416

Whittaker. Island Biogeography. 2007

Embed Size (px)

Citation preview

Page 1: Whittaker. Island Biogeography. 2007
Page 2: Whittaker. Island Biogeography. 2007

Island Biogeography

Page 3: Whittaker. Island Biogeography. 2007

Front cover image: Rapa (Austral Islands, French Polynesia) from Mt. Perahu. Photoby R. G. Gillespie, taken on an expedition funded by La Délégation à la Recherche

and l’Institut Louis Malardé, Tahiti.

Back cover image: Kosrae (Micronesia) from the summit of Mt. Finkol. Photo by G. K.Roderick and R. G. Gillespie, taken on an expedition funded by the National

Geographic Society.

Page 4: Whittaker. Island Biogeography. 2007

IslandBiogeographyEcology, evolution, and conservation

Second Edition

Robert J. WhittakerSchool of Geography, University of Oxford, UK

and

José María Fernández-PalaciosGrupo de Investigación de Ecología Insular, Universidad de La Laguna, Tenerife

1

Page 5: Whittaker. Island Biogeography. 2007

3Great Clarendon Street, Oxford OX2 6DPOxford University Press is a department of the University of Oxford.It furthers the University’s objective of excellence in research, scholarship,and education by publishing worldwide in

Oxford New YorkAuckland Cape Town Dar es Salaam Hong Kong KarachiKuala Lumpur Madrid Melbourne Mexico City NairobiNew Delhi Shanghai Taipei Toronto

With offices inArgentina Austria Brazil Chile Czech Republic France GreeceGuatemala Hungary Italy Japan Poland Portugal SingaporeSouth Korea Switzerland Thailand Turkey Ukraine Vietnam

Oxford is a registered trade mark of Oxford University Pressin the UK and in certain other countries

Published in the United Statesby Oxford University Press Inc., New York

© Robert J. Whittaker and José María Fernández-Palacios 2007

The moral rights of the authors have been assertedDatabase right Oxford University Press (maker)

First published 2007

All rights reserved. No part of this publication may be reproduced,stored in a retrieval system, or transmitted, in any form or by any means,without the prior permission in writing of Oxford University Press,or as expressly permitted by law, or under terms agreed with the appropriatereprographics rights organization. Enquiries concerning reproductionoutside the scope of the above should be s ent to the Rights Department,Oxford University Press, at the address above

You must not circulate this book in any other binding or coverand you must impose the same condition on any acquirer

British Library Cataloguing in Publication DataData available

Library of Congress Cataloging in Publication DataApplied for

Typeset by Newgen Imaging Systems (P) Ltd., Chennai, IndiaPrinted in Great Britainon acid-free paper byAntony Rowe Ltd., Chippenham, Wiltshire

ISBN 0–19–856611–5 978–0–19–856611–3ISBN 0–19–856612–3 978–0–19–856612–0

10 9 8 7 6 5 4 3 2 1

Page 6: Whittaker. Island Biogeography. 2007

Island biogeography is an important subject forseveral reasons. First, it has been and remains afield which feeds ideas, theories, models, and testsof same into ecology, evolutionary biology, and bio-geography. This is because islands provide naturalscientists with model systems—replicated and sim-plified contexts—allowing us to isolate particularfactors and processes and to explore their effects.Secondly, some of these theories have had greatweight placed upon them in applications in natureconservation, as scientists and conservationistsattempt to understand, predict and manage the bio-diversity impacts of habitat loss and fragmentation.Thirdly, in our modern age of anthropogenic extinc-tions, islands qualify as ‘hotspots’: combining theattributes of high levels of unique biodiversity, ofrecent species extinctions, and of likely futurespecies losses. The protection of the unique biolog-ical features of island ecosystems presents us with aconsiderable challenge, not only ecologically, butalso because of the fragmented nature of theresource, scattered across all parts of the globe andall political systems, and generally below the hori-zons of even global media networks. It is our hopethat this book will foster an increased interest inisland ecology, evolution, and conservation andthat it will be of value for students and researchersworking in the fields of the life and environmentalsciences.

This second edition is built upon the foundationsof the first edition but has been substantially reor-ganized and updated to reflect what we considerthe most important developments in island bio-geography over the last decade. As will becomeevident to those who dip into this volume, we covera great deal more than the biology of the systems.Indeed, we have expanded our coverage of thedevelopmental history and environmental dynamics

of islands in this second edition. Much fascinatingnew work has been published in this arena, and itis proving to be fundamental to improving ourunderstanding of island evolution and ecology.

Another feature of this revision is the inclusionof a great deal of material on the island region ofMacaronesia (the Happy Islands), and particularlyof the Canaries. These islands are the Atlantic equiva-lent of Hawaii and the Galápagos, providing a richmix of geological and evolutionary–ecologicalinsights on the one hand and biodiversity conser-vation problems on the other. Much new and excit-ing work has been published on these islands sincethe first edition of this book was written, and wewere keen to bring some of this work to the atten-tion of a wider audience of students and scholars.

Island biogeography is a dynamic field. Whilstmany ideas and themes have long pedigrees, newideas, and insights continue to be generated, oftenbuilding on long-running debates. We haveattempted to reflect the diversity of viewpoints andinterpretations within the field, although inevitablythe selection of material reflects our own biases andinterests.

There are many people we would like to thank,not least our students and the members of ourresearch groups, with whom we have enjoyed illu-minating discussions on many island themes. IanSherman, our editor at OUP, provided encourage-ment, help, and good advice at all stages of theproject, and we thank him, Stefanie Gehrig, andtheir colleagues at the Press, for all their efforts. Wethank the following colleagues for variously com-menting on draft material, supplying answers toqueries, and discussion of ideas: Gregory H. Adler,Rubén Barone, Paulo Borges, Pepe Carrillo, JamesH. Brown, Juan Domingo Delgado, LawrenceHeaney, Scott Henderson, Paco Hernán, Joaquín

v

Preface and acknowledgements

Page 7: Whittaker. Island Biogeography. 2007

Hortal, Hugh Jenkyns, Richard Ladle, MarkLomolino, Águedo Marrero, Aurelio Martín, BobMcDowall, Leopoldo Moro, Manuel Nogales,Pedro Oromí, Jonathan Price, Mike Rosenzweig,Dov Sax, Ángel Vera, and James Watson. All errorsand omissions are of course our own to claim. Mostof the figures were drawn by Ailsa Allen. SueStokes helped in compilation of material. We aregrateful to those individuals and organizationswho granted permission for the reproduction of

copyright material: the derivation of which is indicated in the relevant figure and table legends,supported by the bibliography. Finally, we thankour families, Angela, Mark, and Claire and Neli,José Mari, and Quique, for all their support and tolerance during the preparation of this book.

R.J.W.J.M.F.P.

Oxford and La Laguna, 28 April 2006

v i P R E FAC E A N D AC K N O W L E D G E M E N T S

Page 8: Whittaker. Island Biogeography. 2007

vii

Preface and acknowledgements v

Part I: Islands as Natural Laboratories 1

1 The natural laboratory paradigm 3

2 Island environments 102.1 Types of islands 102.2 Modes of origin 12

Plate boundary islands 16Islands in intraplate locations 20

2.3 Environmental changes over long timescales 22Changes in relative sea level—reefs, atolls, and guyots 22Eustatic changes in sea level 24Climate change on islands 26The developmental history of the Canaries, Hawaii, and Jamaica 27

2.4 The physical environment of islands 32Topographic characteristics 32Climatic characteristics 34Water resources 36Tracks in the ocean 37

2.5 Natural disturbance on islands 38Magnitude and frequency 40Disturbance from volcanism and mega-landslides 41

2.6 Summary 45

3 The biogeography of island life: biodiversity hotspots in context 463.1 Introduction: the global significance of island biodiversity 463.2 Species poverty 493.3 Disharmony, filters, and regional biogeography 50

Filtering effects, dispersal limits, and disharmony 50Biogeographical regionalism and the vicariance/dispersalism debate 52Macaronesia—the biogeographical affinities of the Happy Islands 60

3.4 Endemism 63Neo- and palaeoendemism 63Endemic plants 65Endemic animals 67

3.5 Cryptic and extinct island endemics: a cautionary note 713.6 Summary 73

Contents

Page 9: Whittaker. Island Biogeography. 2007

Part II: Island Ecology 75

4 Species numbers games: the macroecology of island biotas 774.1 The development of the equilibrium theory of island biogeography 79

Island species–area relationships (ISARs) 81Species abundance distributions 82The distance effect 83Turnover, the core model (EMIB), and its immediate derivatives 84

4.2 Competing explanations for systematic variation in island species–area relationships 874.3 Island species numbers and ISARs: what have we learnt? 89

Area and habitat diversity 89Area is not always the first variable in the model 90Distance and species numbers 91Species–area relationships in remote archipelagos 92Scale effects and the shape of species–area relationships 93Species–energy theory—a step towards a more complete island species richness model? 97

4.4 Turnover 99Pseudoturnover and cryptoturnover 100When is an island in equilibrium? 101The rescue effect and the effect of island area on immigration rate 102The path to equilibrium 103What causes extinctions? 104

4.5 Summary 106

5 Community assembly and dynamics 1075.1 Island assembly theory 107

Assembly rules 108Incidence functions and tramps 108The dynamics of island assembly 110Chequerboard distributions 111Combination and compatibility—assembly rules for cuckoo-doves 111Criticisms, ‘null’ models, and responses 113Exploring incidence functions 118Linking island assembly patterns to habitat factors 122Anthropogenic experiments in island assembly: evidence of competitive effects? 125

5.2 Nestedness 1265.3 Successional island ecology: first elements 1295.4 Krakatau—succession, dispersal structure, and hierarchies 131

Background 131Community succession 131A dispersal-structured model of island recolonization 134Colonization and turnover—the dynamics of species lists 136The degree of organization in the Krakatau assembly process 141

5.5 Concluding observations 1435.6 Summary 143

viii C O N T E N T S

Page 10: Whittaker. Island Biogeography. 2007

6 Scale and island ecological theory: towards a new synthesis 1456.1 Limitations of the dynamic equilibrium model of island biogeography: a reappraisal 1456.2 Scale and the dynamics of island biotas 147

Residency and hierarchical interdependency: further illustrations from Krakatau 148

6.3 Forms of equilibria and non-equilibria 1506.4 Temporal variation in island carrying capacities 156

The prevalence and implications of intense disturbance events 156Variation in species number in the short and medium term 157Long term non-equilibrium systems 158Implications for endemics? 159

6.5 Future directions 1616.6 Summary 164

Part III: Island Evolution 165

7 Arrival and change 1677.1 Founder effects, genetic drift, and bottlenecks 167

Implications of repeated founding events 1707.2 After the founding event: ecological responses to empty niche space 172

Ecological release 173Density compensation 174

7.3 Character displacement 1767.4 Sex on islands 177

Dioecy and outcrossing 177Loss of flower attractiveness 178Anemophily 178Parthenogenesis 178Hybridization 179

7.5 Peculiarities of pollination and dispersal networks on islands 179The emergence of endemic super-generalists 180Unusual pollinators 181Unusual dispersal agents 181

7.6 Niche shifts and syndromes 181The loss of dispersal powers 182The development of woodiness in herbaceous plant lineages 184Size shifts in island species and the island rule 186Changes in fecundity and behaviour 190The island syndrome in rodents 192

7.7 Summary 194

8 Speciation and the island condition 1958.1 The species concept and its place in phylogeny 1958.2 The geographical context of speciation events 199

Distributional context 199Locational and historical context—island or mainland change? 201

C O N T E N T S ix

Page 11: Whittaker. Island Biogeography. 2007

8.3 Mechanisms of speciation 202Allopatric or geographical speciation 202Competitive speciation 204Polyploidy 205

8.4 Lineage structure 2068.5 Summary 207

9 Emergent models of island evolution 2089.1 Anagenesis: speciation with little or no radiation 2089.2 The taxon cycle 209

Melanesian ants 209Caribbean birds 211Caribbean anoles 215Evaluation 217

9.3 Adaptive radiation 217Darwin’s finches and the Hawaiian honeycreeper-finches 219Hawaiian crickets and drosophilids 225Adaptive radiation in plants 228

9.4 From valley isolates to island-hopping radiations 230Non-adaptive radiation 230Speciation within an archipelago 230Island-hopping allopatric radiations: do clades respond to islands or to habitats? 233Island-hopping on the grand scale 237

9.5 Observations on the forcing factors of island evolution 2389.6 Variation in insular endemism between taxa 2409.7 Biogeographical hierarchies and island evolutionary models 2459.8 Summary 247

Part IV: Islands and Conservation 249

10 Island theory and conservation 25110.1 Islands and conservation 25110.2 Habitats as islands 25210.3 Minimum viable populations and minimum viable areas 253

How many individuals are needed? 253How big an area? 257Applications of incidence functions 257

10.4 Metapopulation dynamics 259The core–sink model variant 261Deterministic extinction and colonization within metapopulations 262Value of the metapopulation concept 263

10.5 Reserve configuration—the ‘Single Large or Several Small’ (SLOSS) debate 263Dealing with the leftovers 266Trophic level, scale, and system extent 266

10.6 Physical changes and the hyperdynamism of fragment systems 268

x C O N T E N T S

Page 12: Whittaker. Island Biogeography. 2007

10.7 Relaxation and turnover—the evidence 26910.8 Succession in fragmented landscapes 27410.9 The implications of nestedness 27510.10 Edge effects 27610.11 Landscape effects, isolation, and corridors 278

The benefits of wildlife corridors 278The benefits of isolation 279Corridors or isolation? 281Reserve systems in the landscape 281Species that don’t stay put 282

10.12 Does conservation biology need island theory? 283A non-equilibrium world? 283Ecological hierarchies and fragmented landscapes 285Climate change and reserve systems 286

10.13 Concluding remarks: from island biogeography to countryside biogeography? 28710.14 Summary 288

11 Anthropogenic losses and threats to island ecosystems 29011.1 Current extinctions in context 29011.2 Stochastic versus deterministic extinctions 29111.3 The scale of island losses globally 29211.4 The agencies of destruction 295

Predation by humans 295Introduced species 295Disease 300Habitat degradation and loss 302

11.5 Trends in the causes of decline 30211.6 A record of passage—patterns of loss across island taxa 305

Pacific Ocean birds and the Easter Island enigma 307Indian Ocean birds 312Reptiles 313Caribbean land mammals 314Island snails 315Plants in peril 316

11.7 How fragile and invasible are island ecosystems? 32011.8 Summary 322

12 Island remedies: the conservation of island ecosystems 32312.1 Contemporary problems on islands 323

Maldives: in peril because of climatic change 323Okino-Tori-Shima: the strategic economic importance of a rocky outcrop 324Nauru: the destruction of an island 324The Canaries: unsustainable development in a natural paradise 325Contemporary problems in the Galápagos: a threatened evolutionary showcase 328

C O N T E N T S xi

Page 13: Whittaker. Island Biogeography. 2007

12.2 Some conservation responses 329Biological control—a dangerous weapon? 331Translocation and release programmes 331Protected area and species protection systems: the Canarian example 333

12.3 Sustainable development on islands: constraints and remedies 33612.4 Summary 340

Glossary 342

References 351

Index 383

xii C O N T E N T S

Page 14: Whittaker. Island Biogeography. 2007

PART I

Islands as Natural Laboratories

The basaltic massif of Teno is one of the oldest parts of theisland of Tenerife (Canary archipelago), dating back about8 million years. It is only within the last 1.5 Ma or so that thegreat acidic volcanic cycle of Las Cañadas unified the oldbasaltics massifs of Teno, Adeje and Anaga to form what weknow today as the island of Tenerife.

Page 15: Whittaker. Island Biogeography. 2007

This page intentionally left blank

Page 16: Whittaker. Island Biogeography. 2007

. . . it is not too much to say that when we have masteredthe difficulties presented by the peculiarities of island lifewe shall find it comparatively easy to deal with the morecomplex and less clearly defined problems of continentaldistribution . . .

(Wallace 1902, p. 242)

These words taken from Alfred Russel Wallace’sIsland life encapsulate an over-arching idea thatcould be termed the central paradigm of island bio-geography. It is that islands, being discrete, inter-nally quantifiable, numerous, and varied entities,provide us with a suite of natural laboratories, fromwhich the discerning natural scientist can make aselection that simplifies the complexity of the natu-ral world, enabling theories of general importanceto be developed and tested. Under this umbrella, anumber of distinctive traditions have developed,each of which is a form of island biogeography, butonly some of which are actually about the biogeog-raphy of islands. They span a broad continuumwithin ecology and biogeography, the end points ofwhich have little apparently in common. This bookexplores these differing traditions and the linksbetween them within the covers of a single volume.

This book is divided into four parts, each of threechapters. The first part, Islands as NaturalLaboratories, sets out to detail the properties ofthese natural laboratories, without which we canmake little sense of the biogeographical dataderived from them. The second part, IslandEcology, is concerned with pattern and process onecological timescales, and is focused on propertiessuch as the number and composition of species onislands and how they vary between islands andthrough time. The chapters making up the thirdpart, on Island Evolution, focus on evolutionary

pattern and process, at all levels from the instanta-neous loss of heterozygosity associated with thecolonizing event on an island, through to the muchdeeper temporal framework associated with thegreat radiations of island lineages on remoteoceanic archipelagos like Hawaii. The final sectionof the book, Islands and Conservation, incorporatestogether two contrasting literatures, concernedrespectively with the threats to biodiversity derivedfrom the increased insularization of continentalecosystems, and the threats arising from the loss ofinsularization of remote islands.

We cannot begin our investigation of islandecology, evolution, and conservation problems,until we have explored something of the origins,environments, and geological histories of theplatforms on which the action takes place. Thisforms the subject matter of the second chapter,Island environments. There are many forms of‘islands’ to be found in the literature, from individ-ual thistle plants (islands of sorts for the arthropodsthat visit them) in an abandoned field, through toremote volcanic archipelagos like the Galápagosand Hawaiian islands (Fig. 1.1). The former arevery temporary islands in an ecological context,and in practice, individual volcanic islands are alsoquite short-lived platforms when judged in anevolutionary time frame. The extraordinaryenvironmental dynamism of these remoteplatforms requires much greater attention than ithas hitherto received in the island evolutionaryliterature. Consider the following example. TheJuan Fernández archipelago consists of two mainislands, Masatierra, an island of 48 km2 and 950 mheight, lying some 670 km from mainland Chile,and Masafuera, an island of 50 km2 and 1300 m

3

CHAPTER 1

The natural laboratory paradigm

Page 17: Whittaker. Island Biogeography. 2007

height, lying a further 180 km further out in thePacific. We might be interested in analysing howbiological properties of these islands, such as speciesrichness and endemism, relate to properties such asisland area, altitude, and isolation. However, whenMasatierra first arose from a hotspot on the Nazcaplate some 4 Ma (millions of years ago), it probablyformed an island of some 1000 km2 and perhaps3000 m in height (Stuessy et al. 1998). Since then, ithas been worn down and diminished in area by sub-aerial erosion, wave action, and subsidence, losingboth habitats and species in the process. Modellingefforts that ignore this environmental history wouldbe liable to produce quite misleading accounts offactors controlling biotic diversification on suchislands (Stuessy et al. 1998).

The third chapter, on The biogeography of islandlife, concentrates on the biogeographical affinitiesand peculiarities of island biotas—a necessary stepbefore the processes of evolution on islands aretackled. The chapter thus begins in the territory ofhistorical biogeographers concerned with tracing

the largest scales from the space–time plot (Fig. 1.2)and with working out how particular groups andlineages came to be distributed as they are. This ter-ritory has been fought over by the opposing schoolsof dispersalist and vicariance biogeography. Islandstudies have been caught up in this debate in partbecause they seemingly provide such remarkableevidence for the powers of long-distance dispersal,whereas rejection of this interpretation requiresalternative (vicariance) hypotheses for the affinitiesof island species, invoking plate movementsand/or lost land bridges, to account for the break-ing up of formerly contiguous ranges. Some of thepostulated land-bridge connections now appearhighly improbable; nevertheless the changingdegrees of isolation of islands over time remains ofcentral importance to understanding the biogeog-raphy of particular islands. As will be seen, the vic-ariance and dispersalist hypotheses have been putinto too stark an opposition; both processes havepatently had their part to play (Stace 1989; Keastand Miller 1996).

4 T H E N AT U R A L L A B O R ATO RY PA R A D I G M

Figure 1.1 There are many different types of islandsin addition to those found in the world’s oceans. Thisfigure illustrates just a few of these (based on anoriginal in Wilson and Bossert 1971).

Page 18: Whittaker. Island Biogeography. 2007

Remote island biotas differ from those of con-tinents in a number of ways, being generally species-poor and disharmonic (peculiar in taxonomiccomposition), yet rich in species found nowhereelse, i.e. endemic to those islands. Although manyof these features are well known through the morepopular illustrations from island groups such as theGalápagos, the global significance of island biodi-versity is not so well appreciated. Data for a num-ber of taxa are available that unequivocallydemonstrate that islands, particularly large andremote islands, contribute disproportionately toglobal biodiversity, i.e. they are biodiversity‘hotspots’. Island biogeographers have, in recentdecades, continued to discover ‘new’ species onislands, some from extant (i.e. living) populations,others from fossils or subfossils. These new discov-eries require a reappraisal of island biogeographi-cal ideas and models. They also underline theaccelerated rate of attrition of island biotas throughhuman action, which qualifies many islands todayas ‘threatspots’ as well as centres of endemism.

Having explored the properties of the laborato-ries, the second and third parts of the book examinethe ecological and evolutionary insights providedby islands, moving from the finer spatial and tem-poral frames of reference, to the coarser scales overwhich evolutionary change operate (Fig. 1.2).Table 1.1 picks out some of the dominant islandtheories or themes and the typical island configura-tions that, as will become clear later, appear tomatch these themes (cf. Haila 1990). The first two,adaptive radiation and the taxon cycle, are eachforms of evolutionary change: typically, the bestexamples of island evolution are from very isolated,high islands. The last three themes fall within whatmay be termed island ecology. Island ecologicaltheories have applications to ‘habitat islands’within continental land masses as well as to realislands and, indeed, form an important part of thecontribution of biogeography to the problems ofconservation science (conservation biogeographysensu Whittaker et al. 2001).

The separation of ecology from evolution is arather arbitrary one: evolution cannot work withoutecology to drive it; ecological communities operatewithin the constraints of evolution. Nonetheless, wefind it helpful to distinguish theory that treats speciesas essentially fixed units of analysis as ‘island ecol-ogy’, and this we consider within the three chaptersof Part II. The first is an analysis of the macroecologyof islands (Chapter 4), focused on what we light-heartedly term Species numbers games. Undoubtedly,the most influential contribution to this literature hasbeen Robert H. MacArthur and Edward O. Wilson’s(1967) The theory of island biogeography. MacArthur

T H E N AT U R A L L A B O R ATO RY PA R A D I G M 5

Populations:demography,

dynamics

Individualbehaviour

Space

Tim

e

Evolutionary and biogeographic

dynamics

(4)(3)

(2)(1)

Figure 1.2 A scheme of different time–space scales of ecologicalprocesses and criteria that define corresponding scales of insularity.(1) individual scale; (2) population scale 1: dynamics; (3) populationscale 2: differentiation; (4) evolutionary scale. (Redrawn from Haila1990, Fig. 1.)

Table 1.1 Some prominent island biogeographical theories and thegeographical configurations of islands for which they hold greatestrelevance

Type of archipelago Prominent theories

Large, very distant Adaptive RadiationLarge, distant Taxon CycleMedium, mid-distance Assembly RulesSmall, near Equilibrium Model of Island

BiogeographySmall, very near Metapopulation dynamics

Page 19: Whittaker. Island Biogeography. 2007

and Wilson were not the first to recognize or theorizeabout the species–area effect—that plots of speciesnumber versus area have a characteristic form, dif-fering between mainland and island biotas—but itwas they who developed a fully fledged theory toexplain it: the equilibrium theory of island bio-geography, based on a dynamic relationshipbetween forces accounting for immigration toislands (afforced on remote islands by speciation)and losses of species from them.

Whether on islands or continents, the form of therelationship between area and species number is offundamental biogeographical importance. Over thepast half century, there have been numerous islandbiogeographical studies that have set out todescribe and understand the form of thespecies–area curve, and the influence of numerousother variables on species richness. What havethese studies shown? Are we any closer to under-standing the factors that control species richness inpatches and islands, and how they vary in signifi-cance geographically? We will show thatMacArthur and Wilson’s equilibrium theory hasactually proved remarkably difficult to test and thatwhile its heuristic influence remains strong, its pre-dictive value appears limited. In large measure thismay be because of insufficient attention to the scaleof the study system.

The study of Community assembly and dynamics onislands (Chapter 5) has interested natural scien-tists since at least the mid-nineteenth century,which explains why the botanist Melchior Treubtook the opportunity to begin monitoring the recol-onization of the Krakatau islands just 3 years afterthey were sterilized in the eruptions of 1883(Whittaker et al. 1989). It emerges that island biotastypically are not merely a ‘random’ sample of main-land pools. There is generally some degree of struc-ture in the data. Important insights into theprocesses producing such structure have comefrom Jared Diamond’s work on birds on islands offthe coast of New Guinea. Diamond (1975a) formu-lated a set of ‘assembly rules’ based largely on dis-tributional data. In this work he invoked a strongrole for interspecific competition in structuringecological assemblages on islands, but he alsorecognized a role for long-term ecological (and

evolutionary–ecological) processes. As we will seelater, this approach became embroiled in heatedcontroversy, focused both on our ability to measurepattern and on the causal interpretation of thepatterns detected. Other approaches have sincebeen developed that aim to detect and interpretstructure in island assemblages. One importantform of structure is nestedness, the term given tothe situation where upon ranking a set of islandspecies lists by species richness, each species set isfound to represent a proper subset of the next largerspecies set. Again, the measurement of nestednessand its causal interpretation turn out to be less thanstraightforward, but it appears clear that a signifi-cant tendency to nestedness is common withinisland (and habitat island) archipelagos.

Much of the research on island assembly hasbeen based around ‘snapshot’ data, capturing dis-tributions at a particular point in time. However,the dynamics of island biotas have also been thefocus of study and debate. For instance, Diamonddeveloped the idea from his New Guinea studiesthat some species are ‘supertramps’, effective at col-onizing but not at competing in communities atequilibrium, whereas others are poor colonists buteffective competitors. This is a form of successionalstructuring of colonization and extinction.Successional structure can be recognized throughother approaches, for instance by analysis of thedispersal characteristics of the flora of the Krakatauislands through time, or by analysis of the habitatrequirements of birds and butterflies appearing onand disappearing from the same islands. Krakatauis unusual in having species–time data for a varietyof taxa, and to some extent they tell different sto-ries, yet stories linked together by hierarchical linksbetween taxa and by the emergent successionalproperties of the system.

Within this book, a recurring theme is the impor-tance of scale, and we open the final chapter of thissection, on Scale and island ecological theory, witha consideration of the scale of relevance of theoreti-cal constructs such as the MacArthur–Wilsonmodel. There has been an increasing realizationthat ecological phenomena have characteristic spa-tial and temporal signatures, which tend to belinked (Delcourt and Delcourt 1991; Willis and

6 T H E N AT U R A L L A B O R ATO RY PA R A D I G M

Page 20: Whittaker. Island Biogeography. 2007

Whittaker 2002). Placing island ecological studiesinto a scale framework, helps us reconcile appar-ently contradictory hypotheses as actually beingrelevant to different spatio-temporal domains(Fig. 1.2, Haila 1990). We suggest that alongsideislands conforming to the dynamic, equilibriumMacArthur–Wilson model, some islands may showdynamic but non-equilibrial behaviour, whileother island biotas persist unchanged for substan-tial period of time, constituting ecologically ‘static’systems that can again be regarded as either equi-librial or non-equilibrial. Island ecological theoriesshould also be capable of accommodating the hier-archical links that form within ecological systems(e.g. with respect to feeding relationships, pollina-tion, dispersal, etc): this represents a considerablechallenge given a literature hitherto focused mostlyon pattern within a single trophic level, or majortaxon (mammals, birds, ponerine ants, etc.).

The three chapters in Part III, Island Evolution,set out to develop the theme of island evolutionfrom the micro-evolutionary changes followingfrom initial colonization through to the full array ofemergent macro-evolutionary outcomes. InChapter 7, Arrival and change, we start with thefounding event on an island and work in turnthrough the ecological and evolutionary responsesthat follow from the new colonist encountering thenovel biotic and abiotic conditions of the island. Aseries of traits, syndromes, and properties emergeas characteristic of islands, including loss of disper-sal powers, loss of flower attractiveness, the devel-opment of woodiness, characteristic shifts in bodysizes of vertebrates, and rather generalist pollina-tion mutualisms. These shifts are detectable tovarying degrees of confidence, and in a wide vari-ety of taxa, sometimes but not always involvingattainment of endemic species status by the islandform.

The attainment of speciation, the emergence ofseparate species from one ancestor, is of centralinterest within biology, and so we devote Chapter8, Speciation and the island condition, to a briefaccount of the nature of the species unit, and of thevarying frameworks for understanding the processof speciation. Thus, we examine first the geograph-ical context of speciation, in which islands

conveniently provide us with a strong degree ofgeographical separation between the originalsource population and the island theatre, but inwhich the extent of intraarchipelago and intrais-land isolation is often harder to discern. Second, weexamine the various mechanistic frameworks forunderstanding speciation events, and, finally, weconsider phylogenetic frameworks for describingthe outcome of evolutionary change. Having devel-oped these frameworks, we go on to look at theEmergent models of island evolution (Chapter 9),which provide descriptions and interpretations ofsome of the most spectacular outcomes of islandevolutionary change, including the classic exam-ples of the taxon cycle and of adaptive radiation.However, as more studies become available of phy-logenetic relationships within lineages distributedacross islands, archipelagos, and even whole oceanbasins, new ways of looking at island evolution areemerging. From this work it is becoming increas-ingly evident that the ever-changing geographicalconfiguration and environmental dynamism ofoceanic archipelagos are crucial to understandingvariation in patterns and rates of evolutionarychange on islands.

Anthropogenic changes to biodiversity are evi-dent across the planet, and we may turn to islandsfor several important lessons if we care for thefuture of our own human societies (Diamond 2005).In the final section of the book, Islands and Conser-vation, we start with the theme of Island theory andconservation (Chapter 10), which debates thecontribution of island ecological thinking withinconservation science. As we turn continents into apatchwork quilt of habitats, we create systems ofnewly insularized populations. What are the effectsof fragmentation and area reduction within conti-nents? Not just the short-term changes, but thelong-term changes?

As concern over this issue grew in the 1970s and1980s, the obvious place to derive such theory wasfrom the island ecological literature, and in particu-lar the dynamic equilibrium model put forwardby MacArthur and Wilson (1967). We review thisliterature, showing that although we can oftendevelop good statistical models for the behaviourof particular species populations, and for how

T H E N AT U R A L L A B O R ATO RY PA R A D I G M 7

Page 21: Whittaker. Island Biogeography. 2007

species numbers may change in fragments follow-ing isolation, we still have a poor grasp of the over-all implications for diversity on a regional scale ofthe ongoing processes of ecosystem fragmentation.Nonetheless, it is clear that habitat loss and frag-mentation threatens many species with local,regional and ultimately global extinction. It is alsoevident that the processes of ecosystem responsecan take considerable periods (often decades) toplay out, meaning that today’s fragmentation isstoring up numerous extinctions for the future.However, taking a more positive approach to theproblem, the fact that responses are often laggedmeans that there is an opportunity for mitigationmeasures to be put in play, to reduce the so-called‘extinction debt’, providing society cares enough toact. Insofar as island effects are driving specieslosses, the key is to prevent previously contiguousand extensive ecosystems from becoming too iso-lated: reserve systems need to be embedded inwildlife friendly landscapes wherever possible.

If the catalogue of known extinctions is drawn upfor the period since AD 1600, it can be seen that for

animal taxa that are relatively well known (e.g.mammals, birds, and land snails), the majority oflosses have been of island species (Fig. 1.3). Today,some of the greatest showcases for evolution are inperil, with many species on the verge of extinction.Why is this? These issues and some conservationresponses are explored in the final two chapters ofthe book, Anthropogenic losses and threats, and Islandremedies. There is mounting evidence thathumans are repeat offenders when it comes toextinguishing island endemics. Wherever we havecolonized islands, whether in the Pacific, theCaribbean, the Atlantic, the Indian Ocean, or theMediterranean, we have impacted adversely onthe native biota, and often on the ecosystem servi-ces on which we ourselves rely (Diamond 2005).Here we ask the question: are islands inherentlyfragile, or are island peoples peculiarly good atextinguishing species? In some senses, island biotasare indeed fragile, but very often the demise ofendemic taxa can be traced to a series of ‘hardknocks’ and the so-called synergistic interactionsbetween a number of alien forces, such that the

8 T H E N AT U R A L L A B O R ATO RY PA R A D I G M

Num

ber

of s

peci

es

100

80

60

40

20

01600–1629

1630–1659

1660–1689

1690–1719

1720–1749

1750–1779

1780–1809

1810–1839

1840–1869

1870–1899

1900–1929

1930–1959

1960–

Islands

Continents

No date

Figure 1.3 Time series of extinctions of species of molluscs, birds, and mammals from islands and continents since about AD 1600.(Redrawn from Groombridge 1992, Fig. 16.5.)

Page 22: Whittaker. Island Biogeography. 2007

nature of the final demise of the endemic may beunrelated to the driver(s) of the initial range col-lapse. Typically, there have been at least two majorwaves of extinctions, one associated with the abo-riginal or prehistoric human colonizations, and thesecond following contact with modern Europeansocieties.

Whereas, on the continents, the increasing insular-ization of habitats is a prime cause for concern, theproblem for oceanic island biotas today is the increas-ing breakdown of the insularization of theirs.The prime agents of destruction are introduced

exotic species (especially mammalian browsers andpredators), habitat loss, predation by humans, andthe spread of disease. The forces involved are nowwell established, and many of the biological manage-ment options are firmly founded in experience, butattaining the twin goals of sustainable developmentand management for conservation requires attentionto fields largely outside the scope of the presentvolume, concerning the politics, economics, sociology,and culture of islands and island states: continentalsolutions may be poorly transferable to islands.

What is an island, anyhow?

T H E N AT U R A L L A B O R ATO RY PA R A D I G M 9

Page 23: Whittaker. Island Biogeography. 2007

. . . from cowpats to South America, it is difficult to seewhat is not or at some time has not been an island.

(Mabberley 1979, p. 261)

The islands move horizontally and vertically and therebygrossly modify the environment on and around them. Lifeforms too must evolve, migrate, or become extinct as theland changes under them.

(Menard 1986, p. 195)

2.1 Types of islands

Islands come in many shapes and sizes, and theirarrangement in space, their geology, environments,and biotic characteristics are each extremely vari-able, simply because there are vast numbers ofthem in the world. If this makes them marvellousexperimental laboratories for field ecologists andbiogeographers, it also means that attempting tomake generalizations about islands holds the dan-ger—almost the certainty—that one will be wrong!

Defining the term ‘island’ for the purposes ofthis book is not as straightforward as it seems.Many biogeographical studies treat isolatedpatches of habitat as de facto islands, but if we takethe simplest dictionary definition then an islandis a ‘piece of land surrounded by water ’. Thisseems simple enough, yet some authors regardland areas that are too small to sustain a supply offresh water as merely beaches or sand bars ratherthan proper islands: the critical size for a supplyto be maintained being about 10 ha (Huggett1995). At the other end of the scale, the distinctionbetween a continent and an island is also fuzzy, inthat larger islands assume many of the character-istics of continents. Indeed, Australia as an entity

is considered an island continent and is rarelytreated in biogeographical analysis as being anisland (for an exception, see Wright 1983).

Where ought the line to be drawn? If NewGuinea is taken as the largest island—somewhatarbitrarily, given that many would considerGreenland an island (in fact it is a set of threeislands unified by an icecap)—then islands in thesea constitute some 3% of the Earth’s land area(Mielke 1989). Much of island biogeography—andof this book—is concerned with islands signifi-cantly smaller than New Guinea, and commonlytreats such substantial islands as New Guinea andAustralia merely as the ‘mainland’ source pools. Tosome extent this reflects a distinction betweenisland evolutionary biogeography, mostly con-cerned with larger (and typically oceanic) islands,and island ecological biogeography, mostly con-cerned with other types of island.

For present purposes we may divide islandsinto two broad types: true islands, being landwholly surrounded by water; and habitat islands,being other forms of insular habitat, i.e. discretepatches of habitat surrounded by strongly con-trasting habitats (Table 2.1). True islands can inturn be subdivided into island continents(Australia), oceanic islands, continental frag-ments, continental shelf islands, and islands inlakes or rivers.

● Oceanic islands are those that have formed overoceanic plates and have never been connected tocontinental landmasses.

● Continental fragments are those islands that bytheir location would pass for oceanic islands but interms of their origin are actually ancient fragments

10

CHAPTER 2

Island environments

Page 24: Whittaker. Island Biogeography. 2007

of continental rock stranded out in the oceans byplate tectonic processes.

● Continental shelf islands are those islandslocated on the continental shelf. Many of theseislands have been connected to mainland duringthe Quaternary ice ages (formally, the last 1.8 million years), although cooling actually beganearlier), as these were periods of significantly lowersea levels. The most recent period of connection forthese so called ‘land-bridge’ islands ended follow-ing the transition from the Pleistocene into theHolocene. The Holocene began about 11 500 yearsago, but seas took several thousand years to rise totheir present levels.

● Finally, islands occurring within freshwaterbodies, both lakes and large rivers, are likely to bemore comparable to islands in the sea than to habi-tat islands, and hence can also be considered ‘true’islands.

Habitat islands are essentially all forms of insularsystem that do not qualify as being ‘real islands’.This means, for terrestrial systems, discrete patches

of a particular habitat type surrounded by a matrixof strongly contrasting terrestrial habitats. Foraquatic systems we can consider similarly discretehabitat types separated by strongly contrastingaquatic environments (e.g. shallow benthic envi-ronments isolated by deep water) also to constitutehabitat islands. Both habitat and lake islands can, ofcourse, occur within the true islands. Moreover, amore refined classification of marine islands has torecognize that there are, for example, many formsand ages of continental shelf islands, and that thereare anomalous islands midway in characterbetween oceanic and continental.

The literature and theory of island biogeographyhas been built up from a consideration of all forms ofisland, from thistle-head habitat islands (Brown andKodric-Brown 1977) to Hawaii (e.g. Wagner andFunk 1995). True islands have the virtue of havingclearly defined limits and properties, thus providingdiscrete objects for study, in which such variables asarea, perimeter, altitude, isolation, age, and speciesnumber can be quantified with some degree of objec-tivity, even if these properties can vary hugely over

T Y P E S O F I S L A N D S 11

Table 2.1 A simple classification of island types distinguishing: (1) classic types of ‘real’ island, being land surrounded by open water, from(2) habitat islands, for which the contrast between the ‘island’ and the surrounding matrix is less stark but still sufficient to represent a barrier orfilter to population movements. Australia, given its huge size, is essentially continental in character and in practice is not treated as an island inthe present work

Type of island Examples

Land surrounded by waterIsland continent AustraliaOceanic islands Hawaii, CanariesContinental fragments Madagascar, New CaledoniaContinental shelf islands British Isles, NewfoundlandIslands in lakes or rivers Isle Royale (Lake Superior), Barro Colorado island (Lake Gatún),

Gurupá island (River Amazon)Habitat islands

Patches of a distinct terrestrial habitat isolated by a hostile matrix Great-Basin (USA) mountain tops surrounded by desert

Woodland fragments surrounded by agricultural landThistle heads in a fieldContinental lake (Baikal, Titicaca)

Marine habitat islands The fringing reef around an isolated oceanic islandCoral reefs separated from other reefs by stretches of seawater Seamounts (submerged or not yet emerged mountains below sea level) Guyots (submerged flat-topped former islands, i.e. a type of seamount)

Page 25: Whittaker. Island Biogeography. 2007

the lifespan of an island (Stuessy et al. 1998). In con-trast, habitat islands exist typically within complexlandscape matrices, which are often rapidly anddramatically changing over just a few years. Matrixlandscapes may be hostile to some but not all speciesof the habitat islands, and isolation is thus of a dif-ferent nature from islands in the sea. It cannot neces-sarily be assumed that what goes for real islands alsoworks for habitat islands and vice versa.

As much of this book is concerned almost exclu-sively with islands in the sea, their biogeographicalpeculiarities, and their problems, it is important toconsider their modes of origin, environmental char-acteristics, and histories. These topics form the bulkof this chapter.

2.2 Modes of origin

Are islands geologically distinctive compared withmost of the land surface of the Earth? If we take thisquestion in two parts, continental shelf islands area pretty mixed bag, a reflection of their variedmodes of origin, whereas oceanic islands are fairlydistinctive geologically, being generally composedof volcanic rocks, reef limestone, or both (Darwin1842; Williamson 1981). The analysis that follows isintentionally a simplified account of an extremelycomplex reality.

The development of the theories of continentaldrift and, more recently, plate tectonics, has revolu-tionized our understanding of the Earth’s surfaceand, along with it, our understanding of the distri-bution and origins of islands. According to the lattertheory, the Earth’s surface is subdivided into someseven major plates, each larger than a continent, anda number of smaller fragments (Fig. 2.1). With theexception of the Pacific plate and its subordinates(Nazca, Cocos, Juan de Fuca, and Philippines), theplates themselves are typically made up of twoparts, an oceanic part and a continental part. The sil-icon/aluminium-rich granitic parts of the platesare of relatively low density and these form thecontinental parts, supporting the continents them-selves (consisting of a highly varied surface geol-ogy), extending to about 200 m below sea level(Fig. 2.2). This zone, from 0 to �200 m, forms thecontinental shelf and supports islands such as the

British Isles and the Frisian Islands (off Germany,the Netherlands, and Denmark), typically involvinga mix of rock types and modes of formation, suchthat any combination of sedimentary, metamorphic,or igneous rocks may be found. As Williamson(1981) noted, about the only generalization that canbe made is that the geological structures of conti-nental shelf islands tend to be similar to parts of thenearby continent.

In places, continental plate can be found at muchgreater depths than 200 m below sea level, and canthen be termed sunken continental shelf. Islandson these sections of shelf (ancient continentalislands sensu Wallace: Box 2.1) are thus formed ofcontinental rocks; examples include Fiji and NewZealand (Fig. 2.3; Williamson 1981). Typically, how-ever, there is a steeply sloping transition zone fromshallow continental shelf down to c.2000 m ormore, where the basaltic part takes over and thetrue oceanic islands occur. These are all volcanic inorigin, although in certain cases they may be com-posed of sedimentary material, principally lime-stones, formed as the volcanic core has sunk belowsea level. True oceanic plate islands have neverbeen attached to a greater land mass. They maygrow through further volcanism, or subside, erode,and disappear. In a geological sense, they tend to betransitory: some may last only a few days, otherssome millennia; relatively few last tens of millionsof years. Thus, in addition to the islands that sus-tain terrestrial biota today, there are also a largenumber of past and future islands, or seamounts,found at varying depths below sea level. Flat-topped seamounts, formed through the submer-gence of limestone-topped volcanoes, are calledguyots, after the Swiss geologist Arnold Guyot(Jenkyns and Wilson 1999).

Islands may originate by existing areas of landbecoming separated from other land masses towhich they were formerly connected, by erosion, orby changes in relative sea level from a variety ofcauses. Many islands have originated from volcan-ism associated with plate movements. The form thisvolcanism takes and the pattern of island genesisinvolved, depend crucially on the nature of thecontact zone between plates, that is whether theplates are moving apart, moving towards each

12 I S L A N D E N V I R O N M E N T S

Page 26: Whittaker. Island Biogeography. 2007

other, or moving past each other laterally. Relativelyfew islands are ancient biogeographically; somehave experienced a complex history of both lateralmovements and alternating emergence and submer-gence, with profound importance to the resultingbiogeography, not just of the particular island inquestion, but also of the wider region (Keast andMiller 1996).

Plate tectonic processes give rise to islands bythree main means. First, the breaking away of

pieces of continent by sea-floor spreading hascarried New Zealand, Madagascar, and a few otherancient islands off into isolation (Box 2.2). Second,in connection with plate boundaries, volcanicislands may arise to form an archipelago of islands,as in the case of the Greater and Lesser Sundas—themany islands of the Indonesian region. Third, vol-canic islands may arise from hotspots (e.g. theHawaiian islands) and certain parts of mid-oceanridges. Hawaii is of hotspot origin, and Iceland is

M O D E S O F O R I G I N 13

Trench

Sea level

Asthenosphere

Trench

(a)

mantle

Volcanic arc

Convergent plate

boundary

Divergent plate boundary

(mid-ocean ridge)Convergent

plate boundary

(b)

Upper

Eurasianplate

Pacificplate Cocos

Nazcaplate

Antarctic plate

South

Eurasianplate

Africanplate

NorthAmerican

plate

90°N90°E 120°E 150°E 150°W 120°W 90°W 60°W 30°W 30°E 60°E 90°E0°180°

60°N

30°N

30°S

60°S

75°S

Indo-Aust.plate

Americanplate

Continentalcrust

Indo-Aust.plate

Upper mantle

Oceanic crustOceanic

crust

Figure 2.1 Model and map of crustal plates. (a) The basic plate tectonics model, exemplified by the South Pacific. New oceanic crust and uppermantle are created along the divergent plate boundary and spread east and west from here. The eastward moving plate (the Nazca plate) issubducted beneath the continental lithosphere of South America. The westward movement of the much larger Pacific plate ends with subductionbeneath the oceanic lithosphere of the Indo-Australian plate. At both trenches, the subducted plate begins to melt when it reaches theasthenosphere. (b) The major plates of the world. Divergent plate boundaries (mid-ocean ridges), at which plates move apart, are represented byparallel lines. Convergent plate boundaries (mostly marked by trenches) are represented by lines with teeth along one side: the teeth point fromthe downward moving plate towards the overriding or buoyant plate. Transverse plate boundaries are shown by solid lines. Broken lines indicateboundaries of an uncertain nature. (Redrawn from Nunn 1994, Fig. 2.4.)

Page 27: Whittaker. Island Biogeography. 2007

14 I S L A N D E N V I R O N M E N T S

12° 10° 8° 6° 4° 2°

Longitude West

Atlantic Irish shelf Ireland Britain North Sea

Met

res

1000

1000

2000

3000

0

NorthChannel

Figure 2.2 A section across the British Isles at 55�N. The section passes through Londonderry in Northern Ireland, Cairnsmore of Fleetin Scotland, and Newcastle upon Tyne in England. (Redrawn from Williamson 1981, Fig. 1.2.)

Box 2.1 Islands in the ocean: an appraisal of Alfred Russel Wallace’s classification

In his seminal book entitled Island life (3rd edition1902, first published 1880) Alfred Russel Wallaceoffered a first classification of ‘true’ islands,according to their geological origin and biologicalproperties. Although derived long before thedevelopment of plate tectonic theory and at atime where scientists were only just beginning tounderstand the importance of the glaciationevents, his classification, illustrated in the tablebelow, remains fundamental.

● Continental islands (or recent continentalislands sensu Wallace) constitute emergent frag-ments of the continental shelf, separated from thecontinents by narrow, shallow waters. This separa-tion is often recent, as a consequence of postglacialsea-level rise (c.130 m), which has isolated thespecies that were already on these islands from theirmainland conspecifics. Pronounced sea-level changeconnected with glaciation has occurred repeatedlyduring the Pleistocene, resulting in reduced isolationof all continental shelf islands and, for many, joiningthem to the adjacent continental land mass. In suchcases, they are termed land-bridge islands, andtheir effective age as an island is some 10 000 yearsor less. They will remain islands only until the nextglaciation event, perhaps in another 10 000 years orso. Because of their origin, continental islands are

very similar in geological and biological features tothe continents.

● Continental fragments or micro-continents(ancient continental islands sensu Wallace): oncepart of the continents, tens of millions of years ago,tectonic drift started to separate these fragmentsfrom the mainland, with the species they carried.Now the waters between them and the continentsare wide and deep and the long isolation hasallowed both the persistence of some ancient line-ages and the development of new species in situ.These continental fragments lose their status onlywhen, after tens of millions of years of drift, theyfinally collide with a different continent, forming anew peninsula. This was the case for the Indian sub-continent, which split from the Gondwanalandsupercontinent during the Early Cretaceous(c.130 Ma) and began a lonely trip northwards,finally colliding with Eurasia c.50 Ma, in the processforming the Himalayan range (Stanley 1999;Tarbuck and Lutgens 2000).

● Oceanic islands originate from submarinevolcanic activity, mostly with basaltic foundations.Oceanic islands have never been connected to thecontinents and so are populated initially by speciesthat have dispersed to the islands from elsewhere—subsequently enriched by speciation. They form in

Page 28: Whittaker. Island Biogeography. 2007

M O D E S O F O R I G I N 15

Examples of islands corresponding to the three types identified by Wallace, from different oceans of the world

Ocean Continental Continental Oceanic islands fragments islands

Arctic Svalbard IcelandNovaya Zemlya Jan MayenBaffinEllesmere

North Atlantic Britain AzoresIreland MadeiraNewfoundland Canaries

Cape VerdeMediterranean Elba Balearic Archipelago Santorini

Rhodos Corsica–Sardinia EolieDjerba Sicily

CreteCyprus

Caribbean Trinidad Cuba MartiniqueTobago Jamaica Guadeloupe

Hispaniola MontserratPuerto Rico Antigua

South Atlantic Falkland South Georgia AscensionTierra del Fuego St Helena

Tristan da CunhaSouth Sandwich

Indian Zanzibar Madagascar RéunionSri Lanka Seychelles MauritiusSumatra Kerguelen Saint PaulJava Socotra Diego García

North Pacific Vancouver AleutianQueen Charlotte KurilSt Lawrence HawaiiSakhalin Marianas

Central and South Pacific Borneo New Zealand GalápagosNew Guinea New Caledonia Society IslandsTasmania MarquesasChiloé Pitcairn

NB: Japan and the Philippines are good examples of islands with a mixed (continental–oceanic) origin

connection with plate boundaries, and in certaincircumstances, within a tectonic plate, leading todifferent volcanic island types. There may today besome 1 million submarine volcanoes, from whichonly several thousands have been able to reach thesea level to form volcanic islands (Carracedo 2003).

Volcanic islands typically are short-lived and evensubstantial ones may only exist for a few millionyears before subsiding and eroding back into theocean. Where sea temperature is adequate oceanicislands may persist despite subsidence and erosion,through the formation of coralline rings, or atolls.

Page 29: Whittaker. Island Biogeography. 2007

16 I S L A N D E N V I R O N M E N T S

thought to have been formed by both mid-oceanridge and hotspot activity.

Setting aside islands of continental origin for themoment, a two-tier classification relating specifi-cally to oceanic islands has been proposed by Nunn(1994). At the first level are the plate boundary andintraplate island types, each of which may also besubdivided into a number of major types, based ontheir geographical configuration in relation to theplate boundaries (Table 2.2). The recognition ofthese distinctive configurations of islands is a help-ful starting point. However, Nunn (1994) warnsagainst an over-simplistic use of the classification,pointing out that, whereas many islands with acommon origin may be found in proximity to eachother, islands of a particular geographical group donot necessarily share a common origin (this is wellillustrated by the Mediterranean islands; Schüle1993). A brief account of this classification (drawnlargely from Nunn 1994) now follows.

Plate boundary islands

Islands at divergent plate boundariesDivergent plate boundaries produce islands in tworather different circumstances, along mid-oceanridges and along the axes of back-arc or marginalbasins behind island arcs that are themselvesassociated with convergent plate boundaries.Although divergent plate boundaries are construc-tive areas involving more magma output thanoccurs in any other situation, seamounts, ratherthan islands, tend to form in connection with theseboundaries. This is believed to be because of the rel-ative youthfulness of the seamounts, the likelihoodthat the supply of magma decreases as theseamounts drift away from the plate boundary, andthe increasing depth of the ocean floor away fromthese boundaries. In some cases, mid-ocean ridgeislands may also be associated with hotspots inthe mantle, providing the seemingly exceptional

Continental islands split from a continent by seafloor spreading

Volcanic islands arising fromsubduction zones as part ofan island archipelago

Volcanic islands arising fromhot spots and mid-ocean ridges

Easter island

Taumotu RidgeArchipelago

Line Islandchain

MacDonaldSeamount

NewCaledonia

Norfolk

New Zealand

Stewart

Austral Seamount

chain

MarshallEllice

Islandchain

Emperor Seamount

chain

Hawaiian ridge

Hawaii

Aleutianislands

Kurils

Figure 2.3 Map of the Pacific Ocean, depicting the major modes of origin of islands. NB: New Guinea and Tasmania are continental islands ofthe Pleistocene Sahul Mainland, which included these islands plus Australia. The New Zealand land mass has been physically isolated fromAustralia and Antarctica by up to 2000 km for 60 million years, as a result of Late Cretaceous to Late Palaeocene sea-floor spreading (Pole1994). (Redrawn from Mielke 1989, Fig. 9.1., slightly modified)

Page 30: Whittaker. Island Biogeography. 2007

M O D E S O F O R I G I N 17

Box 2.2 The splitting of Gondwanaland and the debate between vicariant and dispersalistexplanations in island biogeography

The origin of many contemporary continental-fragment islands is directly linked with thebreakup of Gondwanaland, the southernsupercontinent that split apart from Pangaea, thelast unique landmass, some 160 Ma. At the startof the Cretaceous period (c.140 Ma)Gondwanaland was still intact. By the LateCretaceous (c.80 Ma), however, South America,Africa, and peninsular India were already discreteentities (Stanley 1999). It was during theCretaceous that first Madagascar, and later theSeychelles and Kerguelen micro-fragments, begantheir trip to their present isolated geographic

locations in the south-west, north-west, andsouth of the Indian Ocean, respectively. Australiaand New Zealand rifted from Antarctica about100 Ma, and subsequently, New Zealand brokefirst from Australia and then from Antarcticaabout 80 Ma (Lomolino et al. 2005). While otherelements of Gondwanaland drifted toward thelower latitudes, the Antarctic slowly movedpoleward, so that by about 24 Ma it was locatedover the South Pole, triggering the developmentof a great icecap.

Although Kerguelen is too far south (50�latitude) to possess a diverse biota, other ancient

(a)

(c) (d)

(b)

ArabianPeninsula

ArabianPeninsula

ArabianPeninsula

ArabianPeninsula

200 my

130 my 100 my

160 my

Australia

AustraliaAustralia

Australia

Africa Africa

Africa

SouthAmerica

SouthAmerica

SouthAmerica

SouthAmerica

India

India

India

India

AntarcticaAntarctica

Antarctica

New Zealand

New Zealand New Zealand

New ZealandSouth Pole

South Pole South Pole

South Pole

Madagascar

Madagascar

Madagascar

Madagascar

Antarctica

Africa

The stages in the breaking up of Gondwanaland. Bold lines show zones of sea-floor spreading. (From Storey 1995.)

Page 31: Whittaker. Island Biogeography. 2007

18 I S L A N D E N V I R O N M E N T S

continental fragment islands like Madagascar, NewZealand, and New Caledonia, are characterized byhigh endemism and some interesting ancientlineages, which biogeographers attribute to theirancient origins. Yet, the biotas of these ancientfragments of continent may not necessarily be asancient as their rocks suggest. For instance, severalMadagascan mammal taxa such as the Lemurs,the Tenrecidae, and the Viverridae, althoughrelatively primitive, are nonetheless estimated tohave diverged from their ancestral lineages duringthe Tertiary period, roughly between 45 and26 Ma: much later than the separation fromAfrica. McCall (1997) points to geological evidenceas suggesting that areas of the Mozambiquechannel were dry land during this period of theTertiary, and that subsequent subsidence has

opened up what is currently a wide clear waterchannel. His interpretation has been disputed bye.g. Rogers et al. (2000), who contend that at bestthere were merely a few isolated dots of land inthe channel during this period.

The idea of vanished intercontinental land-bridges and island arcs across ocean basins toexplain disjunct distributions has in general beendiscarded in favour of plate-tectonic mechanisms,afforced by the processes of long-distancedispersal known to populate remote oceanicisland groups. However, the evidence of subaerialexposure on a few seamounts within theMozambique channel suggests the possibility thatMadagascar’s mammals may well have used, ifnot a solid land bridge, at least a set of now-vanished stepping stones, in colonizing the island

Rafting Land-bridges

Jump-dispersal Separation by plate tectonic processes

Four hypothetical means of species reaching islands: rafting, land-bridges, jump-dispersal, and the separation of a fragment of continent byplate tectonic processes (sketch by John Holden from Tarbuck, E. J. and Lutgens, K. (1999) Earth: introduction to physical geology, 6th edn,© 1999, p. 472, Fig. 19.4. Reprinted by permission of Pearson Education, Inc., Upper Saddle River, NJ)

Page 32: Whittaker. Island Biogeography. 2007

M O D E S O F O R I G I N 19

long after this continental fragment began itsocean voyaging. Similarly, changing sea level,especially the reductions associated with majorperiods of glaciation, can turn islands intoextensions of continents (such that for lengthyperiods during the Quaternary, Tasmania,Australia, and New Guinea have formed one landmass, called Sahul), while providing additional‘stepping stone’ islands that will have assisted thecolonization of otherwise remote islands.

Debates between explanations for island biotasthat postulate vicariance (the breaking of a pastland connection) and those based on long-distancedispersal continue to generate considerablecontroversy and are sometimes not easily resolved.Perhaps the most celebrated example is that of

New Zealand. The case that New Zealand’s biotaconstitutes a Gondwanan ‘time-capsule’ entirelyexplicable through long isolation of a fullassemblage of species (a vicariance model) has beencontested on the basis of evidence indicating morerecently colonization of many lineages. As McGlone(2005) has put it ‘ . . . for New Zealanders inparticular, abandoning “Time Capsule of the SouthSeas” for “Fly-paper of the Pacific” will be awrench. But it has to happen . . . ‘ And indeedmany features of New Zealand’s biogeography arecomparable with those of true oceanic islandarchipelagos such as Hawaii, strongly suggestingthat much of its biota has colonized after thebreak-up of Gondwanaland, by long-distancedispersal (Pole 1994; Cook and Crisp 2005).

Table 2.2 P. D. Nunn’s (1994) genetic classification of oceanic islands, with examples

Level 1 Level 2 Examples

Plate boundary islands Islands at divergent plate boundaries Iceland, St Paul (Indian Ocean)Islands at convergent plate boundaries Antilles, South Sandwich (Atlantic)Islands along transverse plate Cikobia and Clipperton (Pacific)

boundariesIntra-plate islands Linear groups of islands Hawaii, Marquesas, Tuamotu

Clustered groups of islands Canaries, Galápagos, Cape VerdeIsolated islands St Helena, Christmas Island (Indian Ocean),

Easter Island

conditions necessary for their conversion fromseamounts to large islands. Iceland is the largestsuch example (103 106 km2), being a product of themid-Atlantic ridge and a hotspot that may have beenactive for some 55 million years. Another context inwhich mid-ocean ridge islands can be found is inassociation with triple junctions in the plate system,a notable example being the Azores, where theNorth American, Eurasian, and African plates meet.The second form of divergent plate boundary island,exemplified by the Tongan island of Niuafo’ou, isthat sometimes formed in back-arc basins, whichdevelop as a result of plate convergence, but whichproduce areas of sea-floor spreading (Nunn 1994).

Islands at convergent plate boundariesWhere two plates converge, one is subducted belowthe other. This results in a trench in the ocean floor

at the point of subduction. Beyond the trench, a rowof volcanic islands develops parallel with the trenchaxis on the surface of the upper of the two plates(Fig. 2.4). This mechanism accounts for some of theclassic island arcs in the Pacific and Caribbean. Mostcommonly, where islands are formed in connectionwith subduction zones, the edges of the two con-verging plates are formed exclusively of oceaniccrust. The magma involved in island arc volcanismderives from the melting of the subducted oceancrust, complete with its sedimentary load. Its com-position thus owes much to the nature of the sub-ducted crust. It is believed that the subduction ofbasaltic crust and water-bearing sediment leads toexplosive andesitic volcanism. The Sunda islandarc, including the Indonesian islands, is predomi-nantly of this form. Basaltic volcanism in island arcsis less common, possibly because of a relative

Page 33: Whittaker. Island Biogeography. 2007

paucity of sedimentary load in the subducted crust.The Sandwich arc in the south Atlantic involvesboth basaltic and andesitic volcanism.

Islands along transverse plate boundariesThis is a fairly rare context for island formation as,by definition, less divergence or convergence ofplates is involved. However, strike–slip movement,compression, or both, can occur between adjacentparts of plates. An example of an island believed tohave been produced by these forces is the Fijianisland of Cikobia, in the south-west Pacific.

Islands in intraplate locations

This class of islands contains some of the most bio-geographically fascinating islands of all, notablythe Hawaiian group. They are equally fascinatinggeologically.

Linear island groupsThe Hawaiian chain is the classic example of aseries of islands with a significant age sequencearranged linearly across a plate (Table 2.3, Fig. 2.3).Among the larger islands, Kauai is the oldest atmore than 5 million years, whereas Hawaii itself isless than 1 million years old. Islands older than

20 I S L A N D E N V I R O N M E N T S

Sea level

Volcanicislands

Trench

PLATE A PLATE B

Figure 2.4 Simplified representation of the formation of an island arc, by the subduction of one plate under another (from an original inWilliamson 1981).

Table 2.3 Approximate ages of selected islands and of the Meijiseamount. This table illustrates the wide variation in the ages ofoceanic islands, the simple age sequence of members moving north-west along the Hawaii–Emperor chain, and the lack of asimilarly simple sequence moving south in the Austral–Cook cluster.For fuller listings, error margins, and original sources see Nunn(1994); for Hawaii, in addition see Wagner and Funk (1995)

Island group Age (Ma)

Hawaii–Emperor chainMauna Kea, Hawaii (the Big Island) 0.38Kauai 5.1Laysan 19.9Midway 27.7Meiji Seamount 74.0

Austral–Cook clusterAitutaki 0.7Rarotonga 1.1–2.3Mitiaro 12.3Rimatara 4.8–28.6Rurutu 0.6–12.3

Isolated intra-plate islandsAscension 1.5St Helena 14.6Christmas Island (Indian Ocean) 37.5

Kauai stretch away to the north-west, but throughcombinations of erosion, subsidence, and coralformation, they are now reduced to forming coral

Page 34: Whittaker. Island Biogeography. 2007

atolls, or guyots. The chain extends from the LoihiSeamount, which is believed to be closest to thehotspot centre, through the Hawaiian islands them-selves and along the Emperor Seamount Chain, adistance of some 6130 km (Keast and Miller 1996).The oldest of the submerged seamounts is morethan 70 million years old, so that there must havebeen a group of islands in this part of the Pacific forfar longer than is indicated by the age of the pres-ent Hawaiian islands. This may have allowed forcolonization of the present archipelago from popu-lations established considerably earlier on islandsthat have since sunk. However, recent geologicalfindings indicate that there was a pause in islandbuilding in the Hawaiian chain, and molecularanalyses support the notion that few lineagesexceed 10 million years—that is, the pre-Kauaisignal in the present biota is actually rather limited(Wagner and Funk 1995; Keast and Miller 1996).

The building of linear island groups such as theHawaiian chain is explained by J. T. Wilson’s (1963)hotspot hypothesis, which postulates that ‘station-ary’ thermal plumes in the Earth’s upper mantle liebeneath the active volcanoes of the island chain(Wagner and Funk 1995). A volcano builds over thehotspot, then drifts away from it as a result of platemovement and eventually becomes separated fromthe magma source, and is subject to erosion bywaves and subaerial processes, and to subsidenceunder its own weight. It has been calculated thatthe Hawaiian islands have been sinking at a rate ofapproximately 2.6–2.7 mm a year over the last475 000 years (cited in Whelan and Kelletat 2003).As each island moves away, a new island begins toform more directly over the hotspot. This process isthought to have been operating over some75–80 million years in the case of the Hawaiianhotspot.

The change in orientation of the Hawaiian chain(Fig. 2.3) has been attributed to past changes in thedirection of plate movement about 43 Ma,although recent work has questioned whether infact hotspots are really such fixed reference points,raising the possibility of a combination of platemovement and hotspot migration to explain thedistribution of hotspot island chains (Christensen1999). The Society and Marquesas island groups,also in the Pacific, provide further examples of

hotspot chains, and Nunn (1994) provides a critiqueof several other postulated cases.

Clustered groups of islandsMany island clusters, once regarded as hotspotisland chains that had become slightly less regularthan the classic examples, have since been realizedto differ significantly from the hotspot model, suchthat different groups and different islands withinthe groups may require distinctive models. The tec-tonic-control model, attributed to Jackson et al.(1972), postulates that, instead of lying along a sin-gle lineation, the islands lie along shorter lines ofcrustal weakness (termed en echelon lines), whichare sub-parallel to each other. This line of reasoningmay explain clusters of islands and chains in whichthere is no neat age–distance relation along thelength of the chain, an example being the LineIslands in the central Pacific (Nunn 1994).

Two of the largest intraplate island clusters are theCanary and Cape Verde island groups, both in thecentral Atlantic and including a number of activevolcanoes. Until recently, it was held possible that theCanaries were of mixed origins, with the eastern-most—Lanzarote and Fuerteventura—being land-bridge islands, once connected to Africa (Sunding1979). However, it is now established beyond doubtthat the entire archipelago is oceanic in origin, andthat the gap of 100 km and more than 1500 m depthbetween the eastern islands and the African continenthas never been bridged. Although thus clearlyoceanic in the strict sense (Box 2.1), the Canaries donot appear to conform with the classical hotspotmodel as (1) they lack a clear lineal geographic distri-bution, (2) they do not present a simple age sequence,and (3) all the islands, with the exception of LaGomera, have been volcanically active in the last fewthousand of years, with Lanzarote, Tenerife, and LaPalma active within the last two centuries.

One explanation for their origin is the propagat-ing fracture model, which relates the origin of theCanaries to fractures generated in the oceanic crustof this region when the Atlas chain began to formsome 70–80 Ma as a result of the collision of theAfrican and Eurasian plates (Anguita and Hernán1975). However, some authors continue to favourthe hotspot concept, suggesting a locus somewherebetween the two youngest islands (La Palma and

M O D E S O F O R I G I N 21

Page 35: Whittaker. Island Biogeography. 2007

El Hierro) (Carracedo et al. 1998). It seems thatneither idea on its own can explain the origins ofthis particular archipelago, and that the roles of aresidual hotspot and the Atlas tectonics must be con-sidered simultaneously (Anguita and Hernán 2000).

The Cape Verde islands are also of uncertain ori-gin, but Nunn (1994) favours the tectonic-controlmodel. The Galápagos deserve a mention in thissection as being a group of great significance to thedevelopment of island evolutionary models. Theylie in the western Pacific just south of the divergentplate boundary separating the Nazca and Cocosplates. Again, Nunn regards them as an intraplatecluster rather than mid-ocean ridge islands.

Isolated islandsAs knowledge of the bathymetry of the sea floor hasimproved, a number of islands believed to be iso-lated have been shown rather to be part of aseamount-island chain or cluster. In general, it can betaken that truly isolated intraplate islands form onlyat or close to mid-ocean ridges. They result from asingle volcano breaking off the ridge with part of thesub-ridge magma chamber beneath it. Nunn (1994)identifies this condition as essential if the island is tocontinue to grow once it is no longer associated withthe ridge crest. Examples include Ascension, Gough,and St Helena in the Atlantic, Christmas Island in theIndian Ocean, and Guadalupe in the Pacific. Isolatedintraplate islands or island clusters may also, in cer-tain cases, be the product of small continental frag-ments being separated from the main continentalmass. The granitic Seychelles provide the best exam-ple, where there is good evidence of a continentalbasement affiliated to the Madagascar–India part ofGondwanaland (Box 2.2).

2.3 Environmental changes over longtimescales

In the case of these islands we see the importance of tak-ing account of past conditions of sea and land and pastchanges of climate, in order to explain the relations of thepeculiar or endemic species of their fauna and flora.

(Wallace 1902, p. 291)

The separation of environmental changes fromisland origin is artificial. Islands may be built up

by volcanism over millions of years; in such cases,active volcanism is thus a major part of the envi-ronment within which island biotas develop andevolve. The preceding section may give theimpression that all islands have been formed byvolcanism, or at least fairly directly by the actionof plate-tectonic processes. However, the forma-tion of an island may come about either byconnecting tracts of land disappearing underrising water, or by land appearing above thewater’s surface, either by depositional action or bysome other process, of which tectonic forms are alarge but not exclusive class. Indeed, Nunn (1994)offers the observation that changes of sea level areperhaps the most important reason why islandsappear and disappear. This section therefore dealswith long-term changes in the environments ofislands, focusing particularly on changes in sealevel, but also on locational shifts and changes inclimate.

Changes in relative sea level—reefs, atolls,and guyots

As already noted, islands may come and go as aconsequence of sea-level changes. Some of thesesea-level changes are eustatic (i.e. they are due tothe changing volume of water in the sea), and oth-ers are due to relative adjustment of the elevationof the land surface (isostacy). This can be broughtabout by the removal of mass from the land caus-ing uplift, as when an icecap melts, or by tectonicuplift. Subsidence of the lithosphere can be due toincreased mass (e.g. increased ice, water, or rockloading) or may be due to the movement of theisland away from mid-ocean ridges and otherareas that can support anomalous mass. In theright environment, coral reefs build around sub-siding volcanoes, eventually forming atolls—animportant category of tropical island.

Darwin (1842) distinguished three main reeftypes: fringing reefs, barrier reefs, and atolls. Heexplained atolls by invoking a developmentalseries from one type into the next as a result of sub-sidence of volcanic islands: fringing reefs are coralreefs around the shore of an island, barrier reefs fea-ture an expanse of water between reef and island,

22 I S L A N D E N V I R O N M E N T S

Page 36: Whittaker. Island Biogeography. 2007

and the final stage is the formation of an atoll,where the original island has disappeared, leavingonly the coral ring surrounding a shallow waterlagoon (Fig. 2.5). As Ridley (1994) noted, Darwinthought out the outline of his theory on the westcoast of South America before he had seen a truecoral reef! Although the theory may not be globallyapplicable and requires modification, for instancein the light of contemporary understanding of sea-level change, his basic model remains viable formost oceanic atolls and can account for most of themassive coral reefs (Steers and Stoddart 1977).

Coral reefs are built by small coelenterate ani-mals (corals) that secrete a calcareous skeleton.Within the tissues and calcareous skeleton, numer-ous algae and small plants are lodged. The algaeare symbionts critical to reef formation, providingthe food and oxygen supplement necessary toaccount for the energetics of coral colonies, whileobtaining both growth sites and nutrients from thecoral (Mielke 1989). Reef-building corals generallygrow in waters less than 100 m deep (exceptionallythey can be found as deep as 300 m); they requirewater temperatures between 23 �C and 29 �C, and arethus found principally in tropical and subtropicalareas, notably in the Indo-Pacific Ocean and in theCaribbean Sea (Fig. 2.6).

Coral growth has been found to vary between 0.5and 2.8 cm/year, with the greatest growth rates

occurring in water of less than 45 m depth (Mielke1989). Historically, these growth rates have beengreat enough to maintain reefs in shallow water aseither the underlying sea floor subsides or the sealevel has risen (or both). Drilling evidence fromislands with barrier reefs demonstrates this ability,just as predicted by Darwin’s (1842) theory. Forexample, the thicknesses of coral that have accumu-lated on the relatively young islands of Moorea(1.5–1.6 Ma), Raiatea (2.4–2.6 Ma), and Kosrae(4.0 Ma) vary between 160 and 340 m, whereas therather older islands of Mangareva (5.2–7.2 Ma),Ponape (8.0 Ma), and Truk (12.0–14.0 Ma) have accu-mulated depths of 600–1100 m of coral (Menard1986). However, it should be noted that massive reefaccretion also occurs in the western Atlantic on conti-nental shelves that have not experienced long-termsubsidence (Mielke 1989). This lends support toanother theory: that reefs established on wave-cutplatforms during periods of lowered sea level associ-ated with glaciation, and that reef development thenkept pace with rising sea levels in the interglacialperiod. This illustrates that several processes—subsi-dence, eustatic sea-level changes, temperaturechanges, and wave action—must each be consideredwith respect to the influence that they have had onthe growth of coral reefs (cf. Nunn 1997, 2000).

Given the vagaries of tectonic processes and sea-level changes, coral-capped islands can become

E N V I R O N M E N TA L C H A N G E S OV E R L O N G T I M E S C A L E S 23

Island

High water

Low water

(a) Fringing reef

Island

High water

Low water

Lagoon

(b) Barrier reef

(c) Atoll

Submerged island

High water

Low water

Lagoon Figure 2.5 The developmental sequence of coralreefs as a result of subsidence, from (a) to (b) to (c), ashypothesized by Darwin (1842). (Redrawn from Mielke1989, Fig. 7.10.)

Page 37: Whittaker. Island Biogeography. 2007

elevated. Christmas Island in the Indian Oceanexemplifies this. It has been elevated to form aplateau of 150–250 m above sea level, but in partsreaches 360 m. The general aspect of the island is ofcoastal limestone cliffs 5–50 m high, which carrieswith it the implication that the area currently avail-able for seaborne propagules to come ashore is con-siderably less than the island perimeter, a factor incommon with many islands of volcanic origin.

It is becoming apparent that the relative elevationof particular islands may be subject to influence bythe behaviour of others in the vicinity. The loadingof the ocean floor by a volcano appears to causeflexure of the lithosphere, producing both near-volcano moating and compensatory uparchingsome distance from the load; thus one repeatedpattern in the South Pacific is the association withinisland groups of young volcanoes (� 2 millionyears) with raised reef, or makatea islands, as exem-plified respectively by Pitcairn and HendersonIslands (Benton and Spencer 1995). There are manyother examples of islands that contain substantialamounts of limestone as a product of uplift: one thatwill be considered below is Jamaica.

Over time, sea levels have fluctuated markedly,particularly during the Quaternary glacial–inter-glacial cycles. These fluctuations, in combination withisland subsidence, wave action, and subaerial erosion,have resulted in many islands declining below thecurrent sea level. In nautical parlance, they are knownas banks if they are less than 200 m from the surface(Menard 1986). Once they have sunk below about200 m, they are effectively below the range of eustaticsea level and in this situation will rarely re-emergeas islands. Flat-topped sunken islands are termedguyots, and although they were once thought to bepredominantly erosional features, most are in fact flat-topped through accretion of carbonate sediments(Jenkyns and Wilson 1999). The distribution of guy-ots, and particularly of banks, may be crucial to anunderstanding of the historical biogeography ofcontemporary islands (Diamond and Gilpin 1983; andcompare McCall 1997 with Rogers et al. 2000).

Eustatic changes in sea level

From a biogeographical viewpoint, it does not mattergreatly whether the changing configuration of an

24 I S L A N D E N V I R O N M E N T S

Region of water temperature over 20˚C (68˚F) in the coldest month

Figure 2.6 Region of water temperatures exceeding 20 �C in the coldest month, which corresponds roughly to the major reef regions of theworld’s oceans. (Redrawn from Mielke 1989, Fig. 7.9.)

Page 38: Whittaker. Island Biogeography. 2007

archipelago is primarily a result of isostatic or ofeustatic changes. However, it is important to recog-nize these complexities because of the need to under-stand past land–sea configurations (Keast and Miller1996). That isostatic effects associated withQuaternary glaciations have not been confined tohigh-latitude continents and their margins, but alsoextended to low latitudes and ocean basins, wassomething that became clear only in the 1970s, neces-sitating a cautious approach to the construction ofregional eustatic curves (Nunn 1994). Stratigraphicdata from isolated oceanic islands have been ofparticular value in such analyses (Fig. 2.7).

Given the general emphasis on Quaternary events,it is noteworthy that certain of the sea-level oscilla-tions in the Tertiary appear to have been of greateramplitude, principally as a consequence of highstands at 29, 15, and 4.2 Ma (Nunn 1994)—the widerbiogeographical consequence of which are onlybeginning to be explored (Nores 2004). Within theQuaternary the pattern has been one of glacial

episodes in the northern latitudes corresponding withlowered sea levels, and interglacials associated withlevels not dissimilar to those of the present day, yetsuperimposed on a falling trend. At a more detailedand precise level of analysis it has, however, proveddifficult to construct global or even regional models ofsea-level change for the Quaternary (Stoddart andWalsh 1992). Data for the previous and current inter-glacial indicate that sea-level maxima have varied inmagnitude and timing across the Earth’s surface(Nunn 2000). One intriguing, if controversial, expla-nation for some of the irregularities in sea-levelchanges relates to the configuration of the oceanicgeoid surface (i.e. the sea surface itself), which, ratherthan being perfectly ellipsoid, is actually rather irreg-ular, with a vertical amplitude of about 180 m relativeto the Earth’s centre. Relatively minor shifts in theconfiguration of the geoid surface, which might beproduced by underlying tectonic movements, wouldbe sufficient to cause large amounts of noise in theglacio-eustatic picture (Nunn 1994; Benton and

E N V I R O N M E N TA L C H A N G E S OV E R L O N G T I M E S C A L E S 25

Stratigraphic age (Ma)25 20 15 10 5 0

Late Early Middle LateMioceneOligocene

Atoll at seasurface

Material removed by erosion

Solutionunconformity

formed buteroded sub-

sequently

Reefhole

A

B

C

DReeflimestone

Volcanicbasement

Subsidence path Solutionunconformity

preserved

Sea-levelcurve

Ele

vati

on (m

)

+200

+100

0

–100

–200

–300

–4000

Time since island formation (million years)

5 10 15 20 25

–169

–381

0

Depth(m)

Solutionunconformity

Pliocene-Pleistocene

Figure 2.7 The complex sequence of sea-level changes proposed for Midway Island for the last 25 million years. Solution unconformitiesformed during low stages A–C were all removed during lowering of the atoll surface by subaerial processes during stage D (low sea level). Thesolution unconformity formed at this stage has remained preserved because subsidence carried it below the reach of subsequent periods of atollsurface lowering. (Redrawn from Nunn 1994: original sources given therein.)

Page 39: Whittaker. Island Biogeography. 2007

26 I S L A N D E N V I R O N M E N T S

0

5

10

15

20

25

30

Bristol ChannelEnglish ChannelCardigan BaySomerset LevelsNorth Wales

3510

Dep

th b

elow

pre

sent

sea

leve

l (m

)

9 8 7 6 5Radiocarbon years B.P. ('000s)

4 3 2 1 0Figure 2.8 Holocene sea-level rise in southernBritain. (Redrawn from Bell and Walker 1992,Fig. 4.10, after Shennan 1983).

Spencer 1995). However, there is some measure ofagreement that stand levels have not exceeded pres-ent levels by more than a few metres within the past340 000 years. The most widely accepted figure forPleistocene minima is of the order of �130 m,although in places it may have been greater than this(Bell and Walker 1992; Nunn 1994). This order of sea-level depression, given present lithospheric configu-ration, is sufficient to connect many present-dayislands—such as mainland Britain—to continents,thus allowing biotic exchange between land areas thatare now disconnected. Equally important, manyislands existed which are now below sea level.However, as will be clear, simply drawing lines onmaps in accord with present-day �130 m contourlines is not a sufficient basis for reconstructing pastisland–mainland configurations (Kayanne et al. 1993).

The rise from the late glacial minima to presentlevels was not achieved overnight, nor was it asteady or uniform pattern. As a broad generaliza-tion, at around 14 500 BP (Before Present, wherePresent is by convention designated as AD 1950), sealevels stood at about �100 m, rising by some 40 mover the next millennium (Bell and Walker 1992).A second major phase of glacial melting around11 000 BP caused an eustatic rise to about �40 m bythe beginning of the Holocene, when ice volumeshad been reduced by more than 50%. The pattern forthe British Isles for the past 9000 years can be seen inFig. 2.8. Both eustatic and isostatic elements were

involved, leaving a legacy in raised shorelines anddrowned valleys. In the North Sea, the Dogger Bankwas breached by 8700 BP and the Straits of Dover by8000 BP. The present configuration of the southernNorth Sea coastlines was more or less established by7800–7500 BP and that of the British coastline byc.6000 BP, although slow adjustments continue today.The severing of Britain from Ireland took place some2000–3000 years before that of Britain and the rest ofEurope. In short, the switch from glacial to inter-glacial conditions took place some 2000–3000 yearsbefore ‘mainland’ Britain became an island onceagain, with different parts of the British Isles beingseparated from one another at quite different times.These events left a legacy in the biotic composition ofthe islands that has long been noted by biogeogra-phers (Williamson 1981). Oscillations in effective sealevel over the last few thousand years, althoughcomparatively small, have nonetheless been suffi-cient to have significant impacts on human societiesin island regions such as the Pacific (Nunn 2000), andthere is currently considerable concern that futuresea-level rises in association with predicted globalwarming will have serious implications for islandpeoples (see Chapter 12).

Climate change on islands

An important point made by Williamson (1981) isthat short-term variations in climate have lower

Page 40: Whittaker. Island Biogeography. 2007

amplitude than long-term variation, i.e. the variancewithin decades is less than the variance within cen-turies, which is again less than the variance withinmillennia. This can be summed up in the phrase:climatic variation has a reddened spectrum. Withinlarge landmasses one important way in whichspecies respond to high-amplitude climate changeis by range displacement. The possibilities for thismode of response within isolated oceanic islandsare of course limited, with the corollary that per-sistent island endemic forms must find environ-ments they can cope with within the limitedconfines of their island(s). If remote islands haveindeed experienced pronounced climate changewithin the last two or three million years, has thisdriven enhanced extinctions on islands, or has itselected for broad-niched, adaptable genotypes? Atpresent, we don’t have the data to answer thesequestions.

Over the past several million years, the globalclimate system has undergone continual change,most dramatically illustrated by the glacial–interglacial cycles of the Pleistocene (e.g. Bell andWalker 1992). For much of the last glacial period,for instance, large areas of the British Isles resem-bled arctic tundra, while the northern and westernregions supported extensive glacial ice. Subpolarislands such as the Aleutians (north Pacific) andMarion Islands (south-west Indian Ocean) alsosupported extensive icecaps at the last glacialmaximum, and there is evidence that thePleistocene cold phases caused extinction of plantspecies on remote high-latitude islands, such asthe sub-Antarctic Kerguelen (Moore 1979). Theclassical model of four major Pleistocene glacia-tions has long been replaced by an appreciationthat there have been multiple changes betweenglacial and interglacial conditions over the past2 million years (Bell and Walker 1992). Althoughhigh-latitude islands have been the most dramati-cally affected by these climatic oscillations, itwould be unwarranted to assume that low-latitude oceanic islands have been so effectivelybuffered that their climates have been essentiallystable.

The Galápagos Islands are desert-like intheir lower regions, with moist forests in the

highlands; however, palaeoenvironmental datafrom lacustrine sediments demonstrate that thehighlands were dry during the last glacial period.The moist conditions returned to the highlandsabout 10 000 BP, but the pollen data for El Juncolake on Isla de San Cristóbal indicate a lag of some500–1000 years before vegetation similar to that ofthe present day occupied the moist high ground(Colinvaux 1972). This delay may reflect the slowprogress of primary succession after expansionfrom relict populations in limited refugia in moremoist valleys, or the necessity of many plants hav-ing to disperse over great expanses of ocean (thegroup is approximately 1000 km west of mainlandEcuador) to reach the archipelago.

Analyses of pollen cores from subtropical EasterIsland also show the local effects of global climatechange. The data indicate fluctuating climatebetween 38 000 and 26 000 BP, cooler and drierconditions than those of the present day between26 000 and 12 000 BP, and the Holocene being gen-erally warm and moist, but with some drierphases (Flenley et al. 1991). Contrary to the evi-dence from Easter Island and the Galápagos, itappears from studies of snowline changes on trop-ical Hawaii that conditions were both cooler and,quite possibly, wetter during the last glaciation(Vitousek et al. 1995). Given the dominant influ-ence of ocean and atmospheric current systems onthe climates of oceanic islands (below), it isunwise to assume a straightforward relationshipbetween continental and island climate changeover the Quaternary.

The developmental history of the Canaries,Hawaii, and Jamaica

The CanariesThe Canaries constitute a comparatively ancientsystem for an oceanic archipelago. The increasinglywell-specified geological history provides a crucialbackground to understanding the biogeography ofthese islands, and the distributions of the manyregional, archipelagic, and single-island endemicspecies. The oldest basaltic shields (Northern,Betancuria, and Jandía) of the island ofFuerteventura are estimated to have emerged

E N V I R O N M E N TA L C H A N G E S OV E R L O N G T I M E S C A L E S 27

Page 41: Whittaker. Island Biogeography. 2007

above sea level some 20 Ma (Fig. 2.9a), althoughtheir construction process began almost 70–80 Main the Late Cretaceous Era (Anguita et al. 2002).This first island was likely colonized and populatedpredominantly from the nearby African mainland,some 100 km away. Some 4 million years later(c.16 Ma), the basaltic massives of Femés (LosAjaches) emerged, forming the first shield of whatbecame the island of Lanzarote, which was thusavailable for colonization simultaneously from themainland and the older island.

The expansion of the archipelago towards theAtlantic Ocean began 2 million years later, with theemergence of Gran Canaria (Fig. 2.9b), and was con-solidated with the appearance of La Gomera c.12 Maand Famara c. 10 Ma (Fig. 2.9c). These islands, muchlarger and higher than today, have suffered manycatastrophic landslides during their geologicalhistory (below). The next major additions to thearchipelago were the basaltic massifs of Teno,

Adeje, and Anaga, which emerged about 8 Mabetween La Gomera and Gran Canaria, forming thecorners of what was to become the island of Tenerife(Fig. 2.9d). It appears likely that Teno and Adeje weremainly populated from La Gomera, whereas Anagawas mainly colonized from Gran Canaria.Interestingly, some 10% of the endemic plants of theCanaries are restricted to one or more of the palaeois-lands of Tenerife, likely including a mix of earlydiverging and recently diverging lineages (Trustyet al. 2005).

Some 3.5 Ma, the catastrophic Roque Nublo ashflow is thought to have almost completely sterilizedGran Canaria, perhaps with the exception of twolocalized refugia (Marrero and Francisco-Ortega,2001a, b). If so, the recolonization process will havestarted both from those refugia, and, especially forthe mountain and summit ecosystems, from thenearby island massifs of Anaga and Jandía, respec-tively 60 and 90 km away.

28 I S L A N D E N V I R O N M E N T S

AFRICA

F

20 Ma

(a)

AFRICA

FC

L

AFRICA

G CF

L Famara

Femés

AFRICAG

TC

F

L Famara

FemésAnaga

Adeje

Teno

AFRICA

P

G

TC

F

L

AFRICA

Canary IslandsP

H

G

TC

F

L

Present

15 Ma 1 Ma

5 Ma

10 Ma

(b)

(c)

(d)

(e)

(f)

Figure 2.9 Sequence of emergence of the islands of the Canarian archipelago (Modified from Marrero and Francisco-Ortega 2001, Fig. 14.1).L � Lanzarote, F � Fuerteventura, C � Gran Canaria, T � Tenerife, P � La Palma, G � La Gomera, H � El Hierro.

Page 42: Whittaker. Island Biogeography. 2007

The last 2 million years has been a period ofgreat environmental dynamism in the Canaries.Volcanic mountain building activity has con-structed the two westernmost islands, La Palma(1.5 Ma) (Fig. 2.9e) and El Hierro (1.1 Ma). Bothhave been prone to catastrophic landslides, relatedprincipally to the instability of their high volcanicsummits. The last of them (El Golfo in El Hierro)occurred as recently as 15 000 years ago, cuttingaway the northwest half of the island, and distrib-uting it across some 1500 km2 of the ocean floor(Canals et al. 2000).

It is only within the last 1.5 Ma that the greatacidic volcanic cycle of Las Cañadas unified the oldbasaltic massifs of Teno, Adeje, and Anaga intotoday’s Tenerife. The new land surface enabledenhanced biotic exchange between the old massifs.Several landslides (Icod, Las Cañadas, La Orotava,Güímar) followed the construction of this edifice,having catastrophic effects on both Tenerife and thenearby islands. The building of the Teide stratovol-cano is very recent, with its highest tip (3718 m)being built in the prehistoric period, approximately800 years ago. Such events leave their imprint onthe genetic architecture of lineages, throughrepeated subdivision of formally contiguous popu-lations (e.g. Moya et al. 2004).

Finally, some 50 000 BP, the small islets north ofLanzarote (La Graciosa, Montaña Clara, andAlegranza) and Fuerteventura (Lobos) were built,giving the Canary archipelago its present shape(Fig. 2.9f). With the exception of La Gomera, whichhas been dormant for some 2–3 Ma, the islandshave remained active, with some 15 volcanic erup-tions in Lanzarote, Tenerife, and La Palma in thelast 500 years (Anguita et al. 2002).

In addition to the building, collapse, and erosionof the islands, eustatic sea-level changes during thePleistocene have alternately doubled and halvedthe emerged area of the archipelago, from approx-imately 7500 km2 during interglacials to some14 000 km2 during stadials (García-Talavera 1999).During low sea-level stands, such as coincidedwith the peak of the last Ice Age c.13 000 BP, theislands were about 130 m higher in altitude.Lanzarote and Fuerteventura were joined together,along with nearby islets, into a single large island

(Mahan, c.5000 km2 in area). The shortest distancefrom the archipelago to the African mainland wasreduced from about 100 km to 60 km, the present-day submarine bank of Amanay, north of Jandía,formed an island of some 100 km2 (Fig. 2.10), and a‘stepping-stone’ corridor of islands providedenhanced connectivity between the Canaries,Madeira and the Iberian Peninsula (Fig. 2.11).

HawaiiThe Hawaiian chain is characterized by the growthof shield volcanoes that go through a known lifecycle, with four well-defined stages (Wilson 1963;Price and Clague 2002):

E N V I R O N M E N TA L C H A N G E S OV E R L O N G T I M E S C A L E S 29

Salvages

R. Famara

Concepción

Amanay

PALAEOCANARIES (~18kyBp)

Mahan

PRESENT AFR

ICA

AFR

ICA

95 km

60 km

Salvages

Figure 2.10 The Canaries at the time of the sea-level minimum ofthe Last Glacial period and in the present day (Modified fromGarcía-Talavera 1999).

Page 43: Whittaker. Island Biogeography. 2007

1 Volcanoes form over the hotspot and are sub-sequently removed from it by the movement of thePacific plate, producing a linear array of volcanicsummits increasing in age to the north-west.2 After achieving their maximum heights perhaps0.5 Ma after their emergence, volcanoes rapidlysubside as they move away from the hotspot(Moore and Clague 1992).3 Erosive processes reduce volcanic peaks to sealevel over several million years.4 Small atolls may remain while conditions aresuitable for coral growth, or they are reduced toguyots if they drown and sink below the seasurface, carried down by the spreading andsubsiding plate (Jenkyns and Wilson 1999).

Recently, sonar surveys of the sea floor haveenabled the reconstruction of past islandconfigurations (Moore et al. 1994, Clague 1996). AsHawaiian volcanoes grow, the lava depositedbelow the sea level forms a steeper slope than thatdeposited subaerially. The location of the now

submerged breaks-in-slope enables the estimationof the maximum shorelines and thus areas ofislands throughout the archipelago (Fig. 2.12).Furthermore, the maximum altitude achieved byan island can also be reconstructed assuming a7�angle for subaerial lava deposits. Knowing theage, the original altitude of rocky outcrops and thesubsidence rate, it is possible to estimate the rate oferosion of each particular island. This is, of course,a fairly approximate science, as erosion can varythrough time as a result of climatic fluctuations,the occurrence of mega-landslides, and enhancedprecipitation around higher peaks associated withorographic cloud formation.

Using all this information, Price and Clague(2002) have reconstructed the history of theHawaiian Chain, from 32 Ma, which marks theappearance of the first Hawaiian island (the earlierislands of the Emperor seamount chain were eithersubmerged or just atolls by this time). The islandconfigurations at 5 Ma intervals for the Hawaiianchain are given in Fig. 2.13a, and Fig. 2.13bsummarizes the reconstructed distributions ofisland altitudinal range. During the first half ofHawaiian chain history (c.32–18 Ma) a few volca-noes (Kure, Midway, Lisianski, Laysan), brieflyexceeded 1000 m above sea level (ASL) althoughat any given moment most islands have been small.The most important island that pre-dated thepresent high islands was Gardner, which formedaround 16 Ma and was likely comparable in size(10 000 km2) and height (� 4000 m) to today’sBig Island (also known simply as Hawaii). After its -formation, a series of mid-sized volcanoes formed(French Frigate Shoals, La Perouse Pinnacle), culmi-nating with Necker, c.11 Ma. The period between 18and 8 Ma, is described by Price and Clague (2002) asthe ‘first peak period’, because of the existence ofmultiple peaks over 1000 and some over 2000 m,contributing to a substantial archipelago, withseveral zonal (coast to summit) ecosystems.

After the formation of Necker, there was aperiod when only smaller islands formed, leadingto an archipelago diminished in height and areaand thus impoverished in habitats. The emergenceof Kauai (5 Ma) constituted the beginning of the‘second peak period’, which continues today, with

30 I S L A N D E N V I R O N M E N T S

JosephineOrmonde

Gettysburg

Ampere

Seine

Porto Santo

Desertas

Salvages

Canary Islands

DaciaMogador

Concepción

Madeira

AFRICA

0 50 100 150 200 250 km

Figure 2.11 A reconstruction of PalaeoMacaronesia at the time ofthe sea-level minimum of the Last Glacial period (Redrawn fromGarcía-Talavera 1999).

Page 44: Whittaker. Island Biogeography. 2007

multiple volcanic island summits over 1000 and2000 m (Kauai, Oahu, Molokai, Lanai, Maui,Kahoolawe, and Hawaii) and with more emergedland than ever. Only some 1.2 million years ago,the now separated islands of Maui, Lanai, Molokai,and Kahoolawe, as well as the Penguin Bankseamount, constituted the Maui-Nui complex (‘BigMaui’ in Polynesian), an island larger than the cur-rent island of Hawaii, which was divided by subsi-dence some 0.6 Ma (Price and Elliott-Fisk 2004).The Maui-Nui complex has been reunited severaltimes by sea-level falls associated with Pleistocene

glaciation. This increasingly well-specified envi-ronmental history has considerable relevance tounderstanding the biogeography of Hawaii, as willbe discussed in later chapters.

JamaicaSeveral of the themes of this section are well illus-trated by Ruth Buskirk’s (1985) analysis of thehistory of Jamaica in relation to the rest of theCaribbean. The changing configurations ofAntillean land masses during the Cenozoic, assuggested by a variety of authors, are presented in

E N V I R O N M E N TA L C H A N G E S OV E R L O N G T I M E S C A L E S 31

Kure(29.8 Ma) Midway

(28.7 Ma)

Lisianski(23.4 Ma) Laysan

(20.7 Ma)

Gardner(15.8 Ma)

LaPerouse(12.8 Ma) Necker

(11.0 Ma) Nihoa(7.3 Ma)

Hawaii(0.5–0 Ma)

Maui Nui complex(2.2–1.2 Ma)

Oahu(3.0–2.6 Ma)

Kauai(4.7 Ma)

Niihau(5.1 Ma)

0 200 400 600 800 1000 km

pre-subsidencesubaerial

volcanic shicld

sea surface

break-in-slope

post-erosionrock outcrop

submergence/atoll

break-in-slope

post-subsidence

Direction of plate

movement

coral cap

H1

H1

H1

D

DD×tan 7˚

S

H2

1

1

3

3

2

2

4

4

Figure 2.12 Features of the Hawaiian ridge, life stages of volcanoes, and estimates of island altitude over time. The main islands are detailed in theinset. Ages are given in parentheses for selected volcanic peaks. The numbered sequence of boxes depicts life stages of islands and associatedfeatures used to estimate life histories. (1) The break-in-slope, marking the maximum shoreline of a volcano as it formed, is depicted by a blackoutline on the map. (2) The depth of break-in-slope (S) indicates the amount of subsidence that has occurred since shield formation; the presentelevation of an uneroded volcano (H2) is then added to determine its original height (H1). The original height can also be calculated by using thedistance from summit to shore (D) and assuming a 7�angle. Summits of the main islands are shown as white dots on the map over the island areasshown in black. (3) Slower subsidence and erosion reduce volcanic peaks to sea level. Four rocky islets (filled triangles on the map) are near the endof this stage, with the amount of erosion (E) estimated. (4) Finally, with no more rock above sea level, only seamounts (black dots) and small corallineislands (open triangles) remain (Redrawn from Price and Clague 2002, Fig. 1).

Page 45: Whittaker. Island Biogeography. 2007

Fig. 2.14. Beginning in the Eocene, left-lateral fault-ing along the northern Caribbean plate marginslowly moved Jamaica eastward relative to theNorth American plate, the island varying in sizeand becoming increasingly distant from the NorthAmerican continent in the process. More cruciallystill, from a biogeographical perspective, in theLate Eocene, it was entirely submerged, with exten-sive marine deposits being laid down and totalsubsidence in Jamaica amounting to some 2800 m,before its eventual re-emergence as an island in theearly Miocene. The uplift began in the northern andnortheastern coastal region, with general upliftbeginning in the middle Miocene. Maximum upliftand faulting occurred in the Pliocene (the finalstage of the Tertiary), with as much as 1000 m ofuplift in the Blue Mountains area since the middlePliocene. Even after the major period of emergencebegan, the island continued to move away fromCentral America, possibly by as much as 200 km inthe past 5 million years.

The message that must be drawn from this andthe foregoing sections is that such basic environ-mental features as island elevation, area, geology,location, isolation, and climate are each subject to

significant change in the long term. On evolution-ary time scales, volcanic islands in particular areremarkably dynamic platforms. If we are tounderstand the biogeography of such island sys-tems, it is clearly necessary to integrate bothhistorical and contemporary ecological factors(e.g. Stoddart and Walsh 1992; Wagner and Funk1995; Price and Elliott-Fisk 2004; Price andWagner 2004).

2.4 The physical environment of islands

Topographic characteristics

Topographic characteristics for a selection of islandsin the New Zealand area are provided by Mielke(1989), who notes that, although the largest islandshave the highest peaks, smaller islands display noconsistent pattern (Table 2.4). In fact, topographiccharacteristics depend on the type of island.Volcanic islands tend to be steep and relatively highfor their area and, through time, become highly dis-sected through erosion (Fig. 2.15). As long ago as1927, Chester K. Wentworth calculated the age ofthe Hawaiian volcanoes as a function of the degree

32 I S L A N D E N V I R O N M E N T S

25 Ma

(a)

20 Ma

Ku

PH

PH

15Ma

5 Ma

10 Ma

Li La

La

LP

Ma

NeNi Ka

G

G

LPNe

Ma

Ni Present

MNHa

KaOa

G

Li

Figure 2.13 (a) Hawaiian island configurations at 5 Ma intervals. Ku: Kure, PH: Pearl and Hermes, Li: Lisianski, La: Laysan, Ma: Maro, G:Gardner, LP: La Perouse, Ne: Necker, Ni: Nihoa, Ka: Kauai, Oa: Oahu, MN: Maui Nui, Ha: Hawaii. (Redrawn from Price and Clague, 2002, Fig 3).(b) Distribution of estimated island maximum altitudes over time (black areas �2000 m ASL, dark grey: 1000–2000 m, light grey: 500–1000 m,white: �500 m) (Redrawn from Price and Clague 2002, Fig. 2a).

25(b)

20

15

10

5

05 10 15 20 25 30 35

Nu

mbe

r of

vol

can

ic p

eaks

Page 46: Whittaker. Island Biogeography. 2007

of dissection. More recently, Menard (1986) com-pared the values resulting with the ages determinedby potassium–argon dating (up to an age of 1.6 mil-lion years), thus demonstrating that dissection isindeed a function of age of the volcano. Height ofisland can be important in respect to changes in sealevel, receipt of rainfall, and to other climatic char-acteristics. In addition, large, high volcanic islands

are subject to periodic catastrophic landslips, whichcan radically reshape substantial areas of the island(Hürlimann et al. 2004; below).

Coral or limestone islands and atolls tend to bevery low-lying and flat. There are clear implica-tions here with respect to future sea-level changes(Chapter 12). Those which have been uplifted tomore than a few metres above sea level are termed

T H E P H YS I C A L E N V I R O N M E N T O F I S L A N D S 33

Eocene 48Ma Early Miocene 20Ma

Late Miocene 10MaOligocene 38Ma

JA

CA

CUNH

PR

MX

SA

JA

JA

SH

SH

NH

NH

PR

PR

LA

LA

SA

SA

CA

CA

MX

MX

PRNH

SH

CU

JA

MX

CU

CU

(a)

(b)

(e)

(d)

(c)

50 40 30

Cuba

Puerto Rico

20 10Million years before present

Costa Rica - Panama

Jamaica

Honduras - Nicaragua

Mexico - Yucatan

Southern Hispaniola

Northern Hispaniola

Lesser Antilles

South America

Central America

America Western S.

Figure 2.14 Hypothesized configurations of the Caribbean area during the past 50 million years. (a–d) Maps of changing positions. Present-dayland outlines are used for ease of recognition and do not indicate shorelines of the time. CA � Central America, MX � Mexico, JA � Jamaica,CU � Cuba, NH � northern Hispaniola, SH � southern Hispaniola, PR � Puerto Rico, LA � Lesser Antilles, SA � South America. (e) Simplifiedscheme showing the relative positions of the land areas, increased distance apart indicating increased barriers to dispersal. Areas that werelargely inundated are designated with water wave symbols. (Redrawn from Buskirk 1985, Figs 2 and 4.)

Page 47: Whittaker. Island Biogeography. 2007

makatea islands. Examples include Makatea itself(Tuamotu archipelago), Atiu (Cook Islands), andmost of the inhabited islands of Tonga. This is animportant class of islands. They are characterizedby rocky coralline substrates, but some are partlyvolcanic. Many have had commercially exploitabledeposits of phosphate (guano) formed through theages from droppings of seabirds (e.g. Nauru; seeChapter 12). For a discussion of the complexities ofvolcanic, limestone, and makatea (composite)island landscapes, see Nunn (1994).

Climatic characteristics

Island climates have, self-evidently, a strong oceanicinfluence, and quite often are considered anomalousfor their latitude, as a consequence of their locationin the path of major ocean- or atmospheric currentsystems (e.g. the Galápagos; Darwin 1845). Lowislands tend to have relatively dry climates. Highislands tend to generate heavy rainfall, althoughthey may also have extensive dry areas in the rainshadows, providing a considerable environmentalrange in a relatively small space. Even an island ofonly moderate height, such as Christmas Island(Indian Ocean), with a peak of 360 m and generalplateau elevation of only 150–250 m, benefits fromorographic rainfall, allowing rain forest to be sus-tained through a pronounced dry season (Renvoize1979). Islands may also be expected to receive rain-fall that has a generally different chemical contentthan that experienced over continental interiors (cf.Waterloo et al. 1997).

As noted by J. D. Hooker in his lecture to theBritish Association in 1866, the climate and biotaof islands tend to be more polar than those ofcontinents in the same latitude, while intraannualtemperature fluctuation is reduced (Williamson1984). Islands near the equator frequently exhibitannual average temperature ranges of less than1 �C, and even in temperate latitudes, the annualaverage ranges are less than 10 �C: for example, 8 �Cin the case both of Valentia (Ireland) and the ScillyIsles. Some islands do, however, experience quitelarge interannual variations in other features of theirweather. Variability in rainfall, for example related

34 I S L A N D E N V I R O N M E N T S

Table 2.4 Area and relief of New Zealand and neighbouring islands (from Mielke 1989, Table 9.1)

Island Area (km2) General relief

South Island 148 700 50 peaks over 2750 mNorth Island 113 400 Three peaks over 2000 mStewart 1720 Granite; three peaks, up to 987 mChatham 950 Cliffs to 286 mAuckland 600 Volcanic; one peak 615 mMacquarie 118 Volcanic; 436 m, glacial lakesCampbell 113 574 m, glaciatedAntipodes 60 Volcanic; peak of 406 mKermadec 30 Volcanic; peak of 542 mThree Kings 8 VolcanicSnares 2.6 Granite; cliffs to 197 m

(a)

(b) Planeze

(c)

(d)

Figure 2.15 Stages in the erosion of a volcano (after Nunn 1994,based on an original in Ollier 1988). (a) An intact volcano, with radialgrooves (e.g. Tristan da Cunha); (b) Planeze stage, in which some ofthe radial drainage channels have undertaken headwall capture,leaving triangular-shaped remnants relatively untouched by fluvialerosion (examples can be seen on St Helena); (c) Residual volcano,where planezes have been removed by erosion, but the original formof the volcano is still evident (examples can be seen in the Canariesand Cape Verde islands); (d) Volcanic skeleton, where only the necksand dykes remain as prominent features (e.g. Bora-Bora, FrenchPolynesia).

Page 48: Whittaker. Island Biogeography. 2007

T H E P H YS I C A L E N V I R O N M E N T O F I S L A N D S 35

to El Niño–Southern Oscillation (ENSO) phenom-ena, can be of considerable ecological importance.Islands may also experience periodic extremes suchas hurricanes (below) (Stoddart and Walsh 1992).

Contraction, or ‘telescoping’, of altitudinal zonesis another marked feature of smaller islands. Forexample, on Krakatau, Indonesia, the plentifulatmospheric moisture and the cooling influence ofthe sea result in the near permanent presence ofcloud in the upper parts, further lowering tempera-tures and allowing the development of a montanemossy forest at around 600 m above sea level, amuch lower altitude than the continental norm(Whittaker et al. 1989). Taylor (1957) suggested2000 m as the height at which the vegetational tran-sition occurs in interior Papua New Guinea, andWhitmore (1984) observed that clouds normallyform around tropical peaks higher than 1200 m, butthat this altitudinal limit is reduced on offshoreislands. In his review of upper limits of forests ontropical and warm-temperate oceanic islands,Leuschner (1996) cited a range of 1000–2000 m for thelowering of the forest line compared with that forcontinental areas. Factors that have been invokedfor these reductions include steepened lapse rates(and associated trends in atmospheric humiditylevels), droughts on trade-wind-exposed islandpeaks with temperature inversions, the absence ofwell-adapted high-altitude tree species on someislands, and even the immaturity of volcanic soils(Bruijnzeel et al. 1993; Leuschner 1996). Whateverthe precise cause(s) of the telescoping, it is a com-mon feature of tropical islands, and it has the effectof (potentially) increasing the number of species

that an island can support, by compressing habitatsand allowing a relatively low island to ‘sample’additional upland species pools (Fig. 2.16).

The range of climatic conditions of an island isdetermined in large measure by the elevation of thehighest mountain peaks. Thus regression studiesoften find altitude to be an important variable inexplaining species numbers on islands, in somecases ranking only second to, or even ahead of,island area (Fernández-Palacios and Andersson2000). Tenerife, in the Canaries, is an excellent exam-ple of a volcanic island with a steep central ridge sys-tem and a pronounced rain shadow. The Canaries lieon the subsiding eastern side of the semi-permanentAzores anticyclone at about 28�N, 100 km off Africa.The subsidence produces a warm, dry atmospherealoft that is separated at 1500–1800 m from a lowerlayer of moist, southward-streaming air, producingwhat is known as a trade-wind inversion(Fernandopullé 1976). Tenerife’s climate is one ofwarm, dry summers and mild, wet winters: inessence, the island experiences a Mediterranean-type climate, despite its latitudinal proximity to theSahara. The north-east trades bring in moisture fromthe sea and the mountain backbone forces the air torise. It cools, and this produces a zone of orographicclouds, locally known as mar de nubes, between about600 and 1500 m, typically 300–500 m thick, on thewindward slope of the island—a feature that buildsup more or less daily (Fig. 2.17), providing the laurelforest that grow at that altitude with the moisturesurplus (through fog drip and reduced evapotran-spiration) needed to overcome the dry Canariansummers (Fernández-Palacios 1992, Fig. 1). The

3

2

1

0

Alt

itud

e (k

m)

Approximate distance from coast (km)4.5 5 37 53

Figure 2.16 The transition to mossy forest onmountains in Indonesia occurs at lower elevationson small mountains near the sea than on largermountains inland. From left to right: Mt Tinggi(Bawean), Mt Ranai (Natuna Island), Mt Salak(West Java), and Mt Pangerango (West Java).(After van Steenis 1972.)

Page 49: Whittaker. Island Biogeography. 2007

36 I S L A N D E N V I R O N M E N T S

Sea level

Lowervadose

zone

Upper vadose

zone

Key

Secondary fractures

Ground water compartment

Isolated ground watercompartmentSpring

Water level

Flow path

Flow direction

Salt water

Fresh andbrackish water

Interface

Figure 2.18 Scheme of the vertical distribution and flow of groundwater through a volcanic island. The vadose zones contain groundwatercompartments and chains, interspersed with dry zones. The Ghyben–Herzberg lens of fresh water under an oceanic island rests on the densersalt/brackish water that permeates the base of the island. (Modified from Ecker 1976, Fig. 4, with permission from Elsevier.)

southern sector, in the rain shadow, receives far lessprecipitation. As a consequence of the differences inprecipitation and temperature, the lower and upperlimits of forest growth are higher on the southernside. Indeed, the southern sector lacks a dense forestzone at mid-altitude and is much more xerophytic. Afurther consequence of the trade-wind inversion isthe existence of well-defined vegetation belts on thewindward slope, whereas on the leeward slopes,beyond the influence of the orographic cloud layer,the vegetational landscape seems to be closer to acoenocline, i.e. a vegetation continuum (Fernández-Palacios and Nicolás 1995).

Water resources

Availability of water shapes the ecology and humanuse of islands (Whitehead and Jones 1969; Ecker1976; Menard 1986). Most oceanic islands, whetherthey are high volcanoes or atolls, contain largereservoirs of fresh water. Fresh lava flows arehighly permeable, but over time the permeabilityand porosity of the rock decreases as a result ofweathering and subsurface depositional processes.The residence time of groundwater in the fracturedaquifers (Fig. 2.18) of large volcanic islands may befrom decades to centuries. We can divide these

Teide

Sub-alpine zone

SOUTH NORTH

Cloud layer

North Easttrade

winds

Dry southerncurrent Xerophytic

zone

Forestzone

Figure 2.17 Climatic/vegetation zones ofTenerife. (Redrawn from Bramwell and Bramwell1974.)

Page 50: Whittaker. Island Biogeography. 2007

systems into two zones: the vadose zone and thesaturated or basal water zone. The vadose zonecontains groundwater compartments, often linkedinto chains, interspersed with dry zones. The basalzone is characterized by many closely placedgroundwater compartments and by a high percent-age of saturated secondary fractures. Both freshwater and seawater occur in this zone, with tidesinfluencing the water level up to 4–5 km from thecoast (Ecker 1976). In the absence of rain, the sea-water within islands would be at sea level; how-ever, rainwater percolating through an island floatson the denser salt or brackish water that permeatesthe base of the island, in what is termed aGhyben–Herzberg lens (Menard 1986).

As indicated above, the average annual rainfallvaries greatly within the island of Tenerife, fromover 800 mm in the highest parts of the Anagapeninsular (a sum that might be doubled if fogdrip were considered), to less than 100 mm in theextreme south of the island. The rain shadow thushas profound effects both on the ecology and onthe potential human use of the island. The steepnorth side, which benefits from the majority of theprecipitation, is intensively cultivated, whereas

much of the flatter east and west coastal zones isessentially useless for cultivation without irriga-tion. Thus, even on this large, high island, ground-water (which increasingly has been tapped fromaquifers deep in the volcano) is the mostimportant source of water for the human inhabi-tants, and water constitutes a key limitingresource for development (Fernández-Palacioset al. 2004a). Even low-lying atolls maintain afreshwater lens but, as noted in the introduction tothis chapter, very small islets (below the order ofabout 10 ha) can lack a permanent lens. Such habi-tats are liable to be hostile to plants other thanthose of strandline habitats, thus limiting the vari-ety of plant species that can survive on them(Whitehead and Jones 1969).

Tracks in the ocean

One of the intriguing features of the island biogeo-graphical literature is that isolation is, in general,rendered merely as distance to mainland. Thisignores the geography of the oceans. Figure 2.19 dis-plays the pattern of surface drifts and ocean currentsin January. This pattern is, of course, variable on an

T H E P H YS I C A L E N V I R O N M E N T O F I S L A N D S 37

Oya Shiocurrent

North Pacific current

N. equatorial c.

Eq. counter-current

S. equatorial current

West wind drift

PeruorHumboldtcurrent

Falkland

BrazilS. eq. c

Eq. c. c

N. eq. c.

Californiacurrent Florida

Gulfstream c.

Labradorcurrent

North Atlanticdrift

Warm currentsCool currents

Canariescurrent

Guinea c.

Benguelacurrent

Agulhas

Eq. c. c.

S. eq. c.

West wind drift

N. eq. c.

c.

c.

c.

c

Kuroshiocurrent

Alaskacurrent

Figure 2.19 Surface drifts and ocean currents in January. (Redrawn from Nunn 1994, Fig. 4.)

Page 51: Whittaker. Island Biogeography. 2007

interannual and an intraannual basis. However,in some areas, the ocean currents and windcurrents are strongly directional and persistent. Insome cases, knowledge of current systems providesa parsimonious explanation of biogeographical pat-terning (Cook and Crisp 2005), as in the case of thebutterflies of the island of Mona in the Antilles. Theisland is equidistant between Hispaniola and PuertoRico and is 62 km2 in area. Its 46 species of butterfliesfeature 9 subspecies in common with Puerto Ricoand none with the larger Hispaniola, whereas theratio of source island areas is 9:1 in favour ofHispaniola. The explanation appears to lie in aremarkably constant bias in wind direction, which isfrom Puerto Rico towards Hispaniola (Spencer-Smith et al. 1988). However, it should be recalled thatpalaeoenvironmental data from islands such asHawaii and the Galápagos indicate that significantchanges in ocean and atmospheric circulationpatterns occur over time (Vitousek et al. 1995).

2.5 Natural disturbance on islands

One of the themes that we develop in this book is thatthe significance of naturally occurring disturbanceshas not been given due recognition in the develop-ment of island biogeographical theory. We havealready seen that any ancient island will, over thecourse of tens of thousands of years, haveexperienced substantial environmental changes.This section is concerned with shorter-termchanges in environment, such as the individualvolcanic eruptions that may intermittently add tothe bulk of an island while, at least temporarily,reducing its plant and animal populations(e.g. Partomihardjo et al. 1992; Hilton et al. 2003).

Ecologists have defined disturbance as ‘any rela-tively discrete event in time that removes organ-isms and opens up space which can be colonizedby individuals of the same or different species’(Begon et al. 1990, p. 754). Such a definition israther inclusive and, in the context of island bio-geography, it will be recognized that differentscales of perturbation are relevant in differing con-texts (Fig. 2.20). Increasingly, ecologists are recog-nizing that natural systems are largely structuredby disturbance (Pickett and White 1985; Huggett

1995). Often, a single event such as a hurricane(tropical cyclone), will impact on both mainlandand island systems; however, because of the geo-graphical and geological idiosyncrasies of smallislands and their location in oceans, their distur-bance regimes when scaled to island size are atyp-ical of continental land masses.

In order to incorporate the disturbance regimeinto island biogeography, some form of classifica-tory framework is required (Whittaker 1995).Figure 2.20 represents an example of one attempt todevelop such a framework. Another scheme,devised for the Caribbean islands and reproducedin Table 2.5, indicates the main types of disturbancephenomena, the area liable to be affected, the pri-mary nature of the impact, the duration, and therecurrence interval. In an approach owing much toH. T. Odum, Lugo (1988) has classified the distur-bance phenomena into five types, each character-ized by their impact on energy transfers.

● Type 1 events change the nature or magnitude ofthe island’s energy signature before the energy canbe ‘used’ by the island; for example, shifts ofweather systems through ENSO events, leading todroughts or heavy rainfall (Stoddart and Walsh1992; Benton and Spencer 1995).

● Type 2 events are those acting on the majorbiogeochemical pathways of an island, forexample as a result of changes caused by anearthquake.

● Type 3 events are those that remove structurefrom an island ecosystem, but without altering thebasic energy signature, so that recovery can proceedrapidly after the event. Hurricanes are Type 3 events.A well-researched example is Hurricane Hugo,which caused complete defoliation of a large part ofthe Luquillo forest on Puerto Rico in 1989, althoughone might note that, despite a rapid ‘greening-up’ ofthe forest, the signature of such an event will be evi-dent in the unfolding vegetation mosaic for manydecades (Walker et al. 1996).

● Type 4 events alter the ‘normal’ rate of materialexchange between the island and either the oceanor atmosphere. For instance, if the trade winds areinhibited by changes in atmospheric pressure,exchanges may be reduced.

38 I S L A N D E N V I R O N M E N T S

Page 52: Whittaker. Island Biogeography. 2007

N AT U R A L D I S T U R B A N C E O N I S L A N D S 39

Environmental disturbanceregimes

Biotic responses

Evolution ofthe biota

Ecosystem change

Speciation

Extinction

Secondarysuccession

Gap-phasereplacement

CompetitionProductivity*Disturbance events

Soil development

Platetectonics

100100

103

106

109

104 108

Tem

pora

l sca

le (y

ears

)

1012 100

Spatial scale (m2)*Examples:Wildfire, wind damage,flood, earthquake

104 108 1012

Glacial-interglacial

climatic cycles

Human activities

Fire regime

Pathogen outbreak

Species migration

Climatic fluctuations

Figure 2.20 Environmental disturbance regimes and biotic responses, viewed in the context of four space–time domains (shown here boundedby dashed lines), named micro-, meso-, macro-, and mega-scales by the scheme’s authors. (Redrawn from Delcourt and Delcourt 1991, Fig. 1.6;from an original in Delcourt and Delcourt 1988, with kind permission from Kluwer Academic Publishers.)

Table 2.5 Disturbance phenomena affecting Caribbean islands (after Lugo 1988). See text for explanation of the five types

Disturbance Type Area Primary Duration Recurrencephenomena affected impact

Hurricanes 3, 5 Large Mechanical Hours–days 20–30 yearsHigh winds 3, 4, 5 Large Mechanical Hours AnnualHigh rainfall 4 Large Physiological Hours DecadeHigh-pressure systems 1 Large Physiological Days–weeks DecadesEarthquakes 2, 5 Small Mechanical Minutes 102 yearsVolcanism All Small Mechanical Months–years 103 yearsTsunamis 3, 4, 5 Small Mechanical Days 102 yearsExtreme low tides 1 Small Physiological Hours–days DecadesExtreme high tides 3, 4, 5 Small Mechanical Days–weeks 1–10 yearsExotic genetic material 2, 3 Large Biotic 102 years DecadesHuman, e.g. energy 1 Small Biotic Years 1–10 yearsHuman, e.g. war 5 Small Mechanical Months–years ?

Page 53: Whittaker. Island Biogeography. 2007

● Type 5 events are those that destroy consumersystems, by which is meant human systems,possibly with subsequent repercussions for therest of the island’s ecology; examples againinclude hurricanes and earthquakes. The eruptionsof Montserrat in 1995–1998 provide a classicillustration of the extensive cross-cutting impactthat within-island volcanic action can have in transforming the physiography, ecology andhuman use of islands (Hilton et al. 2003; Le Friantet al. 2004).

Magnitude and frequency

Between 1871 and 1964, an average of 4.6 hurri-canes per year was recorded in the Caribbean,resulting in, for example, a return time of 21 yearsfor the island of Puerto Rico (Walker et al. 1991a;Fig. 2.21). With minimum wind speeds of120 km/h and paths tens of kilometres wide, hur-ricanes can have a profound impact, fundamentalto an understanding of the structure of naturalecosystems in the region. Moreover, the Caribbeanby no means corners the market in tropical storms.Hurricanes develop in all tropical oceanic areaswhere sea surface temperatures exceed 27–28 �C,although they are generally absent within5�north and south of the equator, where persistenthigh pressure tends to prevent their development(Nunn 1994). Wind defoliation and large

blow-downs are important and frequent distur-bances for forested tropical islands throughout thehurricane belts, centred 10–20�north and south ofthe equator. They may have particularly destruc-tive impacts on high islands, an example being thedevastation of large areas of forest on the Samoanisland of Upolu by tropical cyclones Ofa, Val, andLyn in 1990, 1991, and 1993, respectively (Elmqvistet al. 1994). Although storm damage to lowerislands can also be severe, storms can be importantto island growth by throwing up rubble ramparts(Stoddart and Walsh 1992). Thus some very smallislands, such as the so-called motu (sand islands),may be in a kind of dynamic equilibrium withboth extreme and normal climatic phenomena(Nunn 1994). A comprehensive analysis of distur-bance regimes requires quantification of bothmagnitude and frequency of events (Stoddart andWalsh 1992). Lugo (1988) attempted this for a sub-set of the Caribbean disturbance phenomenagiven in Table 2.5. He concluded that, whereashurricanes have the highest frequency of recur-rence among his major ‘stressors’, the susceptibil-ity of Caribbean islands to hurricanes isintermediate, partly because they do not changethe base energy signature over extensive periods.Also, with the larger Caribbean islands, hurricanedamage is unlikely to impact catastrophically onthe entire island area. Given their recurrenceinterval, these island ecosystems should be

40 I S L A N D E N V I R O N M E N T S

18991928

1956

1893

1835

18251807 1837

1876

1932

1989

1766, 1772?

18671931

N

40 km0

SAN JUAN

LuquilloForest

Figure 2.21 Paths of hurricanes that have crossed Puerto Rico since AD 1700. (Redrawn from Scatena and Larsen 1991, Fig. 1.)

Page 54: Whittaker. Island Biogeography. 2007

evolutionarily conditioned by such storm events(Whittaker 1995; Walker et al. 1996).

Not all extreme weather phenomena take theform of storms. ENSO phenomena are large-scaleinterannual events that are the product of varia-tions in air pressure and associated rainfall patternsbetween the Indonesian and South Pacific regions,coupled with ocean-current and temperaturevariation. In the Asia–Pacific region, ENSO eventsare associated with heightened amplitude of inter-annual climatic variation—that is, more intensedroughts and wet periods. As an illustration, theisland of Aututaki (Cook Islands) received3258 mm rainfall in December 1937, which is inexcess of the average annual total (Nunn 1994).That events of the magnitude of ENSO phenomenahave a detectable ecological impact, even on popu-lations conditioned evolutionarily to the island wayof life, was shown by the significant increase innumbers of four endemic Galápagos bird species(three finches and a mockingbird) as a consequenceof the increased food supply resulting from theexceptionally high rainfall during 1982–1983 (Gibbsand Grant 1987). Analyses of historical data forhurricanes show that the frequency of climatic phe-nomena such as hurricanes and ENSO eventsvaries over decadal time scales, further complicat-ing analysis. This goes to illustrate that, despite thereddened spectrum of climatic variation, short-term fluctuations deserve the attention of ecologistsand biogeographers (Stoddart and Walsh 1992;Benton and Spencer 1995; Schmitt and Whittaker1998).

There is a premise that, because of the compara-tive simplicity of geology and climate of manyoceanic islands, their landscapes should attain acondition of dynamic equilibrium. Nunn (1994)argues that this notion cannot be supported as agenerality:

. . . as is evident from the often catastrophic impacts ofcertain irregular climatic phenomena on oceanic islandlandscapes, the degree to which particular environmentscan accommodate the impacts of particular uncommon orextreme events without fundamental alteration, whilevariable, is generally low. For those oceanic island land-scapes in a state of dynamic equilibrium, the effects ofsuch events may be to cause a threshold of landscapedevelopment to be crossed . . . . The effects of irregular

climatic phenomena are so variable, so site-specific, that itis pointless to attempt a broad generalization. (pp. 159–160)

These observations might prompt us to ask ourselves, first, with what portion of the magnitude–frequency spectrum of climatic variation mightoceanic island biotas best be considered to be indynamic equilibrium and, second, might not extremeclimatic events have caused ecological or evolution-ary thresholds to be crossed in the same manner asdescribed above for their landscapes?

Disturbance from volcanism and mega-landslides

In contrast to the geological stability of continentalfragments or continental islands, oceanic islandsare characterized by the repetitive incidence of geo-logical catastrophes, amongst which volcanic erup-tions and gravitational flank collapses or landslides(with the tsunamis they generate affecting nearbyislands) are the most frequent.

Volcanic eruptionsVolcanic eruptions are the main constructive forcesof oceanic islands: they bring them into existence,and subsequently make them larger and higher,whereas the processes of erosion (by rain, wind, andsea) and subsidence provide the opposing forcesthat reduce them to sea-level or below (e.g. Figures2.11, 2.15). However, the same volcanic processesthat enlarge islands simultaneously damage ordestroy island ecosystems. The impacts can be dif-fuse, whereby volcanic ash is deposited in thin lay-ers over a wide area, or can be intense and entirelydestructive. In some cases, destructive volcanic ashflows can be deposited over the entire island, moreor less completely eliminating the existing biota.This is known as island sterilization, and impliesthat colonization has to begin from nothing. It is, ofcourse, very difficult to demonstrate complete ster-ilization, even for historical events such as the 1883Krakatau event (Backer 1929; Docters van Leeuwen1936; Whittaker et al. 1989; Thornton 1996). The chal-lenge is even greater for events as distant as GranCanaria’s c.3.5 Ma event (Marrero and Francisco-Ortega 2001a,b), although in this case it can beinferred from phylogenetic evidence that some

N AT U R A L D I S T U R B A N C E O N I S L A N D S 41

Page 55: Whittaker. Island Biogeography. 2007

resetting of the biological clock has occurred, withrenewed colonization from other islands. Santoriniin the Aegean Sea, whose collapse in 1628 BC mayhave destroyed the Minoan culture on Crete, or,more recently, Martinique and Montserrat in theCaribbean, provide further examples of the destruc-tiveness of island volcanoes.

It has been established that volcanic islands aretypically active over an extensive period, oftenspanning many millions of years, and that there is avariety of distinctive tectonic situations in whichvolcanic islands are built (above). A corollary of thisis that a number of major types of volcanic eruptioncan be identified. In increasing degrees of explosive-ness (Decker and Decker 1991) they are as follows:

● Icelandic eruptions are fluid outpourings fromlengthy fissures, and they build flat plateaux oflava, such as typify much of Iceland itself.● Hawaiian eruptions are similar, but occur moreas summit eruptions than as rift eruptions, therebybuilding shield volcanoes.● Strombolian eruptions take their name from asmall island off Sicily that produces small explo-sions of bursting gas that throw clots of incandes-cent lava into the air.● Vulcanian eruptions, named after the nearbyisland of Vulcano, involve the output of dark ashclouds preceding the extrusion of viscous lava flows,thus building a stratovolcano or composite cone.● Peléan eruptions produce pyroclastic flowstermed nuées ardentes, high-speed avalanchesof hot ash mobilized by expanding gases and trav-elling at speeds in excess of 100 km/h.● Plinian eruptions are extremely explosive,involving the sustained projection of volcanic ashinto a high cloud. They can be so violent, andinvolve so much movement of magma from beneaththe volcano, that the summit area collapses, forminga great circular basin, termed a caldera.

Like all generalizations, this classification is over-simplistic; for example, calderas have formed bycollapse in Iceland and Hawaii as well as throughmajor explosive volcanoes. However, this classifica-tion serves to illustrate that there are great differ-ences in the nature of volcanism, both betweenislands and within a single island over time.

The eruptive action within Hawaii over the pastfew centuries has been both more consistent and lessecologically destructive than that of Krakatau, whichin 1883 not only sterilized itself but, through calderacollapse, created a tsunami killing an estimated36 000 inhabitants of the coastlines of Java andSumatra. This death toll was tragically surpassed byan order of magnitude on 26 December 2004, by atsunami that swept across coastal settlements in anarc from north Sumatra, in this case generated not bycaldera collapse but by submarine faulting. Evenfairly small volcanic eruptions can have very impor-tant consequences for island ecosystems. Forinstance, eruptions on Tristan de Cunha in 1962 cov-ered only a few hectares in ejecta, but toxic gasesaffected one-quarter of the island area. The evacua-tion of the human population left behind uncon-trolled domestic and feral animals, whichtransformed the effects of the eruption and produceda lasting ecological impact (Stoddart and Walsh 1992).

Mega-landslidesThe progressive accumulation of volcanic materialupon young oceanic islands tends to build up rela-tively steep slopes within a short geologic time span,and at a faster rate than ‘normal’ subaerial erosionprocesses can bring into dynamic equilibrium (cf. LeFriant et al. 2004). Exceeding critical slope values pro-duces gravitational instabilities that lead to the col-lapse of the slopes through debris avalanches thattransfer hundreds of cubic kilometres into the sea(Carracedo and Tilling 2003, Whelan and Kelletat2003). Such collapses may occur unexpectedly andsuddenly and may lead to the disappearance of a sig-nificant part of an island, as much as a quarter of it,in just minutes (Carracedo and Tilling 2003), achiev-ing mass movements with velocities that have beenestimated as over 100 km/h (Lipman et al. 1988).

Although they were introduced into modern vul-canology by Telesforo Bravo (1962) almost half acentury ago to explain the origin of the LasCañadas caldera in Tenerife, our knowledge of cat-astrophic landslides has developed relativelyrecently, with improvements in the three-dimen-sional bathymetric analysis of the ocean floor. Suchanalyses have led to the discovery of the blocks anddebris avalanches resulting from the landslides col-lapses, often extending over areas of hundreds of

42 I S L A N D E N V I R O N M E N T S

Page 56: Whittaker. Island Biogeography. 2007

square kilometres around the islands. Lipman et al.(1988) and Moore et al. (1989) were amongst the firstauthors contributing indisputable evidence of theexistence of many giant landslides in the Hawaiianarchipelago. Later, such giant landslides have beenidentified in several volcanic islands around theworld (Table 2.6). Extensive deposits in the seaaround oceanic islands can also be produced bypyroclastic flows (Hart et al. 2004), or indeed by acombination of landslips and volcanic ejecta.

Major volcanic slope failures are fairly commonglobally, occurring at least four times a century andthey are a particular feature of volcanic islands inexcess of 1000 or 1500 m height (Hürlimann et al.2004). As many as 20 mega-landslides, with volumesvarying between 30 and 5000 km3, have been recog-nized both for the Canaries (Canals et al. 2000) and forHawaii (Fig. 2.22; Moore et al. 1994). These events arereiterative in actively forming islands. For example,El Hierro, the youngest of the Canaries, with an ageslightly greater than 1 million years, has already suf-fered four such catastrophic events, the last of them(El Golfo landslide c.15 000 BP), carried more thanhalf the island into the sea (Carracedo et al. 1999).Similarly, the La Orotava valley slide on Tenerife is

estimated to have been up to 1000 m in thickness,and created an amphitheatre up to 10 km wide and14 km long and an offshore prolongation several kilo-metres in length (Hürlimann et al. 2004). The fre-quency of recurrence of large landslides on theCanaries has been about one per 100 000 years acrossthe whole archipelago, or once every 300 000 yearsfor each single island (Masson et al. 2002).

Hürlimann et al. (2004) suggest that there are fourkey factors controlling where mega-landslidesoccur on volcanic islands:

● the development of deep erosive canyons on theflanks of the volcano● the presence of high coastal cliffs eroded at thebase by wave action● the presence of widespread residual buried soilsfrom an earlier phase of the development of the vol-cano, providing a line of weakness● the orientation of structural axes.

Such landslides can modify the subsidence regimeof the affected islands, as well as promoting newvolcanic activity, mainly centred in the summit scars(Ancochea et al. 1994). When not eroded or buriedby the emission of new volcanic material, the scars

N AT U R A L D I S T U R B A N C E O N I S L A N D S 43

Table 2.6 Examples of large debris-avalanche landslides occurring around young volcanic islands (slightly modified from Canals et al. 2000 andWhelan and Kelletat 2003)

Island Landslide Vertical distance Sea floor Volume Event agename moved (m) area covered (km2) (km3) (ky)

Réunion Ralé-Poussée 1700 200 30 4.2Réunion Eastern Plateau 3000 – 500 15–60

100–130Fogo (C. Verde) 1200 – 100 �10El Hierro El Golfo 5000 1500 400 15–20

and 100–130El Hierro El Julan 4600 1800 130 �160El Hierro Las Playas II 4500 950 �50 145–176La Palma Cumbre Nueva 6000 780 200 125–536Tenerife Güímar 4000 1600 120 780–840Tenerife La Orotava (Icod) 2100 �500 540–690Tenerife Las Cañadas �4000 1700 150 150–170Kauai South Kauai �4000 6800 – �13Kauai North Kauai �4000 14000 – –Oahu Nuuanu 5000 23000 5000 –Hawaii Pololu �4000 3500 – 370Hawaii Alika 1–3 4800 4000 2000 247

Page 57: Whittaker. Island Biogeography. 2007

created by such collapses are easily recognized.They include calderas (Taburiente in La Palma orLas Cañadas in Tenerife), valleys with knife-cutcliffs—called laderas in Spanish—delimiting thecollapse-formed valleys (Laderas de Güímar orLadera de Tigaiga, in Tenerife) or concave cliff-archssuch as those in the Anaga (Tenerife), Tamadaba(Gran Canaria), or Jandía (Fuerteventura) massifs.

The sudden collapse of volcanic island slopesand consequent debris avalanches has the potentialto generate tsunamis of extreme magnitude (mega-tsunamis), which may not only have highlydestructive impacts on the coasts and low areas ofnearby islands but also impact far-distant land-masses. For example, it is estimated that the ElGolfo collapse (El Hierro) produced a mass

44 I S L A N D E N V I R O N M E N T S

100 km

100 km

La Palma

La Gomera Tenerife

EL Hierro

Niihau

Maui

Kahoolawe

Hawaii

Lanai

MolokaiOahu

Kauai

Gran Canaria

CANARY ISLANDS

HAWAIIAN ISLANDS

Lanzarote

Fuerteventura

Figure 2.22 Mega-landslides in two oceanic island archipelagoes: (a) Landslides recognized in the Canaries (b) Landslides recognized in Hawaii(Redrawn from Carracedo and Tilling 2003, Fig. 2.18).

Page 58: Whittaker. Island Biogeography. 2007

movement with a volume of c.400 km3, generatinga mega-tsunami that transported boulders weigh-ing more than 1000 t on to elevations of 11 m on theBahamas Archipelago, located more than 3000 kmwestwards (Hearty 1997, in Whelan and Kelletat2003). Ward and Day (2001) have highlighted thegreat instability of the Cumbre Vieja massif, theactive southern summit of La Palma, where sevenvolcanic eruptions have been concentrated in thelast five centuries. It has been predicted thatCumbre Vieja will collapse to the west in the comingmillennia, generating a mega-tsunami likely todevastate the east coast of North America on the farside of the Atlantic.

2.6 Summary

Two broad classes of island are identified, trueislands surrounded by water, and habitat islandssurrounded by a contrasting matrix. True islands canbe subdivided into: island continent, oceanic islands,continental fragments, continental shelf islands, andislands within lakes. This chapter is concernedalmost exclusively with islands located in the oceansof the world. These islands are rarely ancient in geo-logical terms, and in many cases are significantly lessancient biologically than geologically. Oceanicislands have volcanic foundations, are concentratedin a number of distinctive inter- and intraplatesettings and have commonly experienced a dynamichistory involving varying degrees of lateral andvertical displacement, the latter confounded by, butoften in excess of, eustatic sea-level changes. In thetropics and subtropics, upward growth of reef-form-ing corals at times of relative or actual subsidencehas led to the formation of numerous islands of onlya few metres elevation, contrasting with the gener-ally steep topography of the high volcanic islands, to

a large degree according with Darwin’s 1842 theorieson coral reef formation.

Some of the more obviously important environ-mental features of islands are directly related tothese geological factors and include characteristictopographic, climatic, and hydrological phenom-ena. Subsurface water storage is a particularlyimportant feature of both volcanic and low,sedimentary islands: the possession of a subsurfacefreshwater lens is a characteristic of all but thesmallest islets. Island environments might bethought to have been shielded from the full ampli-tude of continental climatic fluctuations, but thepalaeoecological record indicates significant climaticchanges within the late Pleistocene and Holoceneon a range of islands, including examples from thetropics and subtropics. With limited scope for rangeadjustments within the island setting, the biogeo-graphical significance of such environmentalchanges should not be ignored.

Shorter term environmental variation, perturba-tion, or disturbances (broadly overlapping cate-gories) characterize many island environments, notleast for those islands which lie in the tropicalcyclone belt and for those which remain volcani-cally active. Recent bathymetric, geomorphological,and volcanological advances have revealed justhow dynamic oceanic island environments are.Island surfaces are constantly in flux, not only builtby volcanism, and reduced by subaerial and coastalerosion, but also subject to transforming cata-strophic slope failures. Some appreciation of thedistinctive nature of the origins, environmentalcharacteristics, and history of islands is almost self-evidently important to an understanding of thebiogeography of islands, although, in this scene-setting chapter, the biogeographical content hasintentionally been kept within limits.

S U M M A RY 45

Page 59: Whittaker. Island Biogeography. 2007

. . . the scarcity of kinds—the richness in endemic formsin particular classes or sections of classes,—the absence ofwhole groups, as of batrachians, and of terrestrial mam-mals notwithstanding the presence of aerial bats,—thesingular proportions of certain orders of plants,—herba-ceous forms having developed into trees, &c.,—seem tome to accord better with the view of occasional means oftransport having been largely efficient in the long courseof time, than with the view of all our oceanic islands hav-ing been formerly connected by contiguous land with thenearest continent . . .

(Darwin 1859, p. 384)

3.1 Introduction: the global significanceof island biodiversity

The term biodiversity is a contraction of biologicaldiversity, and broadly defined refers to the variabil-ity of life from all sources, including within species,between species, and of ecosystems (Matthews et al.2001). We may at times be concerned with higher orlower points in the taxonomic hierarchy, such thatfamilial, generic, subspecific, or even gene fre-quency data may be analysed under the header‘biodiversity’, but the most commonly used diver-sity unit is undoubtedly the species. As a first-ordergeneralization, islands are species poor for theirsize but rich in forms found nowhere else, i.e.endemic to that island or archipelago. They arethus ‘hotspots’ of biodiversity (see Box 3.1). Giventhat the total number of living species on the planetis not yet known to within an order of magnitude(Groombridge 1992), it is difficult to provide a firmmeasure of the relative contribution of islands to

global biodiversity. Yet, for particular taxa there aresufficient data to demonstrate that, taken collectively,islands contribute disproportionately for their areato global species totals.

In order to appreciate the special significance ofisland biotas it is important to consider some oftheir general properties, and in what ways they arepeculiar. This chapter sets out to do that, distin-guishing the notion of their compositional distinc-tiveness or ‘disharmony’ from simple speciespoverty. We also explore aspects of the historicalbiogeographical context and the importance of dis-persal limitations in determining the balance ofisland assemblages.

The affinities of island biotas to continental sourceareas have fascinated biogeographers for more thana century, providing some key insights for evolu-tionary theory (e.g. Darwin 1859; Wallace 1902).First steps are to establish how island biotas arerelated to continental biotas, where they ultimatelyderive from, and how far the unique (i.e. endemic)forms found on islands have developed from thecolonizing forms. Once this context has been devel-oped, we can move on to look at the mechanismsand processes involved in the evolution of islandbiotas, which we do in later chapters. In addition, asdiscussed later in this volume, islands are reposito-ries for many of the world’s threatened species andare worthy of special attention in conservation pri-oritization decisions. Scientific guidance as to whereparticular islands fit in global and regional planningframeworks is to be derived from the application ofbiogeographical techniques and theories, under therecently recognized (but fairly long-developing)

46

CHAPTER 3

The biogeography of island life:biodiversity hotspots in context

Page 60: Whittaker. Island Biogeography. 2007

I N T R O D U C T I O N : T H E G L O B A L S I G N I F I C A N C E O F I S L A N D B I O D I V E R S I T Y 47

Box 3.1 Islands as hotspots and their place in Conservation International’s hotspots scheme

There are at least three separate hotspot conceptsemployed within this book (and there are othersaround in the literature). First, is a geologicalusage, whereby hotspots are localized but long-lived areas of volcanism overlying plumes in themantle (Chapter 2). Second, in the introduction tothis chapter we have applied the termbiodiversity hotspot to signify the concentrationof endemic species on isolated islands. This usageis exemplified in BirdLife International’s EndemicBird Area (EBA) scheme, which delimits areassupporting two or more species of birds withbreeding ranges of <50 000 km2. Of the 218 EBAsdesignated in the 1996 version of the scheme, 113are island EBAs (21 continental islands, 21 largeoceanic islands, 55 small oceanic islands) (Longet al. 1996). It is beyond doubt that the world’sislands, and especially the tropical islands, hold adisproportionately high degree of global diversity.

The third usage stems from an approach toconservation prioritization put forward by theBritish environmentalist Norman Myers in the late1980s and recently promoted by the internationalenvironmental NGO Conservation International(Myers et al. 2000). Myers’ approach was tocombine a measure of the concentration ofbiodiversity, with a simple index of threat, i.e. it isreally a hotspot–threatspot scheme. The version ofthe CI hotspot scheme published in 2000identified as ‘hotspots’ 25 areas of the world thatmet two criteria: (1) the area should possess atleast 0.5% (1500) of the world’s plant species asendemics, and (2) the area should have lost 70%or more of its primary vegetation. Islands featuredprominently among the 25 CI 2000 hotspots,which is appropriate given the high proportion ofglobal species losses over recent centuries thathave been island species. The CI hotspotapproach attracted unprecedented levels offunding following the 2000 paper, and hasbecome extremely influential in strategicconservation planning.

Partly as a result of the phenomenal fund-raising success of the organization (aided bythe publication of the 2000 scheme in Naturemagazine), partly in response to new data ondiversity and habitat loss, and partly in an effort

to improve the fit of their scheme with anothermajor conservation planning effort (the WWFecoregions approach), CI revisited the boundariesof their hotspots in a review carried out in 2004,which resulted in a new map published on theirweb site (CI 2005). We term this version the CI 2005 hotspots. It depicts 34 hotspots, andincreases the coverage of the world’s islands,particularly in the Pacific.

Although the stated reason for the addition ofthe East Melanesian islands as one of the newhotspots is recent habitat loss, to a large degreethe increased attention to islands representsmerely a widening of their cast to include islandsof high biodiversity value (e.g. the Galápagos andJuan Fernández islands) that, as they put it, mightotherwise have slipped through the net ofconservation priorities. The approach isacknowledged on their web site (CI 2005) to bepragmatic, ‘with full recognition that the floristicaffiliations of these islands with their associatedland masses are often tenuous at best.’

Within the 34 CI 2005 hotspots, 9 areas (seemap) are composed exclusively of islands.Organized by geographical region these are:(1) the Caribbean islands (American ‘region’); (2)Madagascar and adjacent islands (the Comoros,Mascarenes, and Seychelles) (African region); (3)East Melanesia, (4) Japan, (5) New Caledonia, (6)New Zealand, (7) Philippines, (8) Polynesia-Micronesia, and (9) Wallacea (all Asia-Pacificregion). Three more have a substantial proportionof their diversity within islands: (10) theMediterranean basin (including the Atlanticislands of Macaronesia) (Europe and Central Asiaregion); (11) the Western Ghats and Sri Lanka,and (12) Sundaland (both Asia-Pacific region). Inthis expanded version of the scheme, almost alltropical and subtropical islands of high biodiversityvalue have been included, although it is evidentthat were many of the individual archipelagosconsidered as separate biogeographical entitiesthey would not have reached the qualifyingdiversity-plus-threat criteria. Examples of newlyincluded islands in the 2005 scheme include theGalápagos and Malpelo islands, lumped into aTumbes–Chocó-–Magdalena hotspot; the Juan

Page 61: Whittaker. Island Biogeography. 2007

Map of the CI-2005 hotspots.

48 T H E B I O G E O G R A P H Y O F I S L A N D L I F E

Fernández and Las Desventuradas islands,grouped with the Chilean WinterRainfall–Valdivian Forests; Revillagigedo, Cocosand Las Tres Marías islands, included in theMesoamerica hotspot; and the Macaronesianislands, which form an intriguingly shapedintrusion of the Mediterranean region into theAtlantic ocean. Biogeographically, this finalexample is not as peculiar as it might first seem(e.g. Blondel and Aronson 1999), although theislands—especially the Cape Verde islands—alsohave African affinities (see text).

CI claim that the CI 2005 hotspots schemecovers just 2.3% of the Earth’s surface, butincorporates as endemics over 50% of the world’splant species, and 42% of terrestrial vertebrates.The fashion in which the hotspots have been

delimited in both the 2000 and 2005 versions ofthe scheme may be rather arbitrary, but long-recognized biogeographical patterns arenonetheless detectable in their map products,which now serve to guide their conservationinvestment strategy. The world’s islands not onlyrepresent disproportionate amounts of diversity,but they also account for a high proportion ofrecorded global extinctions over the last fewhundred years, and a high proportion of globallythreatened species. A stress on islands is thus anappropriate part of any global conservationassessment based on the currency of species andindeed is common to schemes promoted by allthe major international conservation NGOs at thepresent time (e.g. the WWF, ConservationInternational, BirdLife International).

Polynesia - Micronesia

Polynesia-Micronesia

Polynesia- Micronesia

CaliforniaFloristic

Province

MadreanPine-OakWoodlands

MadreanPine-OakWoodlands

MesoamericaMesoamerica

CaribbeanIslandsCaribbeanIslands

Tumbes-Chocó-Magdalena

Tumbes-Chocó-Magdalena

Tropical Andes

ChileanWinter

Rainfall-Valdivian

Forests

ChileanWinter

Rainfall-Valdivian

Forests

Cerrado

Atlantic Forest

CapeFloristicRegion

SucculentKaroo

Guinean Forests of West Africa

Maputaland-Pondoland-Albany

Madagascarand theIndian OceanIslands

Madagascarand theIndian Ocean Islands

enatnomorf

Anr et s a

E

Coastal Forests of Eastern Africa

Horn ofAfricaHorn of Africa

MediterraneanBasin

MediterraneanBasin

CaucasusCaucasus

Irano-Anatolian HimalayaHimalaya

Mountains ofCentral AsiaMountains of Central Asia

Mountains ofSouthwestChina

Mountains of SouthwestChina

Western Ghats and Sri Lanka

WallaceaWallacea

SundalandSundaland

PhilippinesPhilippines

JapanJapan

NewCaledoniaNewCaledonia

New ZealandNew Zealand

EastMelanesianIslands

EastMelanesianIslands

SouthwestAustralia

NewZealand

New Zealand

Indo-BurmaIndo-Burma

HOTSPOTSWilderness AreasConservation International

February 2005

subfield of conservation biogeography (Whittakeret al. 2005). We illustrate some of the issues for con-servation biogeographers involved in assigningisland systems within faunal and floral regions byconsidering the affinities of the Macaronesianislands. Of course, in practice, political and eco-nomic frameworks and affinities may be as or moreimportant in developing and realizing strategic

conservation planning, but here we lay these prag-matic concerns to one side.

Not only do islands have many threatenedspecies today, but they have also contributed dis-proportionately to species extinction in historic andprehistoric times, such that in places much of theunderlying biogeographical structure may alreadyhave been lost or distorted. This is problematic for

Page 62: Whittaker. Island Biogeography. 2007

biogeographers, in that we may misinterpret frag-mentary data and reach erroneous conclusions.Before we embark on a consideration of speciationpatterns and processes, we provide at the end ofthis chapter some brief illustrations of island lossesas a cautionary note.

3.2 Species poverty

Islands typically have fewer species per unit areathan the mainland, and this distinction is moremarked the smaller the area of the island, i.e.interarchipelago species–area curves are steeperthan curves constructed by subdividing a largemainland area (Rosenzweig 1995). This will beexamined more fully in Chapter 4, and the follow-ing examples should suffice to introduce the themehere. First, Figure 3.1 illustrates that no matterwhether the Californian mainland plant data arebuilt up in spatially nested sets or non-nested data areconsidered, the points lie above the regression linefor the Californian island data. A second example(after Williamson 1981) is provided by Jarak islandin the Straits of Malacca (96 km from Sumatra,64 km from Malaya, and 51 km from the nearestisland). This 40 ha island is forested but lacksdipterocarps, the family that provides the mostcommon dominant species in the forests of Malaya,but whose members have poor long-distancedispersal ability. Within a 0.4 ha plot on the island,

Wyatt-Smith (1953) recorded 34 species of treeshaving a trunk diameter greater than 10 cm,compared with figures of 94 and 102 species,respectively, for two Malayan mainland plots.These mainland figures actually exceeded the totalspermatophyte (not just tree!) flora noted by Wyatt-Smith in his survey of Jarak, and although it wouldbe a surprise if this was found to be a completeinventory of all the species on the island (anotoriously difficult feat to achieve), it nonethelessgives an indication of the extent of impoverishmentof a not particularly remote island. A third exampleis St Helena (15�56�S, 5�42�W), a remote island ofabout 122 km2, and 14.5 million years old, whichhas a known indigenous flora of just 60 species, ofwhich 50 are endemic, with 7 of them extinct, andothers endangered (Davis et al. 1994).

The successful spread of species transported toislands such as St Helena in recent centurieswould appear to indicate that such islands areundersaturated with species and could supportmore in total (D’Antonio and Dudley 1995; Saxet al. 2002). However, some turnover is ofteninvolved, with endemic plant and especiallyanimal species becoming extinct as part of thisprocess. As discussed later in this volume, theassessment of such changes in diversity is prob-lematic, as, for instance, the losses and gains arealso tied up with other human influences, such asforest clearance and horticultural and agricultural

S P E C I E S P OV E RT Y 49

Figure 3.1 Species–area curves for Californian plants,showing the steeper slope for islands (log S1 opensquares), compared with two alternative nested setsfrom within the mainland (log S1 dots, San Francisco area;log S2, open circles, Marin County). The diamondsrepresent other areas from the mainland, which lie inneither nested set. (Redrawn from Rosenzweig 1995,Fig. 2.9.)

Page 63: Whittaker. Island Biogeography. 2007

50 T H E B I O G E O G R A P H Y O F I S L A N D L I F E

activities. Moreover, it may well be the case forSt Helena, as for some other island systems (e.g.Easter Island—Flenley 1993), that some native andendemic species became extinct before botanicalinvestigations were undertaken (Cronk 1989), sothat the known native biota is an incompleterecord of what once was present.

3.3 Disharmony, filters, and regionalbiogeography

This section deals with the regional biogeograph-ical setting of islands, beginning with some fairlysimple patterns, and progressing through increas-ingly complex scenarios.

Filtering effects, dispersal limits, anddisharmony

Islands tend to have a different balance of speciescompared to equivalent patches of mainland: theyare thus said to be disharmonic. There are twoaspects to this disharmony (Williamson 1981). First,as Hooker noted, both the climate and the biotaof islands tend to be more polar than those of nearbycontinents (Williamson 1981). Thus, the Canaries,located west of the Sahara, have a generallyMediterranean flora, Kerguelen in the Indian Oceanis bleak and Antarctic-like for its latitude, and theGalápagos, although equatorial, are closer to desert(i.e. subtropical) islands, being influenced byupwellings of cool subsurface waters. Hence, there isa climatic ‘filter’ involved in the assemblage ofspecies on islands. Secondly, islands are disharmonicin that effectively they sample only from the disper-sive portion of the mainland pool: i.e. there is anisolation or dispersal filter. This effect has of course,to be distinguished from simple impoverishment,i.e. it should not be merely a random subset of apotential mainland pool that is missing.

It is hard to quantify the maximum dispersalcapability of particular taxa across oceans, althoughwe attempt just that in Table 3.1. At the core of theproblem are the following complications:

● Present-day distances may not reflect distances atthe time of colonization (Chapter 2, Myers 1991;Galley and Linder 2006).

● There is much variation in dispersal powersbetween related species.● Dispersal powers change through time followingisland colonization.

Readers should be aware that some historical bio-geographers regard extreme long-distance dispersalwith enormous scepticism, instead arguing for ‘vic-ariant’ events, i.e. the past formation of barriers (bya variety of geotectonic and eustatic processes) sep-arating islands from previously connected land-masses with which they share biogeographicalaffinities (e.g. contrast the analyses of the biogeog-raphy of the Falklands offered by Morrone 2005 andMcDowall 2005). However, the evidence for theoccurrence of long-distance dispersal is clearlyestablished. What remains at issue is not whether itoccurs at all, but exactly how far can it reach forparticular taxa, and what is the relative importanceof vicariance and dispersal for particular biogeo-graphically challenging island groups like NewZealand (see Winkworth et al. 2002)?

Table 3.1 provides an analysis at a coarse taxo-nomic level, and it is important to realize thatwithin taxa such as birds, or higher plants, there isconsiderable structure to the dispersal reach ofsubtaxa (Fig 3.2). Thus, for instance, Krakatau,sterilized in 1883 and isolated by only c.40 km ofocean, is already home to one quarter of the ferns ofWest Java, but lacks altogether certain tree familiesand genera that are common on the mainland,but have no obvious ocean-going dispersalmechanisms (Whittaker et al. 1997). The partialsampling of mainland pools is also exemplified bythe flora of Hawaii, which, relative to other tropicalfloras, has few orchids, only a single genus ofpalms (Pritchardia), and is altogether lacking in gymnosperms and primitive flowering plant fami-lies. The largest eight families of flowering plantson Hawaii—Campanulaceae, Asteraceae, Rutaceae,Rubiaceae, Lamiaceae, Gesneriaceae, Poaceae, andCyperaceae—together constitute over half of thenative species (Davis et al. 1995).

Many other examples could be given. Forinstance, the Azores and the islands of Tristan daCuhna possess no land mammals or amphibians. Inthis case, the disharmony is clearly related to dis-persal ability. Tristan da Cuhna also lacks birds of

Page 64: Whittaker. Island Biogeography. 2007

D I S H A R M O N Y, F I LT E R S, A N D R E G I O N A L B I O G E O G R A P H Y 51

Table 3.1 Apparent limits to long-distance dispersal of selected taxonomic groups based on water gapdistances. Considerable caution should be observed when inferring dispersal capability from water gapdistances for the following reasons: (a) For continental fragments there is the possibility of a vicarianceexplanation due to geotectonic processes rather than a dispersal event; (b) eustatic sea level changes may inthe past have diminished the distances between islands and the mainland, in the case of many continentalshelf islands removing the barrier altogether. (Data compiled from various sources, including: Carlquist 1965,1974; Gorman 1979; Menard 1986.)

Taxonomic group Water gap (km) Example

Freshwater fishes* 5 British Columbia–Vancouver IslandLarge mammals 70–150 Pliocene Mediterranean megafaunaSmall mammals (excepting rodents) 410 Africa–MadagascarRodents, land tortoises, and snakes 920 South America–GalápagosAmphibians 1265 Africa–Seychelles

(1641) (Australia–New Zealand(Palaeoendemisms?) )

Gymnosperms 1356 Europe–AzoresLizards (geckos) 1641 Australia–New Zealand

(Palaeoendemisms?)Bats, land birds, snails, arthropods, fungi, 3650 North America–Hawaiimosses, ferns, angiosperms

*Excluding diadromous fishes, for which almost anything appears possible (R. McDowall, personalcommunication).

Animal-external

Animal-internal

Aquatic-active

Aquatic-passive

Aerial-active

Aerial-passive

Figure 3.2 The various modes of dispersal by which species may naturally disperse to and colonize a remote island through or across asea-water gap (redrawn from Aguilera et al. 1994, based on an original in Rodríguez de la Fuente 1980). Alternative nomenclature for the topthree categories as they apply to plants: seed dispersal by wind = anemochory; in the gut of an animal = endozochory; attached to the fur orfeathers = exozoochory; flotation in the sea = thalassochory.

Page 65: Whittaker. Island Biogeography. 2007

prey: with very few species of land birds, no landmammals, and a small land area, it is doubtfulwhether a predator population could maintainitself, were one to reach the island (Williamson1981). This interpretation suggests that impoverish-ment among one group may lead through food-chain links to disharmony in another.

The notion of disharmony has been criticized byBerry (e.g. 1992) who has written:

It is a somewhat imprecise concept indicating that therepresentation of different taxonomic groups on islandstends to be different to that on the nearest continent, andcarrying the implication that there is a proper—or har-monious—composition of any biological community.This is an idea closer to the ‘Principle of Plenitude’ ofmediaeval theology than to modern ecology. (Berry 1992,pp. 4–5)

Berry has a point: that island environments have‘sampled’ distinctive subsets of the mainland pooldoes not necessarily imply that there is somethingwrong with island assemblages and that they arefurther from balance with their physical environ-ment than those of continents (this is a questionworth posing, but it can be distinguished as a sepa-rate question). Nonetheless, consideration of remoteoceanic islands shows that they are distinctive, andthere seems little point in developing a longer phraseto capture the same idea as the term disharmony.

Another classic example of the filtering out ofparticular groups of animals with increased isola-tion is provided by Fig. 3.3, which shows the distri-butional limits of different families and subfamiliesof birds with distance into the Pacific from NewGuinea. Such patterns have sometimes been termedsweepstake dispersal. The data in Fig. 3.3 seem tobe interpretable broadly in terms of the differingdispersal powers of the groups of birds concerned,illustrating that insular disharmony can be identi-fied not only at the level of orders but also belowthis at the family level (cf. Whittaker et al. 1997).However, as pointed out by Williamson (1981),within Fig. 3.3 there is a general thinning out ofislands and decline in size to the east. This raisesthe question of whether dispersal difficulties alonehave caused the filter effects—a problem that, usingpresently available data, appears insoluble (but seeKeast and Miller 1996).

The New Guinea–Pacific study provides a basi-cally unidirectional pattern, but in other cases atwo-way filter appears to operate (Carlquist 1974),best exemplified for certain linearly configuredarchipelagos. A classic example is shown in Fig. 3.4,which quantifies the decline in reptiles and birds oforiental affinity and the increase in Australianspecies going from west to east along the LesserSunda Islands. A similar sort of two-way patterncan also be detected in certain circumstances onmainlands, particularly on peninsulas (e.g. Florida;Brown and Opler 1990), but island archipelagosgenerally provide the clearest filter effects.

Biogeographical regionalism and thevicariance/dispersalism debate

These two-way filter effects evidence a mergingof faunas or floras between two ancestral regions.They indicate the biogeographical imprint ofextremely distant events in the Earth’s history,intimately connected with the plate-tectonicprocesses that have seen the break-up of super-continents, parts of which have come back intocontact with one another many millions of yearslater. Early students of biogeography, notablySclater and Wallace, recognized the discontinu-ities and, on the basis of the distribution patternsof particular taxa, were able to divide the worldinto a number of major world zoogeographic orphytogeographic regions. One such scheme isshown in Fig. 3.5.

The boundary between the Oriental andAustralian zoogeographic regions is marked bywhat has long been known as Wallace’s line(Fig. 3.6), which marks a discontinuity in thedistribution of mammals, and divides the Sundaislands between Bali and Lombok—not an alto-gether obvious place to draw the line from the pointof view of the present-day configuration of theislands. Modifications, and additions in the form ofWeber’s line, distinguishing a barrier for Australianmammals like Wallace’s line for Oriental mammals,have also been proposed (Fig. 3.6). However, theprecise placing of lines is problematic, in partbecause the area between the Asian and Australiancontinental shelves (‘Wallacea’) actually contains

52 T H E B I O G E O G R A P H Y O F I S L A N D L I F E

Page 66: Whittaker. Island Biogeography. 2007

relatively few Asian or Australian mammals (Coxand Moore 1993), and in part because other groupsof more dispersive animals, such as birds, butter-flies, and reptiles, show the filter effect (see Fig. 3.4)rather than an abrupt line. Another problem for thedrawers of lines is that the insects of New Guineaare mainly of Asian origin. There are several suchenigmas of distribution in relation to these lines:for a fuller discussion see Whitmore (1987) andKeast and Miller (1996). Perhaps the best option is torestrict the Oriental and Australia regions to the

limits of the respective continental shelves, and totreat the intervening area of Wallacea as essentiallyoutside the framework of continental biogeography(Cox 2001).

The placing of distant oceanic islands into thissort of traditional biogeographical framework hasdrawn criticism from Carlquist (1974, p. 61) whohas argued that because of the isolation of islands,and their differing histories from continents, theirbiotas provide very poor fits with regional biogeo-graphical divisions. Commonly, oceanic islands

D I S H A R M O N Y, F I LT E R S, A N D R E G I O N A L B I O G E O G R A P H Y 53

Hawaii

Marquesas

Pitcairn EasterIsland

Cook Is.

Society Is.

Tonga

Samoa

FijiNewCaledonia

New Guinea

1 23

45 6

7

120°E 140° 160° 180° 160° 140° 120°W

40°S

20°

20°

40°N

Figure 3.3 Eastern limits of families and subfamilies of land and freshwater breeding birds found in New Guinea. The decline in taxa is fairlysmooth, and shows both differences in dispersal ability and that there is a general decline in island size to the east. Not beyond: 1, New Guinea(14 taxa), pelicans, snakebirds, storks, larks, pipits, logrunners, shrikes, orioles, mudnesters, butcherbirds, birds of paradise, bowerbirds, Australiannuthatches, Australian tree-creepers; 2, New Britain and Bismarck islands (2 taxa), cassowaries, quails, and pheasants; 3, Solomon Islands (10taxa), owls, frogmouths, crested swifts, bee-eaters, rollers, hornbills, pittas, drongos, sunbirds, flower-peckers; 4, Vanuatu and New Caledonia (7taxa), grebes, cormorants, ospreys, button-quails, nightjars, wren warblers, crows; 5, Fiji and Niuafo’ou (4 taxa), hawks, falcons, brush turkeys,wood swallows; 6, Tonga and Samoa (7 taxa), ducks, cuckoo-shrikes, thrushes, whistlers, honeyeaters, white-eyes, and waxbills; 7, Cook andSociety islands (3 taxa), barn owls, swallows, starlings; 8 (beyond 7), Marquesas and Pitcairn group, herons, rails, pigeons, parrots, cuckoos, swifts,kingfishers, warblers, and flycatchers. Others: owlet-nightjars, one species from New Caledonia, otherwise limit 1; ibises, one species from theSolomon Islands, otherwise limit 1; kagu, endemic family of one species from New Caledonia. (From Williamson 1981, Fig. 2.5. We have notattempted to update this figure in the light of more recent fossil finds.)

Page 67: Whittaker. Island Biogeography. 2007

will have gained significant proportions of theirspecies from more than one land mass and morethan one biogeographical province. For instance,Fosberg (1948) calculated that the flowering plantsof Hawaii could have descended from some 272original colonists, of which 40% were believed to beof Indo-Pacific origin; 16–17% were from the south(Austral affinity); 18% were from the American con-

tinent; less than 3% were from the north; 12–13%were cosmopolitan or pantropical; and the remain-der could not be assigned with confidence. Can theHawaiian islands logically be put into an Orientalprovince when appreciable portions of their floraand fauna are American in origin? Most phytogeo-graphers, in fact, assign the flora to its own floristicregion (Davis et al. 1995).

54 T H E B I O G E O G R A P H Y O F I S L A N D L I F E

Damma

Kei Is.

TanimbarWetar

AlorSumbawaFlores

BaliJava

Western (Oriental) species

Eastern (Australian–New Guinean) species

Bir

ds

(%)

100

50

0

Rep

tile

s (%

)

100

50

0

Lom-bok

AruIs.

Figure 3.4 The two-way filter effect, showing trends in the proportion of Oriental and Australian species of birds and reptiles across the Sundaislands, South-East Asia. In contrast to the gradual changes shown here, mammals ‘obey’ Wallace’s line, a sharp line of demarcation between Baliand Lombok (see Fig. 3.5). (Modified with permission, from Carlquist, S. (1965). Island life: a natural history of the islands of the world. NaturalHistory Press, New York, Fig. 3.7.)

Palaearctic60°

30°

60°

30°

African

Neotropical

Nearctic

Oriental

Australian

Figure 3.5 Based on the distributionof extant mammals, A. R. Wallace in1876 divided the world into zoogeographicalregions. (The map shown here has beenredrawn from Cox and Moore 1993,Fig. 8.1.)

Page 68: Whittaker. Island Biogeography. 2007

Renvoize’s (1979) analysis of the dominantregional phytogeographical influences within theIndian Ocean (Fig. 3.7) tells a similar story. Hispaper illustrates that many island biotas are not theproduct of simple one- or two-way filter effects, nordo they enjoy full affinity with a particular conti-nental region, but instead they consist of a mix ofspecies of varying sources and origins. For instance,Christmas Island, close to Indonesia, is dominatedby colonists from South-East Asia, whereas theFarquhar Group has an African component, areflection of the location of these low-lying islandsnorth-east of the northern tip of Madagascar.

Renvoize’s (1979) analysis of Christmas Islandwas based on a list of 145 indigenous vascular plantsfrom the end of the nineteenth century. The break-down of this flora was: 31 pantropical, 21palaeotropical, 76 tropical Asian (Madagascar toPolynesia and Australia), and 17 endemic species.This contrasts with the indigenous flora of theFarquhar Group, which consisted of 23 pantropical,18 Indo-Pacific, and 5 African species—which seems

to be a fairly typical sort of mix for the low islands ofthe Indian Ocean. Leaving aside introduced species,the pantropical species constitute a far greater pro-portion of the flora of the Farquhar Group than ofChristmas Island (respectively 50% and 21%).Although several factors may be involved in thiscomparison, the most obvious differences are loca-tion and in the range of habitats available on the highas opposed to low-lying islands (Renvoize 1979).

Christmas Island is an elevated coral-cappedisland, having a plateau at 150–250 m ASL, a peak of361m, and a land area of 135 km2. On low islandsfew areas are beyond the direct influence of the sea.Strandline species thus dominate the floras and,since they are nearly all sea-dispersed and since themost constant agent of plant dispersal is the sea, thecoastal habitats receive a steady flow of possible col-onizers and a steady gene flow, not conducive to theevolution of new forms. High islands do havecoastal habitats, but these habitats and their charac-teristically widespread sea-dispersed species do notdominate the floras as they do on the low islands.

D I S H A R M O N Y, F I LT E R S, A N D R E G I O N A L B I O G E O G R A P H Y 55

INDO-CHINA

NEWGUINEA

AUSTRALIA

Huxley'sline 1868

PHILIPPINES

Wallace's line 1863–1880

Wallace's line

JAVA

BORNEOSUMATRA

Weber's line

IndianOcean

Weber's line 1904

Weber 1894/Lydekker 1896

PacificOcean

Figure 3.6 Map of South-East Asia and Australasia showing some of the boundaries proposed by A. R. Wallace and others, dividing the twofaunal regions. The continental shelves are shown lightly shaded. In contrast to the sharp discontinuity first noted by Wallace in the distribution ofmammals between Bali and Lombok—marked by the original version of Wallace’s line—birds and reptiles show a more gradual change betweenthe two regions (see Fig. 3.3). (From various sources.)

Page 69: Whittaker. Island Biogeography. 2007

Those oceanic islands with large proportions ofendemic plants are typically high islands withvaried habitats (Box 3.2; Humphries 1979). Recentwork, based on molecular phylogenies, supportsthe more traditional biogeographical analysesundertaken by Renvoize (1979) in indicating multi-ple influences on particular islands, and a variety ofstepping-stone routes back and forth across theIndian Ocean (e.g. Galley and Linder 2006; Rocha et al. 2006).

Renvoize (1979) declined to attribute the lowislands of the Indian Ocean below the level ofPalaeotropical Kingdom because they contain toohigh a proportion of very widespread species. Thisrestrained approach to demarcation is realistic, andalthough the categorization of islands into particu-lar biogeographical provinces or regions is clearlyproblematic (Cox 2001), such phytogeographicaland zoogeographical analyses have some value.They establish the broad biogeographical contextwithin which the compositional structure of a par-ticular island or archipelago needs to be placed(cf. van Balgooy et al. 1996).

Distance from a larger source pool can be a poorindicator of biogeographical affinity. First, there maybe historical factors, as in the cases of Madagascarand the granitic islands of the Seychelles, which are

ancient continental fragments from Gondwanalandand so began their existence as islands millions ofyears ago, complete, we assume, with a fullcomplement of species. In the case of Madagascarthe split from Africa has been estimated to haveoccurred as long ago as 165 million years (Davis et al.1994), but the effective isolation date may beconsiderably more recent (see Box 2.2). Secondly,ocean or wind currents (or even migration routes)may have determined colonization biases (Cook andCrisp 2005), such as apparently in the case of thebutterflies of the Antillean island of Mona(Chapter 2). Another classic example is that ofthe flora of Gough Island (Moore 1979). Thepredominant wind circulation system around theSouth Pole provides a plausible explanation forthe affinities with New Zealand and South America(Fig. 3.8). Changes in relative sea level, landmasspositions, and oceanic and atmospheric circulationover geological time (Chapter 2) can mean thatsuch analyses are difficult to bring to a definiteconclusion. We illustrate the problems involved inmore detail below, with reference to Macaronesia.

As already mentioned, within historical biogeog-raphy two opposing camps have formed up overrecent decades, the dispersalist and vicariancebiogeographers, each concerned with how disjunct

56 T H E B I O G E O G R A P H Y O F I S L A N D L I F E

Gloriosa

AFRO-ARABIAN SUB-KINGDOM

PALAEOTROPICAL KINGDOM

MADAGASCANREGION

S.E. ASIA-MALAYANREGION

DECCAN-MALAYSIANSUB-KINGDOM

Seychelles

Agalega

Amirantes

Coetivy

ComorosTromelin Cargados

CarajosMauritius

ReunionRodrigues

Addu

Farquhar

MaldiveIslands

LaccadiveIslands

AldabraGroup

ChristmasIsland

Cocos orKeeling

Island

ArchipelagoChagos

AndamanIslands

NicobarIslands

AFRO-ORIENTALDOMAIN

DECCANREGION

Figure 3.7 Phytogeographical classification of the Indian Ocean islands. (Redrawn from Renvoize 1979, Fig. 1.)

Page 70: Whittaker. Island Biogeography. 2007

D I S H A R M O N Y, F I LT E R S, A N D R E G I O N A L B I O G E O G R A P H Y 57

Box 3.2 Islands in the ocean: general features

In Box 2.1 we outlined the differing origins ofthree classes of island. Here we summarizeaspects of their geology, geography, and biology.● Oceanic islands are built over the oceanicplate, are of volcanic or coralline formation, areremote and have never been connected tomainland areas, from which they are separatedby deep sea. They generally lack indigenous landmammals and amphibians but typically have afair number of birds and insects and usuallysome reptiles.● Continental islands are more varied geologi-cally, containing both ancient and recent stratifiedrocks. They are rarely remote from a continentand always contain some land mammals andamphibians as well as representatives of the other classes and orders in considerable variety(Wallace 1902), typically having been connected

to other land masses before the Holocene transgression.● Continental fragment islands are in somerespects intermediate between the other twoclasses, but being typically ancient and long-isolated they are in other respects biologically peculiar.

The biogeographical distinction between islandsof differing geological origins of course dependsgreatly on the ability of potential inhabitants of anisland to disperse to it, whether over land or sea.As a simplification, we may conceptualize anoceanic island as one for which evolution is fasterthan immigration, and a continental island as onewhere immigration is faster than evolution(Williamson 1981, p. 167). Thus, a particular islandmay be thought of as essentially continental forsome highly dispersive groups of organisms (e.g.ferns (Bramwell 1979) and some types of birds),

General features of islands of different origin of relevance to their biogeography (modified from Fernández-Palacios 2004)

Features Continental islands Continental fragments Oceanic islands

Origin Interglacial sea-level rise New mid-ocean rift creation Between or within platesubmarine volcanic activity

End Glacial sea-level drop Collision with a continent Erosive break up and subsidenceDegree of isolation Small Variable; small for Cuba, Large (except Canaries)

large for New ZealandSize From very small to very large Some small, mostly large Small (except Iceland)Longevity Short (20–30 Ka) Long (50–150 Ma) Variable (hours–20 Ma)

as separate entitiesWater gap depth Small (�130 m) Large (�1000 m) Large (�1000 m)Parent rocks Granites Granites BasaltsErodibility Low Low HighFragmentation Low Low (except Balearics and High (except lonely islands

Seychelles) such as Ascension or St. Helena)Original biota Present Present AbsentRelictualism Lacking High ModerateSpeciation process Limited Vicariance Founder event, etc.Endemism Variable according to latitude Very high High

(low in high latitudes,often high in low latitudes)

Page 71: Whittaker. Island Biogeography. 2007

140°E

120°E

100°E

80°E

60°E

40°E 20°E 20°W 40°W

60°W

80°W

100°W

120°W

Days elapsedfrom time oflaunch

5100

90

753040

50

15

20

65

80

95

3555

45

35

25

10

7060

0

5

102

160°E 180° 160°W

Figure 3.8 Floristic affinities of Gough Island (Tristan da Cunha) in relation to predominant wind direction as revealed by the flight of aballoon: areas sharing 20–25 species, cross hatch; 15–20 species, single hatching; 10–15 species, heavy stipple; 5–10 species, light stipple.(After Moore 1979, Figs 1 and 2.)

yet essentially oceanic in character for organismswith poor powers of dispersal through or acrossseas (e.g. conifers, terrestrial mammals, andfreshwater fish). Moreover, there are some groupsof islands of mixed origins and types, e.g. theSeychelles, or which appear to have been

partially connected at times of lowered sea levels, such that they have biological characteristics part way between oceanic and continental, land-bridge islands (as discussed for the avifauna of the Philippines by Diamond and Gilpin, 1983).

58 T H E B I O G E O G R A P H Y O F I S L A N D L I F E

Page 72: Whittaker. Island Biogeography. 2007

distributions arise (see Box 2.2). In essence, dispersalhypotheses have it that descendent forms are theproduct of chance, long-distance dispersal across apre-existing barrier, whereas vicariance biogeogra-phers envisage species ranges being split up by phy-sical barriers, often followed by speciation in the nowseparate populations (Myers and Giller 1988). Somevicariance biogeographers seem remarkably dismis-sive of long-distance dispersal (e.g. Heads 2004),given the evident importance of the process in islandcolonization (Wagner and Funk 1995, McGlone et al.2001; Winkworth et al. 2002; Cook and Crisp 2005,McGlone 2005; Rocha et al. 2006) and indeed in post-separation exchanges between continental land-masses (e.g. Galley and Linder 2006). Although thereare contexts in which the relative contribution of bothprocesses can be exceedingly difficult to discern, pro-viding plenty of room for genuine debate, the polar-ization of debate (and especially the dismissal oflong-distance dispersal) is unnecessary and hasbecome a real hindrance to progress. Available dataconcerning island origins and environmental histo-ries illustrate that both types of process operate, andboth contribute to the complexities of biogeographi-cal patterning (Stace 1989; Keast and Miller 1996;Winkworth et al. 2002; Galley and Linder 2006).

This point can be exemplified with reference tothe geological history of the Caribbean, introducedin Chapter 2. Buskirk (1985) set out to employ thisframework in interpreting the phylogenetic rela-tionships of modern beetles, reptiles, amphibians,and other terrestrial animals in the Greater Antilles.The evidence indicates that the Antilles have beenislands throughout most of the Cenozoic and thattherefore most of their faunas were derived fromover-water colonists and not from the fragmenta-tion of original populations on a large land mass.However, Buskirk suggests that tectonic eventshave affected dispersal and colonization distances.The shear along the northern Caribbean plateboundary has moved Jamaica further away fromCentral America, and thus modern Jamaican speciesthat are relicts or radiations from relatively earlycolonists have more Central American affinitiesthan do the endemic species of neighbouringAntillean islands. Jamaica was largely submerged inthe early to mid-Tertiary, and its modern endemic

fauna lacks the groups that invaded the otherAntilles at about that time. As well as the change insea level, the accompanying climatic changes havebeen important, and in Jamaica have apparentlyselected against members of some animal groupsthat required more mesic habitats (Buskirk 1985).Thus, Buskirk shows that to explain the uniquenessof Jamaica’s fauna, it is necessary to refer to tectonicevents (horizontal movement and extensive lateTertiary uplift), combined with Pleistocene cycles ofclimate and sea-level change. Both dispersal andvicariance processes are involved in this history.

Over the past few million years, vertical changesof land and sea have been of more general signifi-cance to Caribbean biogeography than horizontalland movements. An example is provided byWilliams’ (1972) studies of anoline lizards. Anolesare small green or brown lizards that are more orless the only diurnal arboreal lizards found in mostof the region. Much of the distributional patternwithin this group is related to the submarine bankson which today’s islands stand. Each bank has manyislands on it, but as recently as about 7000 BP as ageneralization, many of these islands were joinedtogether, as they would have been for lengthy peri-ods during the Pleistocene. The present-day distri-butions of anoline lizards thus owe much to eventsduring the low sea-level stands (Williams 1972).

The unravelling of past species movements inorder to determine how and when a speciesreached its present insular distribution presentsconsiderable challenges. Much can depend onassumptions concerning the powers of long-distance dispersal of particular species. Many ques-tions in island biogeography come back to the issueof dispersal, as it is inherently extremely difficult todetermine the effective long-distance range ofpropagules (‘potentially viable units’) (but seeHughes et al. 1994; Whittaker et al. 1997). We willreturn to this theme again; suffice to note for themoment that although dispersal limitations mayprovide parsimonious interpretations of data suchas the anoline lizard distributions, it is important toremain open to new data and insights into disper-sal powers of particular species. This point wasdriven home by Johnson’s (1980) documentation,complete with photographs, of the ability of

D I S H A R M O N Y, F I LT E R S, A N D R E G I O N A L B I O G E O G R A P H Y 59

Page 73: Whittaker. Island Biogeography. 2007

elephants to swim across ocean gaps, using thetrunk as a sort of snorkel. It had been generallyassumed that this behaviour was beyond the abilityor, if not the ability, then the inclinations of ele-phants. Thus, the existence of elephants on anisland was formerly taken to be proof of a formerland connection. As we discuss in the next section,the increasing application of modern phylogeo-graphical methods is gradually bringing improvedresolution to such matters.

Macaronesia—the biogeographical affinities ofthe Happy Islands

From the beginning of the nineteenth century it wasalready evident to the European naturalists whocalled in at what we now term the Macaronesianislands in the course of their transatlantic voyaging,that these islands possessed an interrelated biota.The term Macaronesia (derived from the classicGreek macarion � islands and nesoi � happy) wasfirst used by the British botanist Philip Barker Webbin 1845, with reference to the Canaries (in turn fol-lowing their Roman name Insulae Fortunatae;Stearn 1973). Today, however, the term is morebroadly utilized for the designation of a biogeo-graphical entity including all the north-westAtlantic archipelagoes off the European and NorthAfrican coasts, thus comprising the Azores,Madeira with the Salvage Islands, the Canaries,and the Cape Verde Islands (Fig. 3.9). To the initialcore formed by these archipelagos, some authorshave suggested adding a narrow coastal strip of theAfrican continent, extending approximately fromAgadir (Morocco) to Nouadhibou (Mauritania),including the valleys and wadis (dry riverbeds) ofthe Anti-Atlas mountain chain (Sunding 1979).

With the exception of the Azores, where the vol-canic activity relates to the Mid-Atlantic oceanridge, the origins of the Macaronesian islands liewithin the African plate. Their formation beganperhaps 80 Ma, gradually building the submarineplatforms for the East Canarian and SalvageIslands. Their emergence above sea level begansome 27 Ma in the Salvage Islands, which today arealmost completely eroded back to sea level. Later,the Canaries (from 20 Ma) and Madeira (15 Ma)

emerged as the nucleus of the region (CentralMacaronesia), and finally the southern and north-ern archipelagos, the Cape Verde and Azoreanarchipelagoes, arose some 10 and 8 Ma, respec-tively. Eruptions in recent decades attest to theactive nature of these volcanic systems (e.g.Capelinhos, Faial in the Azores in 1957; Teneguía,La Palma in the Canaries in 1971; and Caldeira doPico, Fogo, in the Cape Verde archipelago in 1995).

The Macaronesian archipelagoes range across aconsiderable latitudinal (40 �N (Corvo, Azores) to15� N (Brava, Cape Verde), and climatic gradient,from the cool-oceanic climate of the Azores to theoceanic tropical monsoon-drift climate of the CapeVerde islands, with the Mediterranean climates ofMadeira and the Canaries in between (Lüpnitz 1995,Fernández-Palacios and Dias 2001). Moreover, thedistances from the African and European mainlandsvary hugely. For example, only 96 km separatesFuerteventura (Canaries) and Stafford Point

60 T H E B I O G E O G R A P H Y O F I S L A N D L I F E

Azores

Madeira

Salvage

Canaries

MA

CARO

NESI

A

Cape Verde

Figure 3.9 The biogeographical area known as Macaronesiaconsists of the island groups as shown and a narrow coastal strip ofnorth-west Africa (after García–Talavera 1999). The affinities with theIberian peninsula are also recognized by some as warranting a link tothe peninsula.

Page 74: Whittaker. Island Biogeography. 2007

(Western Sahara), but Boa Vista (Cape Verde) is570 km from the nearest mainland at Dakar,Senegal, and Porto Santo in Madeira is 630 km fromthe nearest mainland at Cape Sim (Morocco). SaoMiguel in the Azores is some 1370 km from main-land landfall at Lisbon, in south-west Europe.Furthermore, the islands of Corvo and Flores (in theAzores), although occurring within the NorthAmerican plate, are virtually equidistant from CapeRace (Newfoundland) and Lisbon.

Macaronesia has been considered a distinct phyto-geographical region within the Holarctic Kingdomfor more than a century. Floristically, although thereare shared elements, it is now recognized that thereis a transition from strong Eurosiberian–Atlanticaffinities for the Azores, to a Mediterranean flavourfor Madeira and the Canaries and, finally, to aSaharan–Sudanian character for the Cape Verdeislands (Sunding 1979; Nicolás et al. 1989; Kunkel1993), leading some to question the validity of theMacaronesian region (Lobin 1982; Lüpnitz 1995).

Kunkel’s (1993) solution is a hierarchical frame-work recognizing Central Macaronesia (includingthe Canaries and Madeira), Lauri Macaronesia(including Central Macaronesia, Azores, and thesouth-western part of the Iberian peninsula) andfinally Great Macaronesia (including LauriMacaronesia, the Cape Verde islands, and a strip ofthe African coast). Lobin (1982) prefers droppingthe term Macaronesia altogether, placing theAzores in the Atlantic region of the HolarcticKingdom, Madeira and the Canaries in theMediterranean Region (cf. Blondel and Aronson1999), and finally the Cape Verde islands in theSaharan region of the Palaeotropic Kingdom.Lüpnitz (1995) goes still further, advocating theinclusion of the Canaries in the Palaeotropics. Inaddition, although the nucleus of the Canaries andMadeira are recognized to have zoogeographicalaffinities, on the whole Macaronesia is a phytogeo-graphical rather than a zoogeographical concept(Pedro Oromí, personal communication).

What becomes clear from the above is that thesewidely spaced archipelagos have connections ofvarying strength to different mainland regions.Recent developments in our understanding of theenvironmental history of the islands (Chapter 2)

and the phylogeography of Macaronesian lineagesshed considerable light on these phytogeographicalarguments and provide fascinating scenarios of thecolonization history of the region. The general pic-ture is as follows. The islands emerged morerecently than traditionally thought, at a point whenthe continents (Africa, Europe, and North America)were clearly well separated. So, the islands havebeen colonized by long-distance dispersal acrossopen ocean, in a structured fashion, resulting invarying island–mainland affinities acrossMacaronesia: a scenario some historical biogeogra-phers dismiss as improbable.

Successful, structured long-distance dispersalbecomes easier to understand and accept as reason-able when we consider the following factors(Nicolás et al. 1989):

● First, as a result of continuing volcanic activity,new territory has been provided for colonizationthroughout the history of the various archipelagos.● Secondly, because of the existence in this part of

the North Atlantic Ocean of well-consolidatedwind and marine current systems since the closureof the Panama strait (c.3.5–5 Ma), the north-easterlytrade winds and the cool Canarian marine current,respectively, have favoured the constant arrival ofpropagules from the adjacent continents.

● Thirdly, during the repeated and lengthy lowsea-level stands in the Pleistocene, a number of sub-marine banks have emerged as islands, serving asstepping stones at intervals of 200 km or less, facil-itating the flow of colonists to and among thesearchipelagos (Fig. 2.10, 2.11; Carine et al. 2004).

These stepping stones may have been particu-larly important for birds, which in turn have beenresponsible for introducing many plant species tothe islands. The banks existing to the north ofMadeira (Seine, Ampere, Gettysburg, Ormonde)provided stepping-stone connections betweenMadeira and the Iberian peninsula, and the Daciaand Concepción banks similarly linked Madeirawith Africa (Fig. 2.11). Dispersal opportunitiesbetween the Canaries and Madeira were alsoenhanced by these banks (Dacia and Concepción)and by the Salvage Islands. These connectionsmay play some role in explaining the greater biotic

D I S H A R M O N Y, F I LT E R S, A N D R E G I O N A L B I O G E O G R A P H Y 61

Page 75: Whittaker. Island Biogeography. 2007

richness of Madeira and the Canaries and how it isthat these islands come to be regarded as theauthentic core of Macaronesia (e.g. Kunkel 1993),sharing some 50 Canarian–Madeiran endemicplant species (Sziemer 2000), as well as 8 endemicgenera (Bystropogon, Heberdenia, Isoplexis,Marcetella, Normania, Phyllis, Semele and Visnea)(Santos-Guerra 2001). In contrast, the Azores andCape Verde islands were less well served byformer stepping stones, again consistent withthe pattern of affinities (Dias 1996; Brochmannet al. 1997).

Recent molecular phylogenetic analyses havegreatly advanced our understanding of these rela-tionships, indicating the most likely colonizationpathways for particular taxa both between themainlands and Macaronesia and within the region.For example, it appears that some plants ofCanarian origin (certainly species of Aeonium,Echium, Sonchus, and likely Lavandula and Limoniumas well) dispersed across the c.1300 km from theCanaries to the Cape Verde islands, where endemicforms developed (Santos-Guerra 1999).Furthermore, at least six plant taxa appear to havecolonized Madeira from the Canaries: Bystropogon,Aeonium, Convolvulus, Crambe, Pericallis, and thewoody Sonchus alliance (Trusty et al. 2005). Somelineages (e.g. Pericallis) subsequently colonized theAzores, most likely via the southernmost Azoreanisland, Santa Maria, some 850 km from Madeira. Incontrast, Argyranthemum and Crambe, twoAsteraceae genera that have radiated profusely inMacaronesia, are believed to have colonized theCanaries from Madeira (Santos-Guerra 1999).Amongst birds, it has been suggested that thechaffinches colonized the Azores from the Iberianpeninsula, and then colonized the western Canariesvia Madeira, whereas the Canarian blue tit appearsto have colonized Tenerife, in the centre of theCanaries, and then to have spread to the otherislands of the Canaries (Kvist et al. 2005).

In terms of African links, the recent discovery inthe Anti-Atlas valleys of typical Macaronesianfloristic elements, including the Dragon tree(Dracaena draco) and the laurel (Laurus cf. novoca-nariensis), highlight the role of northwest Africa in

the colonization of Macaronesia. Another very inter-esting piece of the jigsaw is the discovery of NorthAfrican endemic species of genera considered to beCanarian in origin. Specific examples include thefollowing species of Aeonium: A. arboreum, A. kor-nelius-lemsii (both in Morocco), A. leucoblepharum,and A. stuessyi (both in Ethiopia). A further exampleis a species of Sonchus (S. pinnatifidus), still extant inthe easternmost Canaries, from where it has colo-nized the African coast, indicating a return trip forlineages that developed on the Canaries havingoriginally colonized from Africa some 4 Ma (Santos-Guerra 2001).

Summing up current understanding based onrecent molecular studies, Carine et al. (2004) reportthat in the majority of genera investigated so far,Macaronesian endemic taxa form a single mono-phyletic group, in most cases with westernMediterranean sister groups, but also with affinitiesrevealed with North Africa and the Iberian penin-sula, and indeed with more widely separatedregions, such as the New World (Madeiran Sedum),East Africa (Solanum) and southern Africa (Phyllis).They also note cases of multiple congeneric colo-nizations, and plausible cases of back-colonizationfrom Macaronesia to continental areas (e.g. in Tolpisand Aeonium).

How robust these scenarios will prove to be inthe light of future collecting and phylogeography isunknowable at the present time. However, we candraw several conclusions from these studies:

● There have been multiple and indeed conflictingcolonization pathways followed by different taxa.● The Macaronesian flora comprises lineagesdiverging from their closest continental relatives atdifferent times, some ancient and some relativelyrecent.● There may well have been distinct waves of colon-ization related to the greatly changing geologicaland climatic history of the region.● Following from this, although the islands aretruly oceanic, such that long-distance dispersal isnecessary to the exchange of lineages with themainlands, this is not to rule out a role for past vari-ation in the strength of barriers between these

62 T H E B I O G E O G R A P H Y O F I S L A N D L I F E

Page 76: Whittaker. Island Biogeography. 2007

3.4 Endemism

Neo- and palaeoendemism

A fair number of island endemics belong to groupsthat formerly had a more extensive, continental dis-tribution. Since colonizing their island(s) in theremote past (over either land or sea), they havefailed to persist in their original form in their main-land range (Box 3.3). The implication is that not all

islands and mainland area, and in this sense bothdispersal and vicariance reasoning has relevance tothe region’s biogeography.● Because of the complexity of the colonization sce-narios within Macaronesia, assigning these islandsto a particular phyto- or zoo-geographic region, ormainland source area, is inevitably going to beproblematic; nevertheless, there is clearly a strongMediterranean flavour to the biota.

E N D E M I S M 63

Box 3.3 Palaeoendemism of the Macaronesian laurel forest

We illustrate the idea of palaeoendemism with anexample drawn, once again, from the islands ofMacaronesia. The laurel forests of the Azores,Madeira and the Canaries are examples of theevergreen broadleaved subtropical biome that canbe found between 25 and 35� latitude in bothhemispheres (i.e. Texas, Florida, Atlantic islands,Chile, Argentina, South Africa, South East Asia,Japan, South Australia, and New Zealand)growing, at least in Macaronesia, under theinfluence of orographic cloud layers (Santos-Guerra 1990). The Macaronesian laurel forests areconsidered to be the only surviving remnants of aonce rather richer forest flora, which wasdistributed in central Europe (fossils found inAustria and Hungary) and southern Europe (fossilsfound in France, Italy, and Spain) during theOligocene–Miocene (Axelrod 1975; Sunding1979). Today we know that the laurel forest hasbeen present on the Atlantic islands at least fromthe late Pliocene (a minimum of 2 Ma) (Sziemer2000) and, very likely, much longer.

The failure of these forests to persist in Europehas been attributed to a range of dramaticgeological and climatic events that took place inEurope following the transition of the Tethysocean seaway into the largely enclosedMediterranean basin during the Tertiary. Inparticular, around the end of the Tertiary, theconnection between the Mediterranean andAtlantic closed between c.5.6 and 5.3 Ma, leadingto the Mediterranean drying out almostcompletely, with associated drought andsalinization. This spectacular event was followedby the glacial cycles of the last 2–3 million yearsand the desertification of the Saharan region,

which began 5–6 Ma (Blondel and Aronson1999). With the closure of the Panama strait, and the associated shifts in global ocean currents, the present-day Mediterranean climatedeveloped some 3 Ma and with it emerged theMediterranean sclerophyllous forest, which wasbetter adapted to the new prevailing climaticconditions of hot, dry summers and cool, wetwinters.

The explanation for the survival of this ‘relict’forest flora on the Atlantic islands, albeit in ratherspecies-poor form, has been attributed to threefactors.

● The islands may have been buffered from theclimatic extremes of mainland southern and central Europe and the Mediterranean by theirmid-Atlantic location.

● The topographic amplitude of the islands hasenabled vertical migrations of several hundredmetres in response to such changes in climate asthe islands have experienced.

● The existence in the higher islands of a stable orographic cloud layer has counteractedthe aridity of the generally Mediterranean cli-mate, especially so in the northern hemispheresummer.

As a result, the laurel forests of the Atlanticislands possess a number of plant species that areconsidered to be palaeoendemics. These includethe following Madeiran/Canarian species knownfrom Tertiary deposits in Europe (5.3 Ma in SouthFrance): Persea indica, Laurus azorica, Ilexcanariensis, Picconia excelsa. However, this is notto say that they are absolutely identical to theancestral forms (see table), and as Carlquist (1995)

Page 77: Whittaker. Island Biogeography. 2007

64 T H E B I O G E O G R A P H Y O F I S L A N D L I F E

Selected Central and South Europe fossil taxa and corresponding palaeoendemic Macaronesian laurel forest taxa (source: Sunding 1979,slightly modified). Sunding’s interpretation of the Canary Islands biogeography is now regarded as flawed in a number of respects: e.g. thesupposition of past land connections between the eastern islands and the mainland, and the notion of woodiness in normally herbaceousgenera as a relictual trait (Carlquist 1995)

Central–South European fossil taxon Corresponding Macaronesian taxon(if different from the fossil)

Adiantum reniformeApollonias aquensis Apollonias barbujana, A. ceballosiAsplenium hemionitisClethra berendtii Clethra arboreaIlex canariensisLaurus azorica Laurus azorica, L. novocanariensisMaytenus canariensisMyrsine spp. Myrsine canariensisOcotea heerii Ocotea foetensPersea indicaPicconia excelsaSmilax targionii Smilax canariensisViburnum pseudotinus Vinurnum tinus ssp. rigidumWoodwardia radicansOther laurel forest species still extant in Iberia and present in Macaronesia:Culcita macrocarpa, Davallia canariensis, Erica arborea, Myrica faya, Prunus lusitanica, Trichomanes speciosum, Umbilicus heylandianus

cautions, some authors have attributed relictualismto features (e.g. woodiness in herbaceous taxa)and lineages that are actually recent products of insitu evolutionary change on islands. Indeed, theCanaries in particular are characterized by plentyof examples of neoendemic speciation andradiation (Santos-Guerra 1999).

In the last few hundred years, humansettlement and deforestation within Macaronesia has led to the loss of most of the laurel forest, especially in the Azores, andtoday large remnants are only to be found inMadeira and the western Canaries, particularly on La Gomera.

species now endemic to an island or island grouphave actually evolved in situ. Cronk (1992) refers tosuch relict or stranded forms as palaeoendemics,whereas species that have evolved in situ areneoendemics. As with all such distinctions, this isan oversimplification, as it implies a lack of changein either island or mainland forms, respectively,which may not strictly be the case.

It should be noted that the significance of relictforms or relictual ground plans as against in situevolution is a topic that generates considerableheat, and indeed some of the more keenly contestedissues in biogeography (cf. Heads 1990; Pole 1994;Carlquist 1995; Winkworth et al. 2002). Cronk (1992)

argues that ancient, taxonomically isolated, relictspecies should hold a special importance in conser-vation terms, in that their extinction would cause agreater loss of unique gene sequences and morpho-logical diversity than the extinction of a specieswith close relations.

There is no doubt as to the importance ofendemic forms on particular islands, although astaxonomic work continues, estimates of the degreeof endemism for an island group vary. Figure 3.10demonstrates this for the flora of the Galápagos.Since fairly complete inventories became avail-able, the number of known endemics has changedrelatively little: the refinements come from

Page 78: Whittaker. Island Biogeography. 2007

improved knowledge of mainland source areasand as a result of taxonomic revisions. Althoughthe proportion of endemics appears to havedeclined (Fig. 3.10b), this is mostly the result ofincreases in numbers of introduced species. Ifthese species are excluded, the proportion ofendemics in the flora has remained reasonablyconstant. The basic pattern established by Hookerand Darwin (Ridley 1994) has thus proved to berobust. As a proportion of the indigenous flora ofthe Galápagos today, the endemics represent 43%of the indigenous taxa (species, subspecies, orvarieties); allowing for recent extinctions anduncertainties in taxonomy, the figure could fall to37% (Porter 1979). Recently published estimatesfor Hawaii are that the following proportions of

native taxa are endemic: birds, 81%; angiosperms,91%; molluscs, 99%; and insects, 99% (Sohmer andGustafson 1993).

The following sections provide a few estimatesthat set such figures in the broader context ofregional or global diversity.

Endemic plants

Groombridge (1992) provides the following ‘bestestimates’ for the number of higher plant speciesin the world: 12 000 pteridophytes (ferns and theirallies); 766 gymnosperms (conifers, cycads, andtheir allies); and 250 000 angiosperms (floweringplants). This makes for a grand total ofabout 263 000 species. It will be recalled that if

E N D E M I S M 65

200

400

600

300

500

700

800

1001840 1860 1880 1900 1920 1940 1960 1980 2000

Num

ber

of s

peci

es

Year

(a)

1840 1860 1880 1900 1920 1940 1960 1980 200020

25

30

35

40

45

50

55

Per

cent

end

emic

Year

(b)

Flora including introduced species

Flora excluding introduced speciesNumber of endemics

Flora including introduced species

Flora excluding introduced species

Figure 3.10 Changing estimates of the Galápagos flora through time: (a) the size of the flora; and (b) the proportion of endemics. (Data fromPorter 1979 and Davis et al. 1995.)

Page 79: Whittaker. Island Biogeography. 2007

66 T H E B I O G E O G R A P H Y O F I S L A N D L I F E

Table 3.2 Native higher plant species richness and endemism ofselected islands or archipelagos. (From Groombridge 1992; Daviset al. 1995, Sziemer 2000, Izquierdo et al. 2004, Traveset andSantamaría 2004, Arechavaleta et al. 2005.) These figures should betreated cautiously as different values for some islands can be found,in cases even within a single publication (see Groombridge 1992,Tables 8.3, and 14.1)

Island or Number of Number of % Endemicarchipelago species endemic species

Borneo 20 000–25 000 6000–7500 30New Guinea 15 000–20 000 10 500–16 000 70–80Madagascar 8000–10 000 5000–8000 68.4Cuba 6514 3229 49.6Japan 5372 2000 37.2Jamaica 3308 906 27.4New Caledonia 3094 2480 80.2New Zealand 2371 1942 81.9Seychelles 1640 250 15.2Fiji 1628 812 49.9Balearic 1359 89 6.5Canaries 1300 570 44.3Hawaii 1180 906 89.8Mascarenes 878 329 37.5Madeira 793 118 14.9Galápagos 529 211 39.9Cook Islands 284 3 1.1Cape Verde 224 66 29.4Azores 197 67 34.0St Helena 74 59 79.7

New Guinea is taken as the world’s largest,islands constitute about 3% of the land surfacearea of the world. The small selection of 20 islandsand archipelagos in Table 3.2 provides a minimumestimate of about 35 500 species endemic to thosesame islands, amounting to some 13.5% of theworld’s higher plant species, indeed, just 10 ofthese islands/archipelagos account for 12.8%. Ithas been calculated that the Pacific islands containabout 22 000 vascular plant species, nearly half ofwhich may have been introduced by humans. Ofthe 11 000–12 000 native species, about 7000 areconsidered to be endemic, most being foundwithin a single island or island archipelago (Daviset al. 1995).

All such figures come with large error margins.Indeed, estimates of the number of flowering

plant species in the world have varied between240 000 and 750 000 (Groombridge 1992, p. 65).However, island species are as much a part ofthese uncertainties as are continental species (seedata for New Guinea and Borneo, Table 3.2), so therelative importance of island species is unlikely tobe seriously diminished by further refinements ofthe estimates. It is a reasonable estimate that aboutone in six of the Earth’s plant species grows onislands, and one in three of all known threatenedplants are island forms (Groombridge 1992, p. 244). Clearly, in terms of plant biodiversity, theislands of the world make a disproportionatecontribution for their land area and are also suf-fering further disproportionate pressure in termsof the maintenance of that biodiversity.

The percentage of endemics varies greatly amongislands. The highest proportions are often associ-ated with ancient continental islands, such asMadagascar and New Zealand. Large numbers ofendemics are also associated with the larger, higheroceanic islands in tropical and warm-temperatelatitudes. Hawaii, to provide a classic example, hasabout 1180 native vascular plants, and of the 1000angiosperms (flowering plants) about 90% areconsidered endemic (Wagner and Funk 1995). Amore conservative estimate for their numbers is 850endemic species, although the figure could conceiv-ably be as high as 1000 (Groombridge 1992; Daviset al. 1995). Some lineages have radiated spectacu-larly. The silversword alliance comprises 28endemic species, apparently of monophyletic ori-gin (i.e. a single common ancestor), but placed inthree genera: Argyroxiphium, Dubautia, and Wilkesia(Wagner and Funk 1995). The genus Cyrtandra(Gesneriaceae) is represented by over 100 species,and although they may not be monophyletic, theirradiation is no less impressive (Otte 1989; Wagnerand Funk 1995). The Macaronesian flora providesthe best developed Atlantic equivalent to Hawaii,with some exquisite examples of largely mono-phyletic radiations. For instance, there are morethan 60 monophyletic endemic species ofCrassulaceae, distributed across the generaAeonium, Greenovia, Aichryson, and Monanthes; thereare more than 20 species representing mono-phyletic taxa within 4 genera of Asteraceae, namely

Page 80: Whittaker. Island Biogeography. 2007

Pericallis, Cheirolophus, Argyranthemum, and Sonchus;and there are almost equally impressive radiationswithin the genera Echium (Boraginaceae), Lotus(Fabaceae), Sideritis (Lamiaceae), and Limonium(Plumbaginaceae) (Sziemer 2000; Izquierdo et al.2004).

Smaller oceanic islands, even in the tropics (e.g.the Cook Islands), have much smaller floras andtypically lower proportions of endemics. This is duein part to the reduced variety of habitats and to thewide mixing of the typically sea-dispersed strand-line species that dominate their floras. Another fea-ture of the data which must be borne in mind whenexamining statistics for island endemism is that insome areas there are numerous species which areendemic to islands, but which are shared betweendifferent geographical entities. Thus, in the LesserAntilles (the Leeward and Windward Islands), theshared or regional island endemics form a consider-able proportion of the endemic flora as a whole.

Endemic animals

The following constitutes a selection of some of theanimal groups that are sufficiently well-known tobe placed into either a regional or global contextwith some degree of confidence.

Land snailsThere may be as many as 30 000–35 000 land snailspecies in the world (Groombridge 1992), of whicha substantial proportion occur on islands (Wallace1902). Some of the better-known of island snailfaunas are quantified in Table 3.3. Although onlyeight archipelagos are listed in the table, theyaccount for approximately 7.7–9.0% of the world’sland snails by present estimates (some regions ofthe world remain poorly known and the currentpicture could change) (Groombridge 1992). Studiesin the Pacific suggest that, once again, it is thelarger, higher oceanic islands (15–40 km2, >400 mASL) that typically have both most species and mostendemics, whereas low-lying atolls have neitherhigh richness nor high degrees of endemism. Onmost islands with high snail diversity, the snailsappear to be concentrated in the interiors, espe-cially mountainous regions with ‘primary’ forest

cover. For those islands with good data, a markedpositive correlation exists between numbers ofendemic plant species and endemic molluscs, butnot between molluscs and birds (Groombridge1992).

InsectsNo attempt will be made to cover the insects in asystematic fashion, but the following statisticsfrom Hawaii and the Canaries give an idea. In thefamily Drosophilidae (fruit flies), 511 species arecurrently named and described for Hawaii, withanother 250–300 awaiting description (Wagner andFunk 1995). Given the pattern of discovery, it isestimated that there may be as many as 1000endemic species. The radiation of tree crickets onthe archipelago has resulted in 3 endemic generaand some 68 species. The orthotyline plant buggenus Sarona is another endemic Hawaiian genus,with 40 known species. Moving across to theAtlantic Ocean, the Canarian insect fauna isincreasingly well studied, and comprises some5700 species and 500 subspecies, of which almost2200 species and 370 subspecies are consideredendemic (Izquierdo et al. 2004). Among them,perhaps the most explosive example of radiation isoffered by the weevil genus Laparocerus, compris-ing some 65 monophyletic species, with approxi-mately another 40 waiting to be described andmay be as many more waiting to be discovered(Antonio Machado, personal communication).

E N D E M I S M 67

Table 3.3 Land snail species richness and endemism for selectedislands for which the data are complete enough for the proportion ofendemics to be estimated (from Groombridge 1992, Table 14.3)

Island or Number of Number of % Endemicarchipelago species endemic

species

Hawaiian Islands c.1000 c.1000 99.9Japan 492 487 99Madagascar 380 361 95New Caledonia 300 c.299 99Madeira 237 171 88Canary Islands 181 141 77.9Mascarene Islands 145 127 87.6Rapa �105 �105 100?

Page 81: Whittaker. Island Biogeography. 2007

68 T H E B I O G E O G R A P H Y O F I S L A N D L I F E

Unfortunately, with the exception of a few well-studied islands, few groups of insects are suffi-ciently well known for broad biogeographicpatterns to be established. The Lepidoptera providean exception. The tropical Pacific butterfly faunaconsists of 285 species, of which 157 (55%) occur oncontinental land masses, i.e. they have not speci-ated after arrival on the islands. One hundred areendemic to a single island/archipelago and 28 areregional endemics, i.e. they are found on more thanone archipelago but not on the mainlands (Adlerand Dudley 1994). Interestingly, butterfly specia-tion in the Pacific archipelagos is primarily a resultof limited interarchipelago speciation. Intra-archi-pelago speciation appears important only in theBismarcks and Solomons, which contain, respect-ively, 36 and 35 species endemic at the archipelagolevel. Of the other 24 archipelagos in the survey,only New Caledonia makes double figures, with 11endemic species, and the others contribute just 18species between them. The Bismarcks andSolomons are the two largest archipelagos in landarea and the closest to continental source areas.Butterflies are generally specialized herbivores,their larvae feeding only on a narrow range ofplants, or even a single species. On the smaller andmore remote archipelagos, the related host plantssimply may not be available for the evolution ofnew plant associations, thereby impeding theformation of new butterfly species (Adler andDudley 1994).

LizardsIn their survey of the 27 oceanic archipelagos andisolated islands in the tropical Pacific Ocean forwhich detailed herpetological data exist, Adler et al.(1995) record 100 species of skinks in 23 genera, ofwhich 66 species are endemic to a singleisland/archipelago and a further 13 are regionalisland endemics. Of the 23 genera, 9 are endemic tothe islands of the tropical Pacific. A few species arewidespread; for example, Emoia cyanura occurs on24 of 27 archipelagos in the survey. The Bismarcks,Solomons, and New Caledonia appear particularlyrich in skinks, but most archipelagos contain fewerthan 10 species. Hawaii has only 3 species, none ofwhich is endemic, in stark contrast to the birds,

insects, and plants. Some Atlantic competition inthe biodiversity statistics game is provided againby the Canaries, which feature the endemic Gallotializards, an interesting monophyletic groupcomposed of 7 extant species of varying sizes, plusat least one more extinct species (Izquierdo et al.2004). Another important group of island lizardsare the Caribbean anolines, which are typicallysmall, arboreal insectivores. This is one of thelargest and better-studied vertebrate genera: about300 species of anoline lizards have been described,half of which occur on Caribbean islands (Losos1994).

BirdsThe examples of adaptive radiation best known toevery student of biology from high-school days areprobably the Galápagos finches and the Hawaiianhoneycreepers. Therefore it will come as no sur-prise that islands are important for bird biodiver-sity. In excess of 1750 species are confined toislands, representing about 17% of the world’sspecies; of these 402, or 23%, are classified as threat-ened, considerably in excess of the 11% of birdspecies worldwide (Johnson and Stattersfield 1990;Groombridge 1992, p. 245).

Adler (1994) has examined the pattern of birdspecies diversity and endemism for 14 tropicalIndian Ocean archipelagos. Probably the mostfamous island endemics of all, the dodos ofMauritius, Rodrigues, and (possibly) Réunion,although extinct, are included as members of thisdata set, which incorporates both extant speciesand extinct species known only from subfossils (i.e.recently extinct species). Using stepwise linearregression it was found that numbers of each of thefollowing were related positively to the number oflarge islands and total land area: total species num-ber; number of continental species; and number ofregional endemics. In addition, less remote andlow-lying islands tended to have more continentalspecies (of the low islands, only Aldabra had itsown endemics), and higher islands tended to havemore local endemics and proportionately fewercontinental species.

Similar results were obtained for the birds of thetropical Pacific, but the Indian Ocean avifauna

Page 82: Whittaker. Island Biogeography. 2007

rodents occur on the Solomons and Bismarcks,neither archipelago being particularly remote fromNew Guinea. But for the more remote islands,beyond the reach of non-volant terrestrial mam-mals (Table 3.1), mammal representation isrestricted to bats, of which flying foxes arearguably the most important group, althoughother types of bat occur on islands. There arebelieved to be 161–174 species of flying fox(Chiroptera: Pteropodidae) and they are distrib-uted throughout the Old World Tropics (but notthe New World). In number of species, range, andpopulation sizes, flying foxes have been the mostsuccessful group of native mammals to colonizethe islands of the Pacific. The Pacific land regionconsists of about 25 000 islands, most of them verysmall. In addition to occupying small islands, mostflying foxes have restricted distributions. Thirtyeight of the 55 island species occupy land areas ofless than 50 000 km2 and 22 of these species occupyland areas of less than 10 000 km2; 35 occur only ona single island or group of small islands (Wilsonand Graham 1992).

Comparisons between taxa at the regional scaleWithin the figures cited above, those from G. H.Adler and his colleagues provide a commonmethodological approach and enable comparisonsof patterns of endemism of different taxa at theregional scale (Table 3.4). The studies are based on26–30 tropical Pacific Ocean islands or islandarchipelagos, plus in one case 14 Indian Oceanislands/archipelagos. The data include all knownextant species plus others known only from sub-fossils.

Both birds and butterflies are capable of activedispersal over long distances and have colonizedvirtually every archipelago and major island withinthe tropical Pacific. Birds have a higher frequencyof endemism than butterflies. Adler and Dudley(1994) consider that birds have superior dispersalability, and that this might be anticipated to havemaintained higher rates of gene flow than in but-terflies, contrary to the higher degree of endemism.One plausible explanation is that the lower rate ofendemism in butterflies might be a consequence ofthe constraints of the co-evolutionary ties with

contained fewer species, had lower proportions ofendemics, and had several families that were muchmore poorly represented, despite being diverse inthe mainland source areas (Adler 1992). The gener-ally small size of the Indian Ocean islands may besignificant in these differences. The total land area isonly 7 767 km2, compared to 165 975 km2 for thePacific study, and there are also fewer island archi-pelagos in the Indian Ocean and they tend to be lessisolated. The poor representation of certain families,notably the ducks and kingfishers, might be due toa shortage of suitable habitats, such as freshwaterstreams and lakes. Intra-archipelago speciation hasbeen rare in the Indian Ocean avifauna, possiblyoccurring once or twice in the Comoros and a fewtimes in the Mascarenes. Adler (1994) suggests thatthere may not be sufficient numbers of large islandsto have promoted intraarchipelago speciation in theIndian Ocean. In contrast, it has been commonwithin several Pacific archipelagos, accounting formost of the endemic avifauna of Hawaii, beingimportant in the Bismarcks and Solomons, andoccurring at least once in New Caledonia, Fiji, theSociety Islands, Marquesas, Cooks, Tuamotus, andCarolines. In conclusion, Adler (1994) echoesWilliamson’s (1981) remark in respect of bird speci-ation on islands that ‘geography is all important’.

MammalsWith the exception of bats, native mammals are nota feature of the most isolated oceanic islands, butless isolated islands include many that either oncehosted or still do host interesting assemblages ofnative and often of endemic mammals. Perhaps themost impressive example is the Philippines, anarchipelago of 7000 mostly true oceanic islands, fea-turing 170 species of mammals in 84 genera, ofwhich 24 (29%) genera and 111 (64%) species areendemic (Heaney et al. 2005). Particularly impres-sive radiations have occurred in fruit bats andmurid rodents, with patterns of endemism in thenon-volant mammals clearly tied to Ice Agebathymetry, i.e. to the configuration of the islandsas they were during the major low sea-level standsof the Pleistocene.

In the tropical Pacific, Carvajal and Adler (2005)note that a number of species of marsupials and

E N D E M I S M 69

Page 83: Whittaker. Island Biogeography. 2007

70 T H E B I O G E O G R A P H Y O F I S L A N D L I F E

particular host plants required by butterfly larvae.This may limit their potential for rapid evolutionarychange on islands (see above). On the other hand,the capacity of lepidopterans to reach moderatelyremote oceanic islands should not be too lightly dis-missed, as evidenced by Holloway’s (1996) long-termlight-trap study on Norfolk Island, which recorded38 species of non-resident Macrolepidoptera overa 12 year period. Norfolk Island is 676 km fromNew Caledonia, 772 km from New Zealand, and1368 km from the source of most of the migrants,Australia. Most of the arrivals appear to be corre-lated with favourable synoptic situations, such asthe passage of frontal systems over the region.

Lizards are also widely distributed in the tropicalPacific. The proportions of skinks in the three cate-gories of endemic are remarkably similar to theequivalent figures for Pacific birds (Table 3.4). Ingeneral, in each of skinks, geckos, birds, and but-terflies, the proportion of species endemic to anarchipelago is best explained by reference to thenumber of large (high) islands and to total landarea. However, scrutiny of the data on an archipel-ago-by-archipelago basis reveals greater differ-ences. For instance, 100% of New Caledonia’sskinks are endemic to the Pacific Ocean islands,and as many as 93% are endemic to the NewCaledonian islands themselves. The equivalent fig-ures for birds (including subfossils) are 47% and33%, respectively (Adler et al. 1995). These datacontrast with those for Hawaii, on which mostbirds are endemic, but where there are no endemic

lizards. Adler et al. (1995) suggest that these differ-ences are explicable in relation to the differingdispersal abilities of lizards and birds. Birds, beingbetter dispersers, reached Hawaii relatively earlyand have radiated spectacularly, whereas the threespecies of skink may only have colonized fairlyrecently and possibly with human assistance. NewCaledonia, in contrast, may not be sufficiently iso-lated to have allowed such a degree of avifaunalendemism to have developed.

Numbers of birds and butterflies are remarkablysimilar on the less remote Pacific archipelagos eastof New Guinea, but on the more remote archipela-gos the numbers of bird species are far greater thanthose of butterflies (Adler et al. 1995). This isreflected statistically by archipelago area being themost important geographical variable in explainingbird species richness, whereas for butterflies isola-tion is the more important variable. Another pat-tern noted by Adler et al. (1995) is that in Pacificbutterflies, Pacific birds, and Antillean butterfliesthe number of endemic species increases moresteeply with island area than does the number ofmore widely distributed species. This observationsupports the idea that a larger island area providesthe persistence and variety of habitats most con-ducive to radiation on isolated islands (see furtherdiscussion in Chapter 9).

The proportions of Pacific island mammals ineach of the endemism categories falls closer to thefigures for birds and skinks than those for butter-flies. However, the 5 marsupial species make it no

Table 3.4 Degree of species endemism among tropical Pacific and Indian Ocean island faunas. (From Adler1992, 1994; Adler and Dudley 1994; Adler et al. 1995; Carvajal and Adler 2005.)

Group Total number Continental Regional endemics Local endemics

Pacific Ocean butterflies 285 157 (55%) 28 (10%) 100 (35%)Pacific Ocean skinks 100 21 (21%) 13 (13%) 66 (66%)Pacific Ocean birds 592 145 (25%) 59 (10%) 388 (65%)Pacific Ocean mammals 106 42 (40%) 7 (6%) 57 (54%)Indian Ocean birds 139 60 (43%) 10 (7%) 69 (50%)

Continental, species also occurring on continental land masses; regional endemics, species occurring on more than one archipelago within the region but not on continents or elsewhere; local endemics,species restricted to a single archipelago or island. NB: Carvajal and Adler (2005) mistakenly give 292 as thefigure for local endemic Pacific Ocean birds. The correct figure is as given above (G.H. Adler, personalcommunication, 2005).

Page 84: Whittaker. Island Biogeography. 2007

concentrate in less apparent taxa (beetles, and thelike), it is understood that many undiscoveredspecies remain for collectors and taxonomistsamong these relatively cryptic taxa. Only slightlymore surprising is the number of vertebrates scien-tifically discovered only in recent years on under-researched islands such as the Philippines (cf.Heaney et al. 2005). Even in birds, probably the best-studied vertebrate group, ‘new’ species have beendescribed in very recent times in remote islands, e.g.the flightless rail, Gallirallus rovianae, endemic to theSolomon Islands and first collected in 1977(Diamond 1991a). Moreover, in the Canaries, whichmight be expected to be very well known scientifi-cally, two quite sizeable endemic lizards new to sci-ence but of cryptic habits were discovered duringthe 1990s, Gallotia intermedia on Tenerife (Hernándezet al. 2000) and Gallotia gomerana on La Gomera(Valido et al. 2000). Following the first discovery ofG. intermedia in the Teno peninsular of Tenerife, afurther population was found in another locationadjacent to a teeming tourist resort.

Turning to already extinct island species, Schüle(1993) summarizes recent studies of vertebrateendemism for the Mediterranean islands, notingthat although the small-animal faunas have showngood persistence, the larger terrestrial vertebratesof the early part of the Quaternary appear to haveincluded endemic ungulates, carnivores, giantrodents, tortoises, and flightless swans, not one ofwhich has survived to the present day. He regardsit as plausible that some of the losses could beattributed to the arrival of new species drivingolder endemic species to extinction by competition.However, it is increasingly clear that some of thisflux can be attributed to climatic change, sea-levelchange, and other environmental forcing factors(e.g. occasionally volcanism) that exposed the exist-ing island assemblages both to substantial environ-mental change and to new colonists (Marra 2005).From the Late Pleistocene onwards, and especiallyin the Holocene, the extinctions—which includedspecies of pigmy elephants, hippos and cervids,flightless swans, tortoises, and rodents—are oftenattributable to the arrival of humans (Schüle 1993;Blondel and Aronson 1999), although Marra (2005)cautions that we still have an incomplete

further than the Bismarcks, and the 18 species ofnative rodents no further than the Bismarck orSolomon islands where 14 of the 18 species areendemics (Carvajal and Adler 2005). The remain-ing native mammal species indicated in Table 3.4are all bats. Variation in richness in the mammaldata appeared explicable as a function of a com-bination of intraarchipelago speciation in archi-pelagos of large islands, and interarchipelagospeciation, particularly among more isolatedarchipelagos.

To summarize, while different taxa have radiatedto different degrees on particular islands or archi-pelagos, it appears to be the case that the greatestdegree of endemism is found towards the extremityof the dispersal range of each taxon (the ‘radiationzone’ sensu MacArthur and Wilson 1967). Islandsthat are large, high, and remote typically havethe highest proportion of endemics. In total,islands account for significant slices of the globalbiodiversity cake.

3.5 Cryptic and extinct island endemics:a cautionary note

The foregoing account has taken little note of the‘state of health’ of island endemics, and some of theassessments reviewed have included many highlyendangered species and others already believed tobe extinct. Before moving on from the geography ofendemism to an examination of the theories thatmay account for the patterns of island ecology andevolution, it is important to consider both the prob-lem of rare cryptic forms and the extent of thelosses already suffered. Otherwise, there may be adanger of misinterpreting evolutionary patternsthat are just the more resistant fragments of for-merly rather different tapestries (Pregill and Olsen1981; Pregill 1986). The role of humans in reducingisland endemics to threatened status or extinction isimportant enough to warrant a separate chapter(Chapter 11) but for the moment these examplescan serve to place a health warning on attempts toexplain current suites of endemic species onislands.

Given the vast number of islands, especially inthe tropics, and the general tendency for diversity to

C RY P T I C A N D E X T I N C T I S L A N D E N D E M I C S : A C AU T I O N A RY N OT E 71

Page 85: Whittaker. Island Biogeography. 2007

occurred as a result of climatic changes and theassociated sea-level changes and habitat alterationsat the end of the Pleistocene, and could be consid-ered a ‘natural’ part of the biogeographical pattern-ing of the region; others were doubtless due tohuman activities, including hunting, habitat alter-ation and the introduction of exotics. Interpretationof the present-day biogeography of the region thusrequires knowledge not only of past faunas but alsoof the extent to which humans have been involvedin their alteration.

The difficulty of this task is illustrated by a studyfrom Henderson Island. In his investigation of42 213 bird bones, 31% of which were identifiable,Wragg (1995) found that 12% of the fossil birdspecies recorded were accounted for by just 0.05%of the total number of identifiable bones, indicatingthem to be of uncertain status, quite possiblyvagrants. Consequently, biogeographical studiesthat rely simply on a list of fossil birds might,exceptionally, assign resident status to temporaryinhabitants. A different form of bias probablyoccurs in other studies because of the large meshsizes (6 or 13 mm) commonly used in sieving soilsamples. Larger-boned species such as seabirds,pigeons, and rails are usually found, but smallpasserines and hummingbirds are much less likelyto be recovered (Milberg and Tyrberg 1993).

As yet, we have only a fragmentary picture ofpast losses, biased towards groups for which thefossil record has been revealing, notably verte-brates and especially birds. Much is still unknown.What, for instance, has been the impact of thelosses from New Zealand of the giant moas on thedynamics of forests they formerly browsed? Howmight the loss of hundreds of populations and afew entire species of Pacific seabirds have influ-enced marine food webs, in which seabirds are topconsumers? (Steadman 1997a). Many island fruitbats and land birds (including many now extinct)have undoubtedly had crucial roles as pollinatorsand dispersers of plants, and thus their extirpationmust be anticipated to have important repercus-sions for other taxa, many of which may be ongo-ing. That is, unless there is more functionalredundancy within oceanic island biotic assem-blages than most authors currently recognize

72 T H E B I O G E O G R A P H Y O F I S L A N D L I F E

understanding of the relative timing of humanarrival and faunal turnover across the islands of theMediterranean. Thus, on the one hand, the loss ofMyotragus balearicus on the Balearic islands appearsattributable to human settlement, which predatesthe extinction event, while the extinction of hippofrom Cyprus, previously attributed to human hunt-ing, has recently been queried because clear evi-dence of temporal overlap of humans with thehippo is lacking.

Estimates of the numbers of Hawaiian birds thatwere lost following Polynesian colonizationc.1500 years ago, but before European contact, nowstands at a minimum of 40 species (Olson andJames 1982). Each of the 16 Polynesian islands thathas yielded in excess of 300 prehistoric bird bonesapproaches or exceeds 20 extirpated species (i.e.lost from those particular islands, but not necessar-ily globally extinct), pointing to the likelihood of farmore extinct species yet to be catalogued. About85% of bird extinctions during historical times haveoccurred on islands (Steadman 1997a), a rate of loss40 times greater than for continental species(Johnson and Stattersfield 1990).

The cataloguing of insular losses is clearly farfrom complete, and although we have some basisfor estimating how many species of birds may beinvolved, for most taxa such calculations remain tobe made (Milberg and Tyrberg 1993). Morgan andWoods (1986) appraise the problem for West Indianmammals. They calculate an extinction rate for theperiod since human colonization of one speciesevery 122 years, and they emphasize that this is aminimum estimate. The mammals of the WestIndian islands today are thus a highly impoverishedsubset of species that escaped the general decima-tion of the fauna. Morgan and Woods caution thatbiogeographical analyses that fail to incorporate theextinct forms are unlikely to prove reliable.

It is unsurprising, but significant biogeographi-cally, that in addition to species disappearing alto-gether, a number of species have gone extinct fromparticular islands during the past 20 000 years,while still surviving on others. This is particularlyevident in the bats, which supply the majority ofpersisting species of mammal, but is also a patternfound among lizards and birds. Many such losses

Page 86: Whittaker. Island Biogeography. 2007

(compare Rainey et al. 1995; Vitousek et al. 1995;Olesen et al. 2002).

It has been observed that the rapidity of human-induced extinctions of island birds has been influ-enced by the nature of the island. The species ofhigh, rugged islands, and/or those with small orimpermanent human populations, have probablyfared better than those of low relief and dense, per-manent settlement. This is exemplified by the find-ing that a larger percentage of the indigenousavifauna survives on large Melanesian islands thanon small Melanesian islands, or on Polynesian orMicronesian islands. This may be due to the buffer-ing effects that steep terrain, cold and wet montaneclimates and human diseases have had on humanimpact (Steadman 1997a; Diamond 2005).

Steadman (1997a) has pointed out that theKingdom of Tonga did not qualify in a recentattempt to identify the Endemic Bird Areas of theworld because of its depauperate modern avifauna,yet bones from just one of the islands of Tonga,‘Eua, indicate that at least 27 species of land birdslived there before humans arrived, around 3000 BP.Forest frugivores/granivores have declined from12 to 4, insectivores from 6 to 3, nectarivores from 4to 1, omnivores from 3 to none, and predators from2 to 1. As is consistent with predation by humans,rats, dogs, and pigs, the losses have been morecomplete for ground-dwelling species. It is highlylikely that the means of pollination and seed dis-persal, not just of the plants of ‘Eua but of islandsacross the Pacific, have over the past few thousandyears been diminished greatly as a consequence ofthis pattern of loss, replicated, as it has been, acrossthe Pacific. That the attrition of endemic speciesfrom oceanic islands may have so altered distribu-tional patterns, compositional structure, and eco-logical processes serves as a caution on ourattempts in the following chapters to generalizebiogeographic and evolutionary patterns acrossislands.

3.6 Summary

This chapter set out to establish the biogeographi-cal and biodiversity significance of island biotas,particularly remote island biotas, and to provide an

indication of the ways in which island assemblagesare distinctive. In global terms, for a variety of taxa,islands make a contribution to biodiversity out ofproportion to their land area, and in this sense col-lectively they can be thought of as ‘hotspots’. Someof course, are considerably hotter in this regardthan others. Their high biodiversity value is widelyacknowledged, as is the threat to island biodiver-sity, and island thus typically feature prominentlyin global and regional conservation prioritizationschemes.

Accompanying their high contribution to globaldiversity (through the possession of high propor-tions of island endemic forms), islands are, how-ever, typically species poor for their area incomparison to areas of mainland. The more isolatedthe island, and the less the topographic relief, thegreater the impoverishment. Remote island biotasare typically disharmonic (filtered) assemblages inrelation to source areas. This can take the form of aclimatic filter and thus a sort of biome shift on theisland, but more interestingly involves thesampling of only a biased subset of mainland taxa.Such biases may take the form of the absence ofnon-volant mammals, the relative lack of represen-tation within a particular family of plants, and soforth, and are interpretable in terms of the relativedispersal powers of the lineages and taxa underconsideration. An island may be a remote isolate(oceanic in character) for a taxon with poor ocean-going powers, but be linked by fairly frequent pop-ulation flows (continental in character) to a sourcearea for a highly dispersive taxon.

Traditional zoogeographic and phytogeographicanalyses have enabled biogeographers to identifyunidirectional dispersal filters (so-called sweep-stake dispersal routes), filter effects between differ-ent biogeographic regions, and cases where islandshave been subject to the influence of multiple sourceareas. We appraise the issues involved in placingoceanic islands into traditional phytogeographicaland zoogeographical regionalism schemes, showingthat remote islands typically receive colonists frommultiple sources and via multiple routes.

Historical biogeography has been for some timeembroiled in debates between those advocating aprominent role for long distance transoceanic

S U M M A RY 73

Page 87: Whittaker. Island Biogeography. 2007

74 T H E B I O G E O G R A P H Y O F I S L A N D L I F E

dispersal and those largely dismissive of thebiogeographical significance of such processes,favouring tectonic scenarios of barrier creation,involving the splitting of ranges (vicarianceevents). Work reviewed in this and the previouschapter on the environmental history of islands andon molecular phylogenies, argues for both sets ofprocesses playing a role, but in particular providesoverwhelming evidence for long-distance dispersalbeing a biogeographically significant process.

In some cases, island forms may have changedless than the mainland lineages from which theysprung, or have persisted while the mainland pop-ulation has failed, in which case they may betermed palaeoendemics. However, where noveltyhas arisen predominantly in the island context, theyare called neoendemics, although it should be rec-ognized that this is to divide a continuum. Recently,molecular phylogenetic analyses have begun totransform our understanding of island lineages andtheir relationships to continental sister clades, asillustrated herein by extensive reference to thecomplex colonization and evolutionary history ofthe Atlantic islands of Macaronesia.

In this chapter we have avoided discussion of themechanisms of evolutionary change on islands,although noting some correlates with highproportions of endemics in island biotas. Withincreasing isolation, island size, and topographicvariety, the number and proportion of endemicspecies increases. Endemism within a taxon appears

to be at its greatest in regions near the edge of theeffective dispersal range. Thus, although a highproportion of its plants and birds are endemic,Hawaii has only three skinks, none of which isendemic. Their colonization may perhaps have beenhuman-assisted and fairly recent: Hawaii is thusbeyond the effective dispersal range for skinks. Incontrast, the less remote islands of the Pacific Oceanevidence a high proportion of endemism of skinksand, on still less remote islands, of rodents. Ancient‘continental’ island fragments (e.g. Madagascar) canalso be rich in endemics, although this can reflectvarying contributions from (1) stranded ‘vicariant’forms and (2) the operation of long-distancedispersal over long periods of time (e.g. particularlysignificant for the New Zealand flora).

Some lineages have done particularly well onoceanic islands, land snails and fruit bats amongthem. Particular genera, such as the HawaiianDrosophila (fruit flies) and Cyrtandra (a plant exam-ple) have radiated spectacularly. Terrestrial mam-mals have made a good showing only on lessremote islands, often those insufficiently remote tohave escaped the attention of Homo sapiens at anearly stage, and many have thus failed to survive tohistorical times. New discoveries continue to bemade of living (extant) and fossil (extinct) islandforms and this serves as a cautionary note, suchdata must be appraised carefully before we canattempt to explain current biogeographical patternsin terms of evolutionary theory.

Page 88: Whittaker. Island Biogeography. 2007

PART I I

Island Ecology

Volcanic islands can reach considerable heights, and then are subject to process of erosion, dissection and sometimescatastrophic landslides. In the distance, the youthful peak of El Teide (3718 m) is less than 1300 years old, whilst theforeground looks across the much older collapse feature of theOrotava valley. In the foreground, a stand of Canarian pine forest is disappearing into the cloud band that forms daily inassociation with the trade wind inversion, and which is a key regulator of water balance and of vegetation zonation on theisland of Tenerife.

Page 89: Whittaker. Island Biogeography. 2007

This page intentionally left blank

Page 90: Whittaker. Island Biogeography. 2007

Area is the devil’s own variable.

(Anon.)

Coastal islands are notorious for their accumulation, at allseasons, of a staggering variety of migrant, stray, and sex-ually inadequate laggard birds. In the absence of specificinformation on reproductive activity, it is thereforeunwarranted to assume tacitly that a bird species, even ifobserved in the breeding season, is resident.

(Lynch and Johnson 1974, p. 372)

Having established the biogeographical signifi-cance of islands, we now move on to the theoriesthat have been developed from their study. Wetake our lead from Robert H. MacArthur andEdward O. Wilson’s (1967) classic monograph Thetheory of island biogeography, by working froma consideration of ecological patterns andprocesses, through island evolutionary processesto the emergent evolutionary outcomes.

In this first ecological chapter, we focus on aresearch programme within ecological biogeogra-phy in which MacArthur and Wilson provided theseminal influence. This programme centres aroundspecies numbers as the principal response variable,and is concerned with: the nature of species–arearelationships; the factors determining the numberof species found on an island; and the rates ofspecies turnover in isolates. MacArthur andWilson’s (1963, 1967) contribution represented abold attempt to reformulate (island) biogeographyaround fundamental processes and principles,relating distributional pattern to population ecolog-ical processes, and at its heart lay a very simple, butpowerful concept. Their dynamic, equilibriummodel postulates that the number of species of a

given taxon found on an island will be the productof opposing forces leading respectively to the gainand loss of species and resulting in a continualturnover of the species present on each islandthrough time (Box 4.1). Although first proposed asa zoogeographic model (MacArthur and Wilson1963), the model was promoted in their 1967 mono-graph as a general model covering all taxa and allforms of island.

Their theory was developed in large measure toaccount for apparent regularities in the form ofspecies–area relationships, and invokes varyingrates of species gain and loss as a function of isola-tion and area respectively. This they captured inthe famous graphical model, in which immigrationrate declines exponentially and extinction rate risesexponentially as an initially empty island fills uptowards its equilibrial richness value (shown by

77

CHAPTER 4

Species numbers games: themacroecology of island biotas

Box 4.1 The first formal statement of thedynamic equilibrium model

‘We start with the statement that

∆s = M + G – D

where s is the number of species on an island.M is the number of species successfullyimmigrating to the island per year.G is the number of new species being addedper year by local speciation (not includingimmigrant species that mainly diverge tospecies level without multiplying), andD is the number of species dying out per year.’From MacArthur and Wilson (1963, p. 378)

Page 91: Whittaker. Island Biogeography. 2007

the intersection). The immigration rate curve flat-tens with increasing isolation, and extinction ratewith increasing area, thereby generating a familyof curves providing unique combinations of rich-ness and turnover for each combination of areaand isolation (Fig. 4.1). Evolution is subsumed inthe model with the assumption that as an islandbecomes ever more isolated, new forms areincreasingly likely to arise as a result of in situ radi-ation rather than immigration. This elegantly sim-ple core model will be termed here theequilibrium model of island biogeography(EMIB), although as others have stressed it isreally the dynamic equilibrial nature of the modelthat distinguishes it.

At the outset, it is important to emphasize thatThe theory of island biogeography contained a lotmore than this simple model, indeed it was reallyjust the starting point within the 1967 volume, andthe authors went on to develop a series of increas-ingly sophisticated models of modifying effects,and of properties of effective island colonists, end-

ing the book by feeding these ideas into evolu-tionary models. As we discuss in a later chapter, atthe time of their collaboration, Wilson had alreadypublished seminal papers on character displace-ment and the taxon cycle. So although it is fair tocriticize the EMIB for its simplicity (which is alsoits strength), it is inaccurate to claim, as somehave, that their theory gave no consideration tospeciation and evolution.

MacArthur and Wilson’s ideas have provedpivotal to debate in island ecology (and conserva-tion biology) for several reasons. One is that theirtheory has focused attention on two of the mostpervasive, fundamental themes of the subject:

● Do ecological systems operate in an essentiallybalanced, equilibrial state, or do they exhibit non-equilibrium modes of behaviour?● To the extent that they approximate equilibrial behav-iour, how inherently dynamic are they over time?

These themes can be found in subfields of biogeog-raphy as widely spaced as diversity theory,

78 S P E C I E S N U M B E R S G A M E S : T H E M AC R O E C O L O G Y O F I S L A N D B I OTA S

Sfs Sfl Sns Snl

Number of species present

I near

Rat

e

Tns

Tfs

I far

P

E large

E small

Figure 4.1 A version of MacArthur and Wilson’s (1963, 1967) equilibrium model of island biogeography (EMIB), showing how immigrationrates are postulated to vary as a function of distance, and extinction rate as a function of island area. The model predicts different values for S(species number), which can be read off the ordinate and for turnover rate (T ) (i.e. I or E, as they are identical). Each combination of island areaand isolation should produce a unique combination of S and T. To prevent clutter, only two values for T are shown.

Page 92: Whittaker. Island Biogeography. 2007

succession theory, and savanna ecology, to namebut a few. The island laboratories once again havebeen a testing ground for developing and refiningideas of broader relevance (Brown 1981, 1986), andthe EMIB was the starting point.

Hence, in the present chapter, we review theelements that MacArthur and Wilson drewtogether in their ecological model, and how wellthe EMIB has fared in the light of empirical test-ing. We go on to review recent developments inour understanding of the dynamics of speciesnumbers in isolates, some of which build on, andothers of which differ from the MacArthur–Wilsondynamic model. As will become evident, thischapter pays little attention to the compositionalstructure of island biotas; rather we are dealing inthe emergent statistical properties, or macroecol-ogy (Brown 1995) of island assemblages. The reg-ularities of form in these macroecologicalproperties point to the existence of underlyingmechanisms or rules governing ecological sys-tems: the efforts reviewed in the present chapterhave been concerned with identifying what theymight be.

4.1 The development of the equilibriumtheory of island biogeography

Several ingredients were combined in the dynamicequilibrium model (see account by Wilson 1995). Thefirst is the long-known observation that speciesnumber increases in predictable fashion in relation toisland area (e.g. Arrhenius 1921)—termed here theisland species–area relationship (ISAR) (Box 4.2).For instance, the zoogeographer Darlington (1957)offered the approximation that, for the herpetofaunaof the West Indies, ‘division of island area by tendivides the fauna by two’. A second key ingredientwas that in examining data for Pacific birds,MacArthur and Wilson (1963) noted that distancebetween the islands and the primary source area (i.e.island isolation) appeared to explain a lot of the vari-ation. The third ingredient was the notion of theturnover of species on islands, a force that Wilson(1959, 1961) had previously invoked on evolutionarytimescales in the taxon cycle (Chapter 9), and whichother biologists had described as occurring on eco-logical timescales, for instance in analyses of plant(Docters van Leeuwen 1936) and animal

T H E D E V E L O P M E N T O F T H E E Q U I L I B R I U M T H E O RY 79

Box 4.2 Typology of species-number curves and relationships: an attempt to demystify

It is important to distinguish the four frequentlyused macroecological tools discussed in this box.Within the wider literature, a range of acronymshave been used for them. We use the acronymsSAC, SAD, and ISAR for convenience, but thereader should beware: the ‘A’ stands for threedifferent terms.

Species abundance distributions: SADsThe species abundance distribution refers to howthe number of individuals is apportioned across thespecies present in a community. SADs can bedisplayed graphically in several ways, e.g. as a plotof the frequency of each species against its rankimportance in the community, or as a plot of the number of species of increasing classes of

abundance (see Magurran 2004). SADs are not aparticularly key tool in island species-numbersstudies, but assumptions concerning the form ofthe SAD are important to theoretical mechanisticarguments linking patterns of richness to speciesturnover (see text).

Species accumulation curves: SACsThe species accumulation (or collectors) curve is agraph of the cumulative number of species (y axis)with increasing sampling effort (x axis), wheresampling effort could be variously formulated,e.g. (1) recording additional individuals within thechosen sampling area, (2) additional sample plotsof a particular type from within a general locality,(3) additional contiguous sampling plots.

Page 93: Whittaker. Island Biogeography. 2007

80 S P E C I E S N U M B E R S G A M E S : T H E M AC R O E C O L O G Y O F I S L A N D B I OTA S

Researchers use SACs to estimate and comparethe species richness of different localities(Magurran 2004). SACs based on design (3) are frequently used in plant ecological research to determine an appropriate minimum plot size to use for sampling vegetation. As sampling effort is represented byarea, these plots are in fact often calledspecies–area curves.

Species–area curvesThe term species–area curve has been applied totwo very different types of plot of speciesnumber versus area. First, it has been used forspecies accumulation curves where the y axisrepresents the cumulative number of species andwhere the x axis often (but not in all studies)represents contiguous areas (as above). Second,it has been used for plots of the species numberper sample, where the x axis often (but not in all

studies) represents spatially separated areas, suchas islands. The use of the same term for suchvery different phenomena has been and remainsthe source of considerable confusion (e.g.compare Scheiner 2004a,b with Gray et al.2004). To save us from future confusion, Grayet al. (2004) argue for restricting the use of theterm species–area curve to plots of numbers ofspecies per sample against sample area, but thedual use of the term seems too deeplyembedded for this proposal to succeed.Moreover, it is easy to find yourself using thesame acronym for species accumulation andspecies–area curves. Our solution here is to avoid the use of the term species–area curvewhen a more precise term can be used: and we hope that the key distinctions betweendifferent types of analysis can then be morereadily grasped. Just remember in reading the wider literature to check what type ofanalysis is being done under the heading of species–area curve or species–arearelationship. Scheiner’s (2003) scheme shown in the figure provides a useful startingpoint in appreciating differences betweenanalyses.

Island species–area relationships: ISARsSimilar to the species–area curve, different authorsmay mean different things when they talk aboutthe species–area relationship (SAR, or sometimesSPAR). In the present work we attempt to restrictthe use of the term to refer to analyses of thenumber of species found on each of a set ofdistinct islands or habitat islands in relation to thearea of each unit. Thus, the only permittedsampling design is the non-nested type (d) in thegraphic. To make this explicit we have added theterm island, to create the acronym ISAR.

(Dammerman 1948) colonization of the Krakatauislands. MacArthur and Wilson linked these observations to work (by Preston 1948, 1962 and oth-ers) on the distribution of individuals within anassemblage, i.e. the Species–Abundance Distribution(or SAD; Box 4.2), which had established the

prevalence of rare species in virtually all ecologicalassemblages.

MacArthur and Wilson’s theory combines allthese elements. In essence, from among the rarespecies typical of ecological systems, thosespecies present in an island system at very low

(a) (b)

(c) (d)

Classification by Scheiner (2003) of different types of samplinglayout used in constructing species–area curves: (a) strictly nestedquadrats, (b) a contiguous grid of quadrats, (c) a regular but non-continuous grid, and (d) a set of areas of varying size, often islands.

Page 94: Whittaker. Island Biogeography. 2007

T H E D E V E L O P M E N T O F T H E E Q U I L I B R I U M T H E O RY 81

densities would have little chance of persistingand should frequently go locally extinct. And, thefurther they had to travel from the mainland, theless frequently they would recolonize. Hence theISARs should vary in relation to isolation, as afunction of varying rates of immigration to andextinction from the islands: producing predictablepatterns of richness and turnover. There shouldalso be systematic variation in SADs accompany-ing the pattern in ISARs; although samplingproblems in detecting the form of the tail of SADs(and analytical problems in comparing them)make this harder to assess.

It has been noted that K.W. Dammerman (1948),in his monograph on Krakatau, had described sev-eral of the key elements of the eventual equilibriumtheory, although he had failed to put them into amathematical framework (Thornton 1992). Anotherbiologist, Eugene Munroe, went one better thanDammerman. In his studies of the butterflies of theWest Indies, Munroe presented a formulaic versionof the equilibrium model, similar to that independ-ently developed by MacArthur and Wilson in their1963 paper (Brown and Lomolino 1989). Munroe’sformula appeared only in his doctoral thesis (1948)and in the abstract of a conference paper publishedin 1953. Neither Dammerman nor Munroe went onto develop the ideas beyond a fairly rudimentaryform.

In the following sections we explore, in a littlemore depth, the components MacArthur andWilson brought together in their theory.

Island species–area relationships (ISARs)

Building on previous work, MacArthur and Wilson(1963, 1967), concluded that for a particular taxonand within any given region of relatively uniformclimate, the ISAR (Box 4.2) can often be approxi-mated by the power function model:

S = cAz

where S is the number of species of a given taxonon an island, A is the area, and c and z are constantsdetermined empirically from the data, which thusvary from system to system.

MacArthur and Wilson, like others before andsince, were interested in determining species–arearelationships in the most tractable fashion for fur-ther analysis, which is to find the best transforma-tion that linearizes the relationship. FollowingArrhenius (e.g. 1921) they favoured logarithmictransformations of both axes (sometimes termedArrhenius plots). The equation thus becomes:

log S � log c � z log A

which enables the parameters c and z to be deter-mined using simple least squares (linear) regres-sion. In this equation, z describes the slope of thelog–log relationship and log c describes its inter-cept. Thus, a low z value (slope) means that there isless sensitivity to island area than for a system ofhigh z value, while c values reflect the overallbiotic richness of the study system, and thus varywith taxon, climate and biogeographical region.Values of z are easy to compare between study sys-tems, but the parameter c changes with differentscales of measurement used (e.g. km2 versusmiles2), requiring conversion of data to a commonmeasurement scale before they can be analysed forcomparative purposes.

From the data at their disposal, MacArthur andWilson (1967) found that in most cases z fallsbetween 0.20 and 0.35 for islands, but that if youtake non-isolated sample areas on continents (orwithin large islands), z values tend to varybetween 0.12 and 0.17. Thus, the slope of thelog–log plot of the species–area curve appeared tobe steeper for islands, or, in the simplest terms,any reduction in island area lowers the diversitymore than a similar reduction of sample areafrom a contiguous mainland habitat.Wilson’s (1961) own data for Melanesian ants(Fig. 4.2) was one of the data sets used in thisanalysis, although there is a problem in theconstruction of this and other such comparisons,as we explain later.

Although it is possible to find many examplesthat broadly support this island–mainland distinc-tion (e.g. Begon et al. 1986, Table 20.1; Rosenzweig1995), there are also exceptions. Williamson (1988)reviews surveys providing the following slope

Page 95: Whittaker. Island Biogeography. 2007

ranges: ordinary islands, 0.05–1.132; habitatislands, 0.09–0.957; and mainland samples �0.276to 0.925. It is important to establish the reliability ofMacArthur and Wilson’s generalizations on z, andwhether the apparently errant values reported byWilliamson can be explained away. More generally,the simple formulation S� cAz in practice has pro-vided a wealth of opportunity for confusion andcontroversy and we therefore return to the topicand update what we have learnt since MacArthurand Wilson, below.

Species abundance distributions

In most plant and animal communities there arefew species of many individuals and many morespecies of few individuals. This is one of the mostrobust patterns in community ecology. There aretwo foundational theories concerning the distribu-tion of abundance, they differ in relation to theabundance of the rarer species. Fisher et al. (1943)suggested that the largest class of species is of thosethat are individually rarest. This gives rise to thelogarithmic series of abundance. Preston (1948,1962) developed the alternative theory that speciesmore typically fit a log–normal series of abun-dance, i.e. that the most numerous species are

those of middling abundance. He argued thatinsufficient sampling commonly had given rise tothe apparent fit of the logarithmic model. This canbe illustrated diagrammatically for a hypotheticalcommunity if the abundances of individual speciesare plotted on a log scale against number of specieson an arithmetic scale (Fig. 4.3). If the sample issmall, then the sparsest species of the log–normaldistribution will not be sampled, and the abun-dance distribution will be that shown to the right ofthe veil line drawn at point a (Preston 1948). Onincreasing the sampling effort, more of the rarerspecies of the system will be sampled, pushing theveil line to position b. By analogy, a small areaveils’—or rather excludes—the existence of all thespecies whose total abundance falls below a criticalminimum.

To put this into a more familiar ecological context,imagine an area delimited for study within a contin-uous woodland habitat, containing many birds, butjust one or two pairs of a particular bird species—therare end of the distribution. Through chance effects,such as harsh weather or predation, the speciesmight be lost locally, but be replaced almost imme-diately by other individuals from the surroundingwoodland taking over the territory. Repeat thisthought experiment, but for an isolated island: the

82 S P E C I E S N U M B E R S G A M E S : T H E M AC R O E C O L O G Y O F I S L A N D B I OTA S

2.6

2.1

1.6

1.1

0.6

0.10 1 2 3 4 5 6 7

Log

spe

cies

num

ber

Log area (km2)

log S = 1.54 + 0.088 log A

log S = 0.03 + 0.298 log A

Figure 4.2 Species–area relationships for ponerine and cerapachyine ants in Melanesia. Solid dots represent islands; open circles, cumulativeareas of New Guinea up to and including the whole island; triangles, archipelagos (not used in the regression); and the square, all of South-EastAsia. (Adapted from Wilson 1961.)

Page 96: Whittaker. Island Biogeography. 2007

species would be lost and might not be replaced byimmigration for some years (e.g. Paine 1985).Average this across the whole island assemblage andthe island will, at any one time, fail to ‘sample’ a sig-nificant number of (principally) the rarer birds foundin the mainland habitat patch (i.e. they will beabsent). Preston showed that a log–normal series ofabundance should give rise to z values of approxi-mately 0.263, towards the low end of the range ofvalues then available from islands and above thoseof continental patches. These differences thereforepointed to a role for isolation via population migra-tion, and, in short, Preston’s theorizing, combinedwith his analysis of island data, led him inevitably toa focus on turnover in his 1962 article, foreshadow-ing by a few months several of the elements ofMacArthur and Wilson’s (1963) paper.

For present purposes these points of detail are notcritical, yet it remains of compelling interest to theo-reticians to determine the best species abundancedistribution (SAD) model, because SADs are foun-dational to much ecological theory (e.g. Hubbell2001). One crude way of analysing this indirectly isby examination of the form of the ISAR, as on theo-retical grounds it is expected that a plot of S versuslog A should produce a straight line where the loga-

rithmic series applies, whereas a log S versus log Aplot should produce a straight line where thelog–normal series applies. By reference to the argu-ment in the previous paragraph, it might be antici-pated that the former situation would be found toapply for large samples (or areas) and the latter forsmall samples (or areas). While noting exceptions,both Williamson (1981) and Rosenzweig (1995) con-clude that empirical studies from islands show afair degree of support for Preston’s model: thespecies–area plot is more usually linear on a log–logplot than in any other simple transformation.Although this has a wider significance, for themoment it is sufficient to note that the position takenon species–abundance patterns by Preston, and byMacArthur and Wilson, was entirely reasonable.

The distance effect

MacArthur and Wilson (1963, 1967) predicted thatISARs would become steeper with increasing geo-graphic isolation. Unfortunately, the increasedimpoverishment of island biotas with increasingisolation is confounded with variations in otherproperties of islands, particularly their area. Thedistance effect has turned out to be difficult to test;

T H E D E V E L O P M E N T O F T H E E Q U I L I B R I U M T H E O RY 83

Increasing isolation diminishes passivesampling by islands

Num

ber o

f spe

cies

, ari

thm

etic

sca

le

Abundance of individual species, logarithmic scale

b a

Veil line

Increased sampling effort by biologist

Figure 4.3 Preston’s log–normal species–abundance relationship. The portion of this hypothetical abundance distribution that is sampled may be a function of either active sampling effort (by the biologist) or passive sampling effort (by an island). Only the portion to the right of theveil line will be sampled.

Page 97: Whittaker. Island Biogeography. 2007

moreover, those tests that have been conductedhave proved equivocal. At the outset, MacArthurand Wilson (1963, 1967) produced a way of quanti-fying the effect for Pacific birds. First, they showedthat there was a fairly tight, straight-line log–logISAR for the closely grouped Sunda Islands, butthat an equivalent analysis for 26 islands and arch-ipelagos from Melanesia, Micronesia, andPolynesia had both a much greater scatter and atendency away from a straight-line relationship.Secondly, they reasoned that a line drawn throughthe points for near islands could be taken to repre-sent a ‘saturation curve’, and the degree of impov-erishment of islands below the line could then beestimated as a function of area by comparison withthe saturation curve. In general, the degree ofimpoverishment, when corrected for area, wasfound to increase from islands ‘near’ to NewGuinea (less than 805 km) to intermediate, to farislands (greater than 3200 km).

Williamson (1988) suggests that this demonstra-tion is invalid because the example used archipela-gos rather than single islands, and includedarchipelagos of different climate or deriving theiravifaunas from different sources. He reports that,in fact, the line tends to be flatter the more distantthe archipelago, such that (with some caveats)slopes of 0.28, 0.22, 0.18, 0.09, and 0.05 can bederived for birds in the progressively more isolatedarchipelagos of the East Indies, New Guinea, NewBritain, Solomons, and New Hebrides (Vanuatu)archipelagos, respectively. It appears that theassumptions made in analysing species–areacurves in relation to isolation, e.g. whether to lumpseparate islands into archipelagos, whether to cal-culate isolation within or between archipelagos,whether to include islands of rather different cli-mates, etc., are critical to the outcome of the analy-sis (compare Itow 1988; Williamson 1988;Rosenzweig 1995). In any case, the discovery sincethe 1960s of numerous ‘new’ bird species from sub-fossil remains on many of the islands of the Pacific(Chapter 3; Steadman 1997a) renders this particularpart of the MacArthur and Wilson analysis prob-lematic in hindsight, as several points on theirgraph are below their ‘natural’ level. Again, we re-evaluate the distance effect below.

Turnover, the core model (EMIB), and itsimmediate derivatives

It was these observations—of ISARs, SADs and dis-tance—which were combined in the dynamic equi-librium theory, through the mechanism of theturnover of species on islands. The theory postu-lates that there are two ways in which islands gainspecies, by immigration or by evolution of newforms, and that these means of increasing speciesnumber will be balanced in the equilibrium condi-tion by processes leading to the local loss of speciesfrom the island in question. Before immersingourselves in the detail of the model, it is importantto take note of the working definitions of the keyterms involved (Box 4.3).

It is a reasonable generalization that immigrationrate diminishes as a simple function of increasingdistance, hence lowering the equilibrium point formore remote islands. This is only partially coun-tered by increased speciation rates on remoteislands, because of the relatively slow pace of phy-logenesis. For the moment we will set aside the roleof in situ speciation and concentrate solely on immi-gration. The local loss of a species population maybe accomplished either by out-migration or by thedeath of the last representatives on an island, ineither case leading to the local extinction (extirpa-tion) of the species. The greater resource base oflarger islands should mean that extinction rates fora given species richness are lower for larger islandsthan for smaller. Thus the EMIB postulates that thenumber of species found on an island represents adynamic balance between immigration (I) andextinction (E), with immigration varying withdistance from source pool and extinction withisland area. This was presented in simple, accessi-ble, diagrammatic form (Fig. 4.1)—possibly one ofthe keys to the broad uptake of the theory (Brownand Lomolino 1989).

MacArthur and Wilson (1967) also consideredthe effects of slightly more complicated configura-tions, for example where chains of islands(‘stepping stones’) are found strung out from amainland, producing alterations in the expectedimmigration functions and thus in equilibriumvalues. However, the basic ideas may be most

84 S P E C I E S N U M B E R S G A M E S : T H E M AC R O E C O L O G Y O F I S L A N D B I OTA S

Page 98: Whittaker. Island Biogeography. 2007

easily grasped without reference to such com-plicating features.

Figure 4.1 illustrates immigration and extinction ashollow curves, for the following reasons. When thenumber of species on an island is small relative to themainland source pool, P, a high proportion ofpropagules arriving on an island will be of speciesnot resident on the island immediately before thatpoint. As the island assemblage gets larger, fewerarrivals will be of species not already present, and sothe immigration rate (as defined in Box 4.2) has todecline as S nears that of the total mainland pool. Thehollow form of the extinction curve recognizes thatsome species are more likely to die out than others,and that in accord with our knowledge of SADs, themore species there are in a sample or isolate, the rareron average each will be, and so the more likely eachis to die out. At equilibrium, immigration and extinc-tion rates should be approximately in balance andthus equal the rate of species turnover (I � E � T ).

Islands of different size or different isolation mayhave the same turnover, or the same species num-ber, but they cannot have both the same turnoverand species number. Figure 4.1 thus shows how the

theory predicts different combinations of S and T asa function of area and isolation. This reading of thefigure uses it as a representation of space. The dia-gram may also be understood as a representation ofchange through time for a particular island, com-mencing with initial colonization by the first inhab-itants, when immigration rate is high, speciesnumber low, and extinction rate low. The curvesmay be followed towards the equilibrium conditionwhen species number is high and stable, immigra-tion rate has declined, and extinction rate has risento meet it. (However, note that the visual appear-ance of the curves would be different if expressedin normal units of time: as shown in Fig 4.4 theincrease in species richness per annum slowsgreatly as I and E converge.) If we take the case of asmall, near island, species number at equilibriumwill be Sns and the rate of turnover at equilibriumwill be Tns. The portion of the I and E curves tothe right-hand side of their intersection can, ofcourse, be disregarded, as once they have met,neither rate should vary further to any significantdegree; the island has reached its dynamicequilibrium point.

T H E D E V E L O P M E N T O F T H E E Q U I L I B R I U M T H E O RY 85

Box 4.3 Definitions of terms involved in analyses of species turnover

MacArthur and Wilson (1967, pp. 185–191)developed the following formulations:

Immigration. The process of arrival of a propag-ule on an island not occupied by the species. Thefact of an immigration implies nothing concerningthe subsequent duration of the propagule or itsdescendants.Immigration rate. Number of new species arriv-ing on an island per unit time.Propagule. The minimal number of individuals ofa species capable of successfully colonizing a hab-itable island. A single mated female, an adultfemale and a male, or a whole social group maybe propagules, provided they are the minimal unitrequired.Colonization. The relatively lengthy persistenceof an immigrant species on an island, especiallywhere breeding and population increase areaccomplished.

Colonization curve. The change through time ofnumbers of species found together on an island.Extinction. The total disappearance of a speciesfrom an island (does not preclude recolonization).Extinction rate. Number of species on an islandthat become extinct per unit time.Turnover rate. The number of species eliminatedand replaced per unit time.

MacArthur and Wilson’s definitions are fine intheory, but in practice difficult to apply inempirical work. Very often available survey datado not allow researchers to distinguish between aspecies that has immigrated and one that isobserved in transit or in insufficient numbers orcondition to found a population. Similarly, totaldisappearance of a relatively inconspicuous plantor animal from a large, topographically complexisland may be more apparent than real.

Page 99: Whittaker. Island Biogeography. 2007

86 S P E C I E S N U M B E R S G A M E S : T H E M AC R O E C O L O G Y O F I S L A N D B I OTA S

The equation for equilibrium is:

where S is the number of species at time t, St�1 is thenumber at time t � 1, I denotes additions throughimmigration, V denotes additions through evolu-tion (where applicable), and E denotes losses byextinction (it will be noted that as species are inte-gers, continuous representations are approxima-tions). Integration of the rates of I and E asthey vary over time produces the colonizationcurve (Box 4.2). This rises steeply initially, but evermore slowly as eventual equilibrium is neared(Fig. 4.4).

MacArthur and Wilson (1967) recognized thatbiologists can rarely, if ever, be certain of recordingall immigration and extinction events in real-worldsystems. But they provided a little-used way aroundthis, reasoning that the colonization curve might beused to back-calculate indirect estimates of whatthe real I and E rates might be at two stages of thecolonization process. The first point is near thebeginning of colonization, where I may be assumedto be close to the rate of colonization and E to beinsignificant (although both assumptions appearrather bold when successional factors areconsidered: Rey 1985; Chapter 5). The second pointis near the equilibrium condition. In this case theydeveloped a proof for derivation of the extinctionrate at equilibrium as a function of the species num-bers on other similar islands already at equilibriumand the time taken to reach 90% of the equilibrialnumber of species on the target island. They

St�1�St�I�V�E

detailed a number of studies in apparent support ofthis line of reasoning. The first of which was forbird recolonization of the Krakatau islands, whichappeared to have reached a very dynamic equilib-rium by the 1920s (MacArthur and Wilson 1963).This example is still cited in support of the theory(e.g. Rosenzweig 1995), but as MacArthur andWilson (1963, p. 384) speculated might be the case,this was in fact a premature conclusion; indeed datahad already been published which showed a furtherincrease in species number (Hoogerwerf 1953).

The above constitutes the simplest manifestationof their theorizing, the basic dynamic equilibriummodel (the EMIB), which occupies the second andthird chapters of MacArthur and Wilson’s (1967)monograph. It provides an essentially stochasticmacroecological model of biological processes onislands, in which the properties of individualspecies get little attention. Its great virtues werefirst, that it was a dynamic model invoking univer-sal ecological and population processes, and sec-ond, its apparent testability. The predictive natureof the equilibrium theory is important. Although itmay have heuristic value without it, the contribu-tion of the theory to biogeography was in large partto point the way towards a more rigorous scientificapproach to the geography of nature. As Brownand Gibson (1983, p. 449) put it:

. . . like any good theory, the model goes beyond what isalready known to make additional predictions that canonly be tested with new observations and experiments.Specifically, it predicts the following order of turnoverrates at equilibrium: TSN � TSF � TLN� TLF.

Num

ber

of s

peci

es p

rese

nt

Rat

e

Time Time

Colonizationcurve

I

E Figure 4.4 Integration of immigration andextinction curves (left) should, theoretically, producethe colonization curve as shown (right). (Redrawnfrom MacArthur and Wilson 1967, Fig. 20.)

Page 100: Whittaker. Island Biogeography. 2007

The next element of their theory sketched out, inmathematical terms, a portrait of the biologicalattributes of the superior colonist and how islandenvironments should ‘select’ for different speciestypes at early and late stages of the colonizationprocess. They borrowed from the standard nota-tion of population models (r, the intrinsic rate ofpopulation increase; K, the carrying capacity), todescribe early colonists as being r-selected—highly mobile, opportunistic colonizers capable ofrapid growth, maturity and population increase,and later colonists as K-selected—slower dispers-ing and growing, but with greater ability to sustaintheir populations in resource-limited systemsapproaching the environmental carrying capacity.These ideas influenced Jared Diamond who in hiswork on island assembly rules introduced theterms supertramps, tramps, and sedentary speciesfor the continuum from highly r-selected (equiva-lent to pioneer in the succession literature) to K-selected (late successional) species (see Chapter 5).

Another chapter considered additional effects ofstepping stones, allowing enhanced dispersalroutes and thus altering patterns of biotic exchangebetween different biogeographic regions. The laterchapters also incorporated Wilson’s ideas concern-ing niche shifts and the taxon cycle and radiationzones. MacArthur and Wilson (1967) thus built uptheir analyses from a highly simplified island eco-logical model, adding increasing evolutionarydetail through the book. In parallel, there is a move-ment from an essentially stochastic first approxi-mation to increasingly deterministic ideas. As theywrote on p. 121, ‘A closer examination of the com-position and behavior of resident species shouldoften reveal the causes of exclusion, so that randomprocesses in colonization need not be invoked.’

Lynch and Johnson (1974, p. 371), in their criticalanalysis of problems inherent in testing the EMIB,argued that for the theory to apply

‘ . . . turnover should reflect the stochastic nature of anequilibrium condition (i.e. the turnover should not beattributable to some systematic bias such as ecologicalsuccession, human disturbance of habitats, introductionof exotic species, etc.)’.

This is a fair assessment in relation to the sim-plest version of the EMIB, but as we have shown, it

is unnecessarily restrictive in relation to the moreextensive overall body of MacArthur and Wilson’stheory, in which some structure is acknowledgedand even predicted to occur. This is indicative of ageneral problem in assessing MacArthur andWilson’s theory: there is scope to define its limits inrather varied ways. Hence, although individualcomponents and propositions may be tested andrefuted, when it comes to assessing their legacy inthe round, it is more a matter of assessing whethertheir assertion (1967, p. 65) that ‘ . . . only equilib-rium models are likely to lead to new knowledgeconcerning the dynamics of immigration andextinction . . . ‘ has been borne out, and whether theresearch programme they were largely responsiblefor initiating continues to pose good questions andproduce interesting insights.

4.2 Competing explanations forsystematic variation in islandspecies–area relationships

Attempts to evaluate and build on MacArthur andWilson’s theory were initially concerned, first, withthe form of ISARs, and second, with the occurrenceof the equilibrial dynamic at the heart of the theory(Gilbert 1980; Williamson 1988). In the followingsections we consider both aspects. In the dynamicmodel (Fig. 4.1), both area and isolation are pre-dicted to influence island richness, but typically theanalytical approach has been to compute the ISARsfor islands of varying isolation and to comparetheir slopes and intersects. Thus, the standardapproach to the problem has been to treat area asthe primary explanatory variable.

MacArthur and Wilson’s EMIB is not the onlyexplanatory model competing to explain patternsof variation in island species richness (e.g. Connorand McCoy 1979; Kelly et al. 1989). Here we listsome alternative ideas that have been proposedeither as alternatives to or as modifiers of the EMIB:some being of potentially wide application, othersof more limited focus.

● Random placement. If individuals are distributedat random, larger samples will contain more species.An island can be regarded as a sample of such

S YS T E M AT I C VA R I AT I O N I N I S L A N D S P E C I E S – A R E A R E L AT I O N S H I P S 87

Page 101: Whittaker. Island Biogeography. 2007

a random community, without reference to particularpatterns of turnover. Connor and McCoy (1979) termthis passive sampling, and advocate its use as a nullhypothesis against all alternatives (debated inSugihara 1981; Connor et al. 1983).

● The equilibrium model (EMIB) postulates thenumber of species on an island as a dynamicequilibrium between immigration and extinction,dependent on island isolation and area. Only thishypothesis involves a constant turnover of species.

● Habitat diversity. The number of species may bea function of the number of habitats: the larger theisland, the larger the number of habitats.

● Incidence functions. Some species can occuronly on large islands because they need large terri-tories, others only on small islands where they canescape from competition (cf. Diamond 1974).

● Species-energy theory. According to this theory,the capacity for richness is considered a function ofthe resource base of the island, where the latter maybe estimated, for example, using total primary pro-ductivity multiplied by area (Wright 1983). Thismechanism may account for variation in speciesrichness but is essentially neutral on the issue ofturnover.

● Small-island effect. According to this theory, cer-tain species cannot occur on islands below a certainsize (e.g. Whitehead and Jones 1969). This effectmay be more apparent in marine islands than non-marine isolates.

● Small-island habitat effect. In contrast to theprevious idea, it has been suggested that in somesystems, small islands may be different in characterbecause of their smallness, so that they actuallypossess habitats not possessed by larger islandsand thus sample an extra little ‘pool’ of species.Alternatively, they may just have a greater diversityof habitats than anticipated from their area, e.g. viatelescoping of altitudinal zones (Chapter 2).

● The disturbance hypothesis postulates thatsmall islands or ‘habitat islands’ suffer greater dis-turbance, and disturbance removes species ormakes sites less suitable for a portion of the speciespool (e.g. McGuinness 1984, and for a contraryexample Wardle et al. 1997). Disturbance might alsoopen up sites to invasion by new members.

The above list illustrates that simply finding asignificant relationship between species and areais not conclusive in evaluating the EMIB. Kellyet al. (1989) made use of a rather underexploitedfeature of the EMIB: that given knowledge of thespecies pool, it should be possible to predict theproportion of species in common betweenislands, or between equal-sized sample areas on theislands. Kelly et al. (1989) have used the latterapproach to evaluate the first three of the abovehypotheses for 23 islands within Lake Manapouri,New Zealand. It had previously been establishedthat there was a statistically significant ISAR forthe whole flora, with some indication that habitatdiversity might be the cause. In their study, theysampled for richness of vascular plant species intwo vegetation types (beech forest and manukascrub) with a fixed quadrat size. This should elim-inate the effect of island area on observed speciesrichness if either the random placement or thehabitat diversity hypothesis is correct, but shouldleave an effect of island area if the equilibriumhypothesis is correct. The equilibrium hypothesisfailed this particular test. As their samplingdesign was also intended to eliminate the randomplacement and habitat diversity hypotheses, theysuggested that other ideas, such as incidencefunctions or the small-island effect, might haveexplanatory value.

Rosenzweig (1995) has expressed some doubtsas to the meaning of this refutation, arguing thatthe samples were too small, that the restriction tosingle habitat types was inappropriate, and thatthe islands were insufficiently isolated. The plotsused by Kelly et al. (1989) were of 100 m2, which isnot only a fairly standard size for use in NewZealand forests of this character (J. B. Wilson, per-sonal communication), but is also respectable inrelation to islands varying from 242 m2 to2.675 km2 in total area. The restriction of samplingto particular forest types also seems a reasonableprocedure for distinguishing between the threehypotheses as stated above. The point about isola-tion, although it does not invalidate the test, is afair one. Unfortunately for advocates of the EMIB,similar problems apply to many other island eco-logical studies. Arguably, this study exemplifies

88 S P E C I E S N U M B E R S G A M E S : T H E M AC R O E C O L O G Y O F I S L A N D B I OTA S

Page 102: Whittaker. Island Biogeography. 2007

the point we develop later, that the EMIB holds rel-evance to far more restricted geographical circum-stances than many of its advocates have envisaged(Haila 1990).

4.3 Island species numbers and ISARs:what have we learnt?

Area and habitat diversity

Two facets of the relationship between area andspecies number have commonly been recognized:first, a direct area or area per se effect, and second,a habitat diversity effect. Although some authorshave argued to the contrary, MacArthur andWilson’s (1967) theory is not really an ‘area per se’argument. They wrote: ‘Our ultimate theory ofspecies diversity may not mention area because itseldom exerts a direct effect on a species’ pres-ence. More often area allows a large enoughsample of habitats, which in turn control speciesoccurrence.’ (p. 8).

The area per se effect refers to the idea that givena uniform habitat, very small amounts of space cansupport only a limited number of individuals, andthus (through the properties of the SAD), will notbe able to sustain all the species capable of occupy-ing the habitat. As the area increases, more individ-uals can accumulate, sufficient for breedingpopulations of more species, and richness thusincreases. The range in area over which richnesswill continue to rise will vary depending on theecological characteristics of the system. It may bevery small for lichens growing on rock outcrops, orvery large for top carnivores. Gibson’s (1986) datafor habitat islands of grassland plants within anEnglish woodland indicated that an island biogeo-graphical effect of this sort, if present in the herbflora, was detectable only in patches of less than0.1 ha. The habitat diversity effect is that as areaaccumulates, so typically does the variability ofhabitat, such that species specializing in e.g. freshwater, boggy, upland, or low pH habitats are addedas the size and complexity of the island increases.

Numerous studies have since been publishedwhich quantify, via regression, values for z, and thevarying explanatory values of area, isolation, alti-tude, island age, and various other measures of

habitat diversity. Comparative studies of differenttaxonomic groups from the same islands typicallyreturn different explanatory models. For example,multiple regression analyses of Lesser Antilleandata selected habitat diversity and maximumelevation for herptiles, isolation for bats, andisolation, area, and habitat diversity for birds(Morand 2000). Area is very often the most impor-tant explanatory variable because (via area per seand habitat diversity mechanisms) area is a primedeterminant of the resource base available tobiological communities. With larger islands orreserves, it becomes increasingly difficult to sepa-rate the area per se effect from the habitat diversityeffect (but see Rafe et al. 1985; Martin et al. 1995).Depending on the size of the species pool, we cannonetheless appreciate that with increasing area ofa uniform habitat there may come a point wherethere is sufficient area to accommodate breedingpopulations of a large array of species, but wherethe lack of particular microhabitats prevents allspecies from colonizing, e.g. the Krakatau islandslack mangrove habitats and thus mangrove species,although propagules are frequently cast up on thebeaches.

Very often, habitat diversity may be expected toaccrue in fairly simple fashion with increasingisland area. However, as shown in many regressionstudies, particular habitat properties can accountfor significant amounts of variance in ISARs.Moreover, some studies have found evidence ofthresholds whereby a degree of non-linearity isintroduced into the ISAR over particular ranges ofarea. An early example is Whitehead and Jones’(1969) study of plant data from Kapingamarangiatoll (a Polynesian outlier in the west Pacific), inwhich they argued for threshold effects as follows.The smallest islands, up to 100 m2 area, are effec-tively all strandline habitats: their flora is thereforerestricted to the limited subset of strandline species.A little larger, and a lens of fresh water can bemaintained, and the next set of plants can colonize,those intolerant of salt water. A little larger still andthe island would be liable to human occupancyand would be topped up with introduced species.Such modifications to the form of ISARs indicate agreater complexity than captured in the simplest

I S L A N D S P E C I E S N U M B E R S A N D I S A R S : W H AT H AV E W E L E A R N T ? 89

Page 103: Whittaker. Island Biogeography. 2007

version of the EMIB, but are not necessarilyinconsistent with equilibrium theories.

Several authors have developed more formalapproaches to address the habitat effects that con-cerned Whitehead and Jones. One such study isBuckley’s (1982, 1985) habitat-unit model of islandbiogeography. The standard method used by mostauthors has simply been to include indices of habi-tat diversity, along with area, isolation, altitude,and other geographical factors, in multiple regres-sions of species number (e.g. Ricklefs and Lovette1999; Morand 2000). Buckley took a differentapproach, dividing islands into definable habitatsand floristic elements, and relating species richnessto area and isolation independently for each unit.His analysis was based on a series of small islets ininshore waters off Perth, Western Australia. He dis-tinguished three terrain units: limestone, white cal-careous sands, and red sands, and constructedseparate species–area relationships for each. Thevalues were then summed for the whole island. Hefound a significantly better fit for his habitat-unitmodel in predictions of actual species richness thanderived from whole-island regression (Buckley1982). This is unsurprising given the increasedcomplexity of the analysis. He subsequentlydeployed the approach in a study of 61 small isletson bare hypersaline mudflats in tropical Australia,consisting of 23 shell, 19 silt, and 19 mixed-compositionsubstrates (Buckley 1985). The flora comprisedthree elements: ridge species, salt-flat species, andmangrove species. In this second study, area wasfound to be the primary determinant of totalspecies number. Moreover, elevation above mudflatwas a better predictor of plant species richness thanwas substrate composition, probably through itseffect on the soil salinity profile (cf. Whitehead andJones 1969). Underlying these results, differentfloristic elements were found to behave differentlyas, not surprisingly, both island area and maximumelevation above mudflats were more importantdeterminants of numbers of ridge species than ofsalt-flat species.

A similar approach to the habitat-unit model hasbeen developed by Deshaye and Morrisset (1988).Working on plants in a hemiarctic archipelago innorthern Quebec, they found that habitats worked

as passive samplers of their respective speciespools, but that at the whole-island level speciesnumber was controlled by both area and habitatdiversity. They forcibly expressed the view thatdata on habitat effects are crucial to interpretationsof insular species–area relationships (cf. Kohn andWalsh 1994). In response to this need, Triantis et al.(2003) have proposed a simple approach by whichto incorporate habitat diversity within speciesnumber modelling for island data sets. Theyintroduce the term choros (K), apparently anancient Greek word that describes dimensionalspace. K is calculated by simply multiplying thearea of the island with the number of differenthabitat types present. Species richness is thenexpressed as a power function of the choros: i.e.S � cKz, which is of course directly analagous to themore familiar S � cAz. They compared the perform-ance of both these models for 22 data sets ofvarying taxa (e.g. lizards, birds, snails, butterflies,beetles, plants), finding significantly improved fitsin all but two cases. They further suggested thatimprovements in fit might be particularly evidentwhen dealing with smaller islands, and theso-called ‘small-island effect’ (below). There aretwo points to be made concerning this approach,the first of which is the likely sensitivity of the out-come to how habitats are defined (Triantis et al.2005). Secondly, that the choros model includes anadditional parameter (the calculation is reallyS � c(H*A)z (where H is habitat number and A isarea), and so an improvement in model fit is inmany respects unremarkable. The approach maywell prove useful in developing more powerfulfits to particular data sets, but the problems ofstandardization of K across studies means that itwill be less easy to incorporate in comparativeanalyses of multiple data sets.

Area is not always the first variable in the model

Not all studies identify area as the primary variable.Connor and McCoy (1979) speculate that non-significant correlation coefficients between speciesnumber and area are probably published less oftenthan discovered, because they may be perceived by

90 S P E C I E S N U M B E R S G A M E S : T H E M AC R O E C O L O G Y O F I S L A N D B I OTA S

Page 104: Whittaker. Island Biogeography. 2007

authors or reviewers to be uninteresting. Oneexample of no relationship is provided by Dunnand Loehle (1988) who examined a series of uplandand lowland forest isolates in south-easternWisconsin and for each data set reported slopes notdiffering statistically from zero. Their isolates werenot real islands, nor was there an enormous rangeof areas in their data set. Factors other than sizeof isolate, such as disturbance and forest-edgeeffects, were argued to be influences on plantspecies number. Their result was consistent withthe indirect nature of the control exerted by area,but also with the obvious point that meaningfuldepiction of species–area effects requires theselection of a wide range of island areas (Connorand McCoy 1979).

Power’s (1972) study of bird and plant specieson the California islands incorporated 16 islands orisland groups with a broad range (0.5–347 km2) ofareas. The results were presented in a path dia-gram showing the relationships among variablesas indicated by stepwise multiple regression(Fig. 4.5). Although area was an important deter-minant of plant species number, it was a relativelypoor predictor of bird species numbers. The role ofclimate in mediating plant species richness wasindicated by the role of latitude. Power suggestedthat near-coastal islands with richer and struc-turally more complex floras tended to support alarger avifauna. The analysis thus indicates theimportance of hierarchical relationships betweenplants and birds in determining island species

numbers, a line of argument which may also bedeveloped in a successional context (Bush andWhittaker 1991).

Distance and species numbers

The depauperate nature of isolated archipelagoshas long been recognized (Chapter 3). Severalstudies have shown that isolation explains a signif-icant amount of the variation in species numberonce area has been accounted for in stepwise mul-tiple regression (e.g. Fig. 4.5; Power 1972; Simpson1974). Problems arise from the practical difficultiesof measuring the isolation of a target islandbecause of the existence of other islands (the con-figuration of which varies through time: Chapter 2),which may serve as networks for species move-ments across oceans, and because of the multi-collinearity often found with other environmentalvariables (e.g. Kalmar and Currie 2006). Moreover,the effective isolation of islands and habitat islandsis not determined solely by the distance from adesignated source pool. Rather, for real islands itvaries in relation to wind and ocean currents(Spencer-Smith et al. 1988; Cook and Crisp 2005),and for habitat islands it also depends on the char-acteristics of the landscape matrix in which theyare embedded (Watson et al. 2005).

Several authors have made use of the powerfunction model as indirect evidence for the pres-ence or absence of equilibrium, and in particularcases, it may be possible to construct highly plausible

I S L A N D S P E C I E S N U M B E R S A N D I S A R S : W H AT H AV E W E L E A R N T ? 91

Number ofplant species

Isolation

Unknowns

Latitude

Area

Unknowns Number ofbird species

0.579

0.249

0.171

0.668

0.142

0.190

Figure 4.5 Path diagram showing relationships among variables in explanation of species numbers of plants and birds on the California islands(modified from Power 1972). The coefficient associated with each path is the proportion of the variation in the variable at the end of the pathexplained by the independent variable at the beginning of the path, while holding constant the variation accounted for by other contributingvariables.

Page 105: Whittaker. Island Biogeography. 2007

scenarios by such means. Studies of non-volantmammals on isolated mountain tops in westernNorth America by J. H. Brown and colleagues pro-vide an illustration (e.g. Brown 1971; Brown andGibson 1983). The mountains of the Great Basinhave been isolated from comparable habitat sincethe onset of the Holocene (c.8000–10 000 years).Brown assumed that immigration across the inter-vening desert habitats did not occur. It follows thatthe system dynamics must be dominated by extinc-tion events, i.e. they are ‘relaxing’ towards lowerdiversity, ultimately to zero species. Small islandswould have had higher extinction rates and havealready lost most of their species, whereas largeislands still retain most of theirs (Brown 1971). Thisexplained the much steeper z values found for thespecies–area curve for mammals than for birds. Forbirds, the isolation involved was not sufficient toprevent saturation of the habitats. Thus, the mam-mals provided a non-equilibrium pattern and thebirds an equilibrium pattern, although not quite tothe EMIB model, as it appeared that on the rareoccasion that a species went extinct, it tended to bereplaced not by a random draw from the mainlandpool, but by a new population of the same species(Brown and Gibson 1983).

It is possible to test such models by independentlines of evidence, if for example, subfossil remainscan be found of species once found on the moun-tains but which have become extinct since isolation.Some such evidence was found in the Great Basin.The approach has been applied to a number ofother North American mountain data sets, forinstance by Lomolino et al. (1989), again showinghow palaeoecological data can be combined withreasoning based on species–area relationships toderive plausible models of the dynamics of theinsular systems. In this case, however, the montaneisolates in the American South-West (Arizona, NewMexico, and southern Utah and Colorado) wereargued to be influenced by post-Pleistocene immi-grations as well as extinctions.

Connor and McCoy (1979) caution against suchprocedures, arguing that the slope and interceptparameters of ISARs should be viewed simplyas fitted constants, devoid of specific biologicalmeaning, and that it is important to examine the

assumptions involved in such studies critically. Forexample, Grayson and Livingston (1993) observedone of the mountain species, the Nuttall’s cottontail(Sylvilagus nuttallii) within the desert ‘sea’, suggest-ing that immigration across these desert oceansmay occasionally be possible. Lomolino and Davis(1997) provide a re-analysis of data for both theGreat Basin and the American South-west archipel-agos. Their approach is predominantly statistical.They conclude that although the most extremedesert oceans (southern Great Basin) represent aneffective barrier to dispersal, in less extreme cases(the American South-West) they represent immi-gration filters, allowing intermountain movementsof at least some forest mammals on ecologicaltimescales. The scenarios constructed are thus morecomplex than the original, invoking both equilib-rium and non-equilibrium systems and varyingcontributions for immigration and extinction.Although plausible, these deductions ultimatelydemand verification by independent lines ofevidence.

Species–area relationships in remotearchipelagos

It was recognized in the EMIB that species numbercould increase through two means, immigration(I) and evolution (V). In ecological time, onlyimmigration occurs at measurable rates, but inremote islands the evolution of new forms may bemore rapid than the immigration rate from main-lands; the inclusion of V was thus necessary forthe theory to extend to such systems. Wilson(1969) subsequently published a modified ‘colo-nization’ curve, showing how species numbermight rise above the value first achieved via aprocess of ecological assortment (e.g. successionaleffects) and, over long timespans, rise again viaevolutionary additions. No means of quantifyingthe timescale or amplitude of the adjustment wasmade, making the adjusted theory difficult to falsify.

Several authors have applied equilibrium think-ing to evolutionary biotas on remote islands (e.g.MacArthur and Wilson 1967; Juvig and Austring1979). If the islands within a remote archipelago

92 S P E C I E S N U M B E R S G A M E S : T H E M AC R O E C O L O G Y O F I S L A N D B I OTA S

Page 106: Whittaker. Island Biogeography. 2007

exchange species moderately frequently, species–area relationships may be detectable at thearchipelago level. Moreover, taxa capable of rapidrates of speciation may perhaps so rapidly saturateeven geologically young islands that a clear positivecorrelation could result between species richnessand island area. This appears to apply to the highlyvolant Hawaiian Plagithmysus beetles, but not to theflightless Hawaiian Rhyncogonus weevils, whichinstead show a pattern of increasing species numberwith increasing island age, and a decline withisland area (the largest Hawaiian island being theyoungest) (Paulay 1994). Juvig and Austring (1979)have undertaken an analysis of species–arearelationships within the Hawaiian honeycreeper-finches, which produced results they claimedwere consistent with equilibrium interpretations.However, their analysis was undertaken before theextent of the anthropogenic species extinctionsbecame clear. The full distribution of this familybefore Polynesian colonization is unknown, castingdoubt upon interpretations of such species–areacalculations (but see the attempt by James (1995) tocorrect for extinctions).

In addition, the history of introductions to remoteislands suggests that, at least for particular taxa (e.g.plants), the total number of species on such islandscan rise significantly, suggesting that the islands inquestion are effectively unsaturated and thereforenon-equilibrial (e.g. Groombridge 1992, p. 150). Thisis perhaps a weak line of argument because carryingcapacities have undoubtedly been changed byhuman alterations of habitats and this, in tandemwith the anthropogenic increase in immigrationrate, will set a new equilibrium point for the islandin question (Sax et al. 2002). However, this does notmean that a new dynamic equilibrium will beachieved. Gardner (1986) has shown that the lizardsof the Seychelles do not produce a significant ISAR,and that there are unfilled niches on some islands.The patterns of distribution of endemic and morewidespread species were found to be best explainedby reference to historical events, such as periodsof lower sea level in the Pleistocene, and to non-equilibrium hypotheses (see also Lawlor 1986).

The applicability of equilibrium ideas to remotearchipelagos may ultimately depend on the

effective response rate of particular taxa and thescale at which they interact with the island environ-ment. It is not sufficient to consider demonstrationof an ISAR as being proof of dynamic equilibrium inthose island groups in which evolutionary phe-nomena are dominant over ecological phenomena(cf. Williamson 1981, 1988; Lawlor 1986).

Scale effects and the shape of species–arearelationships

ScaleAs a number of authors have pointed out, it isimportant to take account of scale factors inanalysing species–area relationships (e.g. Martin1981; Lomolino 1986; Rosenzweig 1995;Sfenthourakis 1996). The details as to which vari-ables best explain variations in S in a given limiteddata set hold fascination principally in their specificcontext. They are difficult to generalize. This is inlarge part because each group of islands (real orhabitat) has its own unique spatial configurationand range of environmental conditions. The rela-tive significance of such key factors as area andisolation will depend in part on how great therange (how many orders of magnitude) the sampleislands encompass for each variable. Studies ofplant diversity where maximum distance is of theorder of 500 m, but where elevational rangeencompasses distinct habitats, vertically arranged,and where area varies by several orders ofmagnitude, are unlikely to show a significantisolation effect, but are likely to show both area andhabitat effects. This is a description of Buckley’s(1982) Perth islets data set, in which the purpose ofthe study required that isolation be effectivelycontrolled. However, it is not just a matter of howgreat the range is but what the upper and lowerlimits of the variables are relative to the ecologyof the taxon under consideration. This can beappreciated in terms of habitat—the addition of apatch of mangrove habitat will not do the same forplant diversity on a tropical island as the additionof an area of montane cloud-forest.

In terms of isolation, the effective mobilityof different taxa (e.g. birds versus terrestrialmammals) and of different ecological guilds

I S L A N D S P E C I E S N U M B E R S A N D I S A R S : W H AT H AV E W E L E A R N T ? 93

Page 107: Whittaker. Island Biogeography. 2007

(e.g. sea-dispersed vs bird-dispersed plants) candiffer greatly, which may also serve to shape the formof species–area comparisons across archipelagos.For a moderately dispersive group, it might beassumed that very small distances have no impacton colonization and thus on species richness. Atthe other extreme, beyond a certain limit, in theorder of hundreds of kilometres, the group mightbe absent, and thus further degrees of isolationagain have no impact. Similarly, Martin (1981) hasshown that where the range of areas spans smallisland sizes, ISAR slopes are liable to be lowerthan for larger areas. It is thus not surprising thatcomparative studies show a range of forms forspecies–area effects between taxa and islands,with different explanatory variables to the fore(e.g. Sfenthourakis 1996). The collective value ofsuch studies is the promise of refinement of, forinstance, the dispersal limits for particular taxa,and thus of insights into the dynamics of particu-lar insular systems (e.g. Lomolino and Davis1997).

What shape is the species–area relationship?Interest in the form of the species–area relationshipis enduring amongst ecologists. Recently,Rosenzweig (1995) has argued that there are four

different ‘scales’ of SPARs (the point scale, islands,intraprovince, interprovince). But, as we show inBox 4.4, it is not only the scale that varies betweenstudies, but the spatial organization of the sampleunits and the nature of the response variable itself.Hence, the first thing to recognize in attempts tounderstand the form of the species–area relation-ship is that the literature encompasses apples(species accumulation curves: SACs) and oranges(island species–area relationships: ISARs). WhilstSACs take a variety of forms, the distinctionbetween them and ISARs is fundamental (Box 4.4,Gray et al. 2004). To reiterate, an ISAR is simply therelationship for a set of islands between the numberof species occurring on each island and the area ofeach island.

So, what is the shape of the ISAR? A trite answerwould be ‘what shape do you want it to be?’ As setout above, the conventional approach has been tofind a simple transformation of the data that moreor less linearizes the relationship, as straight-linefits make the relationship tractable for furtheranalyses. Tjørve (2003) has perhaps controversiallyargued that our approach to model fitting forSPARs should ‘be based on the recognition of biol-ogy rather than statistics’ (p. 833), i.e. that weshould examine the fit of theoretical (mechanistic)

94 S P E C I E S N U M B E R S G A M E S : T H E M AC R O E C O L O G Y O F I S L A N D B I OTA S

Box 4.4 Combining species accumulation curves and island species–area relationships:apples and oranges?

In Box 4.2 we set out a simple typology ofmacroecological tools used in analysing speciesdiversity phenomena in relation to area. Althoughthere are several ways in which areas of mainlandmay be sampled (graphics a, b, and c in Box 4.2)in order to construct a curve of cumulative speciesnumber with increased sampling effort (an SAC),there is only one way to construct an ISAR, andthat is to plot the number of species per islandversus area of that island. The ISAR approach canalso be used for habitat islands, i.e. isolatedpatches of a distinct type of habitat that aresurrounded by a strongly contrasting ‘sea’ ofother habitats. In principle, it can also be adoptedfor comparing scattered sample plots of

increasing area from within an unfragmentedcontinuous ecosystem, but typically, publishedanalyses use spatially nested sampling systemsinstead. Figure 4.2, derived from Wilson (1961), isa classic example. The mainland New Guinea dataset represents cumulative areas of New Guinea: itis thus a nested sampling design, and hencespecies number is of necessity cumulative.According to the typology in Box 4.2, this makesit essentially a SAC on a regional scale, as distinctfrom the island data also plotted in Fig. 4.2,which we would term an ISAR. By plotting anISAR for the island data and comparing them withan SAC for the mainland, might we be comparingapples and oranges?

Page 108: Whittaker. Island Biogeography. 2007

I S L A N D S P E C I E S N U M B E R S A N D I S A R S : W H AT H AV E W E L E A R N T ? 95

Rosenzweig (1995, 2004) has developed analternative typology of species–area curves,stressing differences in scale (see figure), but whichalso spans these different forms of construct. Toshow how our terms relate to his scheme, see thetable below. Rosenzweig (1995) fashions apowerful argument based on the re-analysis ofnumerous data sets, and with importantimplications for conservation biogeography: henceit is important to understand his typology. Incontrast to our present treatment, he uses a singleterm, species–area curve or SPAR, to refer to asequence of SACs and ISARs applied at different‘scales’. The first level is local species accumulationcurves; next comes species–area relationships for aset (archipelago) of islands; next, the constructionof a species accumulation curve within a mainlandon a regional scale; and finally, plots of overallrichness of two or more regions/provinces ofdifferent area. This switching between differenttypes of relationship within his analysis is deliberateand he comments on the implications of thenested design. He writes (Rosenzweig 1995, p. 10)in relation to intraprovincial SPARS:

‘you must remember to keep your subplotscontiguous when you group them to measure thediversity of larger areas. This is called the nesteddesign. If you do not keep them contiguous, butamass them from scattered subplots, the resultwill have a steeper slope.’

Moreover, being a SAC, the form of such arelationship is typically curved (even in log–log

plots), initially rising swiftly and the rate of increaseslowing as more and more of the species pool areincluded in the cumulative total. Hence we see thatthe three types of species–area curve generalized inRosenzweig’s scale diagram, are each constructedin quite different ways, with two being island orisland-type species–area plots and one a SAC.

Returning to Fig. 4.2, the mainland datarepresent a regional SAC and the island data anISAR. Of necessity the former is a smooth plot ofa single data series, whereas the island data caninclude several islands of identical or nearidentical area, but varying species number,producing scatter around the line of best fit. If wewished to make a more direct comparisonbetween the two data series, it would meancalculating an SAC for the island data, addingislands by order of increasing area, and tallyingcumulative richness versus cumulative area. Eventhen, the spatial structure of the two data serieswould differ. Thus, to some extent, Wilson andRosenzweig are comparing apples and oranges,but in the absence (as far as we are aware) ofcomparative analyses based on multiple mainlandvs. island data sets, the implications of thediffering structures of regional SACs and ISARSremain obscure.

The figure shows three of the four forms ofspecies–area curves and how they relate to eachother in respect of slopes and intersects.Rosenzweig’s (1995, 2004) assertion that largerprovinces will necessarily contain more species thansmaller provinces, thus producing positive slopes forthe interprovincial SPAR, is dependent on carefulselection of provinces being compared so as toeliminate variation attributable to other factors (e.g. climate).

How the alternative typologies of species numbersanalyses compareRosenzweig’s scheme is as given in the figure, inwhich each SPAR type has a characteristic range ofslope values, and which generalizes all but hispoint scale SPAR into a combined graphical model.

Our usage Rosenzweig’s schemeLocal SAC The point scale SPARISAR Archipelagic SPARRegional SAC Intraprovincial SPARInter-Regional ISAR Interprovincial SPAR

Interprovincial

Province A

Province B

Log

spe

cies

num

ber

Log area

A’s islands

B’s islandsIntraprovincial: 0.1–0.2Archipelagic: 0.25–0.45Interprovincial: 0.8–1.0

Typical z-value ranges

An idealized set of species–area curves (from Rosenzweig 2004).

Page 109: Whittaker. Island Biogeography. 2007

96 S P E C I E S N U M B E R S G A M E S : T H E M AC R O E C O L O G Y O F I S L A N D B I OTA S

models against empirical data sets rather thansetting out to find the best straight-line fit throughthe data.

Lomolino (2000c) has also questioned theapproach of using data transformation in search ofthe best linear fit. He argues that there may be goodecological reasons to posit more complex scale-dependent relationships, and that untransformedSPARs should exhibit a sigmoidal form, with aphase across low values of area, where speciesnumbers scarcely increase, followed by a rapidincrease with area and a subsequent flattening asthe number of species approaches the richness ofthe species pool (Fig. 4.6).

At the left-hand end of the plot, the existence ofa ‘small island’ effect appears well established(Lomolino 2000c, 2002; Lomolino and Weiser 2001)although this is not to say it occurs in all data sets(below). Lomolino and Weiser (2001) provide aformal analysis of a large sample of ISARs, com-paring the results of conventional regressionanalyses with those obtained from breakpointregression. These analyses show that the small-island effect is comparatively common in islanddata sets. The threshold area tended to be highestfor species groups with relatively high resourcerequirements and low dispersal abilities, and alsoto be comparatively high for more isolated archi-pelagoes. But with regard to the right-hand end,Williamson et al. (2001, 2002) contend that there isno evidence of the upper asymptote whereby thelog–log ISAR flattens off across the largest islands.As Williamson and colleagues note, just as thespecies pool from which isolated biotas arederived contains a finite number of species, so toodoes the area, i.e. any particular ocean system con-tains a largest island. The shape of the ISAR as itapproaches maximum area might be of a varietyof forms, but they contend, empirically failsto show departure from a straight line relationshipat the right hand end when plotted in log–logspace. Recent analyses by Gentile and Argano(2005) of sets of terrestrial isopod data fromMediterranean islands in which sigmoid modelswere evaluated against linear, semi-logarithmic,

and logarithmic models tend to support this view.Despite the occurrence of a small-island effect, asigmoid model failed to provide the bestrepresentation of the data. This suggests thatthe evolutionary effect posited by Lomolinoon large islands in practice has an indistiguish-able impact on the form of ISARs, which isessentially as suggested by MacArthur and Wilson(1967).

It is noteworthy that the form and biologicalmeaning of ISARS and other forms of species–area plots remain the subject of so much debateand confusion. One important reason is thatempirical data sets each have their own uniquecombination of variation in contributory environ-mental variables, of area, elevation, climate,habitat type, isolation, disturbance histories, etc.This introduces both structure (e.g. Kalmar andCurrie 2006) and some noise into ISARs, whichafter all merely consider one independent vari-able: area. Comparison of published regressionmodels, in which richness has been analysedagainst an array of independent variables, notsurprisingly shows a diversity of answers as towhich factors enter the models in which order,and with what particular relationship to richness.Each of these variables has relevance, but doesnot necessarily have significance over all scaleswithin the empirical data gathered. Perhapsunsurprisingly, few analyses have attempted totackle the possibility of threshold responses inrespect of the complex array of potentially impor-tant, interacting variables: even for area, formaltesting for threshold effects and more complexmodels of the form of the relationship has beenonly a recent and fairly controversial develop-ment. What we can say with respect to simpledata transformations, is that conventional analy-ses have shown that sometimes semi-log andsometimes log-log plots of ISAR data provide thebest fit. The generally preferred approach oflog–log plotting of ISAR data typically provides agood fit over most of the range in area, with mostevidence for systematic divergence being of theform of the small-island effect.

Page 110: Whittaker. Island Biogeography. 2007

I S L A N D S P E C I E S N U M B E R S A N D I S A R S : W H AT H AV E W E L E A R N T ? 97

Species–energy theory—a step towards a morecomplete island species richness model?

The term species-energy theory appears to havebeen coined by Wright (1983), who modified theISAR approach by replacing island area with ameasure of the total energy available to a particu-lar trophic level. He did so both in a graphicalmodel equivalent to Fig 4.1 and operationally inthe regression of island species numbers. Hisapproach was thus very similar to that lateradopted by Triantis et al. (2003) in the chorosmodel, i.e. he replaced area in the species–arearegression with the product of energy and islandarea. As with area, available energy does not esti-mate directly the variety of resource types presenton an island, but it is likely to be correlated with it.Wright argued that, particularly for sets of islandsof variable per-unit-area resource productivity, thespecies–energy model should work better than theoriginal.

Wright provided two empirical demonstrations,first for angiosperms on 24 islands worldwide,

ranging in size from Jamaica (12 000 km2) to theisland continent of Australia (7 705 000 km2), andsecond for land (including freshwater) bird specieson an overlapping set of 28 islands, varying fromSumba (Indonesia) (11 000 km2) to Australia. Forangiosperms he used total actual evapotranspira-tion (total AET), a commonly used, although notideal, parameter (see O’Brien 1993, Field et al. 2005),which estimates the amount of water used to meetthe environmental energy demand, and found that the model S � 123(total AET)0.62 explained 70%of the variation in the logarithm of species number.For birds, he used, total net primary production(total NPP), and the model S � 358(total NPP)0.47

accounted for 80% of the variation in log speciesnumber. In both cases, the original species–areamodel explained significantly less variance.Wright’s data were drawn from a set of islandsranging from the equator (e.g. New Guinea) to thearctic island of Spitsbergen (78� N), and thus were afar more climatically and biogeographically hetero-geneous set than would normally be considered.They were thus well chosen to demonstrate how

Immigration/Extinction Dynamics(MacArthur/Wilson Model)

Speciation

Island area

Spec

ies

rich

ness

t-1 t-2Small islandeffect

(stochastic factors)

Figure 4.6 Might the island species–area relationship be scale-dependent? This model, taken from Lomolino (2000c ), Fig. 6, examines the ideaof thresholds in an arithmetic plot of species richness versus area. Starting at the left-hand end, there is little change in species number until acritical threshold is reached, but beyond area t�1, species number increases rapidly, as a function of the immigration/extinction dynamics of theEMIB. With islands larger than area t�2, species number shows a further increase, as it is afforced by in situ speciation. The ‘small-island effect’is shown by many island data sets, but evidence for the existence of the second threshold and the upward curve towards the right-hand end ofthe plot is lacking. See discussion in text.

Page 111: Whittaker. Island Biogeography. 2007

98 S P E C I E S N U M B E R S G A M E S : T H E M AC R O E C O L O G Y O F I S L A N D B I OTA S

bringing climate-based energy availability intoanalyses of island species numbers may improvemodel fits.

Studies on gamma (regional) scale patterning ofdiversity within large landmasses have shown thatremarkably simple climate-based models canaccount for the first-order pattern of species rich-ness regionally and globally (O’Brien 1993, 1998;Hawkins et al. 2003, Field et al. 2005). These studiesimply that there is a causal connection over timeand space between climate-based water-energydynamics, photosynthesis, subsequent biologicalactivity, and species richness of a wide variety oftaxa—and that it is worth giving more attention toclimate and energy flows in island systems.However, there are also good reasons why a simple,universally applicable formulation may prove elusive.

On theoretical grounds, it is to be expected thatspatial scale will be important to the applicability ofspecies–energy theory (Whittaker et al. 2001). Weknow that the species richness of two samples willbe dependent on how large an area is sampled.Comparison of a 1 m2 patch of chalk grassland witha whole hillside of the same habitat will inevitablyfind the whole hillside to be richer. It follows that insearching for causal explanations of diversity pat-terns between differing locations, area should beheld constant. On coarse spatial scales (comparingareas of tens to hundreds of square kilometres), cli-mate is one of the more prominent sets of variablesinfluencing plant species richness. At much finerscales, factors such as geology, slope, and grazingregime are likely to be proportionally more impor-tant. That is to say, different theories of diversitymay hold relevance to different (albeit overlapping)spatial scales (Wright et al. 1993; Whittaker et al.2001). Island species–energy theory builds bothenergy and area in to the ordinate, as island posi-tion on the energy axis is corrected for total islandarea. Thus it is likely that where island sizes spanseveral orders of magnitude, deviations from thebest-fit relationship may have different causes fordifferent portions of the island size continuum.

Wright (1983) argues that species–energy theoryis consistent with several long-established patterns.For instance, the energy requirements higher up the

trophic pyramid are such that smaller areas may beunable to support top predators, and large-bodiedanimals are often among the first to be lost fromareas newly turned into isolates. Wylie and Currie(1993) have applied the species–energy approach tomammals (excluding bats) on land-bridge islandsfrom across the world. They made use of a slightlydifferent set of explanatory variables, but theapproach was otherwise very similar. They found,as Wright had for birds and angiosperms, thatisland energy explained more of the variation inmammal species richness than did area. However,the improvement was much less marked formammals. The range of island sizes in the mam-mals data set was 0.4 km2 to 741 300 km2, and it ispossible that the greater range, and in particular,the extension of range down to much smallerislands, may have given rise to a stronger relation-ship to area in their data. Latitude was also foundto have a stronger effect on mammal species num-ber than Wright had found. Of 100 log–logspecies–area regressions published in Connor andMcCoy (1979), the average explanatory value ofarea was less than 50%. In the three data setsanalysed by the species–energy method, the figuresranged from 70 to 90%.

It is a curious fact that the early impact of Wright’s(1983) paper was far more evident in the literatureon species-richness gradients within and across con-tinents, than in island biogeography. However, arecent paper by Kalmar and Currie (2006), whichmodels island bird richness globally, marks a keystep forward in the application of these ideas withinisland biogeography. Kalmar and Currie compiled adata set of land-bird species richness for 346 marineislands, ranging from 10 ha to 800 000 km2, and rep-resentative of global variation in climate, topogra-phy, and isolation. Using a fifth-root transformationof richness data (close to a logarithmic transforma-tion) and a variety of simple transformations ofexplanatory variables, they found that as much as85–90% of the variation in bird species numberson islands globally can be accounted for by con-temporary environmental variables. Their explana-tory variables included area, elevation, variousisolation metrics, mean annual temperature, andannual precipitation, and they also examined

Page 112: Whittaker. Island Biogeography. 2007

whether faunal region entered their models. Startingwith the individual variables, they find constraining(triangular) relationships between richness as thedependent variable and area, temperature, precipita-tion, and (inversely) distance from the nearest conti-nent, indicating that these variables may each setupper limits to richness, but that richness frequentlyfalls below these limits because of the effects of theother limiting variables. Using correlation and gen-eral linear modelling, they show first, that the inter-action terms between area and precipitation, andbetween area and temperature, are crucial to push-ing the explained variance above 50%, and second,that the slope of the ISAR increases with rise in meanannual temperature, and with precipitation.Although their simpler models are easiest to inter-pret and readily reach R2 values in the range of0.64–0.70, their final multiple regression modelbased on quite complex manipulations of area, pre-cipitation, isolation, temperature, and elevationattained a cumulative R2 of 0.877. Halting the modelafter the inclusion, in this order, of ‘the area–precipi-tation interaction’, ‘distance to continent’, and ‘tem-perature’, provides a model explain 82% of thevariation, which given that the data encompassglobal variation in island environments, and a largerange of island areas, is a remarkably good fit.Faunal region had only a minor signal in the data,accounted for by lower than expected diversity inthe Sub-Antarctic/Antarctic.

One of the most intriguing findings of theiranalyses relates to the role of isolation. Althoughthey find that isolation has similar explanatorypower to climate and area, the interactions of theexplanatory variables tell a different story.According to MacArthur and Wilson (1967), theslope (z values) of ISARs should vary in relation tothe difficulty of species dispersing to islands, i.e.more isolated sets of islands should producesteeper ISAR slopes. By contrast, in these globalanalyses, Kalmar and Currie (2006) report strongarea–climate interactions, which are consistent withWright’s species-energy ideas, but they do not findarea–isolation interactions. This is particularlyintriguing given that their isolation metrics aloneare capable of explaining just under 50% of the vari-ation in their data set, and together with area 54%.

T U R N OV E R 99

These seemingly contradictory findings may indi-cate that by selecting a globally representative set ofislands, they have assembled a data set in which thevariation in ISAR slope due to climate has largelyswamped the intraregion influence expected as afunction of island isolation. So, it is not that isola-tion has no correlation with richness of island birds,but it is hard to disentangle the independent effectof isolation from variation in area (with which dis-tance is negatively correlated), topography, and climate and the further complications of stepping-stone effects that effectively alter isolation fromcontinents. As Kalmar and Currie are quick to pointout, these findings sit uneasily with the prominentrole asserted for isolation in MacArthur andWilson’s dynamic island model.

David Lack (1969, 1976) argued that given the abil-ity of birds to colonize remote islands comparativelyearly in their existence, and then to speciate on them,the apparent impoverishment of remote islandsmight be less a reflection of immigration–extinctiondynamics than of a relative paucity of resourceswithin their food chains. Kalmar and Currie’s analy-ses seem to lend some support to this argument.

The issue remains, to what extent do the patternsdiscussed above reflect dynamic equilibria of theform proposed in equilibrium theory? To addressthis, it is essential, as Gilbert (1980) advocated, togive close consideration to species turnover.

4.4 Turnover

According to equilibrium thinking, it is a reason-able assumption that most islands are at, or closeto, their equilibrium species number most of thetime. The EMIB postulates that this should be adynamic equilibrium, approached by means ofmonotonic alterations in rates of immigration andextinction. Each of these postulates requiresexamination. Before doing so, it is important toconsider the nature of the evidence involved.Whether or not these postulates are well founded,there is another important issue that must be con-sidered, which is: when turnover does occur atmeasurable rates on ecological timescales, is ithomogeneous (across all species) or heteroge-neous (involving particular subsets of species)?

Page 113: Whittaker. Island Biogeography. 2007

This question is of fundamental importance toapplications of island thinking to conservation(Chapter 10).

Pseudoturnover and cryptoturnover

If we are to assess the rate and nature of turnover ofspecies over time, we have to be able to measure itaccurately. Two forms of error have been described.Cryptoturnover refers to species becoming extinctand re-immigrating between surveys (or viceversa), thus depressing turnover rate estimates: aproblem that will increase with the intervalbetween surveys (Simberloff 1976). A different formof error, pseudoturnover, occurs when censuses areincomplete, and information on breeding statusinadequate, leading to species appearing to turnover when they have actually been residentsthroughout, or alternatively when they have neverproperly colonized. For both these reasons, esti-mates of turnover (and of I and E) are dependent oncensus interval (Diamond and May 1977; Whittakeret al. 1989) as well as on the thoroughness of eachsurvey (Lynch and Johnson 1974; Whittaker et al.2000).

In assessing turnover, it is important to recognizethe distinction between colonization rates andimmigration rates (Box 4.3) and to realize that it isexceedingly difficult to measure real immigrationand extinction rates (Sauer 1969; Abbott 1983). Moststudies of faunal or floral build-up are based onlists of species found at different points in time. Therates presented are perforce observed rates ofchanges in lists: despite commonly being labelledimmigration and extinction rates (e.g. Whittakeret al. 1989, 2000), they are really rather differentfrom the terms as originally defined.

For the extinction of a species to occur, thespecies must first be present. Unfortunately, if weaccept that most extinctions are likely to be of therarer species, which are hard to detect, it followsthat it will be difficult to determine which havereally colonized in the first place. Only ‘proper’colonists should be counted in determining theextinction rate. Examination of the definitions ofterms (Box 4.3) reveals a certain fuzziness at the

centre of this business. The distinction betweenimmigration and colonization is indeterminate, asit relies on what constitutes a propagule for a par-ticular taxon (actually it may vary from species tospecies) and by what is meant by ‘relatively lengthypersistence’. Lynch and Johnson (1974) argue thatpersistence (for a colonist) or absence (for an extinc-tion event) through at least one breeding cycleshould be the minimum requirement. In practice,most island biogeographical studies have lackedadequate data on these phenomena, relyingprincipally on lists of species recorded for differentislands or different points in time, withoutadequate knowledge of breeding status, etc. Theseproblems have plagued island turnover studies,and have led to some pointed exchanges in theliterature.

The study that prompted Lynch and Johnson’scritique and which purported to demonstrateturnover at equilibrium was that of Diamond(1969), who censused the birds of the ChannelIslands, off the coast of southern California,50 years on from an earlier compilation. Diamondfound that among the nine islands 20–60% of thespecies had turned over. However, the interpreta-tion of this turnover was disputed by Lynch andJohnson (1974), who, having first pointed to prob-lems of pseudoturnover, went on to argue that mostof the thoroughly documented changes in the birdscould be attributed to human influence. Thisincluded the loss of several birds of prey due to pes-ticide poisoning, and the spread into the islands ofEuropean sparrows and starlings—both part of acontinent-wide process of range expansion. Theyconcluded that the occurrence of natural turnoverat equilibrium had not been established by thestudy. Notwithstanding these data quality prob-lems, Hunt and Hunt (1974) pointed out an inter-esting feature in the data for one of the islands,Santa Barbara: the predatory land birds were repre-sented by tiny populations, generally just one ortwo pairs, and this level of the trophic hierarchythus underwent greater fluctuations in speciesrepresentation than birds of lower trophic levels.

Nilsson and Nilsson (1985) conducted an interest-ing experiment by re-surveying the plants of a series

100 S P E C I E S N U M B E R S G A M E S : T H E M AC R O E C O L O G Y O F I S L A N D B I OTA S

Page 114: Whittaker. Island Biogeography. 2007

T U R N OV E R 101

of small islets within a Swedish lake. Even with con-sistent survey techniques, they achieved at best only79% efficiency per survey. As different species aremissed on different occasions, a significant propor-tion of the recorded turnover could be attributed topseudoturnover. Having assessed the rate ofpseudoturnover, they were then able to calculate abest estimate of real turnover. However, such proce-dures cannot be applied with any confidence tosurveys by different teams many years apart, whereinthe expertise, experience, special taxonomic interestsor biases, methods, and time spent in surveying arescarcely known or quantifiable. If the islands in ques-tion are large or otherwise difficult to survey, the prob-lems are multiplied. It follows that larger, richerislands will have greater rates of apparent turnoverbecause the degree of error in surveying them is liableto be at least as great as for small islands, and willtherefore involve larger numbers of species. Incircumstances of this sort, resort has to be made tospecial pleading in order to justify rate estimates andtheory verification (e.g. Rosenzweig 1995, pp. 250–8).Such data may tell interesting ecological stories, butscarcely allow unequivocal tests of the EMIB to bemade (Whittaker et al. 1989; 2000)

In his 1980 review, Gilbert argued vigorously thatturnover at equilibrium, a key postulate of theEMIB, and one that set it aside from other compet-ing theories, had not been convincingly demon-strated; the best empirical evidence coming fromSimberloff and Wilson’s mangrove arthropodexperiments (below). Simberloff (1976) hadacknowledged the same point in an earlier article,and in the terms demanded by Lynch and Johnson(1974) one is hard pressed to find convincing evi-dence of stochastic, equilibrial turnover to this day.One paper that comes close is Manne et al.’s (1998)analyses of immigration and extinction rates forbird species on 13 small islands of the British Isles.They use a maximum likelihood method to test ifimmigration rate declines with increasing distance,and extinction rate declines with island size (bothshould be concave functions, as Fig 4.1). Althoughthe relationships were found to be in the predicteddirection, the analyses failed to support the conclu-sions statistically.

When is an island in equilibrium?

MacArthur and Wilson (1967) understood thatmeasuring the real rates of I and E in the field isexceptionally difficult. They therefore suggested ameans of testing turnover using the variance–meanratio. They argued that if a series of islands of simi-lar area and isolation were to be selected, it is rea-sonable to suppose that the variance of the numberof species on different islands will vary with thenumber of species present, i.e. variance should bea function of the degree of saturation. At the earlieststages of colonization the variance should be closeto 1, declining to about 0.5 as islands becomesaturated. This provides an approach to assessingwhether a series of similar islands (i.e. similar in iso-lation, area, etc.) might be viewed as equilibrial (e.g.Brown and Dinsmore 1988), but does not constitutea particularly precise tool.

Simberloff (1983) argued that for equilibriumstatus to be judged it is necessary for researchersto spell out how much temporal variation theywill accept as consistent with an equilibriumcondition. Only by adopting a specific coloniza-tion model, and then testing it against the data,can the equilibrium hypothesis be consideredfalsifiable. He undertook a study using a simpleMarkov model, in which extinction and immigra-tion probabilities were estimated independentlyfor passerine birds of Skokholm island, land birdsof the Farne Islands, and birds of Eastern Wood (ahabitat island in southern England). By compari-son of the simulations with the observed dataseries of between 25 and 35 years, he found a poorfit with the equilibrium model. The data did notshow the regulatory tendencies expected ifspecies interactions cause species richness to becontinuously adjusted towards equilibrium. Foran alternative statistical analysis of these andother data sets, reaching a similar conclusion ofnon-equilibrium dynamics, see Golinski andBoecklen (2006).

As defined by MacArthur and Wilson (1967) intheir glossary (Box 4.3), I, E, and T rates concernchange per unit time, and take no account of the sizeof source pools (although they do elsewhere in that

Page 115: Whittaker. Island Biogeography. 2007

102 S P E C I E S N U M B E R S G A M E S : T H E M AC R O E C O L O G Y O F I S L A N D B I OTA S

publication). A number of analyses of thesephenomena have used formulations in which ratesare expressed as a function of the size of the biota onan island during the course of the study, and/or asa function of the size of the mainland pool (e.g.Thornton et al. 1990). Where P is unknown, orunknowable, then evaluation of the theory must fol-low one of the other routes provided by MacArthurand Wilson (1967). Particular formulations may beoptimal for different contexts (e.g. early periods ofbuild-up, or near-equilibrium condition) andarguably should be given a different nomenclature(see the useful discussion in Thornton et al. 1990).

The rescue effect and the effect of island areaon immigration rate

The EMIB promises a workable, testable modelbecause extinction rate is influenced by area (or itscorrelates) and immigration rate by isolation. If,however, it can be shown that area also influencesimmigration, and isolation influences extinction,then the model shown in Fig. 4.1 is in some jeop-ardy (Sauer 1969).

It will be recalled that immigration rate is theterm given to the arrival of species not already pres-ent on an island. It is logical to assume that a near-shore island for which there is a high immigrationrate will also continue to receive additional immi-grants of the species which are present. This mightbe termed supplementary immigration to distin-guish it from immigration proper. An island popu-lation declining towards extinction might berescued by an infusion of new immigrants of thesame species, thereby lowering extinction rates onnear-shore islands compared with distant islands.This effect of supplementary immigration wastermed the rescue effect by Brown and Kodric-Brown (1977), who censused the number of indi-viduals and species of arthropods (mostly insectsand spiders) on individual thistles growing indesert shrubland in south-east Arizona. Their find-ings conformed quite well with the predictions ofEMIB, with the exception that turnover rates werehigher on isolated plants than on those in closeproximity to others. This was argued to be becauseof increased frequencies of immigration events for

species present on the thistles at time 1, either pre-venting extinctions or perhaps effecting recoloniza-tion before a subsequent survey at time 2. Eitherway, each census was likely to record the species aspresent on the less-isolated thistles. It has beenpointed out that this study was based on rather anotional form of island (individual thistles) andwas not based on breeding populations. Since thephrase was coined, the rescue effect has beeninvoked by a number of authors in explanation oftheir data (e.g. Hanski 1986; Adler and Wilson1989); often, as in the original, this seems to bebased on its good sense rather than on any directproof (but see Lomolino 1984a, 1986).

Similarly, immigration rate may be affected byarea. A large island presents a bigger target for ran-dom dispersers. A simple study of sea-dispersedpropagules on 28 reef islands in Australia showedthat the fraction of the dispersing propagule poolintercepted by each island is proportional tobeach length (Buckley and Knedlhans 1986).Larger islands also present a greater range ofhabitats and attractions to purposeful dispersers,such as some birds, elephants swimming across astretch of sea, or mammals crossing ice-coveredrivers in winter (Johnson 1980; Lomolino 1990).Williamson (1981) points out that MacArthurand Wilson recognized the possible influence ofarea on immigration in their 1963 paper butdiscreetly sidestepped this issue in their 1967monograph.

At least one study has reported both the above‘problems’. Toft and Schoener (1983) undertook atwo-year study of 100 very small central Bahamianislands, in which they recorded numbers of indi-viduals and species of diurnal orb-weaving spi-ders. They found that the extinction rate wasrelated positively to species number and to dis-tance, and was negatively related to area.Immigration rate was positively related to area andnegatively related to species number and distance.In consequence, the turnover rate showed nostrong relationship to any variable, thus the netresult of these interactions is that the effects of areaand distance on turnover tend towards the inde-terminate. Put simply, the EMIB does not work forthis particular insular system.

Page 116: Whittaker. Island Biogeography. 2007

T U R N OV E R 103

The path to equilibrium

One of the few rigorous tests of the process ofcolonization and development of equilibrium wasprovided by Simberloff and Wilson’s classicexperiments on mangrove islets in the Florida Keys(Simberloff and Wilson 1969, 1970; Simberloff1976). These experiments involved, first, a fullsurvey of all arthropods on individual isolatedmangrove trees (11–18 m in diameter); secondly,the elimination of all animal life from islets; andthirdly, the monitoring of the recolonization.Although seemingly a satisfactory demonstrationof the achievement of equilibrium and one inwhich turnover was occurring, there were someproblems. For instance, certain species groupswere regarded as not treating the trees as an island.Some of these groups were included in the analysisand others excluded. Simberloff also had problemsdetermining which species were proper immigrantsas opposed to transients (Williamson 1981). Oneinteresting features was that a degree of overshootoccurred, suggesting that the islands could sup-port more than their ‘equilibrium’ number ofspecies while most species were rare, but as popu-lations approached their carrying capacities, com-petition and predation eliminated the excessspecies. As Simberloff (1976, p. 576) described it:‘more highly co-adapted species sets find them-selves by chance on an island and persist longer assets’. This development was termed an assortativeequilibrium.

Demonstration of equilibrial turnover demandsthe most stringent data. The clearest evidenceinevitably comes from ‘islands’ which are relativelysmall and simple, and for organisms with fast lifecycles. The mangrove ‘island’ experiments are theclassic first test and they fit this description. Thesetiny patches of mangrove consist of a single habitattype (albeit not to an insect), the arthropods con-cerned have very short generation times, and canbe expected to respond very rapidly to changingopportunities and to exhibit relatively little com-munity development. Such features may be atypi-cal of islands in general. It is also notable thatalthough turnover rates were indeed high in theearly stages of the experiment, a couple of years on,

once the assortative process was completed, andpseudoturnover or transient species removed fromthe analysis, the rates were found to have slowedgreatly. This is not expected if area and isolationremain constant.

A similar study was conducted by Rey (1984,1985). His islands were patches of the grass Spartinaalterniflora, varying in size from 56 to 1023 m2, butstructurally simpler than the mangroves. Onceagain, fumigation was used and arthropod recolo-nization was monitored on a weekly basis. Initially,the colonization rate was slow because extinctionrates were high. As the assemblages built up, popu-lations persisted longer and extinction rates fell. Reymakes a useful distinction between two extremes inturnover patterns. If all species are participatingequally in turnover, such that each species isinvolved somewhere in the archipelago in extinc-tion and immigration events, he terms the patternhomogeneous turnover. If, in contrast, only a smallsubset of the species pool is involved in turnover,this would be heterogeneous turnover. If stronglyheterogeneous, the turnover would be inconsistentwith the EMIB. Rey’s data were intermediatebetween these two extremes, neither totally homo-geneous, nor excessively heterogeneous.

Why is heterogeneous turnover problematic forthe EMIB? The most basic equilibrium model ofMacArthur and Wilson (1967) contains the assump-tion that each species has a finite probability ofbecoming extinct at all times. The rate of extinctionwould then depend on how many species therewere. This model assumes entirely homogeneousturnover. However, their favoured model took intoaccount that the more species there are, the rarereach is (on average) and hence an increased numberof species increases the likelihood of any givenspecies dying out. This results in the curved func-tion for extinction. This is clearly correct under sto-chastic birth and death processes, provided youassume that the more species there are on an island,the smaller, on average, each population is. Asalready established, this provides us with the hol-low forms of immigration and extinction curve andthe expectation that equilibrium species numbershould be approached by a colonization curve ofnegative exponential form. It also generates the

Page 117: Whittaker. Island Biogeography. 2007

104 S P E C I E S N U M B E R S G A M E S : T H E M AC R O E C O L O G Y O F I S L A N D B I OTA S

expectation that, on the whole, the rarest species ona particular island are most likely to become extinct.This view, although basically stochastic, is at leastpart way along the axis towards a model of struc-tured turnover. Now, if, in general, turnover at equi-librium involves a subset of fugitive populations,while another, larger group of species, mostly oflarger populations, is scarcely or not at all involvedin turnover, then the turnover is highly heteroge-neous. Both Williamson (1981, 1983, 1989a,b) andSchoener and Spiller (1987) have concluded thatmost empirical data point to turnover mostly beingof such fugitive, or ephemeral species. Whether thisis consistent with the EMIB or not is an arguablepoint: at best, if this really is an accurate descriptionof turnover at equilibrium, then it suggests that thetheory is ‘true but trivial’ (Williamson 1989a).

How is equilibrium approached? Can youassume that the more species an island has, thesmaller the populations? Do the curves for immi-gration rate and extinction rate follow the trendsdescribed by MacArthur and Wilson of smoothdecline and increase, respectively? Actually, asthey put it themselves (1967, p. 22): ‘these refine-ments in shape of the two curves are not essentialto the basic theory. So long as the curves aremonotonic, and regardless of their precise shape,several new inferences of general significance con-cerning equilibrial biotas can be drawn’. Thisacknowledges that it is the monotonicity of the ratechanges which is critical. It is possible to testwhether, indeed, immigration and extinctiontrends are monotonic. The answer in at leasttwo cases appears to be ‘no’. Rey’s study illustratesthis in a simple system. Observations from theKrakatau islands illustrate much the same point ina complex system.

The Krakatau islands were effectively sterilizedin volcanic eruptions in 1883. Recolonization of thethree islands in the group commenced shortly after.It started in a limited number of places and gradu-ally claimed more of the island area. Niche spaceand carrying capacity increased as the systemdeveloped. Successional processes kicked in, andhabitat space waxed and waned as grasslandsdeveloped and then diminished—overwhelmed bya covering of forest. Bush and Whittaker (1991)

demonstrated that, if the lists of species present aretaken at face value, the trends in immigration andextinction from the lists (not the real, unmeasurableI and E on the islands) behave non-monotonicallyfor birds, butterflies, and plants. Bush andWhittaker’s (1991) bird data were later shown to beincomplete for the latest survey periods, but evenafter correction (Thornton et al. 1993), the non-monotonicity of trends remained (Bush andWhittaker 1993). It therefore appears that forcomplex islands, there is more ecological structureto the assemblage of an island ecosystem than isallowed for in the EMIB (see Chapter 5).

What causes extinctions?

What causes extinctions, and thus turnover, is oneof the crunch issues connected with the equilib-rium theory. MacArthur and Wilson (1967) werenot explicit. There is no doubt that they regardedit as, in general, a function of population size, butto what extent does their theory assume it to berandom (and homogeneous), as opposed to adeterministic outcome of species–species interac-tions, particularly competition? The ‘narrow’answer is that the EMIB is essentially a stochasticformulation (above), but the ‘broad view’ answeris that their text acknowledges the importance ofcompetition. For instance, the data for the recolo-nization of Krakatau by plants (up to 1932) failedto fit their expectations. One of the two explana-tions they offered invoked a successional patternin which ‘Later plant communities are dependenton earlier, pioneer communities for their success-ful establishment. Yet when they do become estab-lished, they do not wholly extirpate the pioneercommunities . . .’ (MacArthur and Wilson 1967, p.50). This invocation of competitive effects is acommunity-wide, rather than one-on-one, form ofcompetition, and is sometimes given the label dif-fuse competition. Wilson (1995, p. 265), writingretrospectively about his field reconnaissancebefore the mangrove islets experiments, recollects‘For ants the pattern was consistent with competi-tive exclusion. Below a certain island size, the col-onization of some species appeared to precludethe establishment of others . . . ‘.

Page 118: Whittaker. Island Biogeography. 2007

T U R N OV E R 105

Rosenzweig (1995) invokes both competition andpredation as reasons why extinction rates shouldrise with increasing size of the assemblage. Thedata he cites for a direct role of predation are notentirely convincing of a general effect, as they come(with one exception) from predators introduced byhumans to remote islands, the biota of which hadevolved without them. The exception was for datafor orb spiders on very small islands, where theywere preyed upon by lizards. The support in thiscase is not entirely unequivocal (Toft and Schoener1983) and, indeed, Spiller and Schoener (1995) haveshown that the effects of the lizards on spider den-sities within these islands vary significantly overtime as a function of rainfall variations (data onturnover were not given in the latter study).

Additional evidence for predators causing risingextinction rates have come from studies byThornton (1996) and his colleagues of the coloniza-tion of the newly formed island of Anak Krakatau(the fourth Krakatau island, which emerged fromthe sea in 1930). As the island began to supportappreciable areas of vegetation in the 1980s, frugiv-orous and insectivorous birds were able to colonizein greater numbers, but following the establish-ment of raptors on the island, several species suf-fered considerable reductions. In two cases at least,their disappearance from the island was attributedto predation. This is an intriguing finding, as inother ecological contexts predators have often beeninvoked as preventing competitive exclusion andthereby enhancing diversity at lower trophic levels(Caswell 1978; Begon et al. 1986). In the mid-1980sAnak Krakatau was able to support only a singlepair of oriental hobbies, which in 1989 werereplaced by a pair of peregrine falcons, suggestingthe position of top predator to be rather marginal.Anak Krakatau’s avifauna is set in the context ofthree nearby islands, and individuals found on theisland may well be only partially dependent on itfor food supplies. For instance, Thornton (1996) hasnoted that the home range of the other avian pred-ator on Anak Krakatau, the barn owl, is sufficient totake in all four Krakatau islands. In such cases, apredator may be able to remove one element of itsin situ food supply while supplementing its diet onprey from other nearby islands (equally, new

arrivals or stragglers may be killed by residentpredators on small islands, cf. Diamond 1974). Inthis fashion, Anak Krakatau may be functioning notas a closed but as an open system (cf. Caswell 1978).

The role of predators may also be scaledependent, such that development of the island’secosystems to a greater biomass might diminish theability of predators to cause the total loss of a preyspecies. This suggestion is consistent withDiamond’s (1975a) observations of contrastsbetween Long Island and the more depauperateRitter, both being islands recovering from volcanicdisturbance. Caswell (1978) makes a similar sugges-tion on the basis of his non-equilibrium modellingof predator–prey relationships in open systems.

In order to demonstrate that the extinction curvedoes indeed have a concave, rising form, producedby species interactions, Rosenzweig (1995) had toresort to a complex line of reasoning, based onISARS from two data sets, each of island archipela-gos created at the end of the last ice age andpresumed to be undergoing relaxation. This consti-tutes a rather indirect line of proof, reliant upon aseries of assumptions being taken as read. It isintriguing that he had to work so hard to find sup-port for such a basic element of the EMIB. The gen-erality of the concave, rising trend of extinction doesnot seem to be founded on much empirical evidenceyet. Data can also be found that, at face value, con-tradict this expectation.

There is no reason to assume that there is a singlepath to extinction. In the process of a species declin-ing (perhaps in fluctuating fashion) to extinction,its population must, for some period, be small. Thesmall size of the population may be a good predic-tor of extinction, but why is the population small inthe first place? The explanation could lie in thetrophic status of the organism; in predation, dis-ease, or disaster; in resource shortage; or in compe-tition with other species of its trophic level. It is, asWilliamson (1981) notes, remarkable that the equi-librium theory was widely accepted with almost noevidence on the distribution of probabilities ofgoing extinct and how these vary (cf. Pimm et al.1988). It is arguable that, in many cases, habitatavailability and stability, and food web connections(and continuous availability of resources), are more

Page 119: Whittaker. Island Biogeography. 2007

106 S P E C I E S N U M B E R S G A M E S : T H E M AC R O E C O L O G Y O F I S L A N D B I OTA S

important than competition in determining extinc-tion probabilities (Whittaker 1992).

So, where does all this leave us in evaluatingturnover on islands? The statement in Box 4.1 is ofcourse a truism: the number of species on anisland at a given point in time has to be a functionof the number previously recorded and the num-bers gained and lost in the interim, but the extentto which these rates behave according to theexpectations of the EMIB remains hard to assess.Although we began in this chapter by discussingsome of the core macroecological relationshipswithin ecological systems, by focusing on speciesturnover, we have inevitably been drawn into dis-cussing aspects of composition in the latter part ofthe chapter. We go on in the next chapter to lookmore formally at theories concerned with compo-sitional pattern, and then pull together both sets ofthemes in the final chapter of this section of thebook (Chapter 6).

4.5 Summary

Ecological systems have emergent statistical proper-ties, detectable for example in regularities of form ofthe abundance and number of species, and whichpoint to the existence of some fundamental govern-ing ecological rules or processes. The study of sys-tems in this way is termed macroecology. Islandmacroecology began long before the term wascoined, in the early part of the twentieth century. Itgained its core focus with the publication in 1967 ofMacArthur and Wilson’s equilibrium model ofisland biogeography (EMIB), which postulates thatspecies number is the dynamic product of opposingrates of immigration and local extinction, as inversefunctions respectively of isolation and area.

In this chapter, we first review the buildingblocks that lead to the formulation of the EMIB andthen consider how it has fared alongside subse-quent theoretical developments. Evaluation largelyfalls under two heads: spatial patterns and tempo-ral patterns. We distinguish between severalclosely interrelated properties of ecologicalsystems: species abundance distributions, species–area curves, species accumulation curves andisland species–area relationships (ISARs). Variationin the form of ISARs was a central focus ofMacArthur and Wilson’s theory, and a substantialliterature has since developed, demonstratingvarying degrees of fit with the EMIB—implyingthat at the least a degree of modification isrequired. Variation in form may for instance beattributable to different variables depending on themagnitude of variation in independent variables(area, isolation, elevation, habitat diversity, etc)and their interaction. Yet, there is also the prospectthat combined area-isolation-climate models canbe developed that can account for the first orderpattern of island species richness globally, at leastfor particular taxa.

The EMIB asserts that island richness patternsare dynamic outcomes—the product of perpetualturnover in membership and that once atequilibrium this process continues essentiallyunabated. On this score, the evidence appears atbest equivocal, in part because of the enormousproblems involved in accurately quantifyingimmigration and extinction rates and thusmeasuring turnover. Consideration of evidenceon turnover draws us into discussion of composi-tional structure in the data, which receives muchfuller consideration in the next chapter on islandassembly theory.

Page 120: Whittaker. Island Biogeography. 2007

. . . studies of island biogeography show that a commu-nity is not simply a collection of all those who somehowarrived at the habitat and are competent to withstand thephysical conditions in it . . . a community reflects both itsapplicant pool and its admission policies . . . We need toformulate a new generation of population and commu-nity models with the transport processes explicitly builtin . . .

(Roughgarden 1989, pp. 217–19, with thanks toThornton 1996)

In Chapter 4, we took a macroecological approachto island biotas: species were treated largely asexchangeable units. Such an approach ultimatelyhas its limits. This chapter is therefore concernedwith those aspects of island biogeographic theorydealing with compositional pattern and how it isassembled (mostly) in ecological time. By consider-ing the much more complicated response variableof ‘composition’, it is possible to cast a differentlight on to some of the problems left over from thespecies numbers games (Worthen 1996). Patterninterpretation in this chapter will again invokeforces such as immigration, competition, and pre-dation, but will also pay greater attention to speciesautecology, dispersal attributes, and succession. Inessence then, whereas the previous chapter took amacroecological approach, this chapter is con-cerned with ecological biogeography.

Colonization and ecosystem development of anot-too-distant island arguably constitute just aspecial case of ecological succession, a process(realistically many processes) that in complexecosystems, dominated by long-living organisms,

may be evident over hundreds of years. We illustratehow succession can influence emergent patterns ofspecies numbers through the example of the recol-onization of the Krakatau islands, which is notonly the best-studied ‘natural experiment’ in islandrecolonization, but was also the first test systemused by MacArthur and Wilson (1967; see Wilson1995). However, it is Jared Diamond’s theory thatprovides the starting point for this chapter.

5.1 Island assembly theory

Some readers may have experienced the dailyschool assembly, the rules of which were that youall stood quietly in lines, class by class, andbehaved yourselves while certain school ritualswere observed. Of course, these behavioural ruleswere broken daily; furthermore, the precise orderand number of individuals in each line varied, buton the whole the system approximated to the formlaid down. The casual observer looking in might,however, have failed to recognize this on occasion.This analogy might apply to island ecosystems, buthere detection of the rules, if they exist, requires agreat deal more application. This particular branchof island ecology was given its impetus by anextensive series of studies of the avifauna of NewGuinea and its surrounding islands by Diamond(e.g. 1972, 1974, 1975a). The impetus was lost fora while as the approach became embroiled in atechnical dispute concerning the formulation andtesting of hypotheses, and controversy concerningthe relative significance of competition in pattern

107

CHAPTER 5

Community assembly and dynamics

Page 121: Whittaker. Island Biogeography. 2007

formation. Yet what Diamond (1975a) presentedamounted to a fairly comprehensive island biogeo-graphical theory in its own right, which we termisland assembly theory.

Assembly rules

Diamond adopted the working hypothesis thatthrough diffuse competition, the componentspecies of a community are selected, and co-adjusted in their niches and abundances, so as to fitwith each other and to resist invaders. In his stud-ies he identified the following patterns or assemblyrules (quoted from Diamond 1975a, p. 423):

1 If one considers all the combinations that canbe formed from a group of related species, only cer-tain . . . of these combinations exist in nature.2 Permissible combinations resist invaders thatwould transform them into forbidden combina-tions.3 A combination that is stable on a large or species-rich island may be unstable on a small or species-poor island.4 On a small or species-poor island, a combinationmay resist invaders that would be incorporated ona larger or more species-rich island.5 Some pairs of species never coexist, either bythemselves or as part of a larger combination.6 Some pairs of species that form an unstable com-bination by themselves may form part of a stablelarger combination.7 Conversely, some combinations that are com-posed entirely of stable subcombinations are them-selves unstable.

These assembly rules were derived from severalforms of distributional data (e.g. incidence func-tions, chequerboard distributions) for species andguilds, alongside data for recolonization of defau-nated islands. The ‘rules’ were the descriptions andinterpretations of these emergent patterns.

Incidence functions and tramps

Incidence functions are an essentially simple idea,reliant on the availability only of good data onspecies distributions across a series of isolates. They

show the frequency of a species in relation to valuesof a potentially controlling variable, such as area orisolation (e.g. Watson et al. 2005). However, as orig-inally employed by Diamond (1975a), they wererichness-ordered incidence functions: plots ofisland species number, S, versus the incidence, J, ofa given species on all islands of that value of S.Unless a huge number of islands is available, it isnecessary to group together islands of a givenrange of values of S in order to obtain enoughobservations in each class. Figure 5.1 shows thefunctions calculated by Diamond for a subset of hisbird species. The legend provides a labelling of thedifferent distributional patterns, high-S species,A-tramp, and so on. In order to interpret theselabels some background is needed.

Diamond’s study area lies near the equator, withthe predominant natural vegetation cover beingrain forest. The New Guinea bird species pool con-sists of about 513 breeding non-marine species.Offshore, there are thousands of islands of varyingsizes and degrees of isolation, for many of whichbird data were available, in cases includinginstances of successful and unsuccessful coloniza-tion. Some land-bridge islands can be assumed tohave gained much of their avifauna at times of low-ered sea level, and were thus regarded as supersat-urated. Others, having undergone disruption byvolcanic action (e.g. Long and Ritter islands), wereregarded as displaced below equilibrium. Otherswere regarded as equilibrial. Thus, within the 50islands studied, although a significant species–arearelationship was recorded, there were statisticaldeviations. These deviations appeared intelligiblein relation to historical and prehistorical events,and were therefore interpreted in terms of long-term trajectories in avifauna size and composition.

Some species of birds, such as Centropus violaceus,Micropsitta bruijnii, and Artamus insignis, occur onlyon the largest, most species-rich islands. These weretermed high-S (or sedentary) species (Diamond1974), the precise usage being species found on thefour richest of the Bismarck Islands (islands with81–127 species) plus not more than one from thenext richest category (islands with 43–80 species). Atthe other extreme, a small number of species areabsent from the most species-rich islands and are

108 C O M M U N I T Y A S S E M B LY A N D DY N A M I C S

Page 122: Whittaker. Island Biogeography. 2007

I S L A N D A S S E M B LY T H E O RY 109

1.0

0.8

0.6

0.4

0.2

0

J

1.0

0.8

0.6

0.4

0.2

0

J

1.0

0.8

0.6

0.4

0.2

0

J

0 20 40 60 80 100 120 140S

0 20 40 60 80 100 120 140S

(c)

(b)

(a)

(e)

(d)

(f)

Figure 5.1 Incidence functions for birds of the Bismarck Archipelago. S�number of species on an island, but note that the points representgrouped data for a narrow range of values of S from between 3 and 13 islands per point, except the two largest values, which each represent asingle island. The index J � 1.0 if the species occurs on all islands, and 0.0 if the species occurs on none. (a) Centropus violaceus (cuckoo),a high-S species; (b) Diacaeum eximium (berrypicker), an A-tramp; (c) Pitta erythrogaster (pitta), a B-tramp; (d) Ptilinopus superbus (pigeon), a C-tramp; (e) Chalcophaps stephani (pigeon) a D-tramp; (f) Macropygia mackinlayi (pigeon), a supertramp. See text for explanation of categories.(Redrawn from Diamond, J. M. (1975) In Ecology and evolution of communities (ed. M. L. Cody and J. M. Diamond). Copyright © by the Presidentand Fellows of Harvard College. Reprinted by permission of Harvard University Press.)

concentrated in the smallest, or most remote, andmost species poor islands. An example of thissupertramp category is the pigeon Macropygiamackinlayi. In between, Diamond differentiated

arbitrarily four other categories of lesser tramps, forthose species occurring on some or all of the mostspecies-rich islands and, depending on the categoryof tramp, on other islands of varying size and

Page 123: Whittaker. Island Biogeography. 2007

species richness within the Bismarck Islands: A-tramps 3–9 islands; B-tramps 10–14 islands;C-tramps 15–19 islands; and D-tramps 20–35islands. The smallest or most species-poor islandsgenerally have representatives of both tramp andsupertramp categories.

The conclusion reached from the incidence func-tions was that the island avifaunas representedhighly non-random selections of the species pool.Rather, it appeared that species distributions wererelated to factors such as: competition; the absenceof suitable habitat; island sizes being smaller thanminimum territory requirements; seasonal orpatchy food supply; historical factors such as land-bridge connections; disturbance and changes in car-rying capacity; and interactions of these factorswith population fluctuations.

The dynamics of island assembly

In addition to being found predominantly on thesmaller islands, the supertramps were also charac-teristic of the recently recolonized volcanic islands(Diamond 1974), suggesting that these species areexcellent colonists but poor competitors. The morewidespread tramp species, being found in rela-tively high frequencies on all but the smallestislands, must, in contrast, be both capable colonistsand good competitors. The species restricted tolarge islands might be absent from others becauseof poor dispersal or some form of intrinsic unsuit-ability to small islands. As this high-S set includesspecies of larger body size (e.g. herons), and speciesof large home ranges and strong dispersal potential(e.g. large hawks), the latter appeared important, atleast for some of the species involved. Diamondconcluded that the high-S and A-tramp categoriesactually included a fairly heterogeneous mix ofspecies, some limited by dispersal and some byother factors. About 56% of the high-S species areendemic to the Bismarcks at the semispecies orspecies level, such that the patterns under discus-sion retain at least some signal from evolutionarytimescales.

Some species distributions were interpretable inrelation to habitat requirements. For example, asonly five islands in the data set exceed 3500 feet

(1067 m) elevation, the 13 montane bird species arerestricted to these 5 islands. Of the 5 islands, one isvery small and another was recently defaunated byvolcanic explosion: 12 of the 13 montane birds arelacking from both of these islands. Similar fits tohabitat could be found for other Bismarck birdgroups. The endemic species, which are mostly alsohigh-S species, are mostly lowland forest or moun-tain species. By contrast, the tramp species confinedto scarcer lowland habitats were not found to havedifferentiated beyond subspecies level. This wasthought to be indicative of the greater frequency ofdispersal needed to maintain their lineage in a scat-tered system of habitat patches across the archipel-ago. The supertramp species differ again, in thatwhere they occur on an island, they tend to occur ina greater variety of habitats, and Diamond (1975a,p. 381) characterized their ecology as ‘Breed, dis-perse, tolerate anything, specialize in nothing.’Thus, the lack of supertramps on an island is aproduct of competitive exclusion and not of disper-sal constraints. The gradient between the super-tramps and the high-S species can be viewedas equivalent to that between r-selected andK-selected species, or pioneering and late succes-sional plant species. In each case, the gradient rep-resents an approximation of a more complex reality,with varying degrees of explanatory power fromone case study to another (Begon et al. 1986; and fora critical evaluation of r–K selection see Caswell1989).

As will become apparent in Chapter 9, there areparallels between Diamond’s ideas and E. O.Wilson’s (1959, 1961) earlier ideas of taxon cycles.Both theories seek to explain distributional patternsby means of competitive effects, habitat relation-ships, and evolutionary considerations. Althoughwe have placed them in separate sections of thebook, they are both evolutionary ecological models,both are dynamic models, and both invoke a key rolefor competition. They differ principally in that thetaxon cycle model focuses more on evolutionarychange within the taxon as a biogeographicalprocess, while the assembly rules ideas are con-cerned with identifying and explaining composi-tional regularities across a series of islands, invokingecological forces such as ecological succession.

110 C O M M U N I T Y A S S E M B LY A N D DY N A M I C S

Page 124: Whittaker. Island Biogeography. 2007

Diamond (1975a) himself suggested that the parallelwas quite a close one, although he noted that whilein the taxon cycle ants colonizing a species-poorisland initially occupy lowland non-forest habitatsand subsequently undergo niche shifts to occupy theinterior, many colonizing birds already occupy low-land forests. This suggests that the superior averagedispersability of the birds may be important to thedifferences in the patterns observed.

There can of course be great differences in dis-persal abilities within taxa. This can be deduced forNew Guinea birds by reference to those speciesabsent from non-land-bridge islands. On theassumption that a dispersive species should some-where have found one or more suitable islands inthe non-land-bridge set, only 191 of the 325 low-land bird species have shown a capacity to crosswater gaps greater than 8 km (5 miles) (Diamond1975a). An example of an apparently poor disperseris the flycatcher Monarcha telescophthalmus, whichoccurs on all the large land-bridge islands aroundNew Guinea, but no others. More direct evidence ofdispersal abilities can be derived from the recolo-nization of islands such as Long, Ritter, andKrakatau, which are known to have been defau-nated, or by direct observations of birds movingbetween islands (e.g. Diamond 1975a; Thornton1996; Thornton et al. 2001). This observational infor-mation also played a part in the formulation ofisland assembly theory.

Chequerboard distributions

In developing island assembly theory, the incidencefunctions provided the first step in the reasoning.The second step relied upon a different form ofdistributional patterning, chequerboard distri-butions, found for pairs of congeneric bird speciesin the Bismarcks. A chequerboard (from the gamechequers (US checkers), or draughts) is where two(or more) species have mutually exclusive butinterdigitating distributions across a series of iso-lates, such that each island supports only onespecies. Several examples were given by Diamond(1975a). One example is provided by two species offlycatcher. Pachycephala melanura dahli is found on18 islands, and its congener P. pectoralis on 11, but

they do not occur together on any island. Similarpatterns were found for two species of cuckoo-doves of the genus Macropygia (Fig. 5.2). Macropygiamackinlayi has a weight range of 73–110 g and meanof 87 g, M. nigrirostris has a range of 73–97 g andmean of 86 g. Given these values, they might rea-sonably be expected to have similar resourcerequirements. The interpretation offered for thesedistributional patterns was that either chance deter-mined first arrival, or that slight ecological advan-tages favoured one species over the other on aparticular island. Once established, the residentwould then be able to exclude the congener.

Chequerboards and incidence functions aredifferent facets of the same distributional data sets.For instance, Macropygia mackinlayi was one of thesupertramps identified by the analysis of incidencefunctions. Its congener occurs on all the largerland masses and only a few of the small islands.Again, Diamond invoked competitive exclusion toexplain the failure to find both species resident on asingle island. It will doubtless have occurred to thereader that as some islands lack both these speciesof cuckoo-dove, and as the identification ofM. mackinlayi as a supertramp involves analysis ofentire assemblages, not just other cuckoo-doves,there must be more to explaining the distributionsof such congeners than simply their relationships toone another. The data are certainly consistent witha role for competitive interactions, but other factorsand other biogeographic scenarios (e.g. Steadman1997b) cannot be ruled out on the basis of Fig. 5.2.One means of testing the competition hypothesiswould be provided in the event of one of the pair ofcongeners colonizing an island ‘held’ by the other.According to the interpretation of the incidencefunctions, there ought to be an asymmetry in theircompetitive interactions, and M. mackinlayi shouldgenerally be the loser. Such observations, unfortu-nately, are hard to come by.

Combination and compatibility—assemblyrules for cuckoo-doves

Macropygia mackinlayi, although not overlappingwith M. nigrirostris, does occur on islands with oneof two larger species of cuckoo-dove: namely

I S L A N D A S S E M B LY T H E O RY 111

Page 125: Whittaker. Island Biogeography. 2007

M. amboinensis (average weight 149 g) or a repre-sentative of the Reinwardtoena superspecies (aver-age weight 297 g). This suggests that a certaindegree of niche difference, signified by a weightratio of 1.5–2.0 between members of the same guild,may be necessary to enable coexistence (Diamond1975a). Diamond undertook a more formal analysisof the species combinations to ensure that theywere indeed non-random, and thus evidence forcompetitive effects in island assembly. There are 4Bismarck cuckoo-dove species (represented by theletters AMNR), so 15 possible combinations, butonly 6 combinations were actually observed on the26 islands having one or more cuckoo-dove species.With such a small sample size relative to the num-ber of permutations, it is not expected that eachof the 15 possible combinations would occur. Theanalyses thus relied upon a modelling exercisedesigned to test whether the empirical data exhibiteda significant degree of departure from a randomprocesses of assembly.

The interpretation of the combinational patternsbegins with the incidence rules, which determinethat certain species combinations are ‘forbidden’(i.e. do not occur). The second form of rule wastermed a compatibility rule. The allopatricM. mackinlayi and M. nigrirostris exemplify this rule,as they are so ecologically similar in their resourcerequirements (if not in their incidence characteris-tics) that they are presumed to be unable to co-occurother than on the most temporary basis. Diamondpostulated that the two are probably the product offairly recent speciation from within a former super-species. Thus the combinations MN, MNR, AMN,and AMNR cannot (i.e. do not) occur. The final ruletype, combination rules, in effect mops up theunknowns. For instance, Diamond calculated onthe basis of the incidence functions that the combi-nation AMR should occur frequently on islandswith a medium species number. It is allowed bythe compatibility rules which were deduced, yet itwas not found to occur within the data set. This

112 C O M M U N I T Y A S S E M B LY A N D DY N A M I C S

147° E 150° E 153° E

Key

3° S

6° S

Bismarck Sea

M. mackinlayiM. nigrirostrisNeither

New

GuineaNew Britain

Figure 5.2 Checkerboard distribution of cuckoo-doves (Macropygia) in the Bismarck region. Of those islands for which data are available,most have one, but no island has both species. (Modified from Diamond, J. M. (1975) In Ecology and evolution of communities (ed. M. L. Codyand J. M. Diamond), Fig. 20. Copyright © by the President and Fellows of Harvard College. Reprinted by permission of Harvard University Press.)

Page 126: Whittaker. Island Biogeography. 2007

was taken to imply additional combinational rulespreventing apparently reasonable combinationsoccurring with the frequency to be expected bychance. As most changes to an island’s avifauna arelikely to be produced by single species additions orlosses, recombinations requiring two or more gainsand/or losses simultaneously (to avoid a ‘forbid-den’ combination) might be anticipated to have lowtransition probabilities. Hence, some combina-tions might be viable, but difficult to assemble fromthe ‘permitted’ combinations.

Diamond undertook similar statistical analysesfor the guilds of gleaning flycatchers, myzomelidsunbirds, and fruit-pigeons. The niche relation-ships among the species of Ptilinopus and Duculafruit-pigeons provide a classic example of resourcesegregation. Diamond first established the form ofcommunity structure within the ‘mainland’ low-land forests of New Guinea. The larger pigeonsforage preferentially on bigger fruits, providingone element of resource segregation. If large andsmall species feed within individual trees, thelighter pigeons are able to feed on the smaller,peripheral branches, better able to support theirweight. Thus niche separation can be maintainedwithin a guild of co-occurring species. Yet, withina single locality in New Guinea, no more thaneight species will be encountered, forming agraded size series, each species weighing approxi-mately 1.5 times the next smaller species. Thereare 8 size levels filled by combinations drawnfrom 18 species. On satellite islands off NewGuinea, subsets are drawn in such a way that onsmaller or more remote islands, size levels areemptied as the guild is impoverished according toa consistent pattern. Level 1 (the smallest birds)empties first, followed by levels 2 and 5, followedby level 8. It is intuitively sensible that the small-est birds should be excluded first as they will berestricted to the smallest fruit types, for whichthey would have to compete with slightly largerpigeons that can also make use of larger fruits.This form of structuring could also be identifiedon those islands regarded as being supersaturated,or undersaturated, suggesting that the timerequired to adopt such structure is shorter thanthat required for equilibration of species number

on islands. This observation resonates with subse-quent analyses of ecosystem build-up within theflora and at least some elements of the fauna ofKrakatau (e.g. Whittaker et al. 1989; Zann andDarjono 1992; Thornton 1996).

The above constitutes the bulk of Diamond’sspecies assembly theory as it concerns us here.However, he also applied the ideas to non-insularcommunities, a context in which many of therecent developments in the formulation and test-ing of assembly rules have come (e.g. Wilson andRoxburgh 1994; Weiher and Keddy 1995; Wilsonand Whittaker 1995). Diamond’s studies incorpo-rated distributional data at the island level in thefirst instance, but also involved: data on habitatsin which particular species were observed ortrapped; body weights; feeding relationships;observations of bird movements between islands;and historical records of species colonizations andextinctions, in cases providing apparent evidenceof the unsustainability of certain combinations ofspecies.

Criticisms, ‘null’ models, and responses

Considering how much evidence for prehistoric anthro-pogenic extinctions of island birds has emerged over thepast decade, biogeographers today should hesitate toregard any distributional attribute of modern SouthPacific birds as being uninfluenced by human activity.

(Steadman 1997b, p. 750)

Diamond’s papers provide a coherent theory ofthe biogeography of the New Guinea island birds.This island assembly theory could be seen as adevelopment of several of the lines of argumentto be found in MacArthur and Wilson’s (1967) Thetheory of island biogeography. Yet, the approach,and particularly the assembly rules, drew criti-cism from Daniel Simberloff and colleagues(Simberloff 1978; Connor and Simberloff 1979).The problems were sufficiently troublesome, orintractable, as to hamper subsequent progress,and the approach has arguably failed to receivethe attention that its intrinsic interest warrants(Weiher and Keddy 1995).

One element of the logic employed by Diamondthat might be criticized is the implicit acceptance

I S L A N D A S S E M B LY T H E O RY 113

Page 127: Whittaker. Island Biogeography. 2007

of the equilibrium thinking at the heart ofMacArthur and Wilson’s (1967) monograph. Thus,for instance, he writes,

The fact that the larger land-bridge islands around NewGuinea still have double their equilibrium species number[of birds], more than 10 000 years after the land bridgeswere severed, emphasizes for what long times manyspecies can escape their inevitable doom on a large island’(Diamond 1975a, p. 370).

If it takes in excess of 10 000 years to reach equi-librium, the effect would appear to be a relativelyweak one, potentially or actually to be overriddenby other factors. The inevitability of their doom, thefate ultimately of all living things, as the fossilrecord reveals, is not necessarily in practice theconsequence of their island systems moving closerto a biotically regulated equilibrium. Simberloff(1978) made a similar point when he argued thatthe statistical analyses of distribution were notuniquely interpretable as a function of diffusecompetition.

A second critical paper was more strident. Itclaimed to show that ‘every assembly rule is eithertautological, trivial or a pattern expected werespecies distributed at random’ (Connor andSimberloff 1979, p. 1132). The rules derived inDiamond’s analyses were cast in terms of ‘permit-ted’ and ‘forbidden’ combinations (as rule 2, above)and ‘resistance’ to invasion. Thus, although theywere derived empirically mostly from distribu-tional data, competition was invoked within theirformulation. One element of the critique concernedthe great difficulty of demonstrating a particularrole over time for competition in determiningpresent-day distributional patterns (Law andWatkinson 1989; Schoener 1989). Simberloff (1978)argued that the role of competition had been over-played by several authors, citing Diamond’s stud-ies in illustration. His argument focused largely onthe methods by which non-random distributionsare calculated. He argued that confidence limitswere not given on the combinational rules, whichwere therefore difficult to evaluate, and that therewere too few islands involved to allow rigorousstatistical assessments. He argued that the nullhypothesis should be formulated in terms of

species being considered essentially as similarunits, i.e. without differing dispersal abilities andecologies, and that these species should be mod-elled as bombarding islands entirely at random.The different islands should be accepted as havingdifferent properties, however. While accepting thatthe first of these assumptions is biologically unreal-istic, Simberloff’s argument was that it provided abaseline by which to examine biogeographical dis-tributions. When the data are not consistent withthis null model, ecological phenomena, such ascompetition or dispersal differences, can then beexamined for explanations. Simberloff also arguedthat the biological characteristics of each speciesshould be examined for explanatory value as beinga more parsimonious explanation than competi-tion. Thus one species might be a superior colonistwith respect to another because it has better disper-sal ability and/or better persistence once immigra-tion occurs (as a function of per capita birth anddeath rates).

Thus Simberloff, whose own formative experi-ence was the mangrove islet study (a fast develop-ing, near-source, dynamic system), adopted aposition which is arguably closer to the stochastic,dynamic equilibrium model of island biogeogra-phy (EMIB) than did Diamond in his seminalstudy of New Guinea island avifaunas (a slow,more isolated, and less dynamic system with astrong evolutionary signal). This difference inview was recognized by Simberloff, who notedthat botanists and invertebrate zoologists havemost often sought distributional explanations inresponses of individual species to physical factors,whereas vertebrate zoologists commonly invokecompetition. The generality of competitive effectswas in Simberloff’s view largely unproven, andthe reason was that appropriate statistical testshad yet to be applied to sufficient data sets—indeed such data sets were, by the nature of thestatistical requirements, bound to be difficult toobtain. This line of argument was extended byConnor and Simberloff (1979) who contended thatthe patterns deduced by Diamond as the outcomeof competition could be produced by randomprocesses providing that the following threeconstraints were accepted: ‘(1) that each island has

114 C O M M U N I T Y A S S E M B LY A N D DY N A M I C S

Page 128: Whittaker. Island Biogeography. 2007

a given number of species, (2) that each species isfound on a given number of islands, and (3) thateach species is permitted to colonize islands con-stituting only a subset of island sizes’ (Connor andSimberloff 1979, p. 1132). They based their claimson data for New Hebridean (Vanuatu) birds, WestIndian birds, and West Indian bats, which theyanalysed for evidence of Diamond’s assemblyrules. The statistical part of their analysis failed tosupport the existence of assembly rules in theirdata.

The quote given earlier referred to tautologiesand trivial features within the assembly rules.These arguments are difficult to summarize butamount to an attack on the line of reasoning used toconstruct the assembly rules. In illustration, thethird rule ‘A combination that is stable on a large orspecies-rich island may be unstable on a small orspecies-poor island’ was argued to be reducible to‘A combination which is found on species-richislands may not be found on species-poor islands’because, first, species number is used operationallyby Diamond in lieu of island size and, secondly,because stability is largely dependent on the com-bination of species in question having beenobserved. As large islands contain on average morespecies, it would be expected by chance alone thatsome combinations found on rich islands will notoccur on species-poor islands. By this line of rea-soning, the third rule was deemed to be a trivialoutcome of the definitions used and otherwise aprobabilistic outcome.

Another important line of criticism concernedthe chequerboard distributions, which theyargued could represent ongoing cases of geo-graphic speciation without re-invasion, i.e. theycould be cases of divergence between allopatric orparapatric lineages, rather than being the outcomeof competitive exclusion. The position they adoptat the end of their paper is clear. The assemblyrules were flawed because at no time was a simplenull hypothesis framed and tested, competitiveeffects were invoked without proof and wheresimpler explanations were available. Nonetheless,Connor and Simberloff stressed that they were notimplying that island species are distributedrandomly and that interspecific competition does

not occur, just that neither non-random distributionsnor the primacy of competition in determiningdistributions were convincingly demonstrated byDiamond’s analyses.

Not surprisingly, these criticisms drew aresponse, in the form of consecutive articles byGilpin and Diamond (1982) and by Diamond andGilpin (1983), and indeed generated a flurry ofarticles and comments within the literature of thelate 1970s and early 1980s. Perhaps the core ofDiamond and Gilpin’s (1982) response is a chal-lenge to the primacy of Connor and Simberloff’sconception of a null hypothesis. Diamond andGilpin (1982) take the reasonable position thatbiogeographical distributions may be influencedby many factors, including competition, predation,dispersal, habitat, climate and chance, and that themix is liable to vary with the group of organismsconsidered and with spatial and temporal scale.Further, it follows that no single theory, hypothe-sis, or conceptual position has logical primacy, orspecial claim to be the most parsimonious nullhypothesis (they signify this by placing ‘null’ inquotation marks throughout). Each should beregarded as a competing or contributory hypothe-sis to be evaluated on its merits. This is a pragmaticposition adopted generally in the present book:different answers may be found to similar ques-tions posed in different island biogeographicalcontexts. However, there is, of course, more to thedebate than this. Diamond and Gilpin in fact tookthe opportunity to respond to a series of papers bySimberloff, Connor, Strong, and others, employingnull hypotheses. The problem with null hypothe-ses, i.e. hypotheses of ‘no effect’, is the assump-tions that are made in their formulation (Colwelland Winkler 1984; Shrader-Frechette and McCoy1993). Diamond and Gilpin argue that in generat-ing their null distributions Connor and Simberloff(1979) took the following inappropriate steps:

1 Whereas assembly rule theory posits competitiveeffects within guilds, they diluted their data byincluding data for the whole species pool instead ofonly for the target guild.2 They incorporated hidden effects of competitioninto their constraints. The assumptions employed

I S L A N D A S S E M B LY T H E O RY 115

Page 129: Whittaker. Island Biogeography. 2007

in their null models build in occurrence frequen-cies and incidences of species which can be shownto be influenced by competition by referenceto occurrence frequencies of members of thesame guild in different archipelagos (Bismarck,Solomon, and New Hebrides). These data showthat the fewer competing species of the sameguild that share an archipelago, the higher theoccurrence frequency of particular species, oftenby a factor of 10 or more.3 They produced a statistical procedure incapableof recognizing a chequerboard distribution.4 Simulation procedures used were inadequate toprovide realistic simulations of the matrices.

This is an incomplete listing of the objectionsraised by Diamond and Gilpin (1982), but the gen-eral position they adopt is summed up as follows:‘how can one pretend that one’s “null model” iseverything-significant-except-X, when it was con-structed by rearranging an observed database thatmay have been organized by X?’ (Diamond andGilpin 1982, p. 73). The problem thus afflicts theproponents on both side of the argument, and isone that has since reappeared in other applicationsof null models (papers in Strong et al. 1984; Weiherand Keddy 1995). All null models involve assump-tions; the trick is to recognize what they are, the dis-agreements come over which are the most realisticand appropriate set to use (Box 5.1).

In their second paper, Gilpin and Diamond(1982) developed a test that they regarded as animprovement over the Monte Carlo modellingused by Connor and Simberloff, and affirmed theirearlier conclusion that the distribution patternswere indeed non-random. Comparing results forthe Bismarck Islands and the New Hebrides, theyfound non-random patterns in each system, butthat the richer Bismarck Islands had clearer evid-ence of competitive effects. The New Hebrideshave a smaller avifauna (56 compared with 151species) and thus negative associations (as wouldbe produced by competition) did not exceed arandom expectation: neither were chequerboarddistributions evident for the New Hebrides. Incontrast, they re-affirmed the existence of negativeassociations and chequerboard distributions

amongst ecologically similar pairs of Bismarckspecies. They found that some pairs of species havemore exclusive distributions than expected bychance, and invoked as likely controlling factors,competition, differing distributional strategies,and different geographical origins. Examples of thefirst two have been given above. In illustration ofthe third category, the hawk Falco berigora hasspread from the west, whereas the parrotChalopsitta cardinalis has spread from the east, butin each case to a limited degree. They are not mem-bers of the same guild, obviously, and their cur-rently exclusive distributions relate to theirdiffering geographical origins rather than competi-tion. Other species pairs were found to have morecoincident distributions than would be expectedby chance alone, and this they interpreted in rela-tion to shared habitat, single-island endemisms,shared distributional strategies, or shared geo-graphical origins (Table 5.1).

As Gilpin and Diamond (1982) point out, much ofthe information about non-random co-occurrencesis actually contained in the incidence functions––thefirst element in the whole analysis––which aremuch more intuitively accessible sources of infor-mation than are the more sophisticated modellingexercises. Equally, the chequerboard distributionsprovided unambiguous patterns, strongly sugges-tive of competitive effects: in such cases restrictinganalyses to members within a single guild is clearlynecessary (Diamond 1975a). Thus, where observedwithin guilds of pigeons for which there is excellentevidence of colonizing ability across moderate

116 C O M M U N I T Y A S S E M B LY A N D DY N A M I C S

Table 5.1 Factors invoked in explanation of non-randomco-occurrence of island birds by Diamond and Gilpin in theirarticles on assembly rules

Negative Positive

Competition Shared habitatDiffering distributional strategies Shared distributional

strategiesDiffering distributional origins Shared geographical

originsSingle-island endemics

Page 130: Whittaker. Island Biogeography. 2007

I S L A N D A S S E M B LY T H E O RY 117

Box 5.1 On null models in biogeography

As applied in statistical analysis, a nullhypothesis is the hypothesis of no effect(in respect of a potential controlling variable), orof no difference (between two or morepopulations). The rejection of the null hypothesisleads to the inference that there is an effect or adifference. Null models, on the other hand, arecomputer simulations used to analyse patterns innature, in an effort to achieve predictability andfalsifiability in ecological biogeography. Astypically applied within island biogeography, theapproach taken is aimed to simulate randomcolonization of a set of islands by the members ofa species pool, given particular constraints, suchas how species number per island is constrainedby island area. By repeating the randomsimulation a large number of times, it is possibleto determine whether the observed pattern ofspecies occurrence could have occurred by chanceusing a given level of significance.

The use of null models has been the subject ofintense debate virtually wherever they have beenapplied within biogeography. Sometimes thearguments have been technical in nature,concerning how well the simulations achieve theintended goal of randomness or how well they

match the intended constraints. More intractableis that it is very hard to reach an accord as to thecorrect way to set the constraints upon a system,i.e. which parameters are controlled and whichare allowed to vary. Both sets of problems areexemplified by the assembly rules controversydiscussed in the text. Here, the aim was to test forinterspecific competition effects by buildingmodels that were neutral in respect ofcompetition. But is this possible in the real world?

In practice, unlike for simpler forms ofinferential statistics, we can rarely expectagreement on a single null formulation for use inbiogeographical modelling. It is better to think ofthem not as ‘null models’ with the attachedimplications of a higher form of objectivity, butsimply as simulation models, each involving aparticular set of assumptions and constraints.Such models have value in exploring and testingout ideas, but always leave room for differencesof interpretation.

For further reading and varying perspectiveson the theme see, e.g. Gotelli and Graves(1996), Grant (1998), Weiher and Keddy (1999),Moulton et al. (2001), and Duncan andBlackburn (2002).

ocean gaps, and where an otherwise widespreadspecies is absent from large, ecologically diverseislands offering a similar range of habitats to thoseoccupied by the species elsewhere, but which hap-pen to be occupied by the alternative species, com-petition is the obvious answer.

Other authors have since joined in the debate.Stone and Roberts (1990) re-analysed some of thedata employed in Connor and Simberloff’s (1979)article, but by means of a new form of test thatthey termed the C-score, a means of quantifyingthe degree of ‘chequerboardedness’. Their analysisattempted to avoid some of the problems ofConnor and Simberloff’s methods while maintain-ing the constraints that they assumed. In contrastto Connor and Simberloff, their re-analyses led to

rejection of random distributions for both the NewHebrides and the Antilles avifaunas (see alsoRoberts and Stone 1990). Thus, they found sup-port for chequerboard distributions in both fau-nas, while stopping short of invoking competitiveexplanations for their findings. In a further ana-lysis, Stone and Roberts (1992) examined thedegree of negative and positive relationshipswithin confamilial species of the New Hebrideanavifauna. In this case they found evidence not forchequerboard tendencies, but instead for a greaterdegree of aggregation than expected by chance,i.e. confamilial species tended to occur together.This they interpreted as evidence for similaritiesof ecology within a family overriding competitionbetween confamilial members. However, as they

Page 131: Whittaker. Island Biogeography. 2007

did not restrict analyses to particular ecologicalguilds, this ‘intrafamily’ result is both unsurpris-ing and by no means contradictory to Diamond’s(1975a) arguments.

For the record, Sanderson et al. (1998), Grant(1998), and Gotelli and Entsminger (2001) have con-tinued the debate on how to construct null modelsimulations, again using the New Hebridean birddata as the subject matter. Although differing in thedetails, the overall picture is that some degree ofnon-randomness can indeed be detected.

Such debates are rarely settled by the triumph ofone viewpoint to the complete exclusion of theother, but, as in this case, can sometimes result inthe re-examination and refinement of the originalmodels and positions. Connor and Simberloff’s(1979) intervention led to increased attention to theproblems of statistical analyses in island assemblystructure, which was their stated intent. Even now,it would be a bold move to declare the debate over,but we offer the following conclusions. The original(Diamond 1975a) statements of island assemblytheory laid too much emphasis on competition atthe expense of the many other factors (Table 5.1)that could be involved in structuring species distri-butions across islands. Although there is room fordebate about details of the assembly rules Diamondidentified in his original analyses, the evidence isconvincing that some non-randomness is involved.Moreover, within the avifaunal data (especially thatfrom the Bismarcks), it is possible to find whatappear to be compelling cases for competitiveeffects within guilds in particular cases.

At this point, we have to introduce a final note ofcaution: the patterns in distribution of the avifauna,and especially pigeons and doves may have beenaltered by humans. Steadman (1997b) provides acomprehensive review of prehistoric bone data forthe extinction of columbids (pigeons and doves)from Polynesian islands. He notes that lossesinclude the extinction of at least nine species in sixgenera, as well as the extirpation of numerousisland populations of still extant species. If it werenot for humans, a typically West Polynesian islandwould support at least six or seven species in fouror five genera, instead of the observed one to sixspecies in one to five genera. He comments that one

of Diamond’s supertramp species, Ducula pacifica,which is found in both West and East Polynesia,appears to have expanded its range eastward afterthe arrival of people, which would be consistentwith a response to human induced extirpation ofpopulations of more competitive species. Havingreviewed Polynesian evidence, Steadman notesthat there is by contrast a paucity of prehistoricbone data from Melanesia. Some losses have, how-ever, been documented, and Steadman considers itlikely that many more await discovery. In time, itmay turn out that some of the chequerboards andother assembly rule patterns of columbids at theheart of these debates have been structured not justby the ‘natural’ forces of competition, habitat con-trols, volcanic eruptions, and other historicalprocesses, but also by human interference.

There has been something of a resurgence ofinterest in assembly rules in recent years (e.g.Weiher and Keddy 1995, Fox and Fox 2000). Gotelliand McCabe’s (2002) meta-analysis of 96 previouslypublished matrices found them to be highly non-random. As predicted by island assembly theory,there were fewer species combinations, more che-querboard species pairs, and less co-occurrence inreal matrices than expected by chance. Althoughmany of their data sets are not concerned with orare not limited to insular data and thus lie slightlyoutside the scope of this review, these findings arenonetheless significant in the present context.

Exploring incidence functions

One of the problems of evaluating island assemblytheory as detailed above is that the biogeographicalcontext of the island avifaunas under discussion isin each case quite complex. We now review somesimpler systems, involving non-volant mammalson small, near-shore islands.

Hanski (1986) discusses occurrence and turnoverof three species of shrews on islands in lakes inFinland, a system in which individual species pop-ulations could be studied in some detail. Thelargest species, Sorex araneus, was found to have alow extinction rate except on islands of less than2 ha, on which demographic stochasticity becomesimportant. The smaller S.caecutiens and S. minutus,

118 C O M M U N I T Y A S S E M B LY A N D DY N A M I C S

Page 132: Whittaker. Island Biogeography. 2007

in contrast, were absent from many islands in the2–10 ha range. Temporary food shortages, or‘energy crises’ might be more problematic for thesmaller species than for the larger S. araneus, thuspotentially explaining their lower frequency ofoccurrence as a function of environmental stochas-ticity. Arguably, interspecific competition mightalso have a role to play, but Peltonen and Hanski(1991) subsequently reported no evidence for a sig-nificant effect of competition on extinction rates.Both studies conclude that the body size of specieshas a bearing on their incidence. As the largerspecies do not appear to be affected by interactionswith the smaller species within the guild, they pre-dict that they should occur consistently on islandsgreater than a minimum size set by demographicstochasticity. Small species should have a moreerratic distribution, because of their increased sus-ceptibility to environmental stochasticity. Thisinterpretation combines elements of the incidencefunction and small-island effect hypotheses intro-duced in Chapter 4. Indeed, Hanski (1992) showshow, given knowledge of minimum populationsizes, a simple model of incidence functions cangenerate realistic estimates of colonization andextinction rates for this system.

Incidence functions as originally devised are rel-atively crude univariate tools. Adler and Wilson(1985) provided a more sophisticated multivariateapproach, using multiple logistic regression to esti-mate the probabilities of individual species occur-rences. Their system consisted of records of 9species of small terrestrial mammals for 33 coastalislands off Massachusetts, and their 11 explanatoryvariables provided various measures of island area,length, isolation, habitat, and source pools.Principal components analysis was used to gener-ate two composite environmental variables, termed‘size’ and ‘isolation’. These composite variableswere then used together with a categorical variablerepresenting the dominant habitat type of eachisland (‘domhab’), in multiple logistic regressions.For all but one species, statistically significant func-tions were obtained. In four cases (species ofScalopus, Peromyscus, Microtus, and Zapus), occur-rence on islands was positively related to increas-ing island size, and in four cases (Sorex, Blarina,

Tamias, and Clethrionomys), to decreasing island iso-lation. The habitat variable, domhab, was alsoincluded in the functions of two species(Peromyscus and Microtus). Their study thusshowed another means of quantifying relationshipsbetween species occurrence and a suite of environ-mental variables: their probability functions allow-ing the relative significance of different variables tobe tested within a common framework. One otherfeature of interest was that they also used live-trap-ping over a 2–4 year period to show that the mostwidely distributed species across the islands arealso those which reach the highest population den-sities within them.

The importance of both isolation and of con-stancy of resource availability were shown by Adlerand Seamon’s (1991) study of the population fluc-tuations in just a single species, the spiny ratProechimys semispinosus, on islands in Gatun Lake,Panama. No other species of mouse- or rat-likerodents were captured on the islands. Their datarevealed that colonization and extinction of the ratoccurs regularly to and from the islands. Largerislands with year-round fruit production, regard-less of isolation, have persistent populations of rats.Small, isolated islands only rarely have rats,because they lack a year-round food source andbecause immigration is rare. Small, near islands fre-quently have rats when fruits are present. Theseobservations thus reveal how in some circum-stances (of size, isolation, etc.) fluctuations in islandcarrying capacities influence species compositionand turnover.

A particular land or aquatic system may providedispersal filters which differentiate between differ-ent members of a source biota, and indeed differentspecies within a single taxon (Watson et al. 2005).For instance, as shown by Lomolino (1986), partic-ular functional groups of small mammals may bedifferentially affected by the same immigration fil-ters. In regions with season ice cover, small coastal,inshore, or riverine islands may have their popula-tions supplemented by mammals moving acrossthe snow-covered ice in winter. In some cases, suchmovements have been recorded by tracking stud-ies. In his studies of mammal movements across thesnow-covered St Lawrence River, Lomolino found

I S L A N D A S S E M B LY T H E O RY 119

Page 133: Whittaker. Island Biogeography. 2007

that larger, winter-active species were goodcolonizers, whereas the smaller species and hiber-nators seldom travelled across the ice, and if theydid so, then only for limited distances. In relatedstudies of incidences of species across 8 mainlandand 19 insular sites, it was found that species thatwere known to utilize the ice cover had signifi-cantly higher insular rankings. Furthermore, withina particular functional guild, such as the insecti-vores, the smaller species were found to have lowerinsular rankings.

Although these data are strongly suggestive ofimmigration effects, species interactions may alsobe operative. It is a logical corollary of the EMIBthat in addition to species numbers being a resultof interactive effects of immigration and extinction,so also should species composition. Lomolino(1986) seeks to draw a distinction between com-bined (additive) effects of factors influencingimmigration and extinction, and interactive orcompensatory effects. The latter appears to be, inessence, merely another name for the rescue effectof Brown and Kodric-Brown (1977), whereby sup-plementary immigration of species present but atlow numbers on an island, enables their survivalthrough to a subsequent census. Thus, whereimmigration rates in the strict sense are high, sotoo will be rates of supplementary immigration. Asthe distinction between the two forms of popula-tion movement is effectively abandoned in thisstudy, it might be termed instead the arrival rate.Species may thus have a high incidence on islandseither where low arrival rates (poor dispersersand/or distant islands) are compensated for bylow extinction rates (good survivors and/or largeislands) or where high extinction rates are com-pensated for by high arrival rates. Thus species arecommon on those islands where their rate ofarrival is high relative to their rate of loss. Of itselfthis is a trivial observation, but when combinedwith other ecological features it provides anotheroption for understanding species compositionalpatterns, and one that stresses the dynamisminherent in the EMIB.

Figure 5.3 summarizes five hypothetical insularpatterns for the distribution of particular species.First, a species may be distributed randomly

(panel b) within an archipelago. Secondly, it mayhave minimum area requirements and not bedispersal-limited within the archipelago, in whichcase it will be present on virtually all islands over acritical threshold size (panel c). Thirdly, if it is apoor disperser which has low resource require-ments relative to insular carrying capacities, it mayoccur only on the least isolated islands (panel d).Fourthly, it may depend on both island size and iso-lation, but not exhibit compensatory effects, thusresulting in the block pattern (panel e). Finally, itmay exhibit the compensatory relationships betweenimmigration and area and thus show the diagonalpattern of panel (f) of the diagram. Lomolino (1986)postulated that compensatory effects should be moreevident for archipelagos with a large range of areaand isolation relative to the resource requirementsand vagility (mobility) of the fauna in question. If itis further assumed that within an ecologically sim-ilar group of species, larger species have bothgreater resource requirements and greater vagilitythan smaller species, an interesting predictionarises from this compensatory model. On the leastisolated archipelagos, the smaller species will havea higher incidence on the smaller islands. But, asisolation of a hypothetical archipelago increases,the low persistence of the smaller species combinedwith their poorer vagility means that larger speciesshould be the more frequent inhabitors of smallerclasses of islands, i.e. their incidence relationships‘flip over’ as a function of isolation. In short, ifrecurrent arrivals and losses are important in shap-ing the composition of the islands in question, theincidence of a species as a function of area will alsovary as a function of immigration.

Lomolino developed a form of multiple dis-criminant analysis which enabled him to distin-guish between the block effects and compensatoryeffects of Fig. 5.3. He applied this procedure to 10species of mammals, each occupying at least 2 of19 islands studied in the Thousand Islands regionof the St Lawrence River, New York State.Circularity of argument in using distributionalpatterns to infer causation could be avoided asindependent lines of evidence on species move-ments, body size, and other features of their aute-cology were available.

120 C O M M U N I T Y A S S E M B LY A N D DY N A M I C S

Page 134: Whittaker. Island Biogeography. 2007

It was found that none of the 10 species exhibiteda maximum isolation effect (Fig. 5.3d), which wasnot surprising given the relative lack of variation inisolation in the data set. Only one species, Microtuspennsylvanicus, a small but dispersive generalistherbivore, showed no area or isolation effect. Sixspecies exhibited minimum area effects (Fig. 5.3c)and the remaining three species exhibited compen-satory effects (Fig. 5.3f). Blarina brevicauda is anexample of the latter group, which is consistent withindependent evidence that it is a relatively poor dis-perser. Lomolino (1986) applied his discriminantanalyses to additional data sets for the islands ofLake Michigan, Great Basin mountain tops, and

islands in the Bass Straits. He found general supportfor the model he had developed, the relevance ofscale of isolation to the incidence patterns, and insome cases, for patterns predicted by the compensa-tory model (see also Peltonen and Hanski 1991).

Microtus pennsylvanicus and Blarina brevicauda inthe Thousand Islands region provide an illustra-tion of another important effect structuring islandcommunities, namely predation. The carnivorousshrew Blarina brevicauda preys mainly on imma-ture stages of the vole Microtus pennsylvanicus. Inthe absence of the shrew on more remote islands,Microtus was found to undergo ecological release,occurring in habitats atypical of the species

I S L A N D A S S E M B LY T H E O RY 121

4

3

Are

a

2

1

0

4

3

Are

a

2

1

0

4

3

Are

a

2

1

00 1 2 3

Isolation Isolation4 5 0 1 2 3 4 5

(a) Hypothetical archipelago (b) Random

(c) Minimum area (d) Maximum isolation

(e) ‘Block pattern’ (f) Compensatory

Mai

nlan

d

Figure 5.3 Patterns of insularoccurrence of a species on islands of ahypothetical archipelago (a). Presenceand absence in (b)–(f) are indicated byfilled and open circles, respectively. Theunits for isolation and area are arbitrary.(Redrawn from Lomolino 1986, Fig. 1.)

Page 135: Whittaker. Island Biogeography. 2007

(Lomolino 1984a). When introduced into islands,Blarina brought about drastic declines of Microtusdensities, restricting it to its optimal habitat and,in at least one case, causing its extinction. As thevole is the better disperser, the two species exhibita negative distributional relationship across theislands studied. Lomolino’s studies thus demon-strate the significance both of recurrent arrivals andlosses and of ecological controls such as predationin structuring the mammalian communities of rela-tively poorly isolated islands. As the islands inquestion undergo seasonal variation in their immi-gration filters, such that at times they are more akinto habitat islands than real islands, care must betaken in transferring such findings to other systemswith different scales of isolation. The reliability ofincidence functions is explored further in Box 5.2.

Linking island assembly patterns to habitatfactors

Haila and Järvinen (1983) and Järvinen and Haila(1984) examined the bird communities of the Ålandislands, off Finland in the Baltic. There are thou-sands of islands in this group, ranging from only afew square metres to the main island of 970 km2.Isolation of the islands tends to be slight, the mainisland being only 30 km from Sweden, and 70 km,via a steppingstone chain, from Finland. Twentyspecies of land birds found in the same latitudes inFinland and/or Sweden do not occur on the mainisland. These form three groups:

● species for which habitat is available on Åland,and for which absence is believed to be due to lackof over-water dispersal (e.g. green woodpecker,marsh tit, and nuthatch)● species for which historical explanations can beoffered (e.g. collared turtle dove and tree sparrowhave recently expanded their mainland distribu-tions but have yet to reach the island)● species lacking because of absence of suitablehabitat (e.g. great grey shrike).

They found that species numbers and density ofthe bird communities tended to be similar wherehabitats were of highly similar structure, and

tended to differ when habitats were not directlycomparable. They also noted that small, uninhab-ited islands in the archipelago lack those species ofbirds that are associated either directly or indirectlywith humans. Their overall conclusion was thathabitat factors provided explanations for theassembly of the bird communities of the Ålandislands.

Graves and Gotelli (1983) used a null modellingapproach in analysing the avifaunas of formerland-bridge islands: Coiba, San José, Rey, Aruba,Margarita, Trinidad, and Tobago. They found ingeneral that the avifaunas of these islands could beviewed as a random subset of the mainland‘habitat’ pool when judged at the family level. Theyused families because they felt unable to assignspecies to guilds, and because species within afamily are usually ecologically and morpho-logically similar. While acknowledging that this isnot ideal—families do not necessarily representunits of interspecific competition—they contendthat non-randomness of island avifaunas mightreasonably be expected to be detectable at thefamily level. Using a simple model assuming allsource-pool species to be equiprobable colonists,they found that out of 40 families, 3 are unusuallycommon on land-bridge islands: pigeons(Columbidae), flycatchers (Tyrannidae), andAmerican warblers (Parulidae), and only one, thepuffbirds (Bucconidae), was present less often thanexpected. Thus, although in absolute terms manyspecies and families are absent from the islands, theproportional representation of most families is con-sistent with mainland source pools. Comparison ofthe whole pool of species with the habitat poolshowed that, as expected, the habitat pool is a supe-rior predictor of species richness in each family.This implies that it is important when predictingspecies losses in fragmented landscapes to takeaccount of patterns of habitat change that mayunfold within fragments after isolation. Examinationof the mainland range of the species in the poolsshowed that those species with widespread main-land ranges are disproportionately common onislands. This indicates either that they havepersisted better since sea levels rose to produce thepresent degree of island isolation, or that

122 C O M M U N I T Y A S S E M B LY A N D DY N A M I C S

Page 136: Whittaker. Island Biogeography. 2007

I S L A N D A S S E M B LY T H E O RY 123

Box 5.2 How consistent are species incidence functions?

Incidence functions are simple analyses of thefrequency of occurrence of individual species inrelation to one or sometimes two properties of aset of islands: typically species richness, area,and isolation. Bierdermann (2003) provides aninteresting analysis of area–incidencerelationships of 50 species of vertebrates andinvertebrates in which he demonstrates thatarea requirements increase essentially linearlywith increasing body size on a log–log scale.Bierdermann rightly cautioned that it should notbe assumed that species incidence functions areinvariant. In illustration, Hinsley et al. (1994)have shown for birds of English woodlands thatthey may vary through time in relation todensity independent mortality (extremes ofweather). It might also be the case for habitatislands that they vary across the range of aspecies or as a function of the properties of thelandscape in which the fragments areembedded.

Watson et al. (2005) tested this idea for threefragmented woodland landscapes in south-eastern Australia, all sufficiently close that climateand the regional species pool can be assumedsimilar. The landscapes were: (1) an agriculturallandscape, used for both pastoral and arablefarming; (2) a peri-urban landscape featuringpastures, hobby farms, and small urban areas; and(3) an urban landscape, namely the city ofCanberra. They used logistic regression toestablish sensitivity of species incidence to area,and to isolation in each landscape separately.Results demonstrated considerable variability inthe incidence functions between the threelandscapes and among species.

They may be categorized as follows:8 species showed no sensitivity to area or isolationin any landscape4 species showed minimum area thresholds in alllandscapes4 species showed minimum area thresholds onlyin the urban landscape1 species showed a minimum area threshold butonly in peri-urban and urban landscapes3 species showed maximum isolation thresholdsonly in the agricultural landscape6 species showed area effects in all landscapesand isolation effects (according to the

10, 000

1000

100

10

1

0.1

10, 000

1000

100

10

1

0.1

10, 000

100

10

1

0 1000 2000 3000 4000 5000 6000 7000Isolation (m)

Are

a (h

a, lo

g sc

ale)

Agricultural

0 2000 4000 6000 8000 10,000Isolation (m)

Are

a (h

a, lo

g sc

ale)

Peri-urban

Urban

0 1000 2000 3000 4000 5000 6000Isolation (m)

Are

a (h

a, lo

g sc

ale)

(a)

(b)

(c)

Box Figure 1. Area-isolation incidence plots for the eastern yellowrobin Eopsaltria australis in three fragmented woodlandlandscapes in south-eastern Australia. Open circles are unoccupiedremnants, closed circles are occupied remnants. The speciesdisplays a compensatory pattern in the agricultural landscape (a);and minimum area patterns in both the peri-urban (b) and urban(c) landscapes. Dashed lines represent the threshold where theprobability of occurrence reaches 50%, as determined by logisticregression (From Watson et al. 2005).

Page 137: Whittaker. Island Biogeography. 2007

124 C O M M U N I T Y A S S E M B LY A N D DY N A M I C S

Eopsaltria australis

Area (ha, log scale)

1 10 100 10000.0

0.2

0.4

0.6

0.8

1.0Acanthiza pusilla

Area (ha, log scale)

1 10 100 1000

Inci

den

ceIn

cid

ence

Inci

den

ce

Inci

den

ceIn

cid

ence

Inci

den

ce

0.0

0.2

0.4

0.6

0.8

1.0

Agricultural landscapeUrban landscapePeri-urban landscape

Smicrornis brevirostris

Area (ha, log scale)

1 10 100 10000.0

0.2

0.4

0.6

0.8

1.0

Cormobates leucophaeus

Area (ha, log scale)

1 10 100 10000.0

0.2

0.4

0.6

0.8

1.0

Climacteris picumnus

Area (ha, log scale)

1 10 100 10000.0

0.2

0.4

0.6

0.8

1.0Chthonicola sagittata

Area (ha, log scale)

1 10 100 10000.0

0.2

0.4

0.6

0.8

1.0

Box Figure 2. Incidence curves for six area-sensitive woodland species in the same three landscapes shown in the previous figure. Incidencefunctions were developed using logistic regression (From Whittaker et al. 2005, after Watson 2004).

Page 138: Whittaker. Island Biogeography. 2007

widespread species tend to be better colonists, orpossibly a combination of the two.

Anthropogenic experiments in island assembly:evidence of competitive effects?

It is difficult to assess the importance of competi-tion in the process of community assembly onislands. One reason for this is the difficulty ofobserving immigration events. Human interven-tion in island systems has, however, inadvertentlycreated opportunities to test for island assemblystructure, and in cases provides evidence indicativeof competition as a structuring force.

One such example is a study by Badano et al.(2005), who use the chequerboard index or C-scoreto analyse ant community structure in islands of theCabra Corral dam, Argentina. They found thatnewly created islands showed a random pattern ofspecies co-occurrence, but that on older (forest rem-nant) islands species showed a significant tendencyto co-occur less often than expect by chance. This,they argued, is a consequence of competitive sort-ing on the older islands. Contrasting results wereobtained in another study of newly created (forestremnant) islands from within a South Americanreservoir, this time an analysis of forest-interior

birds nesting on islands within Lake Guri,Venezuela by Feeley (2003). He tested several dif-ferent assembly rule models, finding strong evi-dence of nestedness (below) related to habitatspecificity, such that specialist species tend to beabsent from small, species-poor islands. This indi-cates that the communities have been shaped bystructured losses of species rather than through theeffects of interspecific or interguild competition,although there was some limited evidence of com-petitive limitations in the degree of size similaritypermissible in co-occurring species.

The purposeful introduction of land birds toremote oceanic islands provides another form ofinadvertent experiment on the role of competitiveeffects in structuring insular avifaunas. Such intro-ductions have been for various purposes, includingas game birds, as song birds, and to ‘beautify’ thenative avifauna. Very often the native avifauna hasbeen depleted concurrent with or before the intro-ductions, with the exotic species to varying degreesreplacing the lost species (Sax et al. 2002). Althoughthe natural immigration rate of land birds to suchislands in the absence of human transformation ofthe landscape can be assumed to be negligible, highrates of introductions of species by people havebeen documented in several archipelagos

I S L A N D A S S E M B LY T H E O RY 125

compensatory pattern) only in the agriculturallandscape

1 species (brown thornbill, Azanthiza pusilla)showed an isolation threshold in the agriculturallandscape and area thresholds in the other twolandscapes

None of the species was found to show a blockarea/isolation pattern

The variability in area/isolation is illustratedgraphically for eastern yellow robin (Figure 1).Even where the species showed a common typeof response between landscapes, the minimumarea thresholds were found to vary significantly(Figure 2). These results show that speciesincidence functions can vary substantially, evenwithin the same ecoclimatic region. As thewoodland remnants themselves were similar incharacter across the study system, Watson et al.(2005) concluded that the differences in incidence

functions between the landscapes were beingdetermined by the nature of the matrix of non-woodland habitats in which the woodlandfragments were embedded.

There are many ways in which matrix habitatsmight impact on the incidence of woodlandbirds within a patchy landscape, and it is notpossible to know which were operative in thissystem. Nor is it possible to claim that thepatterns shown, e.g. by the eastern yellow robin,would be replicated in another suite of urban,peri-urban, and agricultural catchmentselsewhere within the range of the species. Thesefindings do, however, imply that althoughspecies incidence functions may show someemergent tendencies (e.g. as reported for bodysize by Bierdermann 2003), they are notconsistent properties within the range of aspecies.

Page 139: Whittaker. Island Biogeography. 2007

(Lockwood and Moulton 1994). Only a proportionof introduced species succeeds in establishing andmaintaining populations over lengthy periods. Intests of data for introductions of passeriform birdsto the Hawaiian islands, Tahiti, and Bermuda, it hasbeen found that the surviving species exhibit whatis termed morphological overdispersion, i.e. theyare more different from one another in characteris-tics that are believed to relate to niche separationthan chance alone would predict (Lockwood andMoulton 1994). Similar results have been reportedfor game birds (galliforms) for the Hawaiianislands and New Zealand (Moulton et al. 2001).These results appear to support the hypothesis thatinterspecific competition shaped the compositionof these anthropogenic communities.

Like other island assembly rules the morpholog-ical overdispersion pattern require careful scrutinyto ensure that it is not an artefact of, for example: (1)the biogeographic region of origin of the intro-duced species, (2) the ecological or taxonomic struc-ture of the introduced species, or (3) the numbers ofintroduced birds (e.g. see Blackburn and Duncan2001; Cassey et al. 2005a). Of these factors, the intro-duction effort has been highlighted as crucial in anumber of data sets (Cassey et al. 2005a,b). On theother hand, in their analysis of game bird introduc-tions, Moulton et al. (2001) argue that evidence forthe introduction effort hypothesis in their data set isweak, citing examples of game birds failing becausethey were introduced into inappropriate environ-ments in rather indiscriminate fashion: in such cir-cumstances mere weight of numbers may notensure survival. Hence, although it is certainly thecase that greater numbers of released birdsprovides a greater likelihood of initial success of aspecies establishing, they argue that the emergentmorphological pattern of the assembled communityof exotic species shows evidence of competitivestructuring.

Just when you think it might be safe to concludein favour of competitive effects, once again the casehas been disputed, by reference to the data derivedfrom the New Zealand game-bird introductions.Duncan and Blackburn (2002) slay the beautifulhypothesis with the traditional ugly fact. Theyargue that competition among morphologically

similar species could not have been responsible forthe failure of most game-bird introductions to NewZealand because most species were released atwidely separated locations or at different times,and would never have encountered other morpho-logically similar introduced game birds, or if theydid would not have done so at the necessary densi-ties to influence the outcome. They conclude thatthe significant pattern of morphological overdis-persion in this instance must therefore result fromsome other cause, such as, for example, greatereffort being expended on introductions of morpho-logically distinct species. Interactions with intro-duced predators, and with other anthropogenicchanges in New Zealand further complicates attri-bution of the overdispersion pattern (Duncan et al.2003). Accepting the criticisms of the New Zealandrequires that the Hawaiian data stand alone, andtaken in isolation the Hawaiian galliform data failto show overdispersion, and so, according toDuncan and Blackburn (2002) the case falls. Fromthis we may conclude that the most predictable fea-ture of island assembly rules is that any claim ofevidence of competitive effects will be contested: atlast some evidence of competition, even if onlyamongst biologists!

5.2 Nestedness

If island biotas are in essence randomly drawnfrom a regional species pool, as per the simplestform of the EMIB, then an archipelago of islandsshould exhibit differences in composition from oneisland to the next and the degree of overlap shouldbe predictable on the basis of a ‘null’ model.Diamond’s assembly rules seek to describe depar-tures from such an expectation, another form ofdeparture is where island biotas exhibit nestedness.A nested distribution describes the situation wheresmaller insular species assemblages constitute sub-sets of the species found at all other sites possessinga larger number of species (Patterson and Atmar1986). Diamond (1975a) intended his assemblyrules approach to be deployed on groups of ecolog-ically related species, or guilds, whereas the analy-ses of nestedness (like incidence functions) providedescriptive tools that can be applied to either

126 C O M M U N I T Y A S S E M B LY A N D DY N A M I C S

Page 140: Whittaker. Island Biogeography. 2007

narrow or broad groups of species. It might thus bepossible for chequerboard distributions and the liketo be detectable within a particular guild of birds,while the avifauna analysed as a whole exhibits atendency towards nestedness.

Tendencies towards nestedness were recognizedby early island biogeographers, such as Darlington(1957), but it was not until the 1980s that statisticaltests were developed for formal detection of thepattern. When the first indices were developed byPatterson and Atmar (1986), they found that insularsystems are commonly nested (Patterson 1990;Blake 1991; Simberloff and Martin 1991). Pattersonand Atmar (1986) took the view that nestednesswas most likely to occur in extinction-dominatedsystems and developed their first metric on thispresumption. However, other authors argued thatimmigration might also produce nestedness pat-terns, requiring different metrics for quantifyingnestedness. This led to much debate about how bestto measure nestedness, a matter that is most likelystill unresolved (e.g. Wright and Reeves 1992; Cookand Quinn 1995; Lomolino 1996; Brualdi andSanderson, 1999; Wright et al. 1998; Rodríguez-Gironés and Santamaría 2006). Although the choiceof metric is important at a detailed level, thegeneral proposition that nestedness is a commonfeature of ecological assemblages (both insular andnon-insular) is robust to the choice of metric(Wright et al. 1998).

The nestedness concept, as applied by Pattersonand Atmar and as generally followed by others, isbased on ordering the data matrix by the size offauna or flora, i.e. it is richness-ordered nestedness.Some authors, however, have used the same term‘nestedness’ when ordering the data matrix not byspecies richness but by island area, which we mightterm area-ordered nestedness, or even by islandisolation, i.e. distance-ordered nestedness (e.g. seeRoughgarden 1989; Lomolino and Davis 1997). Theextent to which island assemblages are found to benested when ordered in these different ways maybe used to infer processes responsible for structur-ing them. However, whichever metric is used, andwhichever variable is used for ordering the datamatrix, inferences of causation ideally requireindependent lines of verification.

As Wright et al. (1998, p. 16) state ‘nestedness isfundamentally ordered composition . . . Any factorthat favours the ‘assembly’, or disassembly . . . ofspecies communities from a common pool in a consis-tent order will produce nested structure’ [italics inoriginal]. They identify four factors or filters thatmay be important contributors in generating nestedstructure: passive sampling, habitat nestedness,distance, and area.

● So called passive sampling of a pool of speciesby a recipient island may produce nested structureif species are drawn randomly from the pool withthe constraint that the availability of propagules isitself strongly non-random, assuming for instance alog-normal species abundance distribution for thepool community. Simulation models on this basisgenerate highly nested model communities, gener-ally in fact more strongly nested than observed.Where passive sampling simulations do comparewell with observed data sets, this may lead to theinference that differential immigration has struc-tured the assemblages. However, Wright et al.(1998) argue that the simulation is not dependenton a particular mechanism of community assembly,and that other scenarios are conceivable. One mightbe that abundant species from within the pool areregionally ‘successful’ species with a low probabil-ity of going extinct from isolates.

● For habitat nestedness to produce biotic nested-ness requires, first, a strong fidelity of species toparticular micro-habitats, and second, for islands toacquire additional habitats in a fairly fixed order.Given which, we would anticipate that it will bemore readily detectable in plant data sets (as plantsoften show high habitat affinity), as exemplified byHonnay et al.’s (1999) study of plants in forestfragments in Belgium.

● If nestedness is clearly related to distance frommainland sources, this suggests that there arepredictable limits to species’ dispersal abilities,such that the system is made up by islands‘sampling’ a series of species’ isolation-incidencefunctions (as Fig. 5.4). In many case studies, whereavailable data are limited to a single time-point, itmay be safest simply to label this pattern a‘distance’ effect (Wright et al. 1998), but there are

N E S T E D N E S S 127

Page 141: Whittaker. Island Biogeography. 2007

cases where the early stages of succession on aseries of islands allow researchers to examine nest-edness during the assembly process, and thusestablish more directly the predictability of thecolonization process. A good example is Kadmon’s(1995) study of seven islands created by the fillingof the Clarks Hill Reservoir, Georgia, USA. Theislands were logged and cleared of woody plantsbefore their separation from the mainland. Thirty-four years later, species number was found to beinversely related to island isolation. The woodyplant floras were found to be nested when rankedby richness, or isolation, but not by area. Theseresults point to the significance of differential colo-nization in structuring the data set. Only a smallproportion of the species included in the analysiscontributed to the observed nestedness, principallybeing those that lacked adaptations for long-rangedispersal. Wind-dispersed species showed noevidence for nestedness. This study thus pointed tothe significance of plant dispersal attributes instructuring the rebuilding of insular assemblages, afeature also of the recolonization data fromKrakatau, as we shall shortly see.

● Nestedness can also result from differential arearequirements across species, again as may be

revealed in the individual species’ area-incidencefunctions of the component species. Where nested-ness within a former land-bridge island data set isstrongly area-related but is unrelated or onlyweakly related to isolation within a data set, this istaken to indicate a strong extinction signal in thestructuring of assemblages. This rationale wasinvoked in classic early studies of nestedness ofmammals on land-bridge islands (e.g. Pattersonand Atmar 1986). This is on the grounds that beforethey became islands, the land-bridge islands werepart of the mainland and can be assumed to havehad a full complement of the mainland pool, withextinction then the dominant process in the sortingof the island biotas. Similarly, extinction-led sort-ing of remnant mountain-top ‘island’ has beeninvoked for the nestedness of mammal assem-blages in the south-western USA (Patterson andAtmar 1986).

Given that nestedness appears to be so prevalent, itis worth considering factors that may constrain it:

● First, consider the implications of supertrampdistributions: species that tend to occur preferen-tially on species poor islands, and to disappear inspecies rich assemblages. Such patterns, along withcertain other of Diamond’s assembly rule patterns(above) will form exceptions to any general system-wide tendency to nestedness.

● Second, nestedness must perforce be diminishedas the geographic extent of the study system isexpanded to incorporate different species pools, i.e.biogeographical turnover in species composition.● Third, Wright et al. (1998) suggest that where thesample system involves mostly very small patches,the degree of nestedness may be diminished byhigh habitat heterogeneity between patches (i.e. theopposite of habitat nestedness).

It is now well established that nestedness is acommon but not universal feature of many insularbiotas. In an applied context, it has also been shownthat within a given landscape, different taxa canrespond in vastly different ways to fragmentation.For instance, Fischer and Lindenmayer (2005),examining data from fragmented forest habitats insouth-eastern Australia, found that whereas birds

128 C O M M U N I T Y A S S E M B LY A N D DY N A M I C S

Islands

Species

Dispersal zones

C DB

A

Mainland

Figure 5.4 Hypothetical scenario for the production of nestedsubsets via differential colonization. From Patterson and Atmar(1986, Fig. 4), who contended, nonetheless, that extinction is a morepotent structural determinant in the majority of insular systems.

Page 142: Whittaker. Island Biogeography. 2007

showed a strong degree of nestedness, the arborealmarsupials were non-nested. In this case, theirexplanation for the nestedness of the avifaunainvoked selective extinction as the dominantprocess.

Table 5.2 provides an interesting classification ofnestedness patterns by Patterson (1990) using oneparticular metric. Opinions have varied as to thebalance of explanatory power among the factorsgenerating nestedness (Pattterson and Atmar 1986;Patterson 1990; Simberloff and Martin 1991; Cookand Quinn 1995; Nores 1995), but Patterson’s con-clusions appear to be broadly supported in a morerecent multimetric, multisystem analysis by Wrightet al. (1998), who have reaffirmed that land-bridgearchipelagos are typically strongly nested, and

immigration experiments least nested. Theircomparative analysis adds further support to theargument that differential extinction is a powerfulforce in producing nested structure, but also pointsto the realization that different types of system maybe dancing to different tunes: sometimes immigra-tion is dominant, sometimes extinction, sometimeshabitat nestedness. The relevance of nestednesswithin conservation biogeography is examined inChapter 10.

5.3 Successional island ecology: firstelements

Consideration of nestedness in this chapter hasled to the point where questions concerning the

S U C C E S S I O N A L I S L A N D E C O L O G Y: F I R S T E L E M E N T S 129

Table 5.2 Summary of significant nested subset structure (according to Patterson 1990), inassemblages affected over various time periods by: mainly extinction; mainly colonization; or bothprocesses in concert. The study was based on the nestedness index, N, of Patterson and Atmar (1986),which has been argued to be suboptimal for the analysis of systems affected by colonization (seediscussion in Cook and Quinn 1995).

Processes Nested Non-nested

Holocene changesExtinction dominated Southern Rocky Mts mammals

New Zealand land-bridge birdsBaja land-bridge mammalsBaja land-bridge herptilesGreat Basin mammals

Colonization dominated Baja oceanic birds New Zealand oceanic birdsBaja oceanic mammalsBaja oceanic herptiles

Both processes Baja land-bridge birdsGreat Basin birds

Long-term changesExtinction dominated Australian wheat-belt mammals Australian wheat-belt lizards

North American park mammalsSão Paulo sedentary birdsBass Strait mammals

Colonization dominated All São Paulo birdsBoth processes Penobscot Bay mammals

Short-term changesExtinction dominated Mangrove area reductionColonization dominated Dispersing shrews and voles Breeding shrews and voles

Weeds of young lots Weeds of old lotsBoth processes Mangrove defaunation

Page 143: Whittaker. Island Biogeography. 2007

colonization process require more detailed attention.Both island biogeography and succession theory goback a long way. Biologists have been studyingthem in combination since the nineteenth cen-tury (e.g. references in Whittaker et al. 1989),although not necessarily within the same frame-works we use today. Ridley may well have beeninfluenced by the early Krakatau data when hewrote in the preface to his classic text on plantdispersal (Ridley 1930, p. xii):

An island rises out of the sea: within a year some plantsappear on it, first those that have sea-borne seeds or rhi-zomes, then wind-borne seeds, then those borne on thefeet and plumage of wandering seafowl, and when thevegetation is tall enough, come land birds bringing seedsof the baccate or drupaceous fruits which they had eatenbefore their flight.

This provides the first elements of a dispersal-structured or successional model of islandassembly.

Successional effects were discussed in relation toequilibrium theory by MacArthur and Wilson(1967), but in a fairly simplified fashion and withlittle empirical data. Following their work, one ofthe more interesting contributions was provided bythe data for the two volcanically disturbed islandsfrom Diamond’s (1974) study of New Guineaislands. Ritter was sterilized in 1888, and Long wasdevastated about three centuries before Diamond’ssurvey of the two islands in 1972. Vegetation recov-ery on each island was found to be incomplete:Ritter supported Pandanus (screw-pines) up to 12 mhigh on its gentler slopes, but was bare in steeperareas, succession being retarded by landslides,strong prevailing winds, and erosion. Long’srecovery was more advanced, but its forestsremained comparatively open and savanna-like.The species–area plot in Fig. 5.5 reveals that bothRitter (4 species) and Long (43 species) remaineddepauperate (Diamond 1974). Observations offailed arrivals of other species on Ritter, includingfeathers at the plucking perch of a resident pere-grine falcon, provided further support for a habi-tat/successional limitation explanation. Diamondcompared his data with a survey from 1933,calculating by the means explained in Chapter 4

that its species number was approximately 75% ofthe equilibrium value. He found that speciesnumbers had increased only slightly in the interim,with a degree of turnover occurring—a result heattributed to the slow pace of forest succession‘arresting’ the progress of the Long avifaunatowards its equilibrium value. As turnover was notgreater than on older islands he termed it a quasi-steady state. This hierarchically determined patternprovides an interesting parallel with subsequentstudies of the build-up of bird numbers from theKrakatau group (below).

Diamond’s line of argument was of muchbroader biogeographical scope, as we have seen,but the successional dynamics represented byRitter and Long held a key place in the overalltheory of the dynamics of the New Guinea islands.Diamond’s reasoning was as follows. Small islands,on which extinction rates should be high, will havea high incidence of the best dispersing species,i.e. they have a high incidence of tramps. Largerdefaunated islands will first be colonized by ther-selected supertramps. Other, less dispersivetramps arrive subsequently and eventually crowd

130 C O M M U N I T Y A S S E M B LY A N D DY N A M I C S

0.001 0.01 0.1

Num

ber

of s

peci

es

Area (km2)1 10 100 1, 000 10, 000 100, 000

Ritter

Long

200

100

50

20

10

5

2

Control islandsExploded volcanoesTidal-wave-defaunatedislets

Key

Figure 5.5 The relationship between log species number and log-island area for resident, non-marine, lowland bird species on theBismarck Islands (redrawn from Diamond 1974). The line of best fit isfor the so-called control islands. The explosively defaunated islands ofLong and Ritter remain significantly (P�0.001) below the line of bestfit, which Diamond interpreted as demonstration of incompletesuccession and failure to attain equilibrium in the period since theirvolcanic disturbance.

Page 144: Whittaker. Island Biogeography. 2007

out the less competitive supertramps. Before theirloss, however, the island acts to supply thesurrounding area with its surplus population, andthe species will therefore survive by having foundanother defaunated island before being excludedfrom the first. Diamond calculated that for somesupertramps the availability of Long island willhave enabled the quadrupling of the entire popula-tion of the species. Long and Ritter thus holdconsiderable interest. In Long’s case, this has beenadded to recently by studies of the emergent islandMotmot, which has formed in the lake within LongIsland, providing a form of nested colonization‘experiment’ (see Thornton et al. 2001).

5.4 Krakatau—succession, dispersalstructure, and hierarchies

Background

Krakatau was the first site used by MacArthur andWilson to test their predictions of an approach to adynamic equilibrium. The natural recolonization(Box 5.3) of the islands by birds between 1883 and1930 appeared consistent with the EMIB: speciesnumber reached an asymptote and further turnoveroccurred (MacArthur and Wilson 1963, 1967).However, the plant data from the islands revealed apoorer fit, as species numbers continued to rise.This was postulated to be due to one of two effects:either that the pool of plant species was sufficientlylarge to prevent a depletion effect on the immigra-tion rate, or successional replacement of the pioneercommunities was incomplete, thereby leading to areduction in extinction rate for a time. The lattereffect was favoured, and illustrated in diagram-matic form, but the EMIB was not modifiedformally to account for the data.

Community succession

Although the first food chains to establish may wellhave been of detritivores, and microorganisms, con-sideration of the real business of community assem-bly begins with the higher plants. Partial survey dataand descriptions of the plant communities areavailable from often brief excursions in 1886,

1896/97, 1905–08, most years between 1919 and1934, 1951, 1979, 1982/83, and 1989–95 (Whittakeret al. 1989, 1992a; 2000). The system must be under-stood both in terms of successional processes and inrelation to constraints on arrival and colonization.The broad patterns of succession will be describedfirst.

The coastal communities established swiftly,most of the flora being typical of the sea-dispersedIndo-Pacific strand flora. In 1886, 10 of the 24species of higher plants were strandline species. By1897, it was possible to recognize a pes-caprae for-mation (named after Ipomoea pes-caprae) of strand-line creepers, backed in places by establishingpatches of coastal woodlands, of two characteristictypes. First, patches which were to become repre-sentative of the so-called Barringtonia association(more generally typified by the tree Terminalia cat-appa), another vegetation type typical for theregion, and secondly, stands of Casuarina equisetifo-lia. The latter is a sea- and wind-dispersed pioneer-ing tree species, which also occupies someprecipitous locations further from the sea. In thecoastal areas it typically lasts for just one genera-tion, as it fails to establish under a closed forest. Itis notable that as early as 1897 the coastal vegeta-tion types were already recognizable as similar tothose of many other sites in the region (Ernst 1908).The coastal communities continued to accruespecies over the following two decades, but havesince exhibited relatively little directional composi-tional turnover, apart from the loss in many areas ofCasuarina. They may thus be described as almost an‘autosuccession’ (Schmitt and Whittaker 1998).

In the interiors, a much more complex sequenceof communities has unfolded. To varying degreesthis can be understood in relation to the consider-able differences in habitats between and withinislands, to differential landfall of plant specieswithin the group, and to the dynamics of thephysical environment, especially the disruptionoriginating from the new volcanic island, AnakKrakatau. In 1886, most of the cover in the interiorwas supplied by 10 species of ferns, accompaniedby 2 species of grass and 2 members of theAsteraceae: all 14 species being wind-dispersed. By1897, the interiors had become clothed in a dense

K R A K ATAU — S U C C E S S I O N, D I S P E R S A L S T R U C T U R E , A N D H I E R A R C H I E S 131

Page 145: Whittaker. Island Biogeography. 2007

132 C O M M U N I T Y A S S E M B LY A N D DY N A M I C S

Box 5.3 The Krakatau eruption of 1883: a dramatic start to a ‘natural experiment’

The Krakatau group has undergone repeatedphases of volcanic activity. In 1883 it consisted ofthree islands, arranged in a caldera which resultedfrom prehistoric eruptive activity. The largestisland of the group became active in May of thatyear, commencing a sequence that ended on27 August in exceptionally destructive eruptions.Two-thirds of the island disappeared, vastquantities of ejecta were thrown into theatmosphere, and a series of huge waves (tsunami)generated in the collapse resulted in an estimated36 000 human casualties in the coastalsettlements of Java and Sumatra flanking eitherside of the Sunda Strait some 40 km or so fromKrakatau. The meteorological and climatologicaleffects of the eruption were observable across theglobe and the islands excited considerable andlasting scientific interest from a number ofdisciplines (Thornton 1996). Each of the threeKrakatau islands was entirely stripped of allvegetation. The main island, now known asRakata, lost the majority of its land area, but allthree islands also gained extensive areas of newland resulting from the emplacement on to thepre-existing solid rock bases of great thicknessesof pyroclastic deposits. Estimates of ash depthsare of the order of 60–80 m and to this dayalmost all of the area of the three islands remainsmantled in these unconsolidated ashes, withrelatively little solid geology exposed at thesurface. No evidence for any surviving plant oranimal life was found by the scientific team led byVerbeek later that year, and in May 1884 the onlylife spotted by visiting scientists was a spider. Thefirst signs of plant life, a ‘few blades of grass’,were detected in September 1884.

The efficacy of the destruction has been hotlydebated (Backer 1929; Docters van Leeuwen1936). It is conceivable that some viable plantpropagules might somehow have survived to beuncovered by later erosion of the ash mantle(Whittaker et al. 1995). However, there is noevidence for survival, and indeed the densestpopulations of early plant colonists were locatedon terra nova. Any viable rain forest propagulesuncovered by erosion of the ash would in anycase have emerged into a vastly different andhostile environment, and their successful

establishment would thus have been entirelyremarkable. The islands can be taken to havebeen as near completely sterilized as makes nopractical difference. The fauna and flora have thuscolonized since 1883 from an array of potentialsource areas, the closest of which, the island ofSebesi, is 12 km distant. All the nearest landareas, including Sebesi, were also badly impactedby the 1883 eruptions.

Here was a group of three islands, eachmantled in sterile ashes and in time receivingplant and animal colonists. The potential of theislands as a natural experiment in dispersalefficacy and ecosystem recovery was appreciated,although, unfortunately, only botanists seized theopportunity at an early stage, with surveys in1886 and 1897. Since then, numerous scientificteams have worked on Krakatau, to varying ends,but along the way accumulating a remarkable(if imperfect) record of the arrival, succession andturnover of plant species and many groups ofanimals (Whittaker et al. 1989; Thornton 1996;Whittaker et al. 2000).

It is important to appreciate the dynamism ofthe platform upon which ecosystem assembly hastaken place. The early pace of erosion of the ashmantle must have been dramatic. It createda deeply dissected ‘badlands’ topography whichremains geomorphologically highly dynamic.The extensive new territories around the coastswere also subject to extreme rates of attrition,and steep cliffs formed rapidly around much ofthe shoreline. Shallow shelving beaches, whichprovide the most favourable points for thecolonization of many plant and some animalspecies, are restricted in their extent. In 1927,after nearly 46 years of inactivity, a new islandbegan to form in the centre of the 1883 caldera,finally establishing a permanent presence, AnakKrakatau, in 1930. Through intermittent activityit grew by the mid-1990s to an island of over280 m in height and 2 km in diameter (Thorntonet al. 1994). Over this period it caused significantand widespread damage to the developingforests of two of the three older islands, Panjangand Sertung, but only indirectly impacted onRakata (Whittaker et al. 1992b; Schmitt andWhittaker 1998).

Page 146: Whittaker. Island Biogeography. 2007

grassland, dominated by Saccharum spontaneum andImperata cylindrica, interspersed with small clustersof young pioneer trees. Ferns dominated only thehigher regions of Rakata and the balance of specieshad also shifted in favour of the flowering plants.By 1906, the woodland species had increasedconsiderably in the interiors, although remainingpatchy, and the fern communities were graduallyreceding upwards. The grasslands were tall anddense and so difficult to penetrate that proper explo-ration of the interiors was greatly hampered. Thewoodlands of the lowlands continued their rapiddevelopment, with Ficus spp., Macaranga tanarius,and other animal-dispersed trees to the fore.

Forest closure took place over most of the interiorof each island during the 1920s, such that by 1930very little open habitat remained. This key phase ofsystem development was fortunately the subject ofdetailed investigations by Docters van Leeuwen(1936). As the forests developed, habitat space forforest-dependent ferns, orchids, and other epiphyticplants became available, and their numbersincreased rapidly in response. Conversely, the pio-neering and grassland habitats were reduced,species populations shrank, and some species disap-peared. Rakata is a high island, c.735 m, and altitudi-nal differentiation of forest composition was evidentas early as 1921. The highest altitudes thereafter fol-lowed a differing successional pathway, in which theshrub Cyrtandra sulcata was for many years a keycomponent. Although most vegetation changesappear to have been faster in the lowlands, spread-ing up the mountain of Rakata, one key species, thewind-dispersed pioneering tree Neonauclea calycina,first established a stronghold in the upper reaches,before spreading downwards. By 1951 it had becomethe principal canopy tree of Rakata from just belowthe summit down to the near-coastal lowlands. Atthe close of the century it was clearly in decline in thelowlands, as the forests developed a more patchycharacter, with a number of other large canopy andsubcanopy species varying in importance from placeto place within the interiors.

The patterns of development on the much lowerislands of Panjang and Sertung were broadlysimilar up to c.1930, although differences in thepresence and abundance of particular forest species

were noted. Since 1930, both islands have receivedsubstantial quantities (typically in excess of 1 mdepth) of volcanic ashes over the whole of theirland areas. Historical records and studies of ashstratigraphies demonstrate that some falls of ashhave been very light, but c. 1932/35 and 1952/53,and possibly on other occasions, the impact hasbeen highly destructive (Whittaker et al. 1992b). Forinstance, Docters van Leeuwen (1936) describedhow in March 1931 the most disturbed forests ofSertung resembled ‘a European wood in winter’,and how grasses re-invaded (possibly resprouted)within the stricken woodlands. Since then, forestsdominated to a considerable extent by the animal-dispersed trees Timonius compressicaulis andDysoxylum gaudichaudianum have become charac-teristic of large areas of both islands. The interplayof slight environmental differences betweenislands, differential patterns of landfall, and distur-bance episodes from the volcano, has produced ashifting successional matrix across each island,influencing diversity patterns at fine and coarsescales in complex ways that are difficult to attributein other than general terms to the array of potentialcausal variables (Schmitt and Whittaker 1998).

Although different forest types have been recog-nized on the Krakatau islands, these ‘communities’are not discrete and forest successional pathwaysare in practice more complex than simple summarysuccessional schema (e.g. Whittaker et al. 1989,Fig. 15) might be taken to imply. It is notable that ofall the vegetation types of the Krakatau islands,only the pes-caprae formation and Barringtoniaassociation were assigned with any confidence tophytosociological types by the earlier plant scien-tists. As Docters van Leeuwen (1936, pp. 262–3) putit, ‘all other associations in the Krakatau islands areof a temporary nature: they change or are crowdedout.’ Although the coastal systems are similar tothose of other locations in the region, those cur-rently recognized from the interior by Whittakeret al. (1989) lack documented regional analogues ofwhich we are aware. The interior forests of theKrakatau islands continue to accrue new species ofhigher plants, and the balance of species in thecanopy is undoubtedly in a state of flux, withstrong directional shifts in the importance of

K R A K ATAU — S U C C E S S I O N, D I S P E R S A L S T R U C T U R E , A N D H I E R A R C H I E S 133

Page 147: Whittaker. Island Biogeography. 2007

particular species being evident over the periodsince 1979 (Bush et al. 1992; Whittaker et al. 1998).The business of forest succession is ongoing.

A dispersal-structured model of islandrecolonization

The community-level changes have been under-pinned by trends within the colonization data

134 C O M M U N I T Y A S S E M B LY A N D DY N A M I C S

(Whittaker et al. 1989, 1992a). Figure 5.6 simplifiesthese trends into three time-slices, restricting thetreatment to the island of Rakata, for which the bestdata (least confounded by volcanic disturbance) areavailable. The figures are for all species recordedfrom the island, whether still present or not(Whittaker and Jones 1994a). Phase 1 represents thepioneering stage during which wind-dispersed pio-neers, first ferns, and then grasses and composites

Sea Animal

Wind

Colonization rate increases as attraction to frugivores increases

Pioneers lose ground

Colonizationcontinues at areduced pace

0.18Ferns

1.18 1.32

0.86Other

Sea Animal

Wind

Habitatrestrictions only

Species with large, heavy,winged seeds, such asmany mature forest canopy trees

Restricted groups:colonization improbable

Large seeded bat-spread or terrestrial mammal-spread

species

Animal

Wind

Delayedcolonization

of interior

Phase 1

Rapid colonization of strand-lineSimple succession

1.64

0.14

1.0Ferns

1.0Other

Wind

0.83Ferns

0.83Other

0.29 1.26

Phase 3

Colonization rate declines Diplochorous speciesspread to interior by birds and bats

More forest trees arrive Local populations of bats

and birds critical to mozaic development

Sea Sea Animal

Phase 2

(ferns, grasses, composites) as habitat availability increasesRapid colonization of interior by pioneers Forest epiphytes increase

Figure 5.6 Plant recolonization of Rakata Island (Krakatau group) since sterilization in 1883. The three phases correspond with survey periodsand represent convenient subdivisions of the successional process. Phase 1, 1883–1897; phase 2, 1898–1919; phase 3, 1920–1989. Arrowwidths are proportional to the increase in cumulative species number in species/year (these values are also given by each arrow). The flora issubdivided into the primary dispersal categories, i.e. the means by which each species is considered most likely to have colonized: animal-dispersed (zoochorous), wind-dispersed (anemochorous), and sea-dispersed (thalassochorous). The model distinguishes between strand-line (outercircle) and interior habitats and, in the fourth figure, identifies constraints on further colonization. (From Whittaker and Jones 1994a.)

Page 148: Whittaker. Island Biogeography. 2007

(Asteraceae), initiated the colonization of the inte-rior. As emphasized above, colonization of thestrandlines was rapid during this phase, graduallydiminishing thereafter.

Phase 2 represents the period during which theextensive grasslands waxed and waned, as, increas-ingly, animal-dispersed trees and shrubs spread outfrom their initial clumps to form woodlands. Theinteriors have been filled almost exclusively byspecies which are primarily either wind- or animal-dispersed. The relative balance between these twogroups shifts dramatically between phase 1 and 2.Few ferns colonized during phase 2 despite the effi-ciency of their dispersal system (microscopicspores). The interpretation offered for this is thatthere are actually relatively few ferns in the regionalspecies pool which typify such extreme, seasonallydroughted pioneer sites, and they arrived veryquickly during phase 1. Wind-dispersed early suc-cessional flowering plants also continued to accrueduring phase 2, including Asteraceae and terrestrialorchids, but very few trees and shrubs. The vastmajority of arboreal species are animal-dispersed,and after a very slow start, in which only twospecies of animal-dispersed species were found by1897, their rate of arrival took off in the early 1900s.

Phase 3 runs from the point when forest closurebecame extensive, to the 1989 datum. During thisphase, the colonization rate of sea-dispersedspecies slackens, which reflects the relativelylimited source pool, but also the limited array ofcoastal habitats available. In the interiors, it wasonly when forest habitats became available that thesecond wave of ferns, those of forest interiors,could build in numbers. Hence their rate ofdiscovery increased rapidly in the period of forestformation, peaking around 1920. The ease of dis-persal of ferns, with their microscopic propagules,can be demonstrated by comparison of the size ofthe Krakatau flora (all islands 1883–1994) with thenative flora of West Java. The ratio for spermato-phytes is 1:10.1 and for ferns 1:4.2, indicatingKrakatau to have a remarkably rich fern flora(Whittaker et al. 1997). Wind-dispersed floweringplants colonizing the newly available forests ofphase 3 were predominantly orchids, many ofwhich are epiphytic, together with a mix of other

epiphytic and climbing herbaceous species. Thesingle largest group recorded for phase 3, however,is the animal-dispersed set, mostly being trees andshrubs.

More detailed analyses of dispersal mechanismssuggest further structural features in the data. Forinstance, wind-dispersed species can be split intothose with dust-like propagules, plumed seeds, andwinged seeds or fruits. The larger, winged propag-ules have limited dispersive ability and, althoughsignificant in the regional pool of forest canopytrees, only one or two species of this category havecolonized Krakatau (Whittaker et al. 1997): they arethus noted in Fig. 5.6 as improbable colonists.Colonization of the animal-dispersed species ofKrakatau, with the exception of a few human-introduced cases, can be attributed to the actions offrugivorous birds and bats. Whittaker and Jones’s(1994b) best estimates of their relative roles aregiven in Table 5.3. Fruit bats swallow only thesmallest seeds (�1 or 2 mm) and it was assumedthat they would be unlikely to carry larger seeds intheir mouths or claws over the many kilometres of

K R A K ATAU — S U C C E S S I O N, D I S P E R S A L S T R U C T U R E , A N D H I E R A R C H I E S 135

Table 5.3 Estimates of the numbers of plant species found onKrakatau between 1886 and 1992 for which birds and bats have adispersal role. Under the heading of animal dispersal two modes oftransport may be distinguished: those seeds which are eaten,swallowed and which pass through the gut or which are eventuallyspat out, and those transported by external attachment to theanimal. The data set include all four Krakatau islands and all records(i.e. including some species which have not maintained a presence)(source: Whittaker and Jones 1994b.)

Dispersal mode Number of species

Animal, gut passage (bird and/or bat)* 124Animal, attached externally (bird) 10Human introduction, but spread by

feeding birds and/or bats 15Sea-colonist, then spread by animals 24

Total animal introduction and/or spread 173

* Of these 124 species, 50 were considered to be bird-dispersed, 31could potentially have colonized either by bird or bat transport, and31 were considered to have arrived by bird but then to have thepotential to be bat-spread within the islands. These estimates shouldbe regarded as first approximations (e.g. see further work onbat-dispersal on Krakatau, by Shilton 1999).

Page 149: Whittaker. Island Biogeography. 2007

sea separating Krakatau from source areas. Thus,small-seeded animal-dispersed species might beintroduced either by birds or bats, but larger-seededspecies can be considered strictly bird-assistedcolonists. Beyond a certain threshold seed size it islikely that only the largest fruit-pigeons can effectan introduction. However, once a species hasestablished on an island, a wider range of dispersalagents may be involved in local dissemination.

Where plant species have two rather contrastingdispersal vectors, they are termed diplochores.Twenty-four sea-colonist, but potentially animal-spread species occur on Krakatau, of which 14 arebat-spread, 4 may be bat- or bird-spread, and 6 arebird-spread. This mixed group of diplochoresincludes several of the earliest colonists of theKrakatau strandlines. They likely had an importantrole in kick-starting the succession of animal-dispersed species. As shown, restrictively animal-dispersed species were laggardly colonists. Therecannot be much advantage in visiting a barrenisland if you happen to be a frugivorous bird or bat.Once the fringing plant communities of sea-dispersed species established and began to fruit,however, the islands would have provided a suit-able food supply in the form of the fruit of thediplochorous species. From the early accounts weknow that many of the first true animal-dispersedplants were concentrated around the coastalfringes. Admittedly the interiors were less wellexplored, and certainly pockets of trees quicklybecame dotted throughout the interiors, yet we canbe fairly confident that the first points of landfall,and the first roosting points for frugivores and thefirst food supply, were provided by the coastal veg-etation.

The first frugivores observed on the islandswere birds, six species being observed in 1908 (thefirst bird survey), whereas it was not until the nextzoological survey in 1919 that bats were noted.Birds may thus have been the more importantgroup in terms both of simply the number ofanimal-dispersed colonists and being earlier inarrival (although survey data are too poor to beconclusive). However, bats have had a differingrole to birds. First, they have spread some speciesthat birds do not transport. For instance, the

important coastal and near-coastal tree Terminaliacatappa only invaded the island interiors after thearrival of the bats. Secondly, Whittaker and Jones(1994b) suggested that they might be more impor-tant to the early seeding of open habitats becauseof differences in their foraging behaviour (in partmediated by interactions with predators). Morerecent detailed work by Shilton (1999; Shilton et al.1999) supports these ideas, and also cruciallydemonstrates the ability of bats to retain viableseed in the gut for over 12 hours between the finalfeed of one night and the start of feeding on a sub-sequent night, by which time they may easily havemoved from mainland Java or Sumatra to theKrakatau islands (or vice versa). Although theirrole in long-distance seed dispersal seems limitedto only a few small-seeded plant genera with theright types of fruit (especially Ficus species), weare now convinced that their role in kick-startingforest succession was more important than earlierresearchers (e.g. Docters van Leeuwen 1936) rec-ognized.

The point of detailing these patterns andprocesses here is to illustrate that on real islands (inthe sea and some kilometres in area) succession iscomplex and demonstrates hierarchical interde-pendency across trophic levels. In order tounderstand how compositional patterns develop inthe vegetation and flora, it is necessary to considerthe ecological attributes of the plant species, theirhabitat relationships, and their hierarchical ecolog-ical links with animals (Bush and Whittaker 1991).Animals effect their dispersal, in cases their polli-nation, and of course also act as seed predators. Inturn the vegetation provides the habitat and foodresources of the animals; no fruit supply means noresident frugivores.

Colonization and turnover— the dynamics ofspecies lists

Given the basic elements of vegetation successionand colonization patterns, broken down into plantdispersal types, it is possible to interpret much ofthe pattern in the rates of immigration andturnover of both plants and animals of theKrakatau islands. What follows here flows on from

136 C O M M U N I T Y A S S E M B LY A N D DY N A M I C S

Page 150: Whittaker. Island Biogeography. 2007

the consideration of turnover in Chapter 4, but it isplaced here because of the structural featuresevident within it. An important caution applies tothese studies. The Krakatau islands are large, topo-graphically complex, and sufficiently difficult toexplore that some parts have never, or only rarely,been penetrated by scientists. For both plants andanimals, surveys span over 100 years of expedi-tions, have involved considerable turnover inpersonnel (and thus in expertise and methods), dif-ferences in intensity and differing areas of search.They have also taken place at irregular intervals.For these reasons, we must be cautious in interpret-ing changes in rates of immigration and extinction.The rates calculated are, in any case, not strictly asdefined in the EMIB, rather they represent ‘arrivalin the lists’ and ‘departure from the lists’.

Whittaker et al. (1989) presented an analysis ofhigher plant colonization for 1883–1983, whichillustrates the smoothing effect of calculating immi-gration rates over greater intervals of time, ignoringthe intervening survey data (Fig. 5.7). The repeatsurveys of Docters van Leeuwen (1936) in 1919 and1922 produce a pronounced spike in the immigra-tion data (another occurs for the same reasons atthe end of the series), but the peak for the 1920sremains even when this survey frequency effect isremoved. The low point, c.1951, reflects surveydeficiency. Bush and Whittaker (1991) therefore cal-culated ‘immigration’ and ‘extinction’ rates forplants using survey data grouped into adjacentsurveys, and ignoring the 1951 datum. Reasoningthat for most plants, especially large forest trees,temporary disappearance from the lists was morelikely to reflect survey inefficiency than genuineturnover, they provided upper and lower estimates:first, the recorded turnover and, secondly,assuming the minimum turnover allowable fromthe data (Fig. 5.8) (see also Whittaker et al. 1989,1992a). The overall trend in species richness is alsoshown in this figure, and although the lack of ade-quate surveys between 1934 and 1979 makes theprecise shape of the curve unknowable, additionalexploration between 1989 and 1997 continued toturn up additional species records. Providingregional source pools are not destroyed by humanaction, it seems probable from our analyses of

forest dynamics (e.g. Schmitt and Whittaker 1998;Whittaker et al. 1998) that species number can con-tinue to rise for some time to come.

Evident in the analyses of ‘immigration’ and‘extinction’ rates is that discovery of new species onRakata peaked during the period of forest forma-tion and closure, and the rate of loss of species fromthe lists also has a peak, lagging behind the‘immigration’ peak, and reflecting the loss of earlysuccessional habitats. Further survey work between1983 and 1994 allowed Whittaker et al. (2000) tore-examine the patterns of species loss in the lightof an improved knowledge of the present day flora.Their approach was to focus on the 325 species

K R A K ATAU — S U C C E S S I O N, D I S P E R S A L S T R U C T U R E , A N D H I E R A R C H I E S 137

01883 1903 1923 1943 1963 1983

2

4

6

8

10

12

14

16

18

Survey points

New

spe

cies

per

ann

um

Rakata raw data,13 intervalsRakata, 6 intervals

Rakata, 2 intervals

Rakata, Sertungand Panjang, 2intervals

Year

Figure 5.7 Numbers of new species of higher plants recorded onKrakatau over a variety of intervals, expressed as an annual rate. Thefigures exclude re-invasions. (Redrawn and corrected from Whittakeret al. 1989, Fig. 16.)

Page 151: Whittaker. Island Biogeography. 2007

138 C O M M U N I T Y A S S E M B LY A N D DY N A M I C S

300

(a) (b)

200

100

1883 1903 1923

Year

Cumulative

Immigration

Extinction

Num

ber

of s

peci

es

Spec

ies

per

annu

m

Nrec

1943 1963 1983 1883 1903 1923

Year

1943 1963 19830

5.0

4.0

2.0

1.0

3.0

0

Figure 5.8 Plant recolonization data for Rakata Island (Krakatau) 1883–1989 (redrawn from Bush and Whittaker 1991). Surveys are groupedinto survey periods, and the mean year of each period is used to estimate rates. Shaded lines represent the difference in rate according towhether minimum turnover is assumed or if ‘recorded’ immigration/extinction is assumed. (a) Cumulative species totals and number of speciesrecorded at each survey (Nrec), curves for higher plants. Nrec figures are on the basis of the assumption of minimum turnover, i.e. species areonly counted as going extinct if they fail to be found in all subsequent surveys. This assumes that temporary absences are artefacts of samplingrather than the true extinction and subsequent re-invasion of the species. (b) Immigration/ extinction curves for spermatophyte species. Rates arecalculated as follows: (i) assuming recorded turnover, i.e. that all extinction and immigration records are real (immigration rate � number ofspecies not recorded at time 1 but recorded at time 2/time 2 � time 1; extinction rate � number of species recorded at time 1 not found at time2/time 2 � time 1); (ii) assuming minimum turnover (immigration rate � number of species not previously recorded, but found at time 2/time2 � time 1; extinction rate � number of species recorded at time 1 not found at or after time 2/time 2 � time 1).

found within the first 50 years of recolonization(1883–1934), and to examine their persistence in thecontemporary flora using first the 1979–1983 datumadopted in their earlier work and second the moreexhaustive 1979–1994 data. The improved surveydata reduced the number of ‘extinct’ species byone-third to 94, of which some 41 species could beregarded as ephemerals that never really estab-lished breeding populations in the first place, i.e.the true number of extinctions could be 50 or fewer.Although the estimates of extinction rate were thusshown to be unreliable, structural features in theextinction data were rather more robust. Lossesrelate to (1) the original abundance of a species asrecorded in the 1883–1934 period (Fig 5.9a), (2) thenumber of islands on which a species occurred(another indicator of abundance) (Fig. 5.9b), and toa lesser degree (3) the primary dispersal mode ofthe species (indicative of the autoecology of thespecies). In short, exactly as we might anticipate, it

is the ephemeral (not proper ‘colonists’) and therare which have failed to persist.

Animal-dispersed species have the lowest overallrate of loss, whilst among sea-dispersed speciesthere is an interesting contrast between core mem-bers of the regional strandline flora, which werefound on all three islands and have shown highpersistence, and those species found initially ononly one island, the majority of which have disap-peared having either never established populationsin the first place, or having been swept away bycoastal erosion (Fig 5.9b). To sum up, the first110 years of the succession have involved hugetransformations in the environment and habitattypes, and in the character of the forest, but even so,few of the species that established themselves in thefirst 50 years have subsequently failed to persist.Those that have failed are a statistically non-ran-dom subset made up mostly of the initially rarespecies.

Page 152: Whittaker. Island Biogeography. 2007

The data quality problems seen for the plantsundoubtedly afflict even the best of the animal datasets for Krakatau, but are less well understood.Here we take them essentially at face value. Thebutterfly data (Fig. 5.10) and land-bird data(Fig. 5.11) show similar trends coincident with for-

est closure, but contrasting trends towards the endof the data series, with butterflies seemingly climb-ing more steeply and birds approaching an asymp-tote. In the first two decades, the poverty of thevegetation arguably presented a limited array ofopportunities for butterfly species, many of which

K R A K ATAU — S U C C E S S I O N, D I S P E R S A L S T R U C T U R E , A N D H I E R A R C H I E S 139

60

(a) (b)

40

20

1883 1903 1923

Year

Cumulative

Immigration

Extinction

Num

ber

of s

peci

es

Nrec

1943 1963 1983 1883 1903 1923

Year

1943 1963 19830

Spec

ies

per

annu

m

2.0

1.0

0

Figure 5.10 Butterfly recolonization data for Rakata Island (Krakatau) 1883–1989 (redrawn from Bush and Whittaker 1991). The peaks inimmigration and extinction appear to correlate with habitat succession as grassland gave way to forest habitat on the islands. (a) Cumulativetotal and number of species recorded at each survey (Nrec), as Fig. 5.10. (b) Immigration and extinction curves, as Fig. 5.8.

60

50

40

30

20

10

0

80(a) (b)

70

60

50

40

30

20

10

0S LR LA WR WA Ferns Wind

Primary dispersal mode

Perc

enta

ge o

f spe

cies

abse

nt in

197

9–94

Num

ber

of s

peci

es

Sea Animal

1 island

Present

Absent

2 island3 island

Abundance category

Figure 5.9 Persisting and failing species from the 1883–1934 flora of Krakatau (Rakata, Sertung, and Panjang islands), as determined bysurvey data for presence on the islands between 1979 and 1994. (a) Presence/absence in relation to abundance categories derived fromdescriptive accounts from earlier botanical work, where S � singleton records; LR � localized and rare; LA � localized but abundant;WR � widespread but rare (low density); WA � widespread and abundant. (b) Losses from the flora in relation to primary dispersal mode andthe number of islands on which each species was recorded in the 1883–1934 period. Wind, sea, and animal refer to spermatophytes only; fernsare all wind-dispersed, in the form of spores. (Source: Whittaker et al. 2000, Figures 2 and 3). Unsurprisingly, initially rare species have a lowerrate of persistence than initially widespread species. Further scrutiny of the raw data allows exclusion of a number of the extinctions documentedin this figure and a lower overall estimate of extinction rate in the flora. Such an adjustment, if made to Fig. 5.8b, would induce a steepernegative trend to the extinction curve than shown there.

Page 153: Whittaker. Island Biogeography. 2007

require the presence of fairly specific food plants.The rate of arrival appears to have peaked duringthe period of most rapid floral and habitat diversi-fication. Extinction also peaked following forestclosure. Bush and Whittaker (1991) attribute four ofthe losses to succession, as open habitats were lost,other losses being possibly due to sampling defi-ciencies and to the inclusion of likely migratoryspecies in the calculations. Thornton et al. (1993)took this view in replotting the data, thus finding alower peak in the extinction rate, although the gen-eral trends in rates were the same.

Figure 5.11 shows the trends in the bird data ascalculated by Thornton et al. (1993). These data aresimilar to the butterfly data but have an additionalsurvey point, for 1951. This survey, by Hoogerwerf(1953), was missed by MacArthur and Wilson(1967) who suggested from the three survey colla-tions for Rakata and Sertung up to 1934 that equi-librium might already have been reached. The datain Fig. 5.11 show that numbers subsequently roseslightly but at a much reduced rate (Thornton et al.1993). There is evidence in these data of succes-sional effects involving different trophic levels. Thedata also indicate a degree of turnover, the precise

figure involved being highly sensitive to assump-tions made as to breeding status of birds (Thorntonet al. 1993). The colonization curve is now fairly flat,but only future survey data can show if the birds ofRakata have achieved a dynamic equilibrium.Rawlinson’s analyses of reptile data also indicateboth an important role for abiotic factors and a lowturnover rate. There have been only two extinctionsof stabilized reptile populations, one due to habitatchanges (canopy closure and coastal erosion) andthe other to catastrophic alteration of habitat byeruptions of Anak Krakatau. No evidence wasfound of an approach to a dynamic equilibrium.

Thus, for Krakatau, the fit of the data (Figs 5.8,5.10, 5.11) to the expectations of the EMIB appearsto be poor, the best relationship being for birds,although even here the fit is imperfect (Fig. 5.11b).MacArthur and Wilson’s (1967) stated position wasthat the precise shapes of the I and E curves couldbe modified within the confines of the EMIB, pro-viding that they remained monotonic, ‘When a newset of curves must be derived for a new situation,the model loses much of its virtue . . . ‘ (p. 64).Departure from monotonicity suggests that theintersects may be unstable or perhaps that there

140 C O M M U N I T Y A S S E M B LY A N D DY N A M I C S

45(a) (b)

35

25

15

5

1883 1903 1923

Year

Cumulative

Immigration

Extinction

Num

ber

of s

peci

es

Nrec

1943 1963 1983 1883 1903 1923

Year

1943 1963 19830

Spec

ies

per

annu

m

0.5

1.0

0

Figure 5.11 ’Resident’ land-bird recolonization data for Krakatau (redrawn from Thornton et al. 1993). (a) Cumulative total and number ofspecies recorded at each survey (Nrec), assuming minimal turnover. (b) Immigration and extinction curves, as species per year for inter-surveyperiods, as Fig. 5.8.

Page 154: Whittaker. Island Biogeography. 2007

may be alternative stable states. These simpleanalyses of the species lists from Krakatau clearlyshow non-monotonic episodes in the early phasesof island successions. This implies that successionis imposing structure that swamps the general sto-chastic processes represented in the EMIB. Much ofthe turnover is successional, and quite a lotinvolves species which may well not have estab-lished breeding populations. If the turnover attrib-utable to the loss and gain of habitats and tosuccession is set to one side, undoubtedly someturnover remains. Precisely how much is real, in thesense of extinction of established populations, isvery hard to quantify. Overall, turnover can bejudged to be heterogeneous rather than homoge-neous sensu Rey (1985).

The degree of organization in the Krakatauassembly process

Is the assembly of the Krakatau system governedby chance, or is it deterministic and predictable inits behaviour? It is difficult to provide definitiveanswers to this loosely phrased question because ofthe deficiencies of the historical data and becausethere are not enough other similar ‘experiments’ tocompare it with. However, some interestinginsights have emerged from studies of the newisland, Anak Krakatau.

The emergence and development of the newisland has modified the conditions for arrival, mor-tality, and survival of plants and animals within theisland group as a whole (Box 5.3; Bush andWhittaker 1991, 1993). It has impinged upon devel-opments on the other islands, not least through itsvolcanic activity. Its own history of environmentaldevelopment has been distinctive, and its colonistscan be assumed to have been drawn predominantlyfrom within the archipelago, rather than from themore distant source pools of the post-1883 period.Indeed, some populations of animals movebetween the islands in foraging. As an experimentin island colonization, it thus cannot be viewed as aproper replicate. Nevertheless, it is interesting tonote that there has been a bias in the early patternsof species assembly on Anak Krakatau towardsthose species which colonized the archipelago in

the early post-1883 phase (Thornton et al. 1992).This applies to several groups of plants andanimals (e.g. birds, reptiles, and bats). The excep-tion to the pattern was provided by the butterflydata set. This was probably due to the dependenceof butterfly species on the availability of particularhabitats and food plants, which differed sufficientlybetween old and new runs of the experiment tobuck the trends.

The recolonization of Anak Krakatau has notbeen an uninterrupted process. Partomihardjo et al.(1992) have analysed the succession of floras fol-lowing, first, the island’s appearance in the centreof the Krakatau caldera in 1930, and, subsequently,following wipe-outs of the vegetation in 1932/3,1939, 1953, and damaging, but not entirely destruc-tive eruptions around 1972 (Thornton and Walsh1992; Whittaker and Bush 1993). Surveys have beencarried out only intermittently, but we have inessence three or four runs of the assembly experi-ment on Anak Krakatau, for each of which a specieslist is available. This rather small-scale naturalexperiment has demonstrated a strong degree ofrepetition, a temporal nestedness, with a core ofconstant species, which make it back each time andare added to (Whittaker et al. 1989; Partomihardjoet al. 1992). The first assemblage was really just aseedling flora, a few members of which were notfound in the second assemblage. Yet, of 32 speciesidentified from these two surveys, 30 recolonizedsubsequently. The eruptions in 1972 are presumedto have severely reduced the ‘third’ flora, yet all but2 of its 43 species were recorded between 1979 and1991.

If the Anak Krakatau flora is broken down intoarbitrary functional guilds, there are basically threesets of species:

1 Strandline species. These are the largest set, repre-senting 58 of 125 spermatophytes found between1989 and 1991. Propagules of these species are sea-dispersed, and are produced in large numberslocally on the other islands. This is a very consistentset. Forty-two of these species have been found onall four Krakatau islands.

2 Pioneers of interior habitats. These species arewind-dispersed ferns and grasses, with a few

K R A K ATAU — S U C C E S S I O N, D I S P E R S A L S T R U C T U R E , A N D H I E R A R C H I E S 141

Page 155: Whittaker. Island Biogeography. 2007

composites. Again they are locally available, arevery dispersive, and produce large numbers ofpropagules. They are adapted to a highly disper-sive life style and are very capable of ‘finding out’such open environments, on which their survivaldepends (they are presumed to be not good com-petitors). As with the strand species, they are inessence a subset of those found on the other islandspost-1883.

3 The third set of species, the second phase colonistsof interior habitats, only managed to become estab-lished in the most recent flora. It will be interestingto establish the extent to which it may repeatfollowing any further wipeout. Comparisons withthe older islands suggest that in relation to thesequence on the older islands this group of speciesis likely to be less predictable in colonizationsequence because of the more variable patterns ofseed production and dispersal (mostly by birds orbats) within the archipelago. For instance, some ofthe contemporarily most abundant tree species onthe other islands (e.g. Arthrophyllum javanicum,Timonius compressicaulis, and Dysoxylum gaudichaudi-anum) were recorded on Anak Krakatau at an earlierstage of successional development than in the post-1883 sequence.

Whittaker and Jones (1994a) have formulatedthese and other observations from Krakatau into aninformal ‘rule table’ in which the dispersal, succes-sional, turnover, and compositional characteristicsof five such functional plant guilds were tentativelyput forward should the opportunity arise to evalu-ate them elsewhere. Within plant successions theremay be both autogenic ‘facilitation’ effects anddiffuse competitive effects between guilds, culmi-nating in later successional communities shadingout earlier ones. Yet much of the structure may beattributable to a form of relay floristics, mediatedthrough dispersal attributes of the plants anddependent, for some guilds, on hierarchical linksbetween plant and animal communities (Bush andWhittaker 1991; Whittaker 1992; Thornton 1996;and compare with Thornton et al. 2001).

Just as in other island systems, as well as therebeing structure as to which species assemble and inwhat sequence, there may be particular sets of

species which fail to colonize. Whittaker and Jones(1994a) highlight those types for which they regardKrakatau as probably undersampling the regionalspecies pool (Fig. 5.6). More formal comparison ofthe Krakatau floras with other sites in the regionhas provided support for these observations(Whittaker et al. 1997). Later successional specieswith poor dispersal adaptations, those which havelarge, winged, wind-scattered propagules, thosedispersed by terrestrial mammals, and large-seeded bat-fruits (lacking diplochory) remaindeficient on Krakatau, whereas highly dispersiveforms such as ferns and orchids have been ‘over-sampled’.

The Krakatau plant and animal recolonizationdata thus suggest a general trend in the degree ofpredictability through time. While the draw ofearly pioneers is fairly predictable as a function inlarge measure of their superior dispersability,precisely which later successional species happento establish breeding populations is much lesspredictable. Thornton (1996) has made a further,interesting point regarding the elements of chanceand determinism. He draws the analogy betweencommunity assembly and the construction of ajigsaw, arguing that the further on in the process aspecies joins the system the less influence it canhave over subsequent events, as it operates withinsuccessively narrow bands set by what has gonebefore. Thus a late-joining species may be unpre-dictable in its identity, yet may be predicted to haverelatively little impact on community trajectories.The parallel with the assembly rules of Diamond isclear. At early stages of the recovery process, anumber of species combinations or communitypathways are possible, but as the system becomesmore complex, and the jigsaw more complete, thenumber of pieces that will fit in to a given gap willdecline, eventually to just one. This form of analogyhas some intuitive value for groups such as birds,but is harder to sustain for plants. Successionalpathways are not as discrete and limited as such apicture implies. The islands are continually subjectto the vagaries of a varied environment, rangingfrom volcanic eruptions, through storms andlandslides, to droughts. In such a disturbed settingcompetitive replacement within natural plant

142 C O M M U N I T Y A S S E M B LY A N D DY N A M I C S

Page 156: Whittaker. Island Biogeography. 2007

communities simply cannot be detected on aone-on-one basis (if it ever can), and returning tothe analogy, the jigsaw is continually beingdisturbed and reformulated in new ways, whileperhaps never quite being completed.

To conclude, the island ecological rates forKrakatau contain a lot of sampling noise, throughwhich clear structural features are evident, reveal-ing island recolonization to be in essence a specialcase of succession. Rates of turnover have, in addi-tion, been affected by the physical environmentaldynamics of the system. Hierarchical featuresacross trophic levels are evident and it has alsobeen argued that, even within a single taxon, not allspecies additions have equivalent effects. The birddata from Anak Krakatau indicate that althoughthe addition of another insectivore or frugivoremay not perturb the existing community structuresignificantly, the addition of the first predatorybirds caused a major community perturbation(Thornton 1996). Similarly, the appearance of thefirst animal-dispersed fruits, and the first birds, andthe first bats, post-1883 must also have constitutedimportant events. A more recent example is thecrossing of a threshold population size of fig treeson Anak Krakatau, in providing for resident popu-lations of fig wasps and of frugivores (Comptonet al. 1994). Once again, we see that different taxareveal a variety of trajectories of their colonizationcurves, and greater and lesser degrees of turnover.

5.5 Concluding observations

This chapter has reviewed a range of forms of struc-ture in the composition and dynamics of islandbiotas. We have paid particular attention to succes-sional processes, as exemplified first within theNew Guinea birds studies of Jared Diamond andcolleagues, and second, through the Krakatau casestudy. Notwithstanding the debates concerning thestatistical tests of such patterns, the evidence fornon-random patterns from a variety of empiricalstudies is overwhelming (Schoener and Schoener1983; Weiher and Keddy 1995). Competition ismerely one of the forces involved, but it does havea role, even if it remains difficult to demonstrate.Progress has been made, as the tools of analysis

of island assembly patterns, incidence functionsand nestedness have been debated, refined andevaluated. In the very nature of the response vari-ables, this body of work encompasses more com-plexity than the previous chapter and in some wayshas been harder to advance. By encompassing morecomplexity, it is correspondingly harder to findgenerality. Yet, theories of island assembly may inthe end be of as much or perhaps more practicalvalue to conservationists (Chapter 10; Worthen1996) than, for example, are attempts to refine geo-metric models of the relationship between coastlineconfiguration and immigration rate. We hope thusto have shown how the study of ecological bio-geography can shed some light on the species num-bers games of the previous chapter. The aim of thenext chapter is to provide a synthesis drawingsome of these ideas together in a re-assessment ofprogress in island ecology.

5.6 Summary

Island biotas are not simply random draws fromregional species pools. Instead, they typicallyexhibit compositional structure: some species,species combinations, or species types, are foundmore frequently, and some less frequently, thanmight be expected by chance. A comprehensiveisland assembly theory was formulated by JaredDiamond, in his studies of birds on islands in theBismarck and other groups around New Guinea.His analyses were founded on several forms of dis-tributional patterns. Incidence functions describethe frequency of a species as a function of islandspecies richness (or sometimes island size). Thosespecies found only on the most species-rich islands,which were also the larger, land-bridge islands,were termed high-S species. Supertramps were, incontrast, found only on small, remote, species-poorislands, including islands recovering from past dis-turbance. Between these two extremes are the ‘lessertramps’, found on varying numbers of the islands.Chequerboard distributions, were also observed,whereby taxonomically and ecologically relatedspecies have mutually exclusive but interdigitatingdistributions across a series of islands. Within par-ticular guilds of birds, Diamond found evidence for

S U M M A RY 143

Page 157: Whittaker. Island Biogeography. 2007

compatibility rules, i.e. that a certain degree of nichedifference is necessary to enable coexistence. Afterthese various effects were considered, it was appar-ent that some combinations of species were morefrequent and others less frequent than expected,suggesting combination rules. In total, seven dis-tinct assembly rules were derived empirically on thebasis primarily of the distributional data. The inter-pretations of these patterns rested on the assump-tion that through competition the componentspecies are selected and co-adjusted in their nichesand abundances, thereby ‘fitting’ with each other toform relatively species-stable communities.

Diamond’s analyses drew pointed criticism fromSimberloff and colleagues, who in a twin-prongedattack questioned both the primacy given to com-petition and the extent to which the distributionalpatterns departed from a random expectation in thefirst place. On the first count, Diamond and Gilpinresponded by emphasizing the extent to whichfactors other than competition (e.g. predation,dispersal, habitat controls, and chance) were in factintegral to the theory. On the second count, theycontested the basis for the null models, demon-strating that the findings depend heavily on thebiological assumptions that particular authorsbuild in to their null models.

The debate has continued and reappeared inother guises. Competition is undoubtedly a force inshaping island biotas but it is extremely difficult todistinguish the precise extent of its effects.Subsequent studies using similar types of distrib-utional ‘tools’ (e.g. incidence functions) havedemonstrated several common features, often find-ing a role for island area, but also involving a rangeof habitat determinants. Some convincing evidence

of a different form of assembly rule invokingcompetitive effects is provided by morphologicaloverdispersion in assemblages of exotic speciesintroduced to oceanic islands. These patterns havebeen reported for several groups of oceanic islandsand for both passeriform and galliform birdgroups.

Nested distributions are where smaller faunas(or floras) constitute subsets of the species found inall richer systems. Nestedness is attributable vari-ously to ‘passive sampling’, nestedness of habitat,differential immigration (frequently using thesurrogate of island isolation), and differentialextinction (area). Statistical tests of the degree ofnestedness have been developed and applied tomany island data sets. Nestedness turns out to be acommon feature of insular biotas, varying inimportance depending on the nature of the islandsand the taxon being considered.

Island recolonization constitutes in essence just aspecial case of ecological succession. This is illus-trated by reference to the recolonization ofKrakatau. In this natural experiment, dispersalattributes, hierarchical relationships across trophiclevels, and habitat changes are each seen as signifi-cant in structuring island species assembly.Turnover from the species lists can be attributedlargely to a combination of succession, habitat loss,in a few cases predation, and to the comings andgoings of ‘ephemerals’, i.e. it is heterogeneous innature. For plants, as might be expected from firstprinciples, it is the rare species that most frequentlyfail. The fit with equilibrium theory appears poor:while some taxa or ‘guilds’ may have reached anasymptote, overall the system has yet to equilibrate,if indeed it is destined ever to do so.

144 C O M M U N I T Y A S S E M B LY A N D DY N A M I C S

Page 158: Whittaker. Island Biogeography. 2007

What appears patchy to a grasshopper may appearuniform to a gnu.

(Wright et al. 1998, p. 19)

In the previous two chapters we have reviewedisland ecological theories from both macroecologi-cal and ecological–biogeographical perspectives,tracing the development of ideas following theappearance of the seminal work of MacArthur andWilson (1963, 1967). We return again to theirequilibrium theory at the outset of this chapter. Theexploration of these ideas by ecologists has pro-duced many insights, sometimes consistent withthe original models, sometimes inconsistent.Departures from predicted behaviours are in manyrespects of greatest interest, as they show how themodels need to be modified. As a researchprogramme, this body of work is characterized by ageneral assumption that the systems can be charac-terized as dynamic and equilibrial: indeed wemight consider this paradigmatic. Yet, as we know,this assumption may not be supported in all cases.Accordingly, we will consider the alternatives inthe present chapter. In so doing we hope to recon-cile the apparently conflicting lessons from 40 yearsof research and to suggest where island ecologicaltheory might be heading.

Several themes run through this chapter. Theyinclude: the need for alternatives to the dynamicequilibrium model, recognizing the possibility ofdynamic but non-equlibrium systems, and oflargely static ecologies, which again may be equi-librial or non-equilibrial; the issue of scale depend-ency; and the need for models to accommodatehierarchical links within ecological systems. Building

in such complexities may lead to more realisticmodels, but we should not readily abandon thesearch for simple, unifying theories and models(Brown 1995), and therefore, we also consider somerecent attempts to develop more satisfactory simpleframeworks for island ecology.

6.1 Limitations of the dynamicequilibrium model of islandbiogeography: a reappraisal

. . . the equilibrium model and its derivatives suffer fromextreme oversimplification by treating islands as func-tional units with no attention to internal habitat diversityand by treating species as functional units with noallowance for genetic or geographical diversity. This is noteven good as a first approximation, because it filters outthe interpretable signal instead of the random noise. Theauthors are in such a hurry to abandon the particulars ofnatural history for universal generalization that they losethe grand theme of natural history, the shaping of organicdiversity by environmental selection . . .

(Sauer 1969, p. 590, on MacArthur and Wilson 1967)

Brown (1981) observed that the EMIB has threecharacteristics which he claimed any successfulgeneral theory of diversity must possess.

● First, it is an equilibrium model, thus historicalfactors, climatic change, successional processes,and the like are acknowledged, but side-stepped.The theory explains rather the ultimate limits, thetheoretical patterns of diversity.

● Secondly, it confronts the problem of diversitydirectly, number of species being the primarycurrency of the model. It also takes account both of

145

CHAPTER 6

Scale and island ecological theory: towards a new synthesis

Page 159: Whittaker. Island Biogeography. 2007

biological processes and some, at least, of the characteristics of the environment.

● Thirdly, it is empirically operational, makingrobust, qualitative predictions that can be tested(although, as we saw in chapter 4, the model turnsout be much more difficult to test than oncethought).

Brown notes that the fact that the EMIB has beenrepeatedly and unequivocally falsified does notdiminish the contribution it has made to advanceour understanding of diversity. In contrast to thejudgement by Sauer (1969, above), he thereforecharacterizes it as a useful beginning:

The model implies that the determination of diversity is avery stochastic process. Species continually colonize andgo extinct, the biota at equilibrium is constantly changingspecies composition, and all of these processes occuressentially at random. MacArthur and Wilson knew thatthis was not so, but ‘ . . . it was a useful simplifyingassumption . . . ‘ (Brown 1981, p. 883).

Perhaps the key virtue of the model as a generalmodel is that it is dynamic. The distinction betweenhistorical and dynamic hypotheses is based onthe repeatability or probability of recurrence of aparticular state or form. Whereas, historical, ortime-bound, knowledge refers to the analysis

of complex states having very small probabilitiesof being repeated (i.e. states of low recoverability),physical, or dynamic, or timeless, knowledgerefers to the analysis of states having a high degreeof probability of being repeated (Schumm 1991).The search for dynamic hypotheses of speciesrichness can thus be likened to the search for thegeneral laws of ecology, from which historicalcontingency supplies the deviations (Brown 1999;Whittaker et al. 2001). Hence, the abiding attrac-tion of the MacArthur and Wilson model and ofthe dynamic, equilibrial framework they put inplay.

Many of the problems with the essentiallystochastic EMIB of MacArthur and Wilson havebeen discussed in the previous two chapters, as wehave focused on the extent to which equilibriummodels apply; our ability to predict island speciesnumbers and identities, and on what forces struc-ture island biotas. Table 6.1 summarizes some of thekey issues. There is no doubting the lasting heuris-tic value of the dynamic equilibrium model, butmore in question is the continuing widespread useof the model, its direct derivatives and assumptionsin predicting species losses in relation to habitatloss and fragmentation: an issue we return to inChapter 10.

146 S C A L E A N D I S L A N D E C O L O G I C A L T H E O RY: TO WA R D S A N E W S Y N T H E S I S

Table 6.1 Some of the limitations identified in MacArthur and Wilson’s equilibrium model of island biogeography (EMIB)

Author Limitations of EMIB

Sauer (1969) Theory ignores autoecology—but species are not interchangeable units (e.g. see Armstrong 1982 on the effects of rabbits introduced onto islands)

Brown and Kodric-Brown (1971) Distance may affect extinction rate (E ) (through the rescue effect), whereas E is treated as afunction only of area in the EMIB

Lynch and Johnson (1974) The data are rarely adequate for testing turnoverHunt and Hunt (1974) Turnover can be confounded by trophic-level effectsSimberloff (1976) Re: mangrove data set, most turnover involves transients, i.e. is pseudoturnoverGilbert (1980) Turnover at equilibrium has not been demonstratedWilliamson (1981) Area may affect immigration rate (I), whereas I is treated as a function only of

isolation in the EMIBWilliamson (1981, 1989a,b) Immigration, extinction, and species pool are each poorly defined. EMIB is imprecise on

reasons for extinction, and most turnover is ecologically trivial (cf. Simberloff 1976)Pregill and Olsen (1981) Ignores historical data and role of environmental changeHaila (1990) EMIB has a narrower domain of applicability than originally thought (may be operational on

the population scale)Bush and Whittaker (1991) Ignores successional effects and pace, and the hierarchical links between taxa

Page 160: Whittaker. Island Biogeography. 2007

There remain those, such as Rosenzweig (1995),who defend the theory and argue that it ‘holds upwell’. He contends that it is inaccessibility thatdefines ‘islandness’ in a biological sense, and that itfollows from this that a biological definition of anisland is preferable to physical definitions such asbeing surrounded by water. His definition is: ‘Anisland is a self-contained region whose speciesoriginate entirely by immigration from outside theregion’ (p. 211). This is a restrictive definitionthat would seem to rule out numerous real islandsthat have high proportions of endemic species.Moreover, the term ‘self-contained region’ is alsoimportant. By it, Rosenzweig means that eachspecies should be a source species, i.e. it has anaverage net rate of reproduction sufficient to main-tain positive population. The alternative conditionis that species are sink species, which are depend-ent on influx from outside to maintain their popu-lations. That is, species which only maintain theirpresence through continual supplementary immi-gration should be discounted, or if they are in abun-dance presumably the study system should not beconsidered insular. This definition, if rigorouslyapplied, would see a high proportion of studysystems excluded from any test of the EMIB. Thisis, in practice, to define islandness by how wellthe EMIB applies to an island. We prefer to state thecase the other way around: the equilibrium modelholds only for a limited subset of islands. Can wegeneralize as to which?

6.2 Scale and the dynamics of islandbiotas

The processes invoked in the EMIB as determininghow richness of an island may change from onepoint in time to the next are immigration, specia-tion, and local extinction (Box 4.1), which of coursehas to be correct. It is also reasonable to supposea general tendency towards equilibrium values,but that this condition will be approximated togreater or lesser degrees for particular islands andarchipelagos as other forces and factors come intoplay. The degree of fit is argued here to be broadlyrelated to scale, and in particular the range of scalein area and isolation of the island system

considered. This line of thinking was developed ina seminal paper by Haila (1990), who argued thatthe key is to clarify what is mean by an ‘island’,relative to the ecological processes at work (note theparallels with Rosenzweig’s thinking, above). Theprocesses themselves form a continuum, with char-acteristic coupled space–time scales, as shown inFig. 1.1. Haila distinguishes (1) the individual scale;(2) the population scale 1—dynamics; (3) thepopulation scale 2—differentiation; and (4) theevolutionary scale. He illustrated this frameworkwith examples taken from northern European arch-ipelagos, although his data provide only indirectevidence of some of the processes at work.

● On the individual scale, a patch of land is anisland if some crucial phase of the life cycle of indi-vidual organisms obligatorily takes places withinits boundaries. Consider those birds which breedon an island but overwinter elsewhere, such thattheir life cycle is influenced by factors external tothe island. Birds of prey, on the other hand, mayinclude several small islands of an archipelagowithin their territory. In such circumstances indi-viduals may be viewed as an integral part of theregional population, such that the insularity of theenvironment on this individual scale has few popu-lation or community-level consequences, otherthan through being inferior or superior places forreproduction.

● When islands are larger and more isolated, theymay support populations that are dynamicallyindependent of those on other land areas. At thispoint the relevance of the EMIB model may beapparent. To be an island on the populationdynamic scale, two criteria are identified: first, thewhole life cycle of the organisms concerned mustbe confined within the island and, secondly, theisland population must be able to demonstrateindependence from the mainland dynamics forseveral generations. For instance, there is noevidence that bird populations on single islandsfrom the Åland archipelago are dynamicallyindependent. Although thse islands are about40 km from the Swedish mainland and 70 km frommainland Finland, and up to some tens of squarekilometres in area, island population dynamics

S C A L E A N D T H E DY N A M I C S O F I S L A N D B I OTA S 147

Page 161: Whittaker. Island Biogeography. 2007

commonly occur in parallel with those of the main-land. Only for a few sedentary bird species on thelargest and most isolated Baltic island is there evi-dence of dynamic independence.

● Should the criteria for independence expressedabove be fulfilled for long periods, the island popu-lations become increasingly independent as geneticunits, and this allows for the second scale of popu-lation processes, differentiation. Exemplification ofthis comes from several studies, two cases being thevarious subspecies of wrens (Troglodytes troglodytes)on different North Atlantic archipelagos (Williamson1981), and the studies of Berry (e.g. 1986) on thefounder effect in small mammals on offshoreislands around Britain. The relevance of the equi-librium theory would seem to be fairly slight insuch circumstances.

● The natural progression in this scheme is to theevolutionary scale, by which Haila (1990) meansprocesses leading to the divergence of species, i.e.taxonomic differentiation to a greater degree thanindicated by the term ‘subspecies’ or ‘variety’ or byslight niche shifts. Empirical support for the propo-sition that remote islands typically reach and main-tain the condition of dynamic equilibrium is patchyat best (Chapter 4).

Of these four scales, the EMIB is restricted mainlyto the population dynamics scale, in that the otherscales do not appear to satisfy the criteria necessaryfor its operation. The scales of Haila’s frameworkgrade one to another, such that particular studiesmay be at the interface, and the applicability of thetheory uncertain. Even within a single taxon, such asbirds, or bats on the Krakatau islands, some speciespopulations may effectively be bound by the confinesof their island, while others may have wider territo-ries inclusive of the islands under study. It is not,then, that the theory holds for all islands within acertain space–time configuration, but that the effectsof the processes it represents may be prominentwithin this conceptual territory, whereas elsewhere(in the remaining subject space) they are subsumedby other dominant processes. It also follows fromthese arguments that it should be considerably easierto develop models based on individual species(e.g. Lomolino 2000a,b) than whole-system models.

Haila’s (1990) scale framework is predominantlya biological one. He makes reference to environ-mental dynamics, but not prominently. It will,however, be apparent that particular forms ofenvironmental change and disturbance will alsohave characteristic scale patterns, which determinetheir relevance to interpretation of the differingscales of island ecological process. In short, it is notsufficient to assume ecological systems to be inequilibrium, this must be demonstrated (Weins1984), and in systems for which abiotic forcingdominates, more complex, non-equilibrium modelsmay be required.

Residency and hierarchical interdependency:further illustrations from Krakatau

One of the features imposed by the equilibriumparadigm has been the restriction of focus tospecies that are resident on islands. Although thishas not always been adhered to, it does result, forexample, in seabirds and (often) migrant birdsbeing excluded from consideration in island eco-logical studies. Such species may be of huge signif-icance to the ecology of an island and may also actto introduce other fauna and flora.

An illustration of the problems concerningnotions of residency comes from the fruit bats ofKrakatau. The largest flying fox species is Pteropusvampyrus. A colony of several hundred individualswas observed intermittently on the islands duringthe 1980s and 1990s. Pteropus have been observed to gather in large roosts on islands but fly to mainland sites in order to feed (and vice versa), andtheir nightly range can be as much as 70 km(Dammerman 1948; Tidemann et al. 1990). Withinmainland areas Pteropus is also known to be highlymobile, exhibiting movements that may representnon-seasonal nomadism in response to variation infood availability, rather than regular seasonalmigration. Krakatau may thus provide just oneroost site of a number used by the same colonywithin a larger geographic area. Their intermittentuse of it may reflect variation in food supply inthe region, human interference at other roostsites, and avoidance of periodic ash falls onKrakatau. These animals and other, smaller fruit

148 S C A L E A N D I S L A N D E C O L O G I C A L T H E O RY: TO WA R D S A N E W S Y N T H E S I S

Page 162: Whittaker. Island Biogeography. 2007

bats, act as important seed vectors for the introduc-tion and/or spread of a particular subset ofKrakatau plant species (Shilton 1999; Shiltonet al. 1999).

The large fruit-pigeon, Ducula bicolor, is anotherimportant species of frugivore, which for islandecological purposes is considered as a Krakatauresident (Thornton et al. 1993), yet which exhibits aroaming behavioural pattern, in which large flocksmove between offshore islands in the region(Dammerman 1948). The two species of KrakatauDucula (the other is D. aena) are believed to haveplayed crucial roles in introducing animal-dispersed (i.e. zoochorous) plants—particularlythose with larger seeds—to Krakatau (Whittakerand Jones 1994b). The particular identities andtiming of the trees introduced by such means mayhave important ramifications for the species ofanimals which may subsequently be supportedwithin the islands.

All these species of frugivores have been highlysignificant to the structure inherent in the assemblyprocess for Krakatau, the interior forests being pre-dominantly composed of bird- or bat-dispersedplants. Arguably, the behaviours of these speciesexhibit characteristics of connectedness to otherlocations which are more akin to metapopulationmodels (Chapter 10) than to the notions of resi-dency demanded by the EMIB. In short, for someanimal taxa and some guilds, there are potentiallycomplex hierarchical links between their turnoverpatterns and floral development and turnover.Other forms of hierarchical links may also befound, such as those linking predator–prey groupsin an area. Although such observations are ofexplanatory value, it is, as we saw with the debatesabout island assembly rules, a considerable chal-lenge to incorporate such processes into generalpredictive models.

Immigration and extinction curves are theorizedin the EMIB to have smooth forms, the formerfalling and the latter rising to a point of intersec-tion: the dynamic equilibrium. The patterns ofarrival and disappearance from the Krakatau listsdo not correspond with this expectation (Chapter 5),in part because of hierarchical, successionalfeatures evident in the data, which are simply

absent from the model. The big early kinks in therates relate to the key switch from open to closedhabitats. It is conceivable, however, that this switchmarks the end of significant autogenically derived(i.e. biotically driven) switches in trends. In whichcase, once beyond the early phases of succession,the EMIB may become more realistic, i.e. at somestage, an island must fill up with species and someform of equilibrium be established. However, giventhe complexity of succcessional processes in tropi-cal lowland forests, Bush and Whittaker (1993)question whether it is reasonable to suppose thatpopulation processes can bring the system to equi-librium before the next major disturbance movesthe system away from that condition again.

Forest succession is a slow process: the lifespanof individual canopy trees can exceed 300 years,and even early successional species may live fordecades. If a true biotic equilibrium in the florawere to be established beyond doubt, it must beafter a period of several generations of bioticallymediated interactions, i.e. hundreds of years.However, such conditions are unlikely to be reachedin very small patches because of varying forms ofenvironmental variability and disturbance. Eventssuch as hurricanes, volcanic eruptions, flooding,landslides, and other high-magnitude phenomenaoccur within continental and island landscapes,and typically have periodicities of less than severalhundred years (Chapter 2).

For this argument to hold force for animalgroups/taxa, there must be a strong pattern ofdependency of animal groups on plants for habitatand food resources, such that their patterns of colo-nization, carrying capacity and turnover are tied tothe dynamics of the plant communities. In turn,animals can have key roles in shaping plant succes-sion, through roles as seed dispersers and pollina-tors (Bush and Whittaker, 1991, Elmqvist et al.1994). If the degree of dependency is weak, it may bethat merely the recovery of forest biomass and pro-ductivity, which is likely to be more swiftly accom-plished (Whittaker et al. 1998), may be sufficient toallow equilibration of particular animal groups. Inpractice, some 120 years on from the initiation of theKrakatau experiment, it remains difficult to providea definitive answer to this question.

S C A L E A N D T H E DY N A M I C S O F I S L A N D B I OTA S 149

Page 163: Whittaker. Island Biogeography. 2007

6.3 Forms of equilibria and non-equilibria

We are not dealing with a well-stabilized situation of longstanding. This may account for some of our difficulties infitting existing situations into a theory which concernsitself only with final equilibrium.

(Preston 1962, p. 429)

We have argued that there is an important dis-tinction between equilibrium and non-equilibriumconditions. There is an equally important distinc-tion between dynamic and static models of islandecology. These notions provide for a variety of com-binations, as shown in Fig. 6.1, in which we high-light four extremes: dynamic equilibrium, staticequilibrium, dynamic non-equilibrium, and staticnon-equilibrium. Some example systems thatappear to match the different conditions are pro-vided in Table 6.2.

The dynamic equilibrium hypothesis corre-sponds to the EMIB, the properties of which havealready been considered at length. The core EMIBmodel is essentially stochastic (i.e. occupying theextreme bottom left corner of Fig. 6.1), butMacArthur and Wilson (1967) recognized in theirbook that some structuring of turnover commonlyoccurred, hence their fuller theory could beassigned a position a little to the right within the

figure. By either view, their ideas are strongly asso-ciated with the dynamic equilibrium corner.Figure 6.1 recognizes a continuum from purely sto-chastic (homogeneous) turnover, through increas-ingly heterogeneous turnover to the ideas ofhabitat determinism with minimal turnover. Thisend of the axis describes the static equilibriumhypothesis, which may be characterized crudelyas where the key controls of species presence/absence appear to be habitat controls (as Martinet al. 1995) and where species turnover measuredon a timescale of generations is insignificant. Thisis not to deny compositional change over time, butthat in the absence of human interference theturnover is unmeasurable, giving the appearanceof stasis.

It is possible to reconcile ideas of habitat deter-minism with the occurrence of systematic variationin species richness related to area and isolation,providing that habitat structure itself is signifi-cantly influenced by the area and isolation of anisland (Martin et al. 1995). Hence, once again, thepattern of turnover is central to distinguishingbetween alternative hypotheses. The emphasis onhabitat controls has often been associated withDavid Lack, who based his analyses on studies ofisland birds (e.g. Lack 1969, 1976). He argued thatthe failure of birds to establish on islands comes

150 S C A L E A N D I S L A N D E C O L O G I C A L T H E O RY: TO WA R D S A N E W S Y N T H E S I S

Environmentaldynamics fasterthan biotic

Biotic dynamics faster than environmental,which are low amplitude

Biotic dynamicsresistant to immigrationand to environmentaldynamics

Where doesyour island fit?

Biotic dynamics laggreatly behindenvironmentaldynamics, but arenot wholly resistant

StaticDynamic

Equilibrial

Non-equilibrial

Figure 6.1 A representation of the conceptual extremesof island species turnover. The dynamic equilibriumcondition corresponds to MacArthur and Wilson’s (1963,1967) theory; the static equilibrium equates to Lack’s (e.g.1969) ideas on island turnover of birds; the static non-equilibrium to Brown’s (1971) work on non-equilibriummountain tops; and the dynamic non-equilibrium to Bushand Whittaker’s (1991, 1993) interpretations of Krakatauplant and butterfly data. Considering a single taxon,different positions in this diagram may correspond todifferent islands or archipelagos, and different taxa in thesame island group may also correspond to differentpositions.

Page 164: Whittaker. Island Biogeography. 2007

from a failure to find the appropriate niche spacerather than from a failure to disperse to the island.He therefore played down the relevance ofturnover. Others have supported this viewpointinsofar as saying that ‘most genuine turnover ofbirds and mammals seems attributable to humaneffects’ (Abbott 1983) or that turnover, where itdoes occur, mostly involves subsets of fugitivespecies and is thus ecologically trivial (Williamson1981, 1988, 1989a). A similar conclusion wasreached in a study of newly isolated Canadianwoodlots. Once successional effects were removedfrom the analysis, turnover within the vascularflora, if occurring at all, was at a very slow pace(Weaver and Kellman 1981).

A similar distinction, between the dynamic andthe static, can be made within non-equilibriumideas. First, we may recognize disturbed islands:those in which equilibrium is reached only rarelybecause environmental dynamics outpace theresponse times of the biota (cf. the disturbancehypothesis of McGuinness 1984). The biotas maytrack changing equilibrium points through time,but always remain a step or two out of phase. This

idea has been given various labels (e.g. Heaney1986; Bush and Whittaker 1993; Whittaker 1995),but within the scheme developed here it is identi-fied as the dynamic non-equilibrium hypothesis.We may distinguish this notion from that intro-duced by Brown (1971) for relictual assemblagesdominated by extinction. In Brown’s system,although isolates may be losing species on a mil-lennial timescale, and are thus in a non-equilibriumcondition, on ecological timescales the system ofisolates appears effectively static, i.e. they fit thestatic non-equilibrium condition.

We regard this diagrammatic model as being ofheuristic value although, of itself, it lacks predic-tive capacity. Moreover, it is often going to bedifficult to assign a study system to a particularcondition such that ready agreement willbe reached amongst ecologists that the dataunequivocally support a particular interpretation(compare e.g. Bush and Whittaker 1991, 1993, andThornton et al. 1993). See Box 6.1 for an illustra-tive discussion.

The difference between the dynamic non-equilibrium hypothesis and the dynamic equilibrium

F O R M S O F E Q U I L I B R I A A N D N O N - E Q U I L I B R I A 151

Table 6.2 Exemplification of classification of studies of island richness and turnover as per Fig. 6.1, showing that different taxa from within asingle island group (Krakatau) support different models, and that in different contexts, data from the same taxa (terrestrial vertebrates, birds,invertebrates, plants) have been interpreted as supporting different models. (After Whittaker 2004.)

Dynamic, non-equilibrial Static non-equilibrial

1. Krakatau plants (Whittaker et al. 1989; Bush and 7. Krakatau reptiles—several species introduced by people, only two species Whittaker 1991) have been lost, both related to habitat losses (data in Rawlinson et al. 1992)2. Krakatau butterflies (Bush and Whittaker 1991) 8. Great Basin mountain tops, North America, small mammals (Brown 1971)3. Bahamas, plants on small islands (Morrison 2002b) 9. Lesser Antilles—small land birds, examined over an evolutionary time-frame

(Ricklefs and Bermingham 2001)

Dynamic, equilibrial Static, equilibrial

4. Krakatau birds are a reasonable fit, although showing 10. Krakatau terrestrial mammals—no recorded extinctions to dateclear successional structure in assembly and turnover (Thornton 1996)(Bush and Whittaker 1991; Thornton et al., 1993) 11. Bahamas, ants on small islands (Morrison 2002a)5. Mangrove islets of the Florida Keys, arthropods 12. Oceanic island birds (Lack 1969, 1976 [but see Ricklefs and (Simberloff and Wilson, 1970) Bermingham 2001]; Walter 1998)6. British Isles, birds on small islands 13. Canadian woodlots plant data (if successional effects are (Manne et al., 1998) removed from the analysis) (Weaver and Kellman 1981)

Page 165: Whittaker. Island Biogeography. 2007

152 S C A L E A N D I S L A N D E C O L O G I C A L T H E O RY: TO WA R D S A N E W S Y N T H E S I S

Box 6.1 Implications of anthropogenic changes in richness of oceanic islands for island theory

In Chapter 5 we considered how introductions of exotic species to oceanic islands providedinadvertent experiments in island assemblystructure. Here we consider the implications ofanthropogenic impacts on island species number.Sax et al. (2002) took the approach of estimatinghistorical richness prior to human-inducedextinctions and colonization events, andcomparing these values with richness subsequentto these processes. They did so for vascular plantsand for land birds on 11 and 21 oceanicislands/archipelagos, respectively. As shown in thefigure, the number of naturalized exotic plantspecies greatly exceeds the number of extinctions(panel A), whereas for birds, exotic additionsapproximately match the number of extinctionsfrom the avifaunas (panel C). Hence, incomparing current with historic richnessestimates, humans have led to an approximatedoubling in size of the floras (panel B), but littlenet change for birds (panel D). In terms of actualnumbers, the number of extinctions variesbetween 0 and 71 for plants, and 1 and 64 forbirds: in both cases, the Hawaiian islandsproviding the highest value.

These results necessarily involve a number ofcaveats concerning data quality, lumping ofislands within archipelagos, and so forth. But,they are intriguing if viewed through the lens ofthe equilibrium paradigm. Taking a simplistic lineof reasoning, we might argue as follows. Thefloras of the oceanic islands have approximatelydoubled in size, as naturalizations have vastlyexceeded extinctions. Hence, the islands mustoriginally have been below their intrinsic capacityfor richness, i.e. they were non-equilibrial. Thebirds, in contrast, have seen a balance ofextinctions and establishments, and must thushave been in equilibrium historically, and havenow re-established their equilibria.

How might we explain away the apparentmismatch between the plant data and theequilibrium condition? Many of the islands in thestudy once possessed extensive areas of forestthat have been cleared for agriculture andsettlements. The removal of forest cover and itsreplacement by predominantly low-stature

vegetation has provided ideal conditions for theestablishment of numerous species of(predominantly) herbaceous weeds, which have amuch smaller modular size than the pre-existingtree flora, i.e. many more species can be fitted in.Moreover, the bombardment of the islands byintroduced plants effectively means that theimmigration rate has been radically altered, and in many cases, continues to be. For example,there is no sign of a slowing down of the pace ofintroductions and establishment of exotic speciesin the Galápagos (S. Henderson, personalcommunication). It is thus evident that we havesimultaneously altered both immigration rate andcarrying capacity. Hence, the enrichment of thefloras may indicate not previous non-equilibrium,but that we are currently en route to some newanthropogenically determined enhancedequilibrium point.

Unfortunately, if we are to allow this argumentto stand for plants then the same argumentsshould apply to birds, in which case their numbersshould also have gone up. We now have todefend the lack of change in bird richness. Birdsdiffer in having a much higher proportion oflosses of native species (from a variety of causes,e.g. Blackburn et al. 2004) than the case forplants, and it turns out that the lost species haveon average been replaced by exotics. As discussedin Chapter 5, not all introductions of birds tooceanic islands are successful, so perhaps for birdswe see an equilibrium condition being re-established despite the hike upwards inimmigration events. This in turn would imply thaton average, the intrinsic carrying capacity of theisland is determined by the resource base itrepresents rather than by the particular vegetation cover.

Have we squared the circle? Perhaps, but if so,it is by special pleading, and we have ended upwith an argument that indicates that a substantialalteration to the immigration rate and tovegetation cover has had great impact on plantnumbers but not on bird richness. Moreover, there is a great deal more to understanding the factors governing anthropogenic changes to island species numbers, whether taking a

Page 166: Whittaker. Island Biogeography. 2007

F O R M S O F E Q U I L I B R I A A N D N O N - E Q U I L I B R I A 153

macroecological modelling approach concerning factors like introduction effort andpredation (e.g. Blackburn et al. 2004, Casseyet al. 2005b) or considering the detailed context of turnover for particular islands (Duncan and Blackburn 2002), such as the impact of disease organisms and their carriers on Hawaii (Pratt 2005).

We hope to have illustrated how difficult it is to bring such arguments about the equilibriumassumption to a clear conclusion. We are left,nonetheless, with some intriguing observations.

First, plant species numbers on oceanic islandsappear to be capable of significant increaseswithout necessarily involving high turnover(extinction) of native species. Second, theproportional relationships seen in the figuresuggest that the slope (z value) of the ISARs(island species–area relationships) have been little changed despite the enrichment of the floras and the turnover involved in the avifaunas.Third, it seems the role of isolation is lesssignificant for determining island land bird species numbers than predicted by the EMIB.

Log (extinct native species + 1) Log (extinct native species + 1)

Log (historic richness of species + 1) Log (historic richness of species + 1)

Lo

g (

exta

nt

na

tura

lize

d s

pe

cie

s +

1)

Lo

g (

curr

en

t ri

chn

ess

of

spe

cie

s +

1)

Lo

g (

curr

en

t ri

chn

ess

of

spe

cie

s +

1)

Lo

g (

exta

nt

na

tura

lize

d s

pe

cie

s +

1)

0

0 0

1

12

2

0

1

2

3

4

1

2

3

4

1 2 0 1 2

0 1 2

3 4

1 2 3 4

A - plants

B - plants

C - birds

D - birds

Species richness of vascular plants and land birds on oceanic islands. Triangles represent islands inhabited before European contact; circlesrepresent those that were not. (From Sax et al. 2002, Fig. 1.)

Page 167: Whittaker. Island Biogeography. 2007

154 S C A L E A N D I S L A N D E C O L O G I C A L T H E O RY: TO WA R D S A N E W S Y N T H E S I S

hypothesis (or EMIB) is of degree, relating to thecausation of turnover and the biotic response times(Whittaker 1995). In fact, MacArthur and Wilson(1967) acknowledged the implications of environ-mental disturbances for particular island systems.For instance, they recognized the following phe-nomena: the occurrence of successional turnover;the influence of storms on immigration rates; andthat much extinction, especially that resulting fromstorms, drought, and invasion of new competitors,is accompanied by a severe reduction in the carry-ing capacity of the environment. Nonetheless, theytended to downplay the general significance of dis-turbance, arguing that only equilibrium models arelikely to lead to new knowledge concerning thedynamics of immigration and extinction. But, asothers have noted, where turnover is the productprincipally of abiotic forcing, turnover patternsmay be poorly predicted by equilibrium models(Caswell 1978; Heaney 1986; Whittaker 1995;Morrison 1997; Shepherd and Brantley 2005).

Figure 6.2 is Bush and Whittaker’s (1993) attemptto generalize the Krakatau island ecological trendsand to set out alternative trajectories, allowing fortaxa of differing ecological roles and responsetimes. It encapsulates the temporally bumpy ratesassociated with successional processes and pro-vides for two possible scenarios whereby equilib-rium might be reached. The data for rate changesindicate that equilibrium might be approached by adeclining immigration rate, combined with twoalternative extinction trends. In the first, extinctionrate is low and turnover rather heterogeneous(interactive). In the second, it is essentiallysqueezed out of the system altogether (non-interactive). Given the problems of collecting ade-quate survey data, and the high degree ofpseudoturnover which is a feature of empiricaldata from such large and complex systems, the twoconditions may not be readily distinguishable inpractice. Close examination of the Krakatau plantdata demonstrated that a high proportion ofapparent turnover involved species best regardedas ephemerals, which had not really colonized inthe first place (Whittaker et al. 2000). The third formof projection recognizes that the islands experienceenvironmental change, which may be dramatic and

destructive. The longer the period before equilib-rium is attained, the greater the likelihood that anevent of intermediate or high magnitude will castthe system away from equilibrium and into the per-petual or dynamic non-equilibrium condition ofFig. 6.1. In this projection, the biota chases perpetu-ally moving environmental ‘goalposts’.

The relevance of the three projections in Fig. 6.2to particular Krakatau taxa depends on features oftheir ecology such as generation times and disper-sal attributes, and their position within the biologicaland successional hierarchy (cf. Schoener 1986).Within the conceptual space of Fig. 6.1, different taxaor guilds (or the same taxa in different island con-texts) may thus occupy different positions: the rep-tiles arguably towards the static non-equilibrium;the plants of Anak Krakatau towards the dynamicnon-equilibrium; the coastal flora at the archipelagolevel towards the static equilibrial position; and thebirds of Rakata, at least at times, relatively close toa dynamic equilibrium but with less homogeneouspatterns of turnover than indicated for a strictEMIB position. The two diagrams are thus comple-mentary, in showing the conceptual space (Fig. 6.1)and temporal features characteristic of differentpoints within it for one exemplar system (Fig. 6.2).

Dynamic non-equilibrium, or multistate modelshave become prominent in several subfields of eco-logy in recent decades (e.g. savannah ecology), butthus far relatively few non-equilibrium modelsappear to have been developed in island ecology(see e.g. Caswell 1978; Weins 1984; Williamson1988; Villa et al. 1992; Russell et al. 1995; Whittaker1995). Incorporating environmental disturbanceinto island ecological models and modelling pres-ents significant but not insurmountable difficulties.An example is provided by Villa et al. (1992). Theyconstructed a model island consisting of a habitatmap of cells containing individuals, a list of thespecies involved, and their life history characteris-tics. The model involved colonization of the island,and varying degrees of perturbing events thatcaused the mortality of individuals. The model washighly simplified, but nonetheless allowed the test-ing of ideas. One intuitively appealing result wasthat the global attainment of equilibrium within thesimulations was strongly dependent on the rate of

Page 168: Whittaker. Island Biogeography. 2007

increase of the species involved. This demonstratedthat equilibrium for slow-growing species (i.e.those with a low intrinsic rate of populationincrease) should be tested over longer periods, apoint equally clear from Schoener’s (1983) reviewof empirical data. Another interesting feature wasthat in the presence of disturbance, turnover wasmostly accounted for by ephemeral populations(arguably trivial ecologically). This held even inequilibrial simulations, a finding again in concor-dance with empirical evidence, in the form of birddata from Skokholm (Williamson 1989a,b).

Figure 6.1 is thus offered as a restatement ofisland ecological theory, in which the ‘narrow’ viewof EMIB is represented in the broader subject space

against other ‘narrow’ view alternatives. The fourextreme positions on turnover are thus dynamicequilibrium, static equilibrium, dynamic non-equi-librium, and static non-equilibrium. In practice,they may be difficult to distinguish with the dataavailable. Moreover, the response times, dispersalabilities, and trophic status of different taxa maymean that comparative studies of different taxafrom the same set of islands adopt differing posi-tions in this framework (Table 6.2; Bush andWhittaker 1993). Once again, following Haila’sargument, the characteristics of the targettaxon/group in relation to the spatial scale of thesystem will largely dictate the applicability of theselabels to case study systems. Similarly, Crowell

F O R M S O F E Q U I L I B R I A A N D N O N - E Q U I L I B R I A 155

HFLM Disturbance

'HurricaneJoe'

LFHM

Sterilizingeruptions

Interactiveequilibrium

Non-interactiveequilibrium

Time

Empirical data for Krakatau, heavily generalized. Includes successional features ('waves')

Projected trends to non-interactive equilibrium

Projected trend to interactive equilibrium, due to declining I rather tha rising E. N.B. Inpractice this and the non-interactive form may not be separable due to pseudoturnover

Role of major environmental disturbances (e.g. a hypothetical 'Hurricane Joe') setting backsuccession and altering I and E projections

High frequency, low magnitude events

Low frequency, high magnitude events

E

HFLM

LFHM

Rat

e

I

Figure 6.2 A model of island immigration and extinction incorporating disturbance. The figure provides a highly generalized representation ofdata for the recolonization of Krakatau and three hypothetical projections. (From Bush and Whittaker 1993.)

Page 169: Whittaker. Island Biogeography. 2007

(1986), in discussing the mammalian fauna of a setof land-bridge islands, aired the notion that it maybe possible to recognize a continuum, from isolatedland-bridge islands having non-equilibrial faunas,to those of intermediate isolation with both relictspecies and those in equilibrium, to islands so near,and/or so small, that all species are in equilibrium.

The options sketched out in Fig. 6.1 are also sim-ilar to the distinctions drawn about 30 years ago byCaswell (1978) (see also Weins 1984; Schoener 1986;Case and Cody 1987). The scale of Caswell’s appli-cation is rather different from those generallyunder discussion here, but the parallel is apparentnonetheless. He set out to develop models ofpredator (or disturbance)-mediated coexistence.He distinguished between open systems (involvingmigration between ‘cells’) and closed systems (nomigration), and between equilibrium and non-equilibrium systems, concluding that it would notdo to suggest that communities as a whole areentirely equilibrium or entirely nonequilibriumsystems. ‘Perhaps a community consists of a core ofdominant species, which interact strongly enoughamong themselves to arrive at equilibrium, surrounded by a larger set of nonequilibriumspecies playing out their roles against the backdropof the equilibrium species’ (Caswell 1978, pp. 149–50).

6.4 Temporal variation in islandcarrying capacities

The dynamic equilibrium paradigm anticipatesthat island species number will in general stabilizeand show little temporal fluctuation. Here we focuson some of the mechanisms that may lead to fluc-tuations in richness through time, relating mostlyto the idea that an island’s carrying capacities maychange, but also noting that both island area anddegree of isolation may also change over time(Table 6.3).

The most obvious cause of change in isolation isalteration in relative sea level, but effective isolationmay be influenced by more subtle environmentalchanges. Many accounts of species turnover rateshave treated ocean waters as a constant barrier to

dispersal. Scott’s (1994) observations of an invasionof San Clemente island, 80 km from the Californianmainland, led him to question this assumption. Inthe late summer of 1984, 26 black-shouldered kites(Elanus caeuleus) took up temporary residence onthe island for a period of several months, althoughpreviously only one or two kites had been recordedduring 19 autumn/winter cycles of bird observa-tion. Scott suggested that the 1984 irruption mayhave been a consequence of an unusual pattern ofCatalina eddies—which are seasonal cyclonicwinds—possibly linked to an El Niño event. Eachpulse of kites arriving on the island coincided withthe first or second day of a Catalina eddy. It appearsthat the eddy system is like a door to San Clementeisland that irregularly opens and closes. The sameEl Niño influenced carrying capacity for Darwin’sfinches in the Galápagos (Chapter 9). From suchsimple empirical observations, it might be reason-able to postulate that fluctuations in the weathermight produce variations in rates of immigration,extinction, and turnover.

The prevalence and implications of intensedisturbance events

Abbott and Black (1980), in their study of vascularplant species on 76 aeolianite limestone islets,attributed slight changes in the turnover patterns toan increase in rainfall and to the passage of acyclone. In discussing other small island studies,they note that ‘ . . . as most cays occur in tropicalregions, they are subject to frequent cyclonic orhurricane disturbances which often cause wavestemporarily to inundate cays and wash vegetationaway’ (p. 405) (see also Sauer 1969). Similar find-ings were reported by Buckley (1981, 1982) in hisanalyses of plants on sand cays and shingle islandson the northern Great Barrier Reef.

Periodic intense disturbance is a natural featureof many islands across the globe (Whittaker 1995),and has increasingly been afforced by humaninfluences. In illustration, hurricanes are frequent,devastating events throughout the region 10–20�

north and south of the equator (which covers a lotof islands) (e.g. Nunn 1994). For the Caribbean,

156 S C A L E A N D I S L A N D E C O L O G I C A L T H E O RY: TO WA R D S A N E W S Y N T H E S I S

Page 170: Whittaker. Island Biogeography. 2007

between 1871 and 1964 an average of 4.6 hurricanesper year was recorded, the island of Puerto Ricohaving a return period of just 21 years (Walker et al.1991a; Chapter 2). Hurricanes are high-energyweather systems, involving wind speeds in excessof 120 km/h, affecting paths tens of kilometreswide. They can have huge ecological impacts onisland ecosystems (e.g. Walker et al. 1991b; Elmqvistet al. 1994).

During post-hurricane succession in theCaribbean, it has been found that the animalcommunity may continue to change long after theinitial ‘greening-up’ of the vegetation, throughinvasions and local extinctions as the vegetationvaries through successional time (Waide 1991).Initially, nectarivorous and frugivorous birds suffermore after hurricanes than do populations of insec-tivorous or omnivorous species, a consequence oftheir differing degrees of dependence on re-establishment of normal physiognomic behaviourin the vegetation.

Turnover patterns in response to disturbances ofthis and other kinds can be anticipated on theoreti-

cal grounds to be dependent upon taxon choice (e.g.Diamond 1975a, p. 369). Schoener (1983) establishedempirically that percentage species turnoverdeclines from ‘lower’ to ‘higher’ organisms, asfollows (approximate figures): 1000%/year inprotozoa, 100%/year in sessile marine organisms,100–10%/year in terrestrial arthropods, and10–1%/year in terrestrial vertebrates and vascularplants. Moreover, across this range of organisms,turnover declines approximately linearly with gen-eration time. Given which, and restricting consider-ation only to islands subject to such environmentaldynamics, the attainment of a dynamic (bioticallybalanced) equilibrium in certain K-selected (long-lived, only moderately dispersive) taxa may bepostulated to be comparatively rare.

Variation in species number in the short andmedium term

It might be argued that islands devastated byevents such as volcanoes and hurricanes areextremes: Krakatau is just a special case. However,

T E M P O R A L VA R I AT I O N I N I S L A N D C A R RY I N G C A PAC I T I E S 157

Table 6.3 How might the capacity for richness of an island vary through time? Left and right boxes provide hypothetical examples of howincrease/decrease might arise from factors under the same general heading (Modified from Whittaker 2004).

Factors leading to an increase Factors leading to a decrease

Primary succession, increasing biomass, system complexity and Late successional stages if passed through simultaneouslyniche space, in part driven by species colonization, in part enabling across an isolate might see competitive exclusion of suites offurther colonization earlier successional species (e.g. Wilson and Willis (1975),

data from Barro Colorado Island, Panama)

Arrival of ‘keystone species’ (e.g. fig trees on Krakatau Arrival of a ‘superbeast’ (e.g., first vertebrate predator),(Whittaker and Jones 1994a,b; Thornton 1996) ) may over-predate naive prey species

Humans may increase habitat diversity, introduce new species, Humans may clear habitat, introduce new pest/predator manipulate ecosystems to raise productivity, etc. or highly competitive species, degrade habitats, etc.

Area/habitat gain, e.g., through coastal deposition, uplift, Area/habitat loss, e.g., through coastal erosion, subsidence,sea-level fall sea-level increase

Moderate disturbances might open up niche space, e.g., for Major disturbance (e.g. hurricane or volcano) may wipe non-forest species on an otherwise forested island out much standing biomass, massively reducing populations

Increasing climatic favourability, increasing biological activity, Climatic deterioration, lower NPPincreased NPP

Evolution, on remote islands, a similar effect to primary succession, In situ co-evolution is unlikely to generate reductions in richness,primarily operating through hierarchical links between organisms but as oceanic islands age they subside and erode (Chapter 2),at different trophic levels such that in parallel with the processes of speciation, carrying

capacity will first peak and then decline through time (Chapter 9)

Page 171: Whittaker. Island Biogeography. 2007

the Indian Ocean tsunami of 26 December 2004, theeruption of Montserrat, in the 1990s, and the recentincrease in hurricane activity in the Caribbeansuggest otherwise. In addition, less dramatic envi-ronmental variation may also perpetuate non-equilibrium conditions. For example, Lynch andJohnson (1974), who may have coined the term‘successional turnover’ in its island ecologicalsense, argued that where turnover is successional innature, such as produced after a fire, then theturnover is determined not primarily byarea–distance effects, but by the timing, extent, andnature of changes occurring in the habitats. Theyregard use of an equilibrium model in suchsituations as inappropriate. They also argue thatinterpretation of faunal change is especiallydifficult if both equilibrium and non-equilibrium(e.g. successional) turnover are involved.

Extreme climatic events, such as freezes, havebeen recognized as important determinants ofisland avifaunas. An illustration is provided byPaine’s (1985) observations of the extinction of thewinter wren (Troglodytes troglodytes) from Tatoosh, a5–6 ha island, 0.7 km from the north-western tip ofthe Olympic Peninsula (Washington, USA). Theloss of the species followed an extremely coldperiod in December 1978, and it took 6 years to re-establish a breeding population, which was pre-sumed to be because of the difficulty of a relativelysedentary species invading across a water gap.Abbott and Grant (1976) found evidence of non-equilibrium in land-bird faunas on islands aroundAustralia and New Zealand, a result that did notappear to be attributable to sampling error or tohumans. They suggested in explanation thatislands at high latitudes (such as these) are gener-ally subject to irregular climatic fluctuations andmay not therefore have fixed faunal equilibria (cf. Russell et al. 1995). Abbott and Grant (1976)argued that the extent to which climate fluctuatesabout a long-term average value determines theextent to which species number does the same. ‘Inthis sense, species number ‘chases’ and perhapsnever reaches a periodically moving equilibriumvalue, hindered or helped by stochastic processes’(Abbott and Grant 1976, p. 525).

Another study that appears to demonstrate thatequilibrium values change in response to modest,non-catastrophic changes in island environmentson timescales of years and decades is that of Russellet al. (1995). They found that a non-equilibriummodel provided improved predictions of observedturnover of birds for 13 small islands off the coast ofBritain and Ireland. They suggested that turnovercould be viewed as operating on three scales: first,year to year ‘floaters’ (trivial turnover); secondly,on a timescale varying between 10 and 60 years, anintrinsic component equivalent to that envisaged inthe EMIB was observable; thirdly, most islands showa change in numbers over time due to so-calledextrinsic factors, such as habitat alteration.

Long term non-equilibrium systems

Extending the time scale, the glacial/interglacialcycles of the Quaternary have involved high-magnitude effects, the boundaries of biomes beingshifted by hundreds of kilometres, sea levels fallingand rising by scores of metres. These changes haveresulted in at least two types of relictual patternstudied by island biogeographers. The first is thatof land-bridge islands, those formerly connected tolarger land masses (their mainlands), which gainedat least some of their biota by overland dispersalbefore becoming islands (e.g. Crowell 1986). Thesecond well-studied context is where habitat‘islands’ have become isolated in a similar fashionwithin continents (e.g. Brown 1971). In both con-texts, data often support a non-equilibrium inter-pretation.

The EMIB would have it that slopes ofspecies–area curves should be steeper (higher zvalues) for distant (oceanic) island faunas than fornearer (often land-bridge) islands in the same archi-pelago. This is due to the combined effects ofextinction and low colonization rates on veryisolated islands, and because the impact of both isgreater on small islands (Fig. 6.3a). However, asLawlor (1986) noted in a review of data forterrestrial mammals on islands, large remoteislands tend not to attain the species richnesspredicted for them on the basis of their area. This is

158 S C A L E A N D I S L A N D E C O L O G I C A L T H E O RY: TO WA R D S A N E W S Y N T H E S I S

Page 172: Whittaker. Island Biogeography. 2007

because immigration is very slow to distantarchipelagos, and is little influenced by island sizein such remote contexts, thus producing relativelyflat intraarchipelago ISARS. Land-bridge islands, incontrast, may have steeper species–area curvesbecause of the predominance of extinction, as overlong time-periods, members of the assemblagestranded by island formation suffer occasionalattrition, supplemented by little immigration(Fig. 6.3b, Millien-Parra and Jaeger 1999).

In such cases, the fauna at the point of isolationhas been termed supersaturated, i.e. it containsmore species than at true equilibrium; thereafter, thedominance of extinction over immigration results(in theory) in relaxation of species number, declin-ing towards eventual, hypothetical equilibrium.Classic illustrations of this effect have also comefrom non-volant mammals occupying montanehabitats in the Great Basin of western NorthAmerica. These studies have been selected for theirillustrative value, and by no means all non-equilibr-ial systems are products of Quaternary climatechange. The distinction between true oceanicislands and land-bridge islands is, as Lawlor (1986)has shown, a crucial one for understandingspecies–area relations of non-volant mammals

of remote islands, and analyses which fail to distinguish the two groups are thus likely to bemisleading.

Implications for endemics?

If environmental fluctuations can be important tosome small to moderate-sized near-continentalislands, why do we not hear more of the adverseeffects of natural disturbance in remote oceanicislands? Might they not be anticipated to be worstaffected, being unable to replenish species stocksswiftly after such events as cyclones? Perhaps theeffects of variability are largely ‘built in’ (evolution-arily) to their ecologies?

Miskelly (1990) reports widespread reproductivefailure and delayed breeding by Snares Island snipe(Coenocorypha aucklandica) and black tit (Petroicamacrocephala dannefaerdi), two endemic land birds,on the Snares Islands, in the New Zealand sub-Antarctic. Snipe also suffered unusually high adultmortality. He attributed these phenomena to thepronounced El Niño–Southern Oscillation (ENSO)event of 1982–83, which produced heavy rainfalland significantly lowered temperatures on theSnares Islands, thereby reducing the invertebrate

T E M P O R A L VA R I AT I O N I N I S L A N D C A R RY I N G C A PAC I T I E S 159

Num

ber

of s

peci

es

Area

Num

ber

of s

peci

es

Area

(a) (b)

Landbridge islands

Landbridge islandsMainland

Oceanic islands

Oceanic islands

Mainland

Figure 6.3 Equilibrium and non-equilibrium representations of island species–area relationships for a hypothetical archipelago, as summarizedby Lawlor (1986). (a) By an equilibrium argument, the effects of recurrent colonization and extinction produce steepening of the species–areacurves with increasing distance, as extinction (arrows) has greater relative impact on remoter islands, for which immigration is increasinglydifficult, (b) In the non-equilibrium model, oceanic islands may be undersaturated, as immigration (open arrows) is too slow to ‘fill’ the islands(extinction is relatively unimportant), whereas land-bridge islands undergo ‘relaxation’, i.e. their richness patterns are extinction dominated (solidarrows).

Page 173: Whittaker. Island Biogeography. 2007

food supply. Neither bird species invested muchenergy in reproduction during the ENSO eventcompared with subsequent years. The poor breed-ing seasons in this study contrast dramatically withthe greatly enhanced reproductive success of landbirds on the Galápagos, which benefited fromimproved plant growth and seed productionenabled by the high rainfall (Gibbs and Grant 1987).Very heavy rainfall stimulated the production ofplants and therefore the caterpillars feeding onthem, important to finches in the breeding season.A significant increase took place in the populationsize of Darwin’s ground finch on Daphne Majorisland. Gibbs and Grant (1987) concluded that rareevents can have a major influence on keypopulation processes, including processes of signif-icance to evolutionary changes, in long-lived birdsliving in temporally varying environments (see alsoGrant 1986). These events did not involve speciesgains and losses, but warranted attention fromecologists because of the importance of the endemicspecies involved.

Whittaker (1995) argued that island endemics areunlikely to be lost very often in response to evenlarge-scale disturbances, such as hurricanes orvolcanic eruptions. First, this is because species andecosystems have evolved within the context of thedisturbance regimes. Secondly, and linked to this,they tend to occur on larger oceanic islands, whichmay well provide refuges unaffected by the pertur-bation somewhere within their land area. A recentillustration comes in the impact of the eruption ofChances peak in the 1990s on the endemicMontserrat oriole. The eruptions led to the loss ofthe species from a substantial part of the range, butsampling indicated that some 4000 individualspersisted in the less badly affected Centre Hills area(Arendt et al. 1999). By this line of thinking, smalloceanic islands, such as cays and atolls, shouldsupport relatively impoverished systems with fewendemics restricted to them.

Support for this general line of reasoning isprovided by a study of the Laysan finch (Telespizacantans), one of the endangered Hawaiian honey-creepers. It occurs on the small coral island ofLaysan in the Hawaiian archipelago, although inprehistorical times it also occurred on Oahu and

Molokai, and a small population has recently beenintroduced to Pearl and Hermes reef. Laysan isonly 397 ha in area and of that only 47% is vege-tated. Morin (1992) found that weather conditionsplayed a crucial role in regulating reproductive suc-cess. In 1986, a severe storm caused almost totalmortality of eggs and chicks regardless of clutchsize. Later that year, the number of fledglings pernest increased as clutch size increased. The speciesis omnivorous and eats some part of almost everyplant on the island, as well as invertebrates, carrion,and eggs. It also has life-history parameters condi-tioned by long experience of the fluctuating condi-tions on the island. Reproduction can be spreadover a long breeding season, with the potential formultiple broods, and the species has a long repro-ductive lifespan. Notwithstanding this flexibility ofresponse, it was concluded that stochastic weatherevents and predation are the two key factors cur-rently limiting the Laysan finch population, in con-trast to most of the other honeycreepers of Hawaii,whose populations are threatened principally andacutely by introduced disease carried by exoticmosquitoes (Pratt 2005).

The South Pacific islands of Samoa have two fly-ing fox (fruit bat) species, Pteropus samoensis(endemic to Samoa and Fiji) and Pteropus tonganus.Pierson et al. (1996) discuss the response of each ofthem to two severe cyclonic storms, Ofa, whichstruck in 1990, and Val, in 1991. Flying foxes areimportant elements of the ecosystem, possibly war-ranting keystone status, because of their roles asseed dispersers and plant pollinators. Activity pat-terns and foraging behaviour were disrupted forboth species. Before the hurricanes P. tonganus wasmuch the commoner, with a population on thesmall island of Tutuila of about 12 000, contrastingwith fewer than 700 P. samoensis individuals, and itwas thought that the more restricted endemicmight be the more vulnerable to storm effects.However, the endemic species was found to havesurvival strategies not observed in the more common and widely distributed P. tonganus, andthe latter experienced the more severe declines, ofthe order of 66–99% per colony. The numbers of P. samoensis seemed to be relatively unaffected. Itssurvival strategies included making greater use of

160 S C A L E A N D I S L A N D E C O L O G I C A L T H E O RY: TO WA R D S A N E W S Y N T H E S I S

Page 174: Whittaker. Island Biogeography. 2007

leaves in the aftermath of the storm and alsofeeding on the fleshy bracts of a storm-resistantnative liane. Pteropus tonganus, on the other hand,seems to have adopted the unfortunate behaviouralresponse of entering villages to feed on fallen fruit,where they were killed by domestic animals orpeople.

Pierson et al. (1996) make the point that islandspecies of the tropical cyclone belt have evolvedwith recurrent cyclones. It appears in this case thatthe species of narrowest geographical range actu-ally has evolved more effective responses tocyclone events. Given that flying foxes can live for20 years or so, individuals may experience severalsevere storms in a lifetime. Such events in them-selves are unlikely to lead to extinction of thespecies, provided that other threats, such as habitatloss and hunting, are constrained. Where the twoare combined, however, extinction becomes morelikely (Christian et al. 1996). Although winddamage is typically patchy, it seems reasonable tosuggest that areas of high topographic complexity,e.g. volcanic cones and deep valleys, are the mostlikely areas to retain patches with some foliage, andso should be given priority in reserve design.

6.5 Future directions

. . . in practice it is next to impossible to find a set ofislands without environmental change or evidence of thesevere impact of man’s activities

(Case and Cody 1987, p. 408)

We have learnt a lot from studying island ecologyunder the ruling dynamic, equilibrial paradigm.We can continue to do so, whether focusing onimproving macroecological models for islandspecies number by incorporating energetics or bynew approaches to nestedness and assembly struc-ture. In this chapter we have reviewed evidence fordepartures from the dynamic equilibrium condi-tion. Of course, some degree of variance can beaccepted within the equilibrium paradigm, as sys-tems of rapid response rates and small amplitudeenvironmental fluctuations may swing rapidlyback in to balance, such that most of the time anisland could indeed be close to equilibrium.Application of dynamic, equilibrial models does

not demand a precise fixed value (Simberloff 1976).However, depending on the relative timescales ofchanges in the island environment and response bythe biota, we contend that islands may variouslybe: always in equilibrium; in equilibrium most ofthe time but occasionally disturbed from equilib-rium; or never in equilibrium, but always laggingbehind changes in environment.

The emphasis on habitat determinism by Lackand others has been opposed by many supportersof dynamic island biogeography. They see it asreturning the subject to a narrative form. The pres-ence of many endemics on very isolated islandsclearly demonstrates that even among a taxon ofgenerally good dispersers, e.g. birds, there comes apoint where isolation is sufficient to render immi-gration and colonization a rare and essentiallyprobabilistic process. However, there is no need todraw such a stark distinction between dynamic andstatic views of islands. Studies reviewed in thischapter suggest that is possible to build bridgesbetween these differing positions and that veryoften the determinants of species number and com-position lie somewhere between the essentiallystochastic, dynamic equilibrium and more ‘static’behaviour, and require a more complex explanationof community organization on islands thanprovided for by the EMIB.

Conceivably, some island systems (large andremote) might respond to a severe El Niño event,for example, by exhibiting considerable flux in pop-ulation sizes, biomass, and productivity, withoutmuch impact on species turnover rates. In othercases (small and near-shore islands), a severe coldsnap in the winter might have little effect on bio-mass and productivity but might greatly impact onpopulations of small birds, leading to measurablealteration of species turnover rates. On the whole,there appears to have been insufficient attentionpaid to systems departing from the assumptions ofthe dynamic equilibrium paradigm. Similarly, asexemplified by Krakatau, where disturbances aresevere, or environmental change is very consider-able, there may be long-term ecological changeswhich unfold over the course of decades orcenturies. In such cases, a successional model ofisland ecology can provide a more appropriate

F U T U R E D I R E C T I O N S 161

Page 175: Whittaker. Island Biogeography. 2007

162 S C A L E A N D I S L A N D E C O L O G I C A L T H E O RY: TO WA R D S A N E W S Y N T H E S I S

Box 6.2 Lomolino’s (2000) tripartite model of island biogeography

In a special issue on island biogeographypublished in the journal Global Ecology andBiogeography in 2000, Mark Lomolino (2000a)argued that given the acknowledged limitationsof the equilibrium theory, a new island model isrequired, and that it should: (1) include feedbacksamong system components (as Bush andWhittaker 1991), (2) acknowledge the role ofevolution (as Heaney 2000), and (3) be species-based (Lomolino 2000b). He stated‘Species vary in many ways that affectimmigration, extinction and speciation, and manybiogeographic patterns derive from, not despite,differences among species. For example, the formof the species-isolation and species–arearelationships may reflect very general patterns ofvariation in immigration abilities and resourcerequirements among species.’ (Lomolino 2000a, p. 3).

Lomolino’s graphical model (see figure)maintains the simplicity of MacArthur and Wilson’s(1967) classic island biogeography scheme, butpulls out a third dimension, explicitly recognizingthe role played by speciation in promoting speciesrichness, as a function of the island’s area andisolation. In many respects this graphic merelysketches out ideas contained within MacArthurand Wilson’s (1967) monograph. As we saw, thefirst statement of their model explicitly includesspeciation, and they went on to discuss howparticular taxa typically show the best patterns ofevolutionary radiation towards the periphery oftheir spread across an ocean’s islands (they termed this the ‘radiation zone’).

The graphic provides a useful summary of someof the more important patterns and processes wehave discussed in the island ecology section, whilealso pointing to the main process we have

Immigration

Islandcharacteristics

Isolation

Biogeographicprocess

Spec

iati

on

Extinction

Area

0

a

b

0

c

d

Lomolino’s (2000a) tripartite model focuses on the relative importance of the three fundamental processes of biogeography: evolutionarychange, extinction, and dispersal. He notes that: (1) immigration rates should increase not only with proximity to a source region but alsowith the vagility of the target species; (2) extinction rates should decrease with island area but increase with the target species’ resourcerequirements (which may in turn be broadly related to body size, Bierdermann 2003); finally, (3) speciation rates should be more importantwhere extinction and immigration rates are lowest, and therefore will be greater on the largest and most isolated islands, but will decreasewith species vagility and resource requirements. The shading represents the relative levels of richness and the relative resistance to andresilience of insular biotas following disturbance. The biotic characteristics of islands in the four corners (a, b, c, and d) are set out in thetable. (From Lomolino (2000a), Fig. 1.)

Page 176: Whittaker. Island Biogeography. 2007

F U T U R E D I R E C T I O N S 163

Island characteristics relating to the positions indicated in the figure

Island feature (a) Large, near (b) Large, far (c) Small, near (d) Small, far

Immigration High Low High LowSpeciation Low High None LowExtinction Low Low High HighSpecies richness High High Moderate to low LowTurnover Low Low High LowIsland type Continental Continental fragments or Small continental Depauperate oceanic

islands large oceanic islands islands islandsExample Newfoundland, Madagascar, Hawaii Skokholm (Wales) Easter Island,

Tasmania Tristan da Cunha

hitherto neglected: evolution. Moreover, althoughLomolino’s model recognizes the patterns ofdiversity and turnover, it doesn’t assumeequilibrium.

So, is the way ahead to take a species-basedapproach? Lomolino’s argument that manybiogeographic patterns derive from, not despite,differences among species echoes many of thepoints discussed throughout our examination ofassembly rules, incidence functions, and theimportance of species-level effects in thedynamics of island biotas. A species-basedapproach is in some respects more flexible, can bemore relevant to the needs of conservationists inparticular circumstances (e.g. modelling theincidence or potential incidence of an endangeredspecies), and opens up a new set of hypotheses

for testing. However, we should also recall thatspecies responses and impacts can vary betweensystems. For example, Watson et al. (2005)demonstrated variation in species incidencefunctions of woodland birds in the Canberra area(Australia), which appeared attributable toproperties of the matrix in which the woodlandsystems were embedded (see Box 5.2). Similarly,many species considered benign and non-invasivewithin their native ranges, upon being introducedinto an exotic island, turn into radicaltransformers of ecosystems (Chapter 10). Thissuggests that their pattern of incidence in onearea may not accurately predict their incidenceelsewhere. Hence, there are likely to be limits tothe predictive power of species-based models, justas there are with system-level models.

representation of turnover and compositionalpatterns than do simpler stochastic models basedonly on parameters such as area and isolation.

These arguments amount to a case for more com-plex models of island ecology. Building in greatercomplexity is fine under two conditions: (1) you aresure of your starting assumptions, i.e. the paradigmyou are working under, or (2) you are attempting todevelop better understanding of detailed case studysystems or areas. However, we should not abandonthe arguably more important scientific task ofsearching out the simple explanation where it willsuffice, and of finding new models of general appli-cability. Box 6.2 reviews one possible direction.

Consideration of scale has been another keytheme of the last two chapters, as we have devel-oped the argument that although some processesmay be universal, their impact is detectable at par-ticular scales of analysis, while at other scalesother processes or confounding variables dominatethe signal. The increasing availability and sharingof data should allow the scale sensitivity of islandecological models to be more thoroughly investi-gated, allowing us to determine the applicability ofparticular theories within the space/time/scalecontinuum.

Throughout the island ecology section of thisbook we have to a remarkable degree disregarded

Page 177: Whittaker. Island Biogeography. 2007

the role of humans, largely reflecting the approachof the research literature in the field. Yet, as clearfrom particular parts of the discussion, human soci-eties have had profound impacts on island ecolo-gies, nowhere more so than on some of the moreremote oceanic islands. We are only beginning tounderstand how this may have impacted on islanddiversities. This caution serves to reinforce the takehome message of this section that it is not enoughto assume that a system is in equilibrium, any morethan to assume that it is not. This should be amatter of empirical investigation, particularly ifother theories are to be constructed on the basis ofequilibrial assumptions (cf. Weins 1984). We needto remember this when we turn to the applicationof island theories to conservation problems in thefinal section of the book.

6.6 Summary

Although the equilibrium model of island biogeog-raphy (EMIB) has served to advance understandingof island ecology, the work reviewed in the previ-ous two chapters demonstrates that the dynamicequilibrium framework is no longer sufficient toencompass the field. We argue in the present chap-ter that the EMIB has a narrower frame of referencethan original envisioned, working for only a subsetof island species data sets. This subset is largelydetermined by the scale parameters of the targetsystem relative to the space usage and dispersalpowers of the target taxon.

Haila distinguishes four coupled space-timescales, which he terms: (1) the individual scale, (2)the population dynamics scale, (3) the populationdifferentiation scale, and (4) the evolutionary scale.Systems that correspond to the population dynam-ics scale are most likely to show dynamics in tunewith the EMIB. We illustrate how even within asingle taxonomic group, such as birds or mammals,different species can in practice have radicallydifferent spatial scales of interaction with an islandsystem. Temporally, the dynamics of different com-ponents of an ecosystem also respond on varyingwavelengths. For instance, whereas bird communi-ties may respond fairly rapidly to changing island

carrying capacity, or disturbance events, the succes-sional dynamics inherent in forested ecosystemsimply that vegetation systems may take decadesto hundreds of years to establish a meaningfulequilibrium condition, always assuming that nofurther disturbance occurs. The importance of tak-ing account of hierarchical relationships acrosstrophic levels is also stressed, such that for instance,the equilibrium point for birds might change overtime in response to longer term dynamics in thevegetation.

To accommodate these complexities, we presenta framework in which the dynamic equilibrium(EMIB) model is placed alongside dynamic non-equilibrium, static-equilibrium, and static non-equilibrium concepts, and we discuss theapplicability of these ideas to case study systems.The lack of long-term high resolution data serieshampers our ability to distinguish between thesecompeting hypotheses, but we go on to evaluateevidence for how and why longer term variation inisland carrying capacities may occur. Natural dis-turbance phenomena, such as tsunami, hurricanes,volcanic eruptions, and even marked fluctuationsin weather conditions linked to ENSO events, mayall impact on island ecologies, with varying poten-tial significance to island species numbers and com-position.

Increasing the complexity of island ecologicalmodels to incorporate such factors will improvethe realism of the models but at the cost of gener-ality. An alternative approach is to developspecies-based models, which may allow us toavoid some of the limitations of the EMIB, but ithas yet to be established that species-based modelshave predictive power across the range of aspecies, or that they will lead to improved modelsfor whole assemblages. Given the importance ofunderstanding the emergent ecological behaviourof assemblages of species, we suggest that furtherefforts be devoted to exploring the scale sensitivityof island ecological models, the degree to whichsystems demonstrate equilibrium dynamics, andhow hierarchical links across trophic levels influ-ence emergent properties, such as species richnessand turnover.

164 S C A L E A N D I S L A N D E C O L O G I C A L T H E O RY: TO WA R D S A N E W S Y N T H E S I S

Page 178: Whittaker. Island Biogeography. 2007

PART I I I

Island Evolution

The Macaronesian clade of Echium (Boraginaceae) consists of 28 endemicspecies (two on Madeira, 23 on the Canaries, and three on the Cape Verdeislands), of which 26 exhibit woodiness. They derive from a single herbaceouscontinental ancestor. The photograph shows Echium wildpretii ssp.trichosiphon, endemic to the summit scrub of La Palma (Canary Islands)(with kind permission of Ángel Palomares).

Page 179: Whittaker. Island Biogeography. 2007

This page intentionally left blank

Page 180: Whittaker. Island Biogeography. 2007

Island forms of the same species [of birds] show everygradation from extremely small differences in averagesize to differences as great as those which separate someof the species. The differences tend to be greater thegreater the degree of isolation. Some of them are adaptiveand others seem non-adaptive.

(Lack 1947a, p. 162)

In the island ecology section we explored patternsand processes while treating species as standardand unchanging units of analysis. This is a usefulsimplifying assumption, but, as we know, speciesare not fixed, nor are they homogenous entities. Inthis section of the book we consider the specialinsights into evolutionary change that islands haveprovided.

We start from the point at which a new colonistarrives on an island, and then consider the micro-evolutionary processes that go to work on suchpopulations and which result in a fascinating arrayof emergent island effects or syndromes. Thechanges with which we are concerned in this chap-ter are expressed in genotypic and phenotypic char-acters, sometimes exemplified by island endemicspecies, but frequently evidenced at subspecificlevel, i.e. they do not necessarily involve speciationand the recognition of distinct island and mainlandspecies. The processes include a mix of essentiallychance effects, resulting in the divergence of islandlineages from their mainland source pools, and ofdirectional evolutionary forces that appear to oper-ate in a consistent fashion across many remoteislands and taxa. These micro-evolutionaryprocesses and patterns are the building blocksinvolved in the phenomena of phylogenesis andthe emergent macro-scale patterns of island evolu-tion described in the following two chapters.

7.1 Founder effects, genetic drift, andbottlenecks

Islands form in two fundamentally different ways:either by arising out of the sea, at which point theyare empty of land organisms, or by once contiguousareas of land becoming separated from one anotherby incursions, as for example when sea levels rise atthe termination of ice ages (Chapter 2). Thus classicoceanic islands start empty, whilst land-bridgeislands have a full complement of species whenthey first become proper islands. However, someislands have such a complex or poorly known envi-ronmental history that this simple dichotomy canbe difficult to apply, or to agree upon (e.g. seeMcGlone et al. 2001; McGlone 2005; McDowall 2005;Morrone 2005). Here we set these issues to one side,mostly taking examples from true oceanic islands,thus allowing us to examine colonization eventsassociated with long-distance dispersal to islandsthat start (and remain for long periods) species-poor for their area, and which develop from abiased subset of the mainland species pool.

The founder principle (Mayr 1954) is that typi-cally a species immigrating to a remote island willestablish by means of a tiny founding population.This contains only a subset of the genetic variationin the source population and it subsequentlyreceives no further infusion. This immediately pro-vides a bias in the island population, on whichother evolutionary processes then operate. Geneticvariability can be increased after establishment bymutation and re-sorting. One important form of re-sorting is genetic drift, the chance alteration ofallele frequencies from one generation to the next,which may be particularly important under sus-tained conditions of low population size. In

167

CHAPTER 7

Arrival and change

Page 181: Whittaker. Island Biogeography. 2007

addition to such non-selective drift there may alsobe distinctive selective features of the novel islandenvironment. The outcome of these phenomenacould, in theory, be a rapid shift to a new, co-adapted combination of alleles, hence contributingto reproductive isolation from the original parentpopulation.

Although it is possible to demonstrate the occur-rence of genetic drift, as Berry et al. (1992) havedone in their study of a small population of housemice on Faray in the Orkneys, the contribution offounding effects and drift to island speciationremains the subject of keen debate (Barton 1989;Clarke and Grant 1996). It is particularly problem-atic that founding events (i.e. colonizations) havebeen theorized to produce a variety of rather differ-ent founding effects, and that the latter are difficultto isolate empirically from other microevolutionaryprocesses. In fact, there are at least three distin-guishable models of how founder effects may leadto speciation:

● Mayr’s peripatric speciation● Carson’s founder-flush speciation● Templeton’s genetic transilience.

These theories place greater degrees of stress,respectively, on:

● the impact of the general loss of heterozygositycombined with random drift● the role of variations in selection pressure (partic-ularly its relaxation immediately after colonization)● changes in a few major loci within the genome(see discussions in Templeton 1981; Barton andCharlesworth 1984; Clarke and Grant 1996).

The founding event represents a form of bottle-neck, a term implying a sharp reduction in popula-tion size. Bottlenecks may occur at other points inthe lifespan of a lineage, e.g. following catastrophicdisturbance of a habitat, or in the context of speciespushed to the brink of extinction by human action.Studies of the Hawaiian Drosophila indicate thatfounder events have occurred and at three levels:in the colonization of the archipelago, of its con-stituent islands, and of patches of forest habitat(kipukas) within islands (Carson 1983, 1992; Carsonet al. 1990). Significantly, this work indicates that

bottlenecks produce a marked loss of heterozygosityonly if they are sustained or are repeated at shortintervals. In contrast, if a bottleneck lasts only onegeneration and recovery to a large population sizeis rapid, then increased genetic variance for quanti-tative traits can occur.

According to theories of founder-effect specia-tion, the narrowing of the gene pool during a bot-tleneck may serve to break up the genetic basis ofsome of the older adaptive complexes, liberatingadditive genetic variance in the new population(Carson 1992). Novel phenotypes might subse-quently result from natural selection operating onthe recombinational variability thus provided inthe generations immediately following the bottle-neck. According to a model proposed by Kaneshiro(e.g. 1989, 1995), an important element in this sce-nario is differential sexual selection. He found thata minority of male drosophilids perform the major-ity of matings, and that females vary greatly in theirdiscrimination, and acceptance, of mates. In nor-mal, large populations the most likely matings arebetween males that have a high mating ability (thestuds), and females that are less discriminating inmate choice. This provides differential selection foropposing ends of the mating distribution and a sta-bilizing force maintaining a balanced polymor-phism. In a bottleneck situation, however, therewould be even stronger selection for reduced dis-crimination in females, and if this was maintainedfor several generations, it could result in a destabi-lization of the previous co-adapted genotypes.

In such bottleneck situations, with reduction inmate discrimination by females, the likelihood of afemale accepting males of a related speciesincreases, thus allowing hybridization to occur. Theterm given to the incorporation in this way of genesof one species into the gene pool of another via aninterspecific hybrid is introgressive hybridization.With the development of molecular tools,hybridization between animal species (especiallyinsects) has been discovered to occur naturallymuch more often than had previously been sus-pected (Grant and Grant 1994; Hanley et al. 1994;Kaneshiro 1995; Clarke and Grant 1996). (This pres-ents a small but perhaps too often overlooked chal-lenge to the assumption that lineage development

168 A R R I VA L A N D C H A N G E

Page 182: Whittaker. Island Biogeography. 2007

occurs solely through branching (Herben et al.2005).) Thus, although bottlenecks can serve toreduce the genetic base of a new colony, in someinstances they may serve to provide enhancedgenetic variability within a population, the verystuff on which natural selection may then go towork!

Laboratory studies of Hawaiian Drosophila haveprovided a number of important insights intofounder effects, but recently these phenomena havebeen studied in an increasing range of island taxa(e.g. Clarke and Grant 1996). For example,Abdelkrim et al. (2005) describe how the use ofcomplementary genetic techniques of analysisallows the detection of varying severities of found-ing events in initial colonization and subsequentinterisland dispersal events for ship rat (Rattusrattus) in the Guadeloupe archipelago. AnotherHawaiian example is provided by endemic mem-bers of the plant genus Silene. Populations appearto be much more polymorphic on the older islandof Maui than on the younger island of Hawaii (i.e.the main or Big Island), suggesting that geneticvariation had been lost in colonization of the latter(Westerbergh and Saura 1994). The process ofrestricted gene flow followed by genetic drift alsoseems to be repeated as younger volcanic terrain iscolonized within the island of Hawaii.

Although some regard founder effects as a keypart of the most likely mode of speciation in mostHawaiian Drosophilidae (e.g. Kaneshiro 1989,1995), others dispute their significance. Forinstance, Barton (1989) observes that enzyme het-erozygosity within Hawaiian drosophilids is simi-lar to that of continental species, and querieswhether, in practice, founder effects cause muchloss in genetic variation. Barton (1989) regardsother elements in the island laboratory as being ofgreater significance than founder effects, and hementions specifically what he terms faunal drift,the random sampling of species reaching an‘empty’ habitat, and thus providing a novel(disharmonic) biotic environment in which selec-tion drives divergence into a variety of new niches(see also Vincek et al. 1997).

Sarre et al. (1990) examined morphological andgenetic variation among seven island and three

mainland populations of the sleepy lizard,Trachydosaurus rugosus, in South Australia. Theirwork supports the significance of random geneticdrift and inbreeding in small isolates. All but two oftheir islands were isolated from the mainlandbetween 6000 and 8500 years ago. Contrary to someother recent studies of vertebrates on SouthAustralian islands, their allozyme electrophoresisanalysis showed that heterozygosity levels did notvary significantly among the populations, i.e. therehas been little erosion of heterozygosity within theinsular populations. Nonetheless, the island popu-lations exhibited reduced allelic diversity, with raremainland alleles tending not to be present on theislands. This may have been due in part to bottle-necks occurring at some point or points since isola-tion of the populations. There was also a greaterdegree of genetic divergence shown between islandpopulations than between mainland populations,and this they attributed to genetic drift on theislands.

Allozymes are different forms of an enzymespecified by allelic genes, i.e. by genes that occupya particular locus. Allozyme electrophoresis hasbecome a conventional method of measuring het-erozygosity levels. It is thus of interest that twomeasures of developmental stability varied signifi-cantly among the island populations in a mannerdistinct from the results of the allozyme study(Sarre and Dearn 1991). The morphological meas-ures used were: fluctuating asymmetry—thedegree of random difference between left and rightsides in bilateral structures (see Box 10.1); and per-centage gross abnormalities, involving featuressuch as missing ear openings, missing toes, clubfeet, and deformed head and labial scales. Three ofthe island populations stood out as having dis-tinctly higher levels of these forms of developmen-tal instability. This was interpreted as being due tothe expression of deleterious genes (in turn due toinbreeding depression) or the genetic disruption ofco-adapted genotypes. One important conclusiondrawn from their study is that reduced levels ofheterozygosity per se are unlikely to be the primecause of the observed developmental instability. Inshort, the allozyme technique was failing to reflectthe genetic basis for the deleterious morphological

F O U N D E R E F F E C T S, G E N E T I C D R I F T, A N D B OT T L E N E C K S 169

Page 183: Whittaker. Island Biogeography. 2007

features. The relative contribution of environmentaland genetic factors to developmental stability in thesleepy lizard cannot at this stage be determined.

In contrast to the emphasis given in this study todrift, Thorpe and Malhotra (1996) stress the impor-tance of natural selection for current ecological con-ditions in their studies of Canarian and LesserAntillean lizards. It may be significant that Thorpeand Malhotra’s islands, unlike Sarre’s, are large andcontain a considerable range of habitats, i.e. the rel-ative importance of stochastic processes versusdirectional selection may turn out to be stronglydependent on the environmental context.

Although opinions differ as to the overall signif-icance of founding effects for the rapidity of evolu-tionary change and ultimately speciation onislands, it has increasingly become evident thatfounding effects can be detected by several differ-ent techniques, and for a range of plant and animaltaxa (e.g. Clarke and Grant 1996; Drotz 2003; Hilleet al. 2003; King et al. 2003). Moreover, in severalwell-specified case studies, it is possible to see anested sequence of events, from the initial coloniza-tion of an archipelago, through interisland move-ments, down even to intraisland colonizationevents (Westerbergh and Saura 1994; Abdelkrimet al. 2005).

Implications of repeated founding events

What are the implications of such scenarios? First,island lineages are likely to contain less geneticdiversity than the mainland source population.Second, because there is a general trend towardsloss of dispersability once established on a remoteisland (discussed below), and because of the oftensignificant interisland distances, the repetition ofsuch founding events is likely to lead to furtherdifferentiation between populations within anarchipelago, especially if there is significant envi-ronmental variation amongst the islands to add aselective component to lineage development.Available data tend to confirm these expectations(e.g. DeJoode and Wendel 1992; Frankham 1997;Hille et al. 2003). DeJoode and Wendel (1992) com-piled data in the form of allozyme variability forapproximately 60 Pacific island endemic plant taxa

(a mix of species and varieties), finding that theyexhibit roughly half the variability previouslyreported for continental plant species, and twothirds of that shown by continental endemics ofrestricted range. Recent analyses of populations ofsilvereyes (Zosterops lateralis) that colonized islandsin the southwest Pacific from mainland Australiaafter AD 1830 allow an unusually well-specifiedquantification of founder effects. The analyses indi-cated that single founder events had little impacton genetic diversity, partly because founding popu-lations have apparently been quite large (24–200 orso), but that repeated island hopping was accom-panied by a gradual decline in allelic diversity(Clegg et al. 2002; Grant 2002).

In general, differences in genetic variabilityamong island and mainland populations have beenfound to be larger: (1) the smaller the size of thefounder event, (2) the larger the differences inpopulation sizes, (3) the smaller the immigrationand dispersion rates, (4) the smaller the islandsize, and (5) the greater the distance from the conti-nent (Frankham 1997). Indeed, analyses of a fewtree species colonizing the Krakatau islands, whichare isolated from mainland Java and Sumatra byonly about 40 km, indicate no loss of geneticvariation in the island populations, and suggestthat for some species the island and mainlandpopulations can be considered panmictic, i.e. asif involving random mating within a singleextensive population (Parrish 2002). Moreover, sim-ilar results were found for two species of Krakataufig wasps (Zavodna et al. 2005); despite the smallsize of these insects, such a distance may bescarcely a barrier to colonization and gene flow.As Grant (2002, p. 7819) commented about thesilvereye study

. . . Perhaps it is more parsimonious to invoke single-immigration events for each island, but I think this islikely to be wrong. If it is wrong, it carries an importantimplication: except for the most isolated islands, repeatedimmigration may obscure or obliterate any founder-effectchanges that take place following the initial colonization.

Reviewing allozyme variation in 69 Canarianendemic plant species, Francisco-Ortega et al. (2000)report an average species-level genetic diversity at

170 A R R I VA L A N D C H A N G E

Page 184: Whittaker. Island Biogeography. 2007

allozyme loci of 0.186, which is more than twice ashigh as the mean reported for Pacific islandendemics (0.064) by DeJoode and Wendel (1992).They suggest that this discrepancy might beexplained through two non-exclusive arguments.The first argument is that the greater age of theCanarian archipelago (e.g. 20 million years forFuerteventura; Anguita et al. 2002), means thatthere has been a longer period of time for geneticvariation to accrue via mutation. However, phylo-genetic studies of several endemic groups do notsupport great ages for the majority of endemic line-ages, hardly exceeding 4–5 million years (Marreroand Francisco-Ortega 2001a,b). Moreover, there isalso evidence of a considerable degree of interfertil-ity among several endemic species belonging toradiating genera, and in cases even for interfertilitybetween members of distinct monophyletic genera,which supports the supposition that they representrelatively recent derivatives from common ances-tors (Francisco-Ortega and Santos-Guerra 2001).Second, the close proximity of the islands to Africa,96 km today but only 60 km during glacial periodsea-level minima (García-Talavera 1999), wouldhave facilitated multiple introductions into theCanaries and thus genetic bottlenecks may havebeen less extreme than for some of the more remotePacific islands. Although a monophyletic origin hasbeen demonstrated for a large majority of Canarianendemics, this does not mean that there could notbeen several dispersal events prior to the radiationprocesses (Francisco-Ortega et al. 2000). These ideasare interesting but speculative: further work isclearly needed to establish how genetic diversitylevels vary between and among archipelagos, andthen to test what factors may be responsible.

It may be relevant to this discussion that thebulk of the genetic diversity displayed by theCanarian endemic plants appears to reflect inter-population, rather than intrapopulation variation.This is consistent with (1) repeated founderevents within the archipelago, and even withinislands, (2) the apparent existence of dispersallimitations of pollen and seed among popula-tions, and (3) the prevalence of autogamic (self-fertilization) reproduction in many of thesespecies. Moreover, when a species has been found

to hold a larger component of intrapopulationthan interpopulation genetic variation, as wasfound for the Gomeran endemic shrub Echiumacanthocarpum, the population disjunction hasbeen interpreted in historical terms as constitut-ing just the surviving remnants of a formerