier, D.E. Bignell, V.K. Brown, W. Didden, P. Folgarait, C. Fragoso, D.W. Freckman, V.V.S.R. Gupta, T. Hattori, D.L. Hawksworth, C. Klopatek, P. Lavelle, D.W. Malloch, J. Rusek, B. Soderstrom, J.M. Tiedje, and R.A. Virginia. 1997. Biodiversity and ecosystem functioning in soil. Ambio 26: 563–570. Evans, R.D., and J.R. Ehleringer. 1993. A break in the nitrogen cycle in aridlands? Evidence from d15 N of soils. Oecologia 94: 314–317. Evenari, M. 1985. The desert environment. Pp. 1–22 in M. Evenari, I. Noy-Meir and E.W. Goodall (eds.). Hot Deserts and Arid Shrublands, Vol. 12A: Ecosystems of the World. Elsevier, Amsterdam. Finzi, A.C., N. van Breemen, and C.C. Canham. 1998. Canopy tree-soil interactions within temperate forests: Species effects on soil carbon and nitrogen. Ecological Applications 8: 440–446. Fließbach A., S. Sarig, and Y. Steinberger. 1994. Effects of water pulses and climatic conditions on microbial biomass kinetics and microbial activity in a Yermosol of the central Negev. Arid Soil Research and Rehabilitation 8: 353–362. Frank, D.A., and P.M. Groffman. 1998. Denitrification in a semi-arid grazing ecosystem. Oecologia 117: 564–569. Freckman, D.W., T.H. Blackburn, L. Brussaard, P. Hutchings, M.A. Palmer, and P.V.R. Snelgrove. 1997. Linking biodiversity and ecosystem functioning of soils and sediments. Ambio 26: 556–562. Friedman, E.I., and M. Galun. 1974. Desert algae, lichens and fungi. Pp. 165–212 in G.W. Brown (ed.). Desert Biology, Vol. 2. Academic Press, London. Giller, P.S. 1996. The diversity of soil communities, the ‘‘poor man’s tropical rainforest.’’ Biodiversity and Conservation 5: 135–168. Groffman, P.M., and P.J. Bohlen. 1999. Soil and sediment biodiversity: cross-system comparisons and large scale effects. BioScience 49: 139–148. Groffman, P.M., C.W. Rice, and J.M. Tiedje. 1993. Denitrification in a tallgrass prairie landscape. Ecology 74: 855–862. Jones, C.G., and M. Shachak. 1990. Fertilization of the desert soil by rock-eating snails. Nature 346: 839–841. Jones, C.G., and M. Shachak. 1994. Desert snails’ daily grind. Natural History 103: 56–61. Kennedy, A.C., and V.L. Gewin. 1997. Soil microbial diversity: Present and future considerations. Soil Science 162: 607–617.

120


Parker, L.W., P.F. Santos, J. Phillips, and W.G. Whitford. 1984. Carbon and nitrogen dynamics during the decomposition of litter and roots of a Chihuahuan desert annual, Lepidium lasiocarpum. Ecological Monographs 54: 339–360. Pastor, J., J.D. Aber, C.A. McClaugherty, and J.M. Melillo. 1984. Aboveground production and N and P cycling along a nitrogen mineralization gradient on Blackhawk Island, Wisconsin. Ecology 65: 256–268. Peterjohn, W.T. 1991. Denitrification: enzyme content and activity in desert soils. Soil Biology and Biochemistry 23: 845–855. Peterjohn, W.T., and W.H. Schlesinger. 1990. Nitrogen loss from deserts in the western United States. Biogeochemistry 10: 67–79. Schlesinger, W.H., P.J. Fonteyn, and G.M. Marion. 1987. Soil moisture content and plant transpiration in the Chihuahuan desert of New Mexico. Journal of Arid Environments 12: 119–126. Schlesinger, W.H., J.F. Reynolds, G.L. Cunningham, L.F. Huenneke, W.M. Jarrell, R.A. Virginia, and W.G. Whitford. 1990. Biological feedbacks in global desertification. Science 247: 1043–1048. Schimel, J. 1995. Ecosystem consequences of microbial diversity and community structure. Pp. 240–254 in F.S. Chapin and A. Ko¨rner (eds.). Arctic and Alpine Biodiversity. Springer-Verlag, Berlin, Heidelberg. Scott, N.A., and D. Binkley. 1997. Foliage litter quality and annual net N mineralization: comparison across North American forested sites. Oecologia 11: 151–159. Service, R.F. 1997. Microbiologists explore life’s rich, hidden kingdoms. Science 275: 1740–1742. Shachak, M., and B. Boeken. 1994. Desert plant communities in human-made patches—implications for management. Ecological Applications 4(4): 702–716. Shachak, M., B. Boeken, and E. Zaady. 1993. Ecological Aspects of the Savannization Project. Annual Progress Report 1993, Jewish National Fund (in Hebrew). Shachak, M., C.G. Jones, and Y. Granot. 1987. Herbivory in rock and the weathering of a desert. Science 236: 1098–1099. Shachak, M., M. Sachs, and I. Moshe. 1998. Ecosystem management of desertified shrublands in Israel. Ecosystems 1: 475–483. Shachak, M., R. Waide, and P.M. Groffman, this volume. Ecosystem processes: a link between species and landscape diversity. Smith, J.L., J.J. Halvorson, and H. Bolton Jr. 1994. Spatial relationships of soil microbial biomass and C and N mineralization in a semi-arid shrub-steppe ecosystem. Soil Biology and Biochemistry 26: 1151–1159. Steinberger, Y., L. Zelles, Q.Y. Bai, M. Von Lu¨tzow, and J.C. Munch. 1999. Phospholipid fatty acid profiles as indicators for the microbial community structure in soils along a climatic transect in the Judean Desert. Biology and Fertility of Soils 28: 292–300. Vitousek, P.M., L.O. Hedin, P.A. Matson, J.H. Fownes, and J. Neff. 1998. Withinsystem element cycles, input-output budgets and nutrient limitation. Pp. 432–451 in: M.L. Pace and P.M. Groffman (eds.) Successes, Limitations and Frontiers in Ecosystem Science. Springer-Verlag, New York. Wall, D.H., and J.C. Moore. 1999. Interactions underground. BioScience 49: 109–188. Wardle, D.A., and K.E. Giller. 1996. The quest for a contemporary ecological dimension to soil biology. Soil Biology and Biochemistry 28: 1549–1554. West, N.E. 1989. Spatial pattern-functional interactions in shrub-dominated plant communities. Pp. 283–305 in: C.M. McKell (ed.). The Biology and Utilization of Shrubs. Academic Press, London.


121

West, N.E. 1990. Structure and function of microphytic soil crusts in wildland ecosystems of arid to semi-arid regions. Advances in Ecological Research 20: 179–22. Yom-Tov, Y. 1971. The biology of the desert snails. Trochoidea seetzenii and Sphincterochia boissieri. Israel Journal of Zoology 20: 213–248. Zaady, E. and M. Shachak. 1994. Microphytic soil crust and ecosystem leakage in the Negev desert. American Journal of Botany 81: 109. Zaady, E., P.M. Groffman, and M. Shachak. 1996a. Release and consumption of nitrogen from snail feces in Negev desert soils. Biology and Fertility of Soils 23: 399–405. Zaady, E., P.M. Groffman, and M. Shachak. 1996b. Litter as a regulator of nitrogen and carbon dynamics in macrophytic patches in Negev desert soils. Soil Biology and Biochemistry 28: 39–46. Zaady, E., P.M. Groffman, and M. Shachak. 1998. Nitrogen fixation in macro- and microphytic patches in the Negev desert. Soil Biology and Biochemistry 30: 449–454.

8 Unified Framework I Interspecific Interactions and Species Diversity in Drylands Gary A. Polis Yoram Ayal Alona Bachi Sasha R.X. Dall Deborah E. Goldberg Robert D. Holt Salit Kark Burt P. Kotler William A. Mitchell

T

he goal of this chapter is to delineate how abiotic conditions, regional processes, and species interactions influence species diversity at local scales in drylands. There is a very rich literature that bears on this topic, but here we focus on mechanisms that promote or constrain local diversity and ask how these factors apply to deserts. We ask, ‘‘What is different about deserts, relative to other habitats, in their patterns of diversity, temporal variability in productivity, and spatial heterogeneity?’’ We assess how such differences might modify extant theory, and sketch relevant examples. Compared with other biomes, productivity, population densities, and community biomass are much lower in deserts, and temporal heterogeneity is typically higher. Do these differences imply distinct ecological processes and patterns in deserts? Or, do processes operate in deserts in similar ways as in tropical forests or grasslands? For example, it is often assumed that abiotic factors are more important in deserts. If so, how do abiotic factors modify biotic interactions? How do we integrate physical and biotic interactions? More generally, we ask what should be the main goals and approaches of a research program to understand the role of species 122

Unified Framework I 123

interactions in determining community structure in drylands, as modified by abiotic factors and regional processes.

What Is Different About Drylands? Relative Diversity of Deserts Deserts are traditionally perceived as relatively simple ecosystems harboring low species diversity. Yet increasing evidence suggests that desert communities can be highly diverse and complex. To our knowledge the only systematic analysis of the relative diversity in desert versus nondesert communities was compiled by Polis (1991a). These data suggest that patterns differ widely among taxonomic groups. In some cases, deserts support high diversity, comparable to or even higher than nonarid areas (see Polis 1991b). For example, while avian (Wiens 1991) and anuran (Woodward and Mitchell 1991) diversities are low compared with other biomes, desert annual plants show extremely high species diversity (Inouye 1991). Ants, succulent plants, lizards, scorpions, and tenebrionid beetles also have relatively high diversity in deserts (Polis 1991a–c, Wiens 1991). But, while very high diversity may occur, local diversity varies greatly in space and time (e.g., ants and annual plants: Danin 1977, Inouye 1991, MacKay 1991). We suggest that deserts lie somewhere in the middle of the spectrum of diversity among biomes of the world, rather than at one extreme as often assumed (Polis 1991a–c). One theme in desert research has been to emphasize clines of diversity along physical and ‘‘aridity’’ gradients. Different patterns may result from the fact that studies compare different taxa, scales, and ecological regions. Studies focusing on trends in species diversity across physical gradients differ in the type of gradient compared. In some cases, precipitation gradients (e.g., birds; MacKay 1991) are studied. Some workers even compare latitudinal clines or a combination of these factors (Ricklefs and Schluter 1993). Thus, conclusions derived as to general patterns of arid-land species diversity may differ depending on the taxon, geographical area, scale (Rosenzweig 1995), and the particular gradient examined. However, in most cases, species richness was not constant across the aridity gradient. In some systems, species richness declines with aridity (e.g., birds), whereas in others, diversity increases (e.g., scorpions). Important Characteristics of Deserts that Affect Diversity 1. Low productivity. By definition drylands differ from other biomes in having low annual precipitation (100–250 mm/year) or net negative evapotranspiration (Polis 1991a). Moreover, desert soils are often nutrient-poor (Anderson and Polis 1999). Low water and nutrient availability translates into very low annual net primary productivity (ANPP averages 5–200 g/ m2/yr) (Lieth 1978).

124


2. Temporal variability. In addition to being unproductive on average, deserts are also the most variable biome in temperature, precipitation, and productivity (Polis 1991a). Annual variation in precipitation is inversely related to mean precipitation, so interannual variation (unpredictability) in rainfall is higher in arid regions than in biomes with higher annual precipitation (Polis and Yamashita 1991, Polis et al. in press). Annual rainfall may vary by 1–2 orders of magnitude in deserts (e.g., southern California: 34mm–301 mm, Polis and Farley 1980; the Namib: 2.2–134 mm, Seely and Louw 1980; Baja California 0–10 mm in dry years to 150–280 mm in El Ninõ years, Polis et al. 1997a). Productivity mirrors precipitation during these periods (Seely and Louw 1980, Noy-Meir 1981). For example, the standing biomass in the Namib increased 600–900% (plants and animals respectively) after heavy rains following a 13-year dry period (Seely and Louw 1980). Above-ground ANPP varied by about 20 times between years in the Chihuahuan desert (Ludwig 1986) and 5–43 times on desert islands in the Gulf of California (Polis et al. 1998). Great variation in rainfall drives variation in productivity (Ludwig 1986, 1987) and community structure (see below). Extreme precipitation events (e.g., due to El Ninõ events) may exert more influence on community composition and ecosystem patterns and processes than do ‘‘mean’’ conditions (Noy-Meir 1973, 1974, Ludwig 1986, Hobbs and Mooney 1995). Longer term changes in precipitation (at scales from 10 to 106 years) also occur. For example, American rain-shadow deserts were produced from more mesic areas during orogenous events that built the Andes and Sierra Nevada Mountains. Large parts of the southwestern deserts of North America were under great lakes as little as 10,000–12,000 years ago (Benson and Thompson 1987). Temporal heterogeneity at geological scales could exert strong historical influences. Desert biota probably experience relatively rapid change in patterns of distribution and abundance during wet–dry cycles. We posit that low mean productivity in deserts may be less important to diversity and species interactions than is high variance in productivity. That is, we speculate that chronic low productivity is more easily accommodated by evolutionary adaptation than is unpredictability in production (Polis 1991a, Wiens 1991). Temporal differences in productivity occur at many different scales and significantly influence communities (Wiens et al. 1986, Giller and Gee 1987, Kotler and Brown 1988, Polis 1991a, Polis et al. 1996). Attempting to elucidate the consequences of temporal variation is one of the most challenging, least resolved, but most important tasks of community ecology. 3. Spatial heterogeneity. Deserts have also been postulated to be more spatially heterogeneous than more mesic environments (Crawford 1986). However, unlike temporal heterogeneity, we are not aware of quantitative comparisons of spatial heterogeneity of deserts with other environments except for information on foliage height diversity (Cody 1974). Comprehensive assessments of spatial diversity that include vegetation,


edaphic, and geological characteristics would be a difficult, but important, endeavor. There seem to be two main lines of reasoning underlying this generalization, both incorporating feedback processes that reinforce heterogeneity over time. First, the paucity of plant cover implies that soil is not well developed. Heterogeneity in the physical nature of the substrate (e.g., rock outcrops, gravel beds, coarse sand, or fine dust) is thus exposed at the surface (rather than buried under soil horizons that may be relatively more uniform in texture, as in more mesic biomes). Surface substrata may differ drastically in water infiltration properties and nutrient storage, and therefore in their potential for plant growth and animal activities. This heterogeneity is reinforced in a positive feedback because of the patterns of run-off of water after rains (Noy-Meir 1981, Yair and Shachak 1987). Run-off sources (e.g., smooth rock, bare soil with a crust) contribute incident water to patches with higher infiltration rate, further increasing soil development in run-on sites (Shachak et al. 1998). Second, the sparse plant cover in drylands causes strong contrasts between local areas where plants occur, and areas where they do not. The ‘‘islands of fertility’’ around individual plants are due to multiple processes that lead to long-term improvements in soil conditions under plants. By forming wind barriers, plants collect finer soil particles and organic debris and add their own litter, thereby increasing nutrient availability. The presence of relatively impermeable plant crusts (lichens, blue-green algae) on bare soil between shrubs provides further horizontal redistribution of water and other materials; crusts lead to run-off, and shrubs collect run-on water, debris, and seeds. These processes increase water infiltration, which further increases microbial activity and nutrient availability. Besides soil modification, the presence of adult plants modifies animal activities, for example, by concentrating burrows or animal urine and feces under plants. These activities again reinforce the initial heterogeneity that facilitated the plant’s presence. Thus, ‘‘hot spots’’ of relatively high primary productivity and biological activity juxtaposed to very unproductive interplant areas characterize deserts. Enriched soils and associated plants, in turn, often cascade to exert strong facilitative effects on other organisms. Two general positive effects arise from plants—a trophic effect from their productivity and a refuge effect from their presence in the physical environment. First, plant production travels up the food web to affect secondary productivity and the distribution and abundance of heterotrophs (Noy-Meir 1985, Polis 1999). For example, some desert shrubs support a high diversity of herbivorous insects (Wisdom 1991); in turn, invertebrate predators are attracted to these areas: 26 spider species occur on creosote (Larrea divaricata) and more than 25 species on saltbush (Atriplex canescens) (Polis and Yamashita 1991). Second, many small animals use plants as refuges from both harsh abiotic conditions and from larger predators. For example, Ayal and Merkl (1994) found that large tenebrionid beetle species were most abundant in habitats with high shrub cover, where they were protected from predatory birds (Groner and Ayal 2001). Ayal et al.

126


(chapter 2) suggest that this pattern may hold for most detritivores and small herbivores in drylands, and that the structural role of plant cover may outweigh the trophic effect of plants in determining dryland food webs. Kotler (1984) has documented the role of shrubs as patches of refuge from predators. Other biotic processes likewise increase spatial heterogeneity. For example, many desert plants secrete salt and increase soil salinity around their canopies (Liphschitz and Waisel 1982). Animal activity such as burrowing (ants, termites, isopods) change soil properties by increasing water absorption, and adding nutrients through secretion (Crawford 1986). Abandoned burrows serve as refuges for small animals (Heth 1991). Termites increase local decomposition rate by burying dry plant material in their burrows, whereas ants concentrate seeds of their food plants in their nests (Whitford 1991). The crested porcupine is very effective in increasing spatial heterogeneity in the Negev Desert in Israel as a result of digging activity while feeding on underground storage organs of plants. Its diggings trap both water and organic material, creating better germination sites for many species (Alkon and Olsvig-Whittaker 1989, Boeken et al. 1995). We are uncertain if these biotic facilitations are more or less important in deserts, compared with more mesic systems. However, it is important to note that some types of spatial heterogeneity are less prominent in deserts. In particular, foliage height diversity in deserts is low. Consequently, whatever effects such architecture exerts on community characteristics (e.g., increasing bird diversity by providing more spatial niches; Cody 1974) must be less important in deserts. While we have postulated that spatial heterogeneity in soil and geological properties is likely greater in drylands than in more mesic closed-cover plant communities, edaphic conditions and sometimes plant composition in these other communities can also be highly heterogeneous. Nevertheless, we argue that the contrast between sites with no plants at all and some plants—as is found in arid systems—is more dramatic and consequential for the lives of other organisms. As discussed in much of the rest of this chapter, this hypothesized higher spatial heterogeneity in more arid environments could have many important consequences for species interactions, population dynamics, and community structure (chapter 12). With the advent of remote sensing, geostatistics, and other techniques of spatial analysis, comparable data on spatial heterogeneity is beginning to accumulate. For example, geostatistics have been used to describe the magnitude and scale of spatial heterogeneity in old fields (Robertson et al. 1988) and in agricultural fields (Robertson et al. 1993). As more such data are published, it will be possible to test with rigor the hypothesis that spatial heterogeneity is greater in arid environments. Consequences of Temporal and Spatial Heterogeneity One key aspect of heterogeneity is the temporally unpredictable ‘‘feast or famine’’ nature of primary productivity and food availability in deserts


(Polis 1991a, Polis et al. 1998). Under ‘‘bad’’ conditions, precipitation is low or nonexistent and plants grow little, if at all. In ‘‘good’’ periods of adequate to heavy rains, a relatively luxuriant plant growth may occur. Dramatic changes in productivity stimulated Noy-Meir (1973, 1974) to propose his ‘‘pulse-reserve hypothesis’’ as a paradigm for arid areas. Noy-Meir argues that plants and animals grow and establish reserves (e.g., seeds, tubers, tissue, eggs) during good times; these reserves then maintain the population or individual during interim dry periods. Several basic features of desert organisms may have evolved in response to unpredictable productivity; for example, life history strategies (Polis and Farley 1980, Louw and Seely 1982, Polis 1991a) and opportunistic diets (Noy-Meir 1974, Brown 1986, Polis 1991c). Differences in life history and trophic opportunism exert great influence on interactions such as competition and predation. For example, opportunistic animals exhibit quick functional responses to prey eruptions/fluctuations but probably cannot tightly regulate particular prey species. Crawford (1986) notes that spatial and temporal patchiness of nutritional reserves in deserts, combined with stochastic arrival of moisture, limit the accuracy of predictions about foraging and the impact of consumers. Spatial patchiness, temporal variation, and aridity clines create a mix of good and bad periods and habitats. Such variation has several consequences at the community level. Perhaps most important, desert systems should be rather dynamic, nonequilibrium communities. ‘‘Hide and seek’’ dynamics in heterogeneous environments allow for local extinctions followed by recolonization (Taylor 1988). Extinctions may be caused by physical disturbances, competitive exclusion, or mortality from predators or pathogens. For example, areas of relatively high productivity (e.g., run-off areas) can serve as refugia in times of severe drought (Noy-Meir 1981). Heterogeneity allows persistence and coexistence among species that are engaged in interactions that could otherwise lead to exclusion (Caswell 1978, Taylor 1988). Resource limitation and competition may only occur during periods of ‘‘ecological crunches’’ (Wiens 1977, 1991). Periods between crunches are marked by little or no competition and relaxed selection. Inferior competitors or prey can escape elimination by being distributed in periods or places that are enemy free. Heterogeneity spreads the risk of extinction and increases population persistence by decreasing overall susceptibility to various mortality factors. These processes, permitted by heterogeneity, promote diversity. Heterogeneity in production between (micro) habitats may also strengthen demographic and trophic interconnections among habitats. High interpatch variability in productivity likely partitions a species’ population into individuals that live in ‘‘source’’ and ‘‘sink’’ habitats (Pulliam 1988). Source habitats produce a net surplus of individuals. Sink habitats are suboptimal areas where populations are not self-sustaining, but persist due to immigration from sources. Similarly, we expect that marked spatial heterogeneity in productivity creates a landscape where consumer–resource and food web interactions in relatively unproductive habitats are strongly affected by subsidies

128


from productive patches (Polis et al. 1997b). In general, nutrients, detritus, prey, and predators should move primarily from productive to less productive habitats. We speculate that strong demographic and trophic spatial interconnections might occur commonly in the patchy environment of deserts. Finally, heterogeneity may slow the speed of evolution and the likelihood that species tightly coevolve. The strength of selection is not constant and gene flow may disrupt locally adaptive changes in gene frequency. A tendency toward trophic generalization suggests that interactions may be relatively less tightly coevolved, leading to the organization of exploiter–victim systems or guilds of potential competitors that are more loosely structured, relative to more homogeneous environments.

Processes Affecting Species Diversity The Relationship Between Regional and Local Diversity Viewed over sufficiently long time scales, all local ecological communities arise via colonization from larger regional and biogeographical species pools, filtered by matches between species autecological requirements and local environments, and interspecific interactions such as predation and mutualism (Zobel 1992, Holt 1993, Belyea and Lancaster 1999). Patterns of species diversity, and in particular the relationship of species richness to local environmental factors such as primary productivity or disturbance rates, ultimately reflect the interplay of multiple factors operating at many spatial and temporal scales (Huston 1999). At a local scale, interspecific interactions such as competition, predation, and mutualism (which directly or indirectly require contacts between individuals) can either enhance or depress species richness. At broad spatial scales, speciation, species migration, and regional extinctions influence the size of the species pool available for colonization into a local focal community (Latham and Ricklefs 1993). For example, the great diversity of tenebrionid beetles in the Namib Desert may be attributed to extensive in situ speciation that occurred in this, the oldest of all extant deserts (Seely 1991). Alternatively, younger or isolated deserts may be depauperate in species diversity because regional pools of desert dwellers may be low or there has been little time for speciation. Species richness at local sites can be viewed as arising from a series of filters relating species pools at large spatial scales to local communities (Zobel 1992). The first stage in explaining diversity at a particular site is simply the size of the regional species pool. All else equal, sites with larger regional pools are likely to have larger local pools—this ‘‘filter’’ is thus a result of biogeographic and evolutionary (speciation and extinction) processes. These processes are important determinants of local diversity (Ricklefs and Schluter 1993), but beyond the scope of this chapter. Instead, we focus on the filters between a given regional species pool and the local community, including abiotic limitations, facilitation, competition, and consumer–resource interactions.


Cornell and Lawton (1992) argued that if local interactions such as competition are important filters limiting membership in local communities, then one should observe a saturating rather than linear relationship on plots relating local to regional richness. Therefore, the observation that linear relationships seem to be the most common result (e.g., Caley and Schluter 1997) suggests that interspecific interactions may not be prime determinants of patterns in local species richness. This conclusion is controversial (Huston 1999), and it is not our goal here to address the various interpretations of these relationships. Rather, we emphasize that most theory and empirical research on species interactions in community ecology has focused on the role of species interactions in explaining patterns within a region with the same species pool, that is, the scatter around the lines, rather than with the shape of the lines themselves. Thus, overall linear relationships between local and regional richness across different regions does not necessarily imply that interactions have no effect on relative richness among sites within a region. Instead, interactions could still strongly influence the extent to which local richness deviates (positively or negatively) from the average expected for a given size of regional pool. It is useful to start with a null hypothesis, which is that local species interactions do not constrain (or facilitate) community membership; instead, abiotic conditions act as the primary filter. We then go on to explore the roles of positive interspecific interactions, competitive interactions, and consumer– resource interactions in determining community membership. Abiotic and Demographic Filters If one views the regional species pool as analogous to a ‘‘continent,’’ and the focal community as an ‘‘island,’’ the classical dynamic model of island biogeography (MacArthur and Wilson 1967) schematically portrays the interplay of regional and local processes in determining species richness and composition. This classical model views local communities as a balance between colonization of new species from the pool, and extinction of resident species. If species are not interacting, by definition the rate of colonization and extinction for any given species should be independent of local community composition. Factors that enhance colonization or lower extinction rates lead to higher local species richness, whereas factors that diminish colonization or aggravate extinction should push down species richness, relative to the size of the species pool. The first filter between the regional pool and the local community is the match between a species’ basic niche requirements and local site characteristics. In deserts, compared with more moderate biomes, is there a stronger abiotic filter because of harsh conditions, low resources, or greater temporal heterogeneity, resulting in lower diversity? If so, this could result in either higher extinction rates or lower colonization rates (depending on the life stage at which the lack of match of requirements and site characteristics occurs) and xeric sites having fewer and mesic sites more species than the average

130


expected, given a certain size regional pool. While this idea of a stronger abiotic filter in extreme environments is old and widespread (e.g., Darwin 1859 for extreme deserts, see Parsons 1996 for more recent discussion), it has also been highly controversial because stronger abiotic filters in more stressful environments are often taken to mean weaker impacts of competition and/or facilitation on community structure. However, this is not necessarily the case. The controversy may result from a confusion between the factors that determine whether a species is present/absent in a habitat (‘‘distribution’’) versus the factors that determine individual fitness and population dynamics of species (‘‘abundance’’) that can persist in the habitat. For example, it is possible that more species are eliminated from xeric than from mesic sites by physiological intolerance of prevailing conditions (i.e., that abiotic filters are stronger) but those species that are able to persist compete just as intensely among themselves, as do species that cohabit in mesic conditions. Most existing comparisons of effects of species interactions along favorability or stress gradients test the ‘‘abundance’’ effect rather than the ‘‘distribution’’ effect (see below for review) because they compare competitive intensities among species that naturally co-occur within a habitat, rather than effects of competition on species that do not naturally persist in the habitat. To rigorously test the hypothesis that abiotic filters are stronger in more xeric environments, one would have to transplant to a habitat species that normally do not occur there and monitor their demography in the absence of potential competitors/facilitators/predators. In addition to complete physiological intolerance, abiotic filters potentially involve several other distinct demographic mechanisms that could control colonization and extinction rates and thus local richness for a given regional pool (table 8.1). First, because deserts have low productivity, many species are expected to be low in abundance, and/or have low maximal growth rates. All else being equal, these demographic factors tend to lower colonization rates into local communities, and enhance local extinction rates. Consider first the effect of low population size. Because of demographic stochasticity, populations with small absolute sizes experience high extinction risk even in favorable environments (Belovsky 1987). If a species has low average abundance in occupied sites, this both reduces the flux of dispersers available for colonizing empty sites, and weakens the ‘‘rescue effect’’ (Brown and Kodric-Brown 1979) in occupied sites. Now consider the effect of low average growth rates. Episodic disturbances can drive species to troughs of low abundance, during which they risk extinction due to demographic stochasticity. Low growth rates lengthen the time period of increased extinction risk. Moreover, chronic mortality factors are more likely to drive species extinct if there is weak compensation with increased growth rates at low densities (e.g., Holt 1985). As noted above, compared with many other biomes, deserts have greater coefficients of variation in production, as well as low overall levels of production. To be present in a local community there obviously has to be a match between a species’ autecological requirements (i.e., its fundamental niche) and

Unified Framework I 131 Table 8.1 Summary of potential effects of low productivity and high temporal and spatial heterogeneity on colonization (C) and extinction (E) rates Characteristic of Drylands Low productivity

High temporal variability

High spatial variability

Processes and Phenomena Affected Low abundance ! increased importance of demographic stochasticity Low abundance ! fewer dispersing propagules Low population growth rate ! low rate of recovery from disturbance

Effect on Demography "E "E #C

Populations at risk under rare ‘‘crunch’’ conditions

"E

Selection for high dispersal rates Selection for demographic buffering mechanisms (e.g., seed bank)

" C** # E**

Habitat specialists have fewer suitable patches ! more at risk in metapopulation scenario Greater range of environmental conditions ! more sink populations More potential for habitat selection

"E "E # E**

Note: Increased extinction (" E) and/or decreased colonization (# C) rates would lead to lower species richness in drylands relative to more mesic conditions. **Effects in the opposite direction, resulting in higher diversity in drylands.

the local environment. In highly variable environments such as deserts, the match between these may be ephemeral, leading to transient and spatially variable matches between niche requirements and local conditions. This could contribute to a high local extinction rate. Even for species that are generally adapted to dryland environments, because of temporal heterogeneity (e.g., occasional exceptionally long bouts between rainfall pulses) there may be enhanced extinction risks. For instance, for an annual plant species without a seed bank, its long-term growth rate is the geometric mean fitness across long time-series of environmental fluctuations. A single year of very low fitness can doom the plant to local extinction, even in an overall favorable environment. Finally, as also noted above, for some organisms desert environments may be very patchy. Metapopulation models (see Holt 1997) suggest that habitat specialists may have trouble persisting on scarce habitats, unless they have very low extinction rates, or high colonization rates. If desert environments overall are experienced as highly patchy, this effect could lower local species richness. Given these demographic considerations alone, one expects a lowering of local species richness, because of increased local extinction and decreased local colonization rates, so that deserts would have lowered slopes in regional–local species richness plots. However, a number of factors may mitigate these demographic effects (table 8.1). For instance, the colonization rate of

132


empty sites does not vary simply with the abundance of a species in those sites it occupies. Colonization rate is likely to increase with an increasing percapita propensity to disperse, and with an increasing probability of successfully traversing unsuitable landscapes separating habitable sites. As noted earlier, temporal variability that is often weakly correlated in space is a hallmark of desert ecosystems. Evolutionary theories of dispersal suggest that such patterns of spatiotemporal variability tend to favor increased per-capita rates of dispersal (Holt and Barfield 2001, Ronce et al. 2000). If species in desert systems are characteristically strong dispersers, this should tend to enrich local communities by increasing colonization rates. (Many deserts are famous for nomadic behaviors in consumers.) Because plant biomass is low, passive dispersal propagules carried by the wind may also travel much farther before coming to rest due to physical obstruction. We are unaware of studies that specifically quantify dispersal rates of desert species, compared with other biomes. Furthermore, many adaptations to desert environments obviously mitigate local extinction risks. For plants, such adaptations include seed banks for annuals, and resource storage structures for perennials. Habitat selection by consumers can permit microhabitats of high or stable fitness to be sought out, reducing the risk of local population extinction. Generalist consumers can benefit from their potential use of a wide range of independently varying resources, which buffer extinction risk. The ideas sketched above suggest that the relationship between local and regional species richness in desert biomes is influenced by many factors, some of which depress diversity, and others which may enhance diversity. The net effect of these factors is unclear. Needed are long-term population studies in deserts that will permit a comparison of local extinction rates with other biomes. While we have described these demographic processes under the rubric of abiotic filters, because they could operate in the absence of interspecific interactions, they all could be moderated or intensified by interspecific interactions, as described in the next three sections.

Positive Interactions Traditional community theory has emphasized how local interspecific interactions make species persistence and coexistence less likely, thus limiting local species richness. In recent years there has been increasing concern with the potential for positive interspecific interactions to influence communities (Bertness and Callaway 1994). Such positive interactions could lead to increasing colonization or decreasing extinction rates with increasing number of species. Can ‘‘diversity beget diversity,’’ and if so, is this process more important in drylands than other biomes? No general theory exists on this theme, so we here outline some initial tentative steps toward a synthesis of facilitation in desert systems.


Examples of positive interactions in drylands abound (see reviews in Callaway 1995). Here we attempt a rough classification of the many distinct mechanisms that have been described to provide a basis for generalizing about their importance in drylands relative to more mesic environments. As with any classification, the categories are not entirely discrete and our assignments are often ambiguous. Nevertheless, it provides a heuristic value in allowing hypotheses to be developed about classes of mechanisms. The most general distinction is between trophic interactions (those mediated entirely through effects on consumable resources) and nontrophic interactions (not mediated through effects on consumable resources). Nontrophic positive interactions are very similar in definition to ecosystem engineering as defined by Jones et al. (1994, 1997), although they emphasize that engineering can have negative influences as well on some species. We prefer the terminology of nontrophic interactions to emphasize the contrast, and allow further parallel classifications within trophic and nontrophic types of interactions. Within each of these categories, we can further divide mechanisms of positive interactions into direct (no intermediary involved) and indirect (either abundance- or traitmediated) interactions. Here we focus on two of the possible types of positive interactions: those in which organisms facilitate other organisms in ways other than providing food (nontrophic facilitations, such as habitat amelioration and habitat creation); and those in which organisms indirectly facilitate other organisms by reducing the probability of their becoming food through predation by a third group of organisms (trophic indirect facilitation). Below, we give examples of each of these and review ideas and evidence about how they are expected to vary in xeric relative to mesic environments. The remaining category, trophic direct facilitations, is simply the consumer side of consumer–resource interactions (i.e., the plus side of a þ= interaction). For nontrophic direct facilitations, the best-documented class of examples in drylands is probably habitat amelioration, defined broadly as when organisms improve the physical habitat in terms of resource availability and/or amelioration of stress in some way that benefits other organisms. In plants, this is often more narrowly referred to as ‘‘nursing’’ or the ‘‘nurse plant phenomenon’’ because the facilitator is often an adult, while the facilitatee is often a juvenile (see Callaway 1995, Callaway and Walker 1997 for comprehensive reviews). For example, larger individuals of sessile species can reduce temperatures due to their shading effect—reducing direct negative effects of high temperature and increasing water availability and decreasing water demand (reviewed in Holmgren et al. 1997). Another example is the ‘‘island of fertility,’’ where isolated plants trap organic matter and gradually increase local nutrient supply for other, usually smaller, plants (e.g., GarciaMoya and McKell 1970). Both of these processes in plants can benefit animals as well. Another class of examples of nontrophic direct facilitation is habitat creation, or the creation of new structures by organisms (allogenic ecosystem engineering, sensu Jones et al. 1994). These are also well-documented in drylands, for example, increased bird diversity with increased plant structural diversity (Cody 1993) and the favorable germination sites in pits

134


created by porcupine diggings (Gutterman 1982). Nontrophic indirect facilitations would be more complex examples in which these effects propagate through links of other species. For example, a shrub that casts shade might facilitate burrowing activities by rodents, in turn modifying growth conditions for seedlings of other plant species. Facilitation through habitat amelioration has been argued to be more common in less productive environments (Bertness and Callaway 1994). In demographic terms, that means the presence of organisms should decrease extinction rates or increase colonization rates more strongly in less productive environments. We suggest that facilitation through microhabitat creation is also more common in less productive habitats and that the reason is similar for both mechanisms: in an extremely unfavorable habitat or a site with no suitable habitat at all, even a small improvement in absolute terms makes a large percentage improvement (Bertness and Callaway 1994). A specific model developing this basic argument was presented for nurse plants in drylands by Holmgren et al. (1997). However, Jones et al. (1997) have argued that habitat creation may be more important in environments with extensive plant cover, abundant large animals, and dominant organisms with massive and persistent structures—characteristics of more, rather than less, productive ecosystems. Although previous workers have not always clearly separated effects of habitat creation and habitat amelioration, narrative reviews of facilitation support the idea that nontrophic direct facilitative effects are more common in unproductive environments in plants (Callaway and Walker 1997, Holmgren et al. 1997). The simplest example of direct facilitation in trophic interactions is for a consumer to have a ‘‘donor-controlled’’ relation with its resource population (DeAngelis 1992). This generates the potential for asymmetrical facilitation, given that the consumer clearly requires the resource population to persist, but without reciprocal dependency. This diversity-generating mechanism is usually thought of as operating through chains of specialist consumers, i.e., a greater diversity of resources can support a greater diversity of consumers while keeping degree of resource overlap constant. However, it may be more likely to operate through generalist consumers because of bet-hedging or portfolio effects; if each of various resource populations responds in a species-specific fashion to environmental variability, a generalist consumer may be able to survive in situations where a specialist on any given resource risks extinction. Ritchie (1999) recently observed this effect of trophic generalization on extinction risk for a dryland herbivore (prairie dogs). The higher risk of extinction of trophic specialists should be exaggerated, given chains of trophic specialization (e.g., a specialist herbivore on a single plant species, supporting a specialist parasitoid, supporting in turn a specialist hyperparasitoid; see Holt et al. 1999). The bet-hedging advantage of trophic generalization should be especially important when resource populations are low in average abundance and strongly variable through time. Thus, we expect specialist consumers to be less common in deserts than in other biomes, particularly at higher trophic levels. Diversity at low trophic ranks can


facilitate diversity at higher trophic ranks via the buffering effects of trophic generalization. This advantage of trophic generalization in turn implies that desert food webs are likely to be highly reticulate. There may also be occasional positive top-down effects upon species richness at low trophic levels. For instance, in situations where competitive exclusion might occur at low trophic ranks (e.g., strong resource competition among rodent species), the scarcity of cover in drylands (chapter 2) could magnify the effect of predators, which in turn can facilitate coexistence (if the more efficient competitor is also more vulnerable to predation, e.g., Kotler 1984). Trophic indirect facilitations include both predator- and competitormediated interactions and both trait- and abundance-mediated indirect interactions. In drylands, one documented mechanism is a trait-mediated indirect effect on predator–prey interactions, whereby one group of organisms provides refuges and therefore reduces predation rates on the target organisms. (This can also be regarded as a form of interference, i.e., nontrophic competition from the point of view of the predator). For example, shrubs in drylands often provide refuges from natural enemies for other plants (e.g., Fuentes et al. 1986), for invertebrate prey species (Ayal and Merkl 1994, Groner and Ayal 2001), and for small mammals (Kotler et al. 1993). Two general hypotheses have been proposed about the importance of indirect facilitation involving refuges. Bertness and Callaway (1994) propose that this type of facilitation is most important at high productivity, consistent with Connell’s (1975) much earlier argument that natural enemies are more important at high productivity than at low or intermediate productivity. In contrast, Ayal et al. (chapter 2) have proposed that refuges created by vegetation for herbivores or intermediate predators are much more important at low productivity. A recent quantitative review (meta-analysis) of the results of field removal experiments in plants (Goldberg et al. 1999) compared the frequency and magnitude of positive interactions along productivity gradients. Because it was impossible to extract information on mechanisms of facilitation from most of the studies, the review tested the total summed effects of habitat amelioration, habitat creation, and provision of refuges, even though these different mechanisms are predicted to change in different ways with productivity. Using the same database as Goldberg et al. (1999), we found a higher frequency of facilitative effects on plants in unproductive environments, consistent with the Bertness and Callaway model for habitat amelioration, but only for growth as the measure of plant response (fig. 8.1, top). In addition, despite this more frequent occurrence of positive interactions, the actual magnitude of interactions (degree to which plants were facilitated or inhibited) did not differ significantly for growth responses to neighbors (fig. 8.1, bottom). This suggests that the competitive interactions that do occur at low productivity tend to be stronger than those at higher productivity, that is, that both extremely strong facilitation and extremely strong competition are more likely in unproductive environments. While this is consistent with the higher

136


Figure 8.1 A meta-analysis of the effects of neighbors on growth and survival of plants in low versus high productivity environments, using the database from the summary of published experiments reported in Goldberg et al. (1999). Only herbaceous plants with herbaceous neighbors were included. Productivity was estimated by standing crop with values 350 g/m2 as ‘‘high productivity.’’ Interaction intensity was quantified as ln(performance with neighbors/performance without neighbors); values >0 indicate facilitation and Ri and to ensure that the consumption rate of exclusive resources kr(Pk Pj ÞðRi Rj ) is greater than resource loss rate. (From Ritchie and Olff (1999).)

smaller species occupy proportionately less total volume than larger, resource-poor patches used by larger species. The functional equation for size ratios (eq. (5)) dictates the number of species ranging in size from wmin to wmax that can be ‘‘packed’’ into an environment. The maximum size (wmax ) is determined by whether there is at least one suitable patch of size P and resource concentration R in a finite space of extent x. The number of suitable patches is found by dividing the total volume of resources rxQ by the resource volume contained within suitable patches (P R ). Since P and R depend on w, however, the actual number of patches in the finite space is weighted by the probability of the

214


Figure 12.4 (a) Predicted relationship between the size ratio (larger/smaller) of pairs of adjacent-sized species and ln(size) of the larger species of each pair. Observed relationships for (b) granivorous ants and (c) granivorous vertebrates (birds and mammals) at the Cave Creek Bajada site in the Chihuahuan desert of Arizona, and (d) vascular plants at Deseret Land and Livestock in the Great Basin. Curves in (b)–(d) are fits (showing R2 ) of the data to the nonlinear function ð1 þ b0 wb1 Þb2 using least-squares nonlinear regression.

occurrence of a patch with the length w specified by P and R . This probability is (F 1)wF (Milne 1992, Ritchie 1998). Therefore, (F 1ÞwF rxQ / (P R Þ ¼ 1. Substituting for P and R (eq. (1)) and solving for w yields 1=D wmax ¼ ðF 1ÞxQ krm=L

ð6Þ

Likewise a minimum size is set by whether at least one patch of sufficient resource concentration occurs. This is found by dividing the expected resource concentration within food rxQF by the resource concentration contained within suitable patches (R ). Since R scales with w, the actual number of patches in the finite space of extent x must be weighted by the probability of the occurrence of resources within a patch of length w. This probability is (Q 1)wQ (Milne 1992, Ritchie 1998). Therefore, (Q 1ÞwQ rxQF = ðR Þ ¼ 1. Substituting for R (eq. (4)) and solving for w yields 1=ð3Q=2DÞ wmin ¼ ðQ 1ÞxQF krm=L

ð7Þ

Resource Partitioning and Biodiversity in Fractal Environments

215

Equations (6) and (7) suggest that landscapes with larger extents support both larger and smaller species because this increases the probability of patches that are sufficiently large or resource-rich. This prediction matches patterns observed for mammals, birds, and fish on islands of different size, including mammals on Sonoran desert islands in the Gulf of California (Marquet and Taper 1998). Species richness (S) is then the number of exclusive size niches allowed between wmin and wmax , which yields the solution S Y i¼1

gðwi Þ ¼

wmax wmin

ð8Þ

An approximate solution for S is S ffi lnðwmax Þ=½lnðgðwmax ÞÞ lnðwmin Þ

ð9Þ

The equation for S predicts a left-skewed, unimodal distribution of species richness versus organism size (fig. 12.5a). This distribution reflects the larger size ratios and thus looser species packing required for smaller species (eq. (5))

Figure 12.5 Left-skewed frequency distribution of species in different body size classes (a) predicted by the scaling model, and observed patterns for (b) Chihuahuan desert vertebrate granivores, and (c) Great Basin vascular plants. There were insufficient species to construct a reliable size distribution for ants.

216


and the limitation of the largest species by the maximum patch size in the environment and the smallest species by the maximum resource concentration. The functions implicit in wmax (eq. (3)) and gðwmax ) mean that the model also predicts the effects on species richness of sampling area (x2 ), habitat fragmentation (D), and the amount and distribution of food and resources (m; r; F; Q). However, we will not explore these predictions here.

Community Structure and Diversity in Dry Environments Do these theoretical predictions apply to species assemblages in drylands? We tested the model predictions for three guilds of species that use similar resources in North American deserts. These included two guilds, granivorous ants and vertebrates, at Cave Creek Bajada in the Chihuahuan desert (Brown et al. 1979, Davidson et al. 1985) and vascular plants at Deseret Ranch in the Great Basin desert (Ritchie et al. 1994). We tested for both declining size ratios with increasing body size and right-skewed distributions of diversity versus body size. We used body length as a measure of size for ants and vertebrates, which all consume seeds of various annual and perennial plants at the soil surface. We measured mean crown width of 10 randomly selected individual plants of each of 20 common species growing in a sagebrush–rabbitbrush–western wheatgrass community (Artemisia tridentata, Chrysothamnus viscidiflorus, Pascopyrum smithii). We then examined size patterns for species growing within a single 6.25 m2 plot. Instead of being constant (Hutchinson 1959, Brown 1981), size ratios in these very different assemblages declined significantly with increasing size (fig. 12.4b,c), and the relationships fit the shape predicted by the model (eq. (5), fig. 12.4a). These patterns are consistent with the hypothesis that these granivore communities are structured by resource partitioning to avoid competition, as has been shown in numerous field experiments (Brown et al. 1979, Davidson et al. 1980, 1985). The size differences among species may lead them to use different-sized seeds or seed patches (Brown et al. 1979), thus enabling coexistence despite considerable diet overlap. The species richness–size distributions of both the vertebrate granivores in the Chihuahuan desert and the plants from the Great Basin desert are both significantly left-skewed (fig. 12.5b,c). At small scales, close species packing of similar-sized species may help explain the clusters of similar-sized birds and heteromyid rodent species and clusters of harvester ant species that occur in the Chihuahuan desert (Brown et al. 1979, Davidson et al. 1980, 1985). These distributions differ from the log-normal or right-skewed distributions most commonly reported (Hutchinson and MacArthur 1959, Morse et al. 1985, May 1988, Brown et al. 1993, Siemann et al. 1996). Virtually all these previously published distributions combine diversity–size distributions of separate guilds (e.g., nectarivores, granivores, herbivores, carnivores) or species that use similar resources but different habitats. Our model may not apply to such communities that include species that use different resources or


217

habitats. In addition, communities assembled through different mechanisms (e.g., colonization limitation) may show different patterns (Hubbell 1997). This discrepancy in patterns suggests that diversity–size patterns may be scale-dependent (Brown and Nicoletto 1991). Patterns at smaller scales may be driven by competition, predation, and resource partitioning, while patterns at larger spatial extents may be driven by geographical isolation and speciation coupled with colonization limitation. The application of spatial scaling laws suggests that a theory of biodiversity may emerge from first principles of how organisms find resources in space. The analysis formalizes earlier ideas that diversity depends on the number of spatial niches (Hutchinson 1959, MacArthur 1964, Brown 1981, May 1988) and suggests that coexisting species cannot infinitely partition space (Rosenzweig and Abramsky 1994, Rosenzweig 1995). In addition, the model synthesizes recent ideas about how resource acquisition (Wright 1983, Belovsky 1986, 1997, Brown et al. 1993) and spatial characteristics of habitat (MacArthur 1964, Palmer 1992) influence diversity. Clearly, other factors, including diversity of resource types (Tilman 1982, Jones and Lawton 1991, Huisman and Weissing 1999), disturbance (Huston 1994), colonization limitation (Tilman 1997, Hubbell 1997), and biogeographical history (Schluter and Ricklefs 1993, Rosenzweig 1995, Hubbell 1997) are also important in explaining diversity patterns. Nevertheless, the spatial scaling of resource use by species of different body size may explain many species diversity patterns across a range of spatial scales and taxa. Drylands, which have visible and easily measurable heterogeneity in physical habitats and resources at a variety of scales, may be especially good for applying this scaling approach. The well-studied communities of ants and granivorous mammals and birds in southern Arizona fit the model’s prediction well. Plants from a Great Basin desert site also support the model’s predictions. These results suggest that differentiation in size (scale of resolution) may be a major mechanism structuring local desert communities. However, the discrepancy between left-skewed diversity-size distributions at local scales of observation and right-skewed or log-normal distributions at larger scales (Brown and Nicoletto 1991), suggests that other mechanisms may be more important in structuring communities assessed at larger scales of observation. Likely mechanisms at larger scales include geographic isolation and speciation coupled with colonization limitation (Ricklefs and Schluter 1994, Tilman 1997). The scale-dependent approach we introduce makes these scale distinctions explicit and suggests new hypotheses for assessing how species diversity patterns change across spatial scales.

Acknowledgments We thank B.T. Milne, J.H. Brown, H. DeKroon, J.M. Emlen, S.A. Gripne, A. Guss, N. Haddad, W.A. Kunin, L. Li, S. Naeem, and W.C. Pitt, for comments. The U.S. National Science Foundation, Santa Fe Institute, Netherlands NWO, USU Ecology Center, and the Utah Agricultural Experiment Station supported this work.

218


References Belovsky, G.E. 1986. Generalist herbivore foraging and its role in competitive interactions. Am. Zool. 26, 51–69. Belovsky, G.E. 1997. Optimal foraging and community structure: the allometry of herbivore food selection and competition. Evol. Ecol. 11, 641–672. Brown, J.H. 1981. Two decades of homage to Santa Rosalia: toward a general theory of diversity. Am. Zool. 21, 877–888. Brown J.H. 1995. Macroecology. University of Chicago Press, Chicago. Brown, J.H., and P.F. Nicoletto. 1991. Spatial scaling of species composition: body masses of North American land mammals. Am. Nat. 138, 1478–1512. Brown, J.H., D.W. Davidson, and O.J. Reichman. 1979. An experimental study of competition between seed-eating desert rodents and ants. Am. Zool. 19, 1129–1143. Brown, J.H., P. Marquet, and M. Taper. 1993. Evolution of body size: consequences of an energetic definition of fitness. Am. Nat. 142, 573–584. Davidson, D.W., J.H. Brown, and R.S. Inouye. 1980. Competition and the structure of granivore communities. BioScience 30, 233–238. Davidson, D.W., D.A. Samson, and R.S. Inouye. 1985. Granivory in the Chihuahuan desert: interactions within and between trophic levels. Ecology 66, 486–502. Gibbens, R.P., and R.F. Beck. 1988. Changes in grass basal area and forb densities over a 64-year period on grassland types of the Jornada Experimental Range. J. Range Manage. 41, 186–192. Halvorson, J.J., H. Bolton, and J.L. Smith. 1997. The pattern of soil variables related to Artemisia tridentata in a burned shrub-steppe site. Soil Sci Soc. Am. J. 61, 287–294. Holling, C.S. 1992. Cross-scale morphology, geometry, and dynamics of ecosystems. Ecol. Monogr. 62, 447–502. Hubbell, S.P. 1997. A unified theory of biogeography and relative species abundance and its application to tropical rainforests and coral reefs. Coral Reefs 16, S9-S21. Huisman, J., and F.J. Weissing. 1999. Biodiversity of plankton by species oscillations and chaos. Nature 402, 407–410. Huston, M.A. 1994. Biological Diversity: the Coexistence of Species on Changing Landscapes. Cambridge University Press, Cambridge. Hutchinson, G.E. 1959. Homage to Santa Rosalia, or why are there so many kinds of animals? Am. Nat. 93, 145–159. Hutchinson, G.E., and R.H. MacArthur. 1959. A theoretical model of size distributions among species of animals. Am. Nat. 93, 117–125. Jones, C.G., and J.H. Lawton. 1991. Plant chemistry and insect species richness of British umbellifers. J. Anim. Ecol. 60, 767–777. Kareiva, P., and U. Wennergren. 1995. Connecting landscape patterns to ecosystem and population processes. Nature 373, 299–302. Kunin, W.E. 1998. Extrapolating species abundance across spatial scales. Science 281, 1513–1515. Lightfoot, D.C., and W.G. Whitford. 1991. Productivity of creosotebush forage and associated canopy arthropods along a desert roadside. Am. Midl. Nat. 125, 310–322. MacArthur, R.H. 1964. Environmental factors affecting bird species diversity. Am. Nat. 98, 387–397. Mandelbrot, B.B. 1983. The Fractal Geometry of Nature. Freeman, New York.


219

Marquet, P.A., and M.L. Taper. 1998. On size and area: patterns of mammalian body size extremes across landmasses. Evol. Ecol. 12, 127–139. May, R.M. 1988. How many species are there on earth? Science 241, 1441–1449. Milne, B.T. 1992. Spatial aggregation and neutral models in fractal landscapes. Am. Nat. 139, 32–57. Milne, B.T. 1997. Applications of fractal geometry in wildlife biology. In Wildlife and Landscape Ecology: Effects of Pattern and Scale (ed. Bissonette, J.). Springer, New York, pp. 32–69. Morse, D., N.E. Stork, and J.H. Lawton. 1985. Fractal dimension of vegetation and the distribution of arthropod body lengths. Nature 314, 731–732. O’Neill, R.V., B.T. Milne, M.G. Turner, and R.H Gardner. 1988. Resource utilization scales and landscape pattern. Land. Ecol. 2, 63–69. Palmer, M.W. 1992. The coexistence of species in fractal landscapes. Am. Nat. 139, 375–397. Peterson, G., C.R. Allen, and C.S. Holling. 1998. Ecological resilience, biodiversity, and scale. Ecosystems 1, 6–18. Ritchie, M.E. 1998. Scale-dependent foraging and patch choice in fractal environments. Evol. Ecol. 12, 309–330. Ritchie, M.E., and H. Olff. 1999. Spatial scaling laws yield a synthetic theory of diversity. Nature 400, 557–560. Ritchie, M.E., M.L. Wolfe, and R. Danvir. 1994. Predation of artificial sage grouse nests in treated and untreated sagebrush. Great Basin Nat. 54, 122–129. Rosenzweig, M.L. 1995. Species Diversity in Space and Time. Cambridge University Press, Cambridge. Rosenzweig, M.L., and Z. Abramsky. 1994. How are diversity and productivity related? In Species Diversity in Ecological Communities: Historical and Geographical Perspectives (ed. Schluter, D., and R. Ricklefs). University of Chicago Press, Chicago, pp. 52–65. Rundel, P.W. 1996. Ecological Communities and Processes in a Mojave Desert Ecosystem. Cambridge University Press, Cambridge. Schluter, D., and R. Ricklefs. 1993. Species diversity: regional and historical influences. In Species Diversity in Ecological Communities: Historical and Geographical Perspectivs (eds. Ricklefts, R., and D. Schluter). University of Chicago Press, Chicago, pp. 350–364. Schoener, T.W. 1973. Population growth regulated by intraspecific competition for energy or time: some simple representations. Theor. Pop. Biol. 6, 265–307. Siemann, E., D. Tilman, and J. Haarstad. 1996. Insect species diversity, abundance and body size relationships. Nature 380, 704–706. Tilman, D. 1982. Resource Competition and Community Structure. Princeton University Press, Princeton, NJ. Tilman, D. 1997. Community invasibility, recruitment limitation, and grassland biodiversity. Ecology 78, 81–92. Wright, D.H. 1983. Species-energy theory: an extension of species-area theory. Oikos 41, 496–506.

13 Unified Framework II Ecosystem Processes: A Link Between Species and Landscape Diversity Moshe Shachak Robert Waide Peter M. Groffman

T

he discipline of ecology can be subdivided into several subdisciplines, including community, ecosystem, and landscape ecology. While all the subdisciplines are important to the study of biodiversity, there is great variation in the extent to which their contributions have been analyzed. For example, the role of community ecology in biodiversity studies is well established. In community ecology, the entities of study are species that differ in their properties and generate a web of interactions that, in turn, organize the species into a community. Similar to community ecology, the contribution of landscape ecology to biodiversity is apparent. The entities of study, definable ‘‘patches,’’ are tangible. They differ in their properties and generate a web of interactions that organize the patches into a landscape mosaic. In contrast to community and landscape ecology, the role of ecosystem ecology in biodiversity is less apparent. In ecosystem ecology, it often is not clear what the entities are, and how they are organized. To the extent that ecosystem ecology focuses on energy flow and nutrient cycling, we can define fundamental entities as compartments and vectors in models that depict the flows of water, energy, and nutrients through communities. If we apply diversity criteria to these entities, we can use the term ecosystem diversity to refer to the number of compartments and vectors, the differences among them in type and size, and their organization in promoting energy flow or nutrient cycling. To our knowledge, ecosystem scientists have not yet developed criteria for ecosystem diversity similar to those used for species and landscape diversity. There has been some use of the term ‘‘ecosystem diversity’’ to refer to a diversity of ecosystems, implying a variety of habitats, landscapes, or biomes. 220

Unified Framework II

221

As discussed above, we suggest that to define the role of ecosystem ecology in biodiversity studies, the approach should be to study the relationships among species, landscape, and ecosystem diversities (chapters 1 and 13). However, since the concept of ecosystem diversity awaits further development, we adopt a different approach for understanding the role of ecosystem science in biodiversity studies. In this chapter, we examine relationships among ecosystem processes, species diversity, and landscape diversity. Our thesis is that ecosystem processes related to the fluxes of energy, water, soil, and nutrients are important links between species and landscape diversity and are fundamental to the development and maintenance of biodiversity. We start with an example of the web of interactions among ecosystem processes, organisms, and landscape mosaic that control the development of landscape and species diversity in the Negev. We then discuss in a more general way the various ways that ecosystem processes link species and landscape diversity.

Ecosystem Processes and Biodiversity in the Negev To illustrate the relationship between ecosystem processes and biodiversity in drylands, we discuss the role of ecosystem processes in the development of species and landscape diversity in the Negev. Figure 13.1 shows how ecosystem processes related to the fluxes of energy, water, soil, and nutrients are important links between species and landscape diversity and are fundamental to the development and maintenance of biodiversity in the Negev. The first stage in the development of ecological systems in the Negev is the formation of a rocky watershed by geological processes. At this stage, landscape diversity (one patch type rocks) and species diversity (only a few species of algae and lichens) are low. This is because the only substrate available for colonization by organisms is rocks and the main resource for sustaining species diversity, water, leaks out of the system as surface run-off (fig. 13.1 (II)). Over time, species diversity increases as a result of a set of ecosystem processes that increase landscape diversity, that is, the formation of soil patches in the rocky watershed. This process of landscape diversification is controlled by the ecosystem processes of dust deposition, the redistribution of water among landscape patches and the growth of cyanobacteria (Eldridge et al. 2000). The formation of soil patches depends on dust deposition and accumulation (Offer et al. 1997, Shachak and Lovett 1998). Soil accumulation in the Negev is often slow, because the rate of erosion by run-off water is high on bare bedrock. The process of soil accumulation is greatly facilitated by the colonization of cyanobacteria on the soil surface. The cyanobacteria, by secretion of polysaccharides, form a soil crust, which is highly resistant to erosion (Yair and Shachak 1987). As dust accumulates, soil forms and the landscape becomes composed of two patch types, rocks and soils that exist in

222


Figure 13.1 Species diversity, landscape diversity, and ecosystem processes. (I) Ecosystem processes as a link between species and landscape diversity. (II)–(IV) Changes in trajectories, through time, of the relationships among ecosystem processes, species, and landscape diversity in the Negev desert. (See text for explanation.)

a source–sink relationship. The rocky patches (source) generate run-off and the soil patches (sink) absorb the run-off water. This new landscape mosaic opens new opportunities for colonization and establishment of higher plants. High numbers of annual plant species colonize and grow on the soil patches, increasing species diversity. These links between landscape diversity, ecosystem processes, and species diversity are shown in fig. 13.1(III). Over time, the process of changing landscape and organism diversity continues as shrubs colonize the soil patches. Shrubs increase the interception of surface run-off and the accumulation of soil materials in these patches, resulting in the formation of soil mounds (Shachak et al. 1998). As in the case of the formation of soil patches on the rocky substrate, the formation of an additional patch type (soil mounds) by ecosystem processes increases the complexity of source–sink interactions in the landscape. The crusted soil, due to its low infiltration capacity, is a source for run-off


223

water, and associated sediments and nutrients. The soil mound is a sink for water, sediments, and nutrients. The consequence of landscape diversification by soil mound formation is a significant increase in plant species diversity (figs. 13.1(I–IV)). This example shows how an organism (shrub) can influence landscape diversity (forming a soil mound) affecting the ecosystem processes (water, nutrient availability) that control species diversity. Over time, landscape diversification in the Negev continues with the formation of mounds and pits by animals such as porcupines, isopods, and ants, further increasing landscape diversity. This leads to further increases in species diversity (Boeken and Shachak 1994). Our example from the Negev shows how the development of biodiversity is a complex process, involving interactions between two types of entities: organisms and patches. Interactions between the two types of entities are controlled by ecosystem processes. The organisms (cyanobacteria, shrubs, and animals) control the number of patch types and their distribution in space and time. The landscape mosaic controls the ecosystem processes of redistribution of water, soil, and nutrients. The resulting increase in the heterogeneity of resource distribution promotes high species diversity.

Ecosystem Processes and Biodiversity We suggest that there is a set of basic interactions that link species with landscape diversity through ecosystem processes. The link shown in the Negev example occurs when the landscape mosaic affects the flow of resources among patches. This flow creates heterogeneity in the distribution of resources and allows for coexistence of a diversity of species. The first link is landscape–ecosystem–species. The Negev example becomes more complex via two additional linkages involving species effects on ecosystem and landscape processes. The second link (species–ecosystem–species) is driven by the effects of specific species on water, energy, and material fluxes in the ecosystem that lead to increases in species diversity. The third link (species– landscape–ecosystem–species) is driven by the ability of specific species to alter landscape diversity (e.g., ecological engineers), which alters ecosystem processes and ultimately species diversity. Below we elaborate on these links, giving examples from ecosystems other than the Negev, to emphasize the role of ecosystem processes in the production and maintenance of biodiversity. Landscape–Ecosystem–Species The landscape–ecosystem–species link encompasses the redistribution of resources in space by the landscape mosaic and its effect on species diversity. It is rare that resources will remain in a single landscape patch. Rather, ecosystem processes move water, soil, organic matter, and nutrients across the landscape from one patch to another. These processes control the distribution of resources in space and thus are fundamental controllers of species

224


diversity. The creation of landscape and species diversity by ecosystem processes can be characterized by a web of source–sink interactions among landscape patches. Materials and propagules move from one patch and are then absorbed by another. This movement is controlled by patch properties and the vectors (e.g., water, wind, animals) that transport materials across the landscape. While the landscape–ecosystem–species link is obvious in drylands like the Negev, it can also be seen in other ecosystems. Didier and Poudevigne (2000) identified three community types along a landscape gradient in northwestern France. The species diversity of the communities was correlated with physical patchiness of the landscape: calcicolous communities on chalk slopes, mesophilous communities on colluvium, and hydrophilous communities on alluviums. These differences in ecosystem parent material influenced ecosystem processes related to water and nutrient availability, which ultimately controlled species diversity. In forests, landscape diversity is increased by the structural complexity formed by trees. This complexity arises from the presence of trees of varying sizes, conditions, and species, as well as standing dead trees, logs, and woody debris on the forest floor, multiple canopy levels, and canopy gaps. This complexity influences the pattern and rate of ecosystem processes related to nutrient and water availability, which increases the diversity of habitats for a wide range of organisms (Baskent and Glenwood 1996). Wardell and Horwitz (1996) found that the southwestern Australia landscape harbors pockets of habitats that show that fine-scale hydrological patterns, persisting at local level patchiness, support high species diversity. The web of interactions among landscape, ecosystem, and species diversities is illustrated by the long-term research on Blackhawk Island in Wisconsin, USA. Varied glacial deposits on this island created landscape diversity through the production of soil types of different texture (Pastor et al. 1982). Driven primarily by the effects of soil texture on water availability, different vegetation assemblages developed on the varied soil types (Pastor and Post 1986). The vegetation differences created variation in ecosystem properties and processes related to soil organic matter quality and nitrogen availability (Pastor et al. 1982, McClaugherty et al. 1985). The differences in nitrogen availability function as a positive feedback, amplifying the vegetation and ecosystem differences. Blackhawk Island is an example of how landscape diversity affects ecosystem processes and organism diversity and how organisms and ecosystem processes feedback to amplify landscape diversity. In a more general sense, much basic ecosystem and landscape ecology research addresses how spatial heterogeneity at the landscape level affects ecosystem processes (Forman and Godron 1986, Turner 1989, Huston 1994, Pickett and Cadenasso 1995), but links between landscape diversity, ecosystem processes, and species diversity have rarely been explicitly addressed. The effects of this heterogeneity can be summarized in a ‘‘state factor’’ model of ecosystems where ecosystem processes are viewed as a function of climate, organisms, topography, parent material, time and humans


225

(Amundson and Jenny 1997). The climate, topography, and parent material state factors have been useful for conceptualizing the landscape–ecosystem– species trajectory (Pastor et al. 1982, Schimel et al. 1985, Groffman and Tiedje 1989, Zak et al. 1991). The evolution of landscapes is well studied in geomorphology (Huggett 1975, Hole 1978) and there have been efforts to develop quantitative indices of landscape and soil diversity (Ibanez et al. 1995). However, there has been little analysis of relationships between landscape and soil diversity and organism and ecosystem diversity. Given the strong controls of landscape on ecosystem processes, these relationships are likely quite important, and should be a focus of future research. Species–Ecosystem–Species Analysis of species effects on ecosystem processes has been a major theme of ecological research over the last 10 years or so. Species influence on organic matter quality, nutrient and water use, and physical structure have been studied in numerous ecosystems (Jones and Lawton 1995). The ‘‘species– ecosystem–species’’ link relates to the ecosystem processes controlled by organisms and their effect on species diversity. In the state factor model of ecosystems and soil development described above (Amundson and Jenny 1997), the ‘‘organism’’ state factor represents the effects organisms have on fundamental soil and ecosystem properties through the accumulation of organic matter, stimulation of chemical and physical weathering, and alteration of water and air flow (Groffman and Bohlen 1999). There has been a great recent increase in interest in species–ecosystem links. This interest has been motivated by concern that human-induced loss of species diversity will influence important ecosystem processes related to nutrient retention, decomposition and production, and more fundamentally, to the stability of these processes, that is, their ability to persist through time in the face of environmental change. Species control ecosystem processes because they differ in the rates and pathways by which they process resources. Thus, changes in species composition are likely to alter ecosystem processes through changes in the functional traits of biota. Traits with profound effects are those that modify the availability, capture, and use of soil resources such as water and nutrients (Vitousek 1990, Chapin et al. 1995, Lawton and Jones 1995). For example, litter production by plants affects soil temperature and moisture (Scholes and Walker 1993, Hobbie 1995), which affects nutrient mineralization. Species influence the stability (resistance and resilience) of ecosystem processes, via differential environmental sensitivity among functionally similar species. The more functionally similar species there are in a community—that is, the greater the diversity within a functional group—the greater is its resilience in responding to environmental change (McNaughton 1977, Chapin

226


and Shaver 1985, Walker 1992, 1995, Lawton and Brown 1993). For example, the presence of drought-tolerant species allowed diverse grasslands to maintain higher productivity in response to drought than grasslands whose diversity had been reduced by experimental addition of nutrients (Tilman and Downing 1994). Although species–ecosystem links have been well studied, the feedbacks where alteration of ecosystem processes by specific species influence species diversity, that is, the full species–ecosystem–species link, have not been well developed (Chapin et al. 1997). For example, while Tilman and Downing (1994) show that species diversity can influence productivity, we know little about the effects of productivity on diversity (Tilman et al. 1996). Given that ecosystem processes fundamentally control the availability of critical resources, the full species–ecosystem–species link is clearly worthy of further study. Species–Landscape–Ecosystem–Species In the former links we show how landscape diversity and species diversity influence ecosystem processes, and how these processes feedback to influence species diversity. However, there is a more complex set of interactions that is initiated by organism activity that affects both landscape and species diversity. Organisms can affect species diversity by changing the landscape mosaic, which in turn changes the distribution of resources. The concept of organisms as ‘‘ecosystem engineers’’ (Jones et al. 1994) suggests that species can control ecosystem processes that underlie resource distribution and availability by modification and creation of patches in the landscape. The species–landscape–ecosystem–species trajectory is frequently driven by patch formation by plants and burrowing animals. Many studies have shown how vegetation pattern strongly influences water redistribution, biomass production, and nutrient dynamics (Wilson and Agnew 1992, Bach 1988a,b, Maubon et al. 1995, McCoy and Bell 1991, Doak et al. 1992, Roland 1993, Kareiva 1985, Turner 1989, Shachak et al. 1998). As described above in our example for the Negev, the diversity of vegetation patterns is particularly pronounced in drylands due to water scarcity, competition, and herbivory. Desert shrubs are examples of plants as ecosystem engineers. Shrublands can be viewed as landscape mosaics composed of shrub and intershrub patches. In a natural state the arrangement of the landscape mosaic is such that resources generated by the intershrub patch are intercepted by and accumulate in the shrub patches. The intershrub patches are sources for soil materials and run-off water (Abrahams et al. 1994, 1995, Eldridge 1993, Snow and McClelland 1990, Stockton and Gillette 1990, Rostango 1989, Yair and Shachak 1987). Shrub patches are sinks and therefore the principal loci of plant species diversity. This is mainly because of the accumulation of soil, water, and nutrients, which promote the growth of rich herbaceous vegetation under the shrubs (Allen 1991, Boeken and Shachak 1994, Garner and Steinberger 1989, Noy-Meir 1985, Schlesinger


227

et al. 1990, Shachak and Lovett 1998, Weinstein 1975, West 1989, Zaady et al. 1996). Under natural conditions the source–sink relationship between the two patch types is critical to the functioning of the system (Ludwig and Tongway 1995). Burrowing animals in deserts, such as porcupines (Alkon 1999), isopods (Shachak and Yair 1984), and ants, are important as links among species diversity, ecosystem processes, and landscape diversity. They modify landscape diversity by adding new patch types, such as pits and mounds. The new patches are capable of accumulating water and nutrients and thus enhance the increase in annual plant diversity. The burrowing animals act as ecosystem engineers which affect biodiversity (Wilby et al. 2001).

Summary This chapter attempts to integrate concepts of ecosystem science with the study of biodiversity. We propose a framework that shows how ecosystem processes link landscape and species diversities and how the development of biodiversity is inexorably intertwined with these processes. Our example from the Negev and other systems demonstrates that these linkages are complex. There are not clear, unidirectional relationships among landscape diversity, ecosystem processes, and species diversity. However, there is a finite suite of linkages among these elements that can be applied to many different ecosystems. We suggest that this framework and set of linkages can be of great value in assessment and management of biodiversity in many areas. If we recognize the inherent linkages between ecosystem processes and species and landscape diversity, we will have a better understanding of the production and maintenance of biodiversity. This understanding will be of great value as we attempt to maintain biodiversity in the face of global climate and land-use change and the demand for a diverse set of ecosystem services by an expanding human population.

References Abrahams, A.D., A.J. Parsons, and J. Wainwright. 1994. Resistance to overland flow on semiarid grassland and shrubland hillslopes, Walnut Gulch, southern Arizona. Journal of Hydrology 156: 431–446. Abrahams, A.D., A.J. Parsons, and J. Wainwright. 1995. Effects of vegetation change on interrill run-off and erosion, Walnut Gulch, southern Arizona. Geomorphology 13: 37–48. Alkon, P.U. 1999. Microhabitat to landscape impacts: crested porcupine digs in the Negev Desert highlands. Journal of Arid Environments 41: 183–202. Allen, E.B. 1991. Temporal and spatial organization of desert plant communities. Pp. 295–332 in J. Skujins, ed., Semiarid Lands and Desert: Soil Resource and Reclamation. Dekker, New York.

228


Amundson, R., and H. Jenny. 1997. On a state factor model of ecosystems. BioScience 47: 536–543. Bach, C.B. 1988a. Effects of host plant patch size on herbivore density: patterns. Ecology 69: 1090–1102. Bach, C.B. 1988b. Effects of host plant patch size on herbivore density: underlying mechanisms. Ecology 69: 1103–1117. Baskent, E.Z., and J. Glenwood. 1996. Designing forest management to control spatial structure of landscapes. Landscape and Urban-Planning 34: 55–74. Boeken, B., and M. Shachak. 1994. Desert plant communities in human made patches—implications for management. Ecological Applications 4: 702–716. Chapin, F.S., III, and G.R. Shaver. 1985. Individualistic growth response of tundra plant species to manipulation of light, temperature, and nutrients in a field experiment. Ecology 66: 564–576. Chapin, F.S., J. Lubchenco, and H.L. Reynolds. 1995. Biodiversity effects on patterns and processes of communities and ecosystems. Pp. 23–45 in V.H. Heywood, ed., Global Biodiversity Assessment. Cambridge University Press, Cambridge. Chapin, F.S. III, B.H. Walker, R.J. Hobbs, D.U. Hooper, J.H. Lawton, O.E. Sala, and D. Tilman. 1997. Biotic control over the functioning of ecosystems. Science 277: 500–504. Didier, A., and I. Poudevigne. 2000. Diversity patterns in grasslands along a landscape gradient in northwestern France. Journal of Vegetation Science 2: 287–294. Doak, D.F., P.C. Marino, and P.M. Kareiva. 1992. Spatial scale mediates the influence of habitat fragmentation on dispersal success: implications for conservation. Theoretical Population Biology 41: 315–336. Eldridge, D.J. 1993. Cryptogram cover and soil surface condition: effects on hydrology on a semi arid and woodland soil. Arid Soil Research and Rehabilitation 7: 203–217. Eldridge, J.D., E. Zaady, and M. Shachak. 2000. Infiltration through three contrasting soil crusts in patterned Negev Desert landscapes. Catena 40: 323–336. Forman, R.T.T., and M. Godron. 1986. Landscape Ecology. John Wiley and Sons, New York. Garner, W., and Y. Steinberger. 1989. A proposed mechanisms for the formation of ‘fertile islands’ in desert ecosystems. Journal of Arid Environments 16: 257–262. Groffman, P.M. and P.J. Bohlen. 1999. Soil and sediment biodiversity: cross-system comparisons and large scale effects. BioScience 49: 139–148. Groffman, P.M. and J.M. Tiedje. 1989. Denitrification in north temperate forest soils: spatial and temporal patterns at the landscape and seasonal scales. Soil Biology and Biochemistry 21: 613–620. Hobbie, S.E. 1995. Direct and indirect species effects on biogeochemical processes in arctic ecosystems. Pp. 213–224 in F.S. Chapin, III, and C. Ko¨rner, eds., Arctic and Alpine Biodiversity: Patterns, Causes, and Ecosystem Consequences. SpringerVerlag, Berlin. Hole, F.D. 1978. An approach to landscape analysis with an emphasis on soils. Geoderma 21: 1–23. Huggett, R. 1975. Soil landscape systems: a model of soil genesis. Geoderma 13: 1–22. Huston, M.A. 1994. Biological Diversity: The Coexistence of Species in Changing Landscapes. Cambridge University Press, New York. Ibanez, J.J., S. De-Alba, F.F. Bermudez, and A. Garcia-Alvarez. 1995. Pedodiversity: concepts and measures. Catena 24: 215–232.


229

Jones, C.G., and J.H. Lawton. 1995. Linking Species and Ecosystems. Chapman and Hall, New York. Jones, C.G., J.H. Lawton, and M. Shachak. 1994. Organisms as ecosystem engineers. Oikos 69: 373–386. Kareiva, P.M. 1985. Finding and losing host plants by Phylotreta: Patch size and surrounding habitat. Ecology 66: 1809–1816. Lawton, B.H., and Brown, V.K. 1993. Redundancy in ecosystems. Pp. 255–270 in E.D. Schulze, and H.A. Mooney, eds., Biodiversity and Ecosystem Function. Springer-Verlag, New York. Lawton, J.H. and C.G. Jones. 1995. Linking species and ecosystems: Organisms as ecosystem engineers. Pp. 141–150 in C.G. Jones, and J.H. Lawton, eds. Linking Species and Ecosystems. Chapman and Hall, New York. Ludwig, J., and D.T. Tongway. 1995. Spatial organization of landscapes and its function in semiarid woodlands, Australia. Landscape Ecology 10: 51–63. Maubon, M., J.F. Ponge, and J. Andre. 1995. The dynamics of Vaccinium myrtillus patches in mountain forests. Journal of Vegetation Science 6: 343–348. McClaugherty, C.A., J. Pastor, J.D. Aber, and J.M. Melillo. 1985. Forest litter decomposition in relation to soil nitrogen dynamics and litter quality. Ecology 66: 266–275. McCoy, E.D., and S.S. Bell. 1991. Habitat structure: the evolution and diversification of a complex topic. Pp. 3–27 in S.S. Bell, E.D. McCoy, and H.R. Musbinsky, eds., Habitat Structure. Chapman and Hall, London. McNaughton, S.J. 1977. Diversity and stability of ecological communities: a comment on the role of empiricism in ecology. The American Naturalist 111: 515–525. Noy-Meir, I. 1985. Desert ecosystem structure and function. Pp. 93–103 in M. Evenari, et al., eds., Hot Deserts and Arid Shrublands. Elsevier Science, Amsterdam. Offer Z.Y., E. Zaady, and M. Shachak. 1997. Aeolian particles input to the soil surface at the northern limit of the Negev desert. Arid Soil Research and Rehabilitation 12: 55–62. Pastor, J., and W.M. Post. 1986. Influence of climate, soil moisture, and succession on forest carbon and nitrogen cycles. Biogeochemistry 2: 3–27. Pastor, J., J.D. Aber, C.A. McClaughertry, and J.M. Melillo. 1982. Geology, soils and vegetation of Blackhawk Island, Wisconsin. American Midland Naturalist 108: 266–277. Pickett, S.T.A., and M.L. Cadenasso. 1995. Landscape ecology: spatial heterogeneity in ecological systems. Science 269: 331–334. Roland, J. 1993. Large scale forest fragmentation increases the duration of tent caterpillar outbreaks. Journal of Animal Ecology 63: 392–398. Rostango, C.M. 1989. Infiltration and sediment production as affected by soil surface conditions in a shrubland of Patagonia, Argentina. Journal of Range Management 42: 382–385. Schimel, D.S., MA. Stillwell, and R.G. Woodmansee. 1985. Biogeochemistry of C, N, and P in a soil catena of the shortgrass steppe. Ecology 66: 276–282. Schlesinger, W.H., J.F. Reynolds, G.L. Cunningham, L.F. Huenneke, W.M. Jarrell, R.A. Virginia, and W.G. Whitford. 1990. Biological feedbacks in global desertification. Science 247: 1043–1048. Scholes R.J., and B.H. Walker. 1993. An African Savannah. Cambridge University Press, Cambridge.

230


Shachak, M., and G.M. Lovett. 1998. Atmospheric deposition to a desert ecosystem and its implication for management. Ecological Applications 8: 455–463. Shachak, M., and A. Yair. 1984. Population dynamics and the role of Hemilepistus reaumuri in a desert ecosystem. Symposium Zoological Society of London 53: 295–314. Shachak, M., M. Sachs, and I. Moshe. 1998. Ecosystem management of desertified shrublands in Israel. Ecosystems 1: 475–483. Snow, J.T., and T.M. McClelland. 1990. Dust devils at White Sands Missile Range, New Mexico. I. Temporal and spatial distributions. Journal of Geophysical Research 95: 13707–13721. Stockton, P.H., and D.A. Gillette. 1990. Field measurement of the sheltering effect of vegetation on erodible land surfaces. Land Degradation and Rehabilitation 2: 77–85. Tilman, D., and J.A. Downing. 1994. Biodiversity and stability in grasslands. Nature 367: 363–365. Tilman, D., D. Wedin, and J. Knops. 1996. Productivity and sustainability influenced by biodiversity in grassland ecosystems. Nature 379: 718–720. Turner, M.G. 1989. Landscape ecology: the effect of pattern on process. Annual Review of Ecology and Systematics 20: 171–197. Vitousek, P.M. 1990. Biological invasions and ecosystem processes: toward an integration of population biology and ecosystem studies. Oikos 57: 7–13. Walker, B.H. 1992. Biodiversity and ecological redundancy. Conservation Biology 6(1): 18–23. Walker, B. 1995. Conserving biological diversity through ecosystem resilience. Conservation Biology 9: 1–7. Wardell J.G., and P. Horwitz. 1996. Conserving biodiversity and the recognition of heterogeneity in ancient landscapes: A case study from south-western Australia. Forest Ecology and Management 85: 219–238. Weinstein, N. 1975. The effects of a desert shrub on its micro-environment and the herbaceous plants. M.Sc. Thesis. The Hebrew University, Jerusalem. West, N.E. 1989. Spatial pattern—functional interactions in shrub dominated plant communities. Pp. 283–305 in C.M. McKell, ed. The Biology and Utilization of Shrubs. Academic Press, London. Wilby, A., M. Shachak, and B. Boeken. 2001. Integration of ecosystem engineering and trophic effects of herbivores. Oikos 92: 436–44. Wilson, J.B., and A.D.Q. Agnew. 1992. Positive-feedback switches in plant communities. Advances in Ecological Research 23: 263–336. Yair, A., and M. Shachak. 1987. Studies in watershed ecology of an arid area. Pp. 146–193 in M.O. Wurtele, and L. Berkofsky, eds., Progress in Desert Research. Rowman and Littlefield, Totowa, NJ. Zaady, E., P.M. Groffman, and M. Shachak. 1996. Release and consumption of nitrogen by snail faeces in Negev Desert soils. Biology and Fertility of Soils 23: 399–404. Zak, D.R., A. Hairston, and D.F. Grigal. 1991. Topographic influences on nitrogen cycling within an upland pin oak ecosystem. Forest Science 37: 45–53.

Part III



14 The Effects of Grazing on Plant Biodiversity in Arid Ecosystems David Ward

C

onventional wisdom views heavy grazing as the major cause of desertification in semiarid and arid areas of Africa, Asia, and Australia (see, e.g., Acocks 1953, Jarman and Bosch 1973, Sinclair and Fryxell 1985, Middleton and Thomas 1997) (table 14.1). Nowhere is the effect of heavy grazing more apparent than in the Sahel of Africa (Sinclair and Fryxell 1985). This land denudation has resulted in a negative feedback loop via decreased soil nutrient status and increased soil albedo (due to lower vegetation cover), causing increased evaporation and decreased precipitation, which in turn reduces the stocking capacity of the land, further exacerbating the negative effects of grazing (Schlesinger et al. 1990). A less dramatic result of overgrazing is a long-term decline in agricultural productivity. For example, the arid Karoo region of South Africa has experienced no climatic change over the last two centuries, yet there has been a 50% decline in stocking rates in seven of eight magisterial districts from 1911 to 1981 (Dean and McDonald 1994). These authors ascribe this decline to heavy grazing that reduced palatable plant populations and hence the carrying capacity of the vegetation in the long term. These examples of the negative effects of grazing in arid ecosystems lie in stark contrast with a large number of African studies that compared the effects of commercial (privately owned) and communal (subsistence, no private ownership) ranching on vegetation and soils (e.g., Archer et al. 1989, Tapson 1993, Scoones 1995, Ward et al. 1999a,b, reviewed by Behnke and Abel 1996). In spite of 5–10-fold higher stocking rates on communal ranches, few studies have shown differences in effects on biodiversity, plant species composition and soil quality between these ranching types (Archer et al. 1989, 233

234


Table 14.1 The main causes of soil degradation in susceptible drylands. Results expressed in millions of hectares of degraded soil (data from Middleton and Thomas 1997) Region

Overgrazing Deforestation Agricultural

Overexploitation

Bioindustrial

Total Degraded

Africa

184.6

18.6

62.2

54.0

0

319.4

Asia

118.8

111.5

96.7

42.3

1.0

370.3

Australasia

78.5

4.2

4.8

0

0

87.5

Europe

41.3

38.9

18.3

0

0.9

99.4

North America

27.7

4.3

41.4

6.1

0

79.5

South America

26.2

32.2

11.6

9.1

0

79.1

Tapson 1993, Scoones 1995, Ward et al. 1999a,b—fig. 14.1). Similarly, studies of grazing in Mediterranean semiarid grasslands (reviewed by Seligman 1996) and Middle Eastern arid rangelands (Ward et al. 1999b) show that the effects of grazing on biodiversity are relatively small.

Parameters of Consensus on Effects of Grazing in Arid Ecosystems A consensus has developed in recent years that arid grazing ecosystems are nonequilibrial, event-driven systems (see, e.g., O’Connor 1985, Venter et al. 1989, Milchunas et al. 1989, Parsons et al. 1997). Ellis and Swift (1988), Tapson (1993), Werner (1994), and Sullivan (1996) contend that rainfall in arid regions is the major driving variable and has the ability to ‘‘recharge’’ a system that suffers heavy grazing pressure. Indeed, it is generally agreed that where pastoralists are able to maintain their activities on a large spatial scale by migrating to areas where key rich resources can be exploited, allowing previously used resources time to recover, negative effects of grazing on plant biodiversity do not develop (Sinclair and Fryxell 1985, Ellis and Swift 1988, Behnke and Abel 1996). Moreover, even where pastoralists are forced to settle in small areas, abiotic variables such as precipitation may be of such overriding importance that these negative effects of grazing on plant cover, plant species richness, and diversity cannot be detected (see, e.g., O’Connor 1985, Ward et al. 1999a,b). An additional factor limiting the effects of grazing in arid ecosystems is the high level of plant resistance to herbivory (Ward and Olsvig-Whittaker 1993, Rohner and Ward 1997, Ward et al. 1997). The cost of regrowth subsequent to herbivory in these ecosystems is high due to low levels of precipitation and low soil nutrient availability (Coley et al. 1985). This has resulted in selection

The Effects of Grazing on Plant Biodiversity in Arid Ecosystems

235

Figure 14.1 Perennial plant percentage cover, species richness, and species diversity (Shannon–Wiener measure) on communal and commercial ranches in arid Otjimbingwe, Namibia. None of the comparisons was significant in spite of the 10fold higher stocking densities on communal ranches (data from Ward et al. 1999a).

in plants for resistance (or tolerance) to herbivory, thereby minimizing the impacts of organ (e.g., branches, leaves) removal on fitness (lifetime reproductive success and survival). Thus, low herbivore numbers and high resistance in plants in arid ecosystems work in concert to reduce the effects of grazing in these ecosystems. Empirical data appear to bear this out (see Ward et al. 2000a for examples from arid and semiarid Namibia). In a global review, Milchunas and Lauenroth (1993) showed that the effects of grazing on species composition and plant biomass increase with increasing precipitation (fig. 14.2). The high spatial and temporal variance in plant cover, biodiversity, and soil quality in arid ecosystems makes comparisons of the effects of grazing on plant biodiversity across habitats and ecosystems rather complex. For this reason, many researchers have emphasized comparisons in the piosphere, the region of heavy grazing around watering points. Clearly, animals in arid ecosystems are limited by the availability of water and hence tend to collect there. As a result, these are usually the most heavily damaged areas in a ranching ecosystem. James et al. (1999) summarized piosphere effects in arid Australian ecosystems as follows: 1. The area near a watering point is usually bare, but supports shortlived, often unpalatable, trample-resistant species after rain. 2. A dense zone of unpalatable woody shrubs usually occurs immediately beyond the denuded area.

236


Figure 14.2 Differences in species composition between grazed and ungrazed lands in arid ecosystems of Americas and Africa plus Asia (data from Milchunas and Lauenroth 1993). There was no significant relationship between species dissimilarity and mean annual rainfall for the American comparison, while there was a significant relationship for the African/Asian comparison ðp < 0:001Þ.

3. Palatable perennial plants decline in abundance and species richness within zones 1 and 2 above. 4. Species richness does not change consistently with increasing distance from watering points.

Effects of Evolutionary History of Grazing I focus here on those aspects of the effects of grazing on plant biodiversity of arid lands that are least understood and that should prove to be the most interesting and exciting new fields of research. Milchunas et al. (1988) predicted that a long evolutionary history of grazing results in selection for regrowth following herbivory and for prostrate growth forms. In such communities, grazing causes rapid shifts between suites of species adapted to either grazing avoidance/tolerance or competition. In their global review, Milchunas and Lauenroth (1993) found that increasing evolutionary history of grazing produced increasing dissimilarity in species composition between grazed and ungrazed sites regardless of the level of precipitation. However, in the Middle East and North Africa, where heavy grazing has occurred for thousands of years, grazing has seldom been shown to affect species composition (Noy-Meir et al. 1989, Perevolotsky 1994, Ward et al. 1999b). A possible reason for this lack of grazing impact is the Narcissus


237

effect (Colwell and Winkler 1984), because selection in the past has resulted in the extinction of all nonresistant/tolerant genotypes. Thus, all extant species are similarly resistant to herbivores, resulting in the absence of an effect of current herbivory on biodiversity (Ward and Olsvig-Whittaker 1993, Perevolotsky 1994). Presumably, in such ecosystems, conditions seldom favor growth-dominated genotypes. Thus, only one (resistant/tolerant) genotype exists in these populations. The conditions under which selection should favor multiple or single genotypes in a population of plants in arid ecosystems has not been adequately investigated. A combination of theoretical and experimental approaches should be followed with this aim.

Period of Rest After Drought and Heavy Grazing There are conflicting opinions and data on the effects of the timing of grazing and the period of rest after drought on plant biodiversity. It has widely been observed that plants require a period of rest after drought and heavy grazing (e.g., Danckwerts 1993, Danckwerts and Stuart-Hill 1987, Van der Heyden and Stock 1995). In ecosystems where plants are not allowed to recover, land degradation may occur as quickly as in ecosystems where stocking rates are 5–10 times as high but plants are allowed to recover. This may be one reason why comparisons of the effects of commercial and communal ranching seldom show large differences in impacts on vegetation in spite of far higher mean stocking rates on communal ranches (Archer et al. 1989, Tapson 1993, Scoones 1995, Ward et al. 1999a,b). Commercial ranchers, who can usually afford to restock their lands soon after drought-induced mortality (or supplement forage to minimize mortality), do not allow the vegetation time to recover lost resources. Communal ranchers, on the other hand, can seldom afford to restock and thus must allow their herds to enlarge through natural reproduction, much in the way that wild herbivore populations recover after drought-induced mortality (Ellis and Swift 1988). During herd regrowth, stocking rates are well below capacity, allowing the plants to recover lost resources. Indeed, the degradation of the Sahel is now considered to be a result of the settlement of pastoralists and the provision of supplementary feed during drought periods, resulting in increased stock survival and greater depletion of plant resources during the postdrought recovery phase (Mainguet 1991, Sinclair and Fryxell 1985). In marked contrast to these studies, many studies comparing continuous and rotational grazing have found that the period of rest after grazing had little independent effect on biodiversity and plant cover in semiarid and arid systems above that induced by differences in stocking density (e.g., Denny and Steyn 1978, O’Connor 1985). Bryant et al. (1989) compared the purported benefits of a rotational short-duration grazing strategy developed by Savory (1978) with conventional continuous grazing. They found that this stocking system produced no positive influence on germination or establishment of plants in arid and semiarid regions of North America. Furthermore, this stocking regimen

238


did not improve range condition at the same or higher stocking rates compared with continuous grazing, and did not increase grass or forb standing crop. They considered stocking density to be the only important factor affecting plant cover and biodiversity. Nonetheless, they recognized that grazing systems that provide plants with rest periods are the only way to restore root reserves and replenish photosynthetic material (Bryant et al. 1989). Controlled studies of the effects of rates of postdrought restocking and herd regrowth on plant biodiversity have not been conducted thus far. It would be particularly useful to examine the effects of grazing on the total nonstructural carbohydrate reserves of plants as a functional index of recovery rate (Van der Heyden and Stock 1995). Another potentially useful tool that bears further investigation in this regard is the traditional range-management classification of grass species according to their abilities to withstand grazing (see, e.g., Stoddart et al. 1975). Species are classified either as increasers (i.e., species that increase in cover and abundance after grazing or disturbance) or decreasers (species that decrease in cover or abundance with grazing). This classification, and its many modifications, was an important first step towards understanding and predicting the effects of grazing on biodiversity. These descriptions also represent an important early attempt at finding assembly rules for grazed plant communities. However, increaser–decreaser species descriptions are rooted in Clements’ (1916) notion that for any given soil/habitat type there is a specific type of community that will prevail at the end of the successional process (i.e., the climax). Thus, the presence of any species or combinations thereof that are inconsistent with the preconceived notion of the climax was used to indicate that the range was not in good condition. Conversely, the presence of a single species (e.g., Themeda triandra in southern Africa) was universally considered to be a sign of a range in good condition regardless of the associated biodiversity or stocking conditions (O’Connor and Bredenkamp 1997). More recently, it has been shown that such classifications depend on genotype-by-environment interactions and on the particular competitors in the environment (see Kirkman 1988, Milchunas and Lauenroth 1993, O’Connor and Bredenkamp 1997, among others). In addition to the need for jettisoning the Clementsian notions involved, there is a need for a more mechanistic understanding of what makes a species either increase or decrease with grazing. Ecophysiological studies of the effects of grazing on regrowth, competitive ability, and fitness of key species under a variety of abiotic conditions and with different suites of competitors are needed. Genotypic variation in widespread species may result in different responses to grazing within the same species (Kirkman 1988). Thus, studies of genotype by environment interactions under different grazing intensities will be particularly important. Furthermore, because grazing may affect plant quality and the induction of physical and chemical defenses in plants (see, e.g., McNaughton and Tarrants 1983, Rohner and Ward 1997), which will affect the probability of subsequent grazing, more studies of the effects of grazing on palatability of plants under natural conditions are required.


239

Rate of Degradation Several studies have shown that degradation as a result of overgrazing may take 80–100 years to become manifest, particularly in shrub-dominated habitats (Dean and MacDonald 1994, Wiegand and Milton 1996, Ward and Ngairorue 2000). Examples from arid Namibia are illustrative at two spatial scales. If one compares grass production at the end of the wet season near watering points that have been in use for at least 150 years with those that have been in use for only 10 years, one can see that there is a difference in grass production. However, one new watering point with particularly high current grazing is in as poor a condition as the old sites (fig. 14.3). At a larger spatial scale at 31 sites along a rainfall gradient from 100–450 mm per annum, we found that there was no correlation between the residuals of grass production (regressed against mean annual rainfall) and stocking density in the current season or averaged over the past 11 years (Ward and Ngairorue 2000). However, when we compare data along the same gradient between 1939 and 1997, grass production in 1997 was approximately 50% less than what it was in the earlier period (Ward and Ngairorue 2000). Thus, some claims that grazing may have little effect on ecosystems may depend on the

Figure 14.3 Comparison of mean slopes of grass height against distance from watering points in March 1998 (end of wet season) at old (>150 years of use) and new (

Biodiversity in Drylands: Toward a Unified Framework (Long-Term Ecological Research Network Series)

Biodiversity in Drylands: Toward a Unified Framework (Long-Term Ecological Research Network Series)

Toward a national research network

Ecological Applications: toward a sustainable world

Climate Variability and Ecosystem Response at Long-Term Ecological Research Sites (Long-Term Ecological Research Network Series)

Ecological Economics Research Trends

Ecological Networks, Volume 42 (Advances in Ecological Research)

Gender and Kinship: Essays Toward a Unified Analysis

Neurophilosophy: Toward a Unified Science of the Mind-Brain

An essay toward a unified theory of special functions

Advances in Ecological Research, Volume 12

Advances in Ecological Research, Volume 44

Advances in Ecological Research, Volume 8

Advances in Ecological Research, Volume 28

Advances in Ecological Research, Volume 6

Advances in Ecological Research, Volume 4

Advances in Ecological Research, Volume 25

Advances in Ecological Research, Volume 23

Advances in Ecological Research, Volume 20

Advances in Ecological Research, Volume 27

Advances in Ecological Research, Volume 19

Advances in Ecological Research, Volume 29: Estuaries

Advances in Ecological Research, Volume 2

Advances in Ecological Research, Volume 16

Advances in Ecological Research, Volume 1

Advances in Ecological Research, Volume 7

Advances in Ecological Research, Volume 21

Advances in Ecological Research, Volume 10

Advances in Ecological Research, Volume 41

Biodiversity in Drylands: Toward a Unified Framework (Long-Term Ecological Research Network Series)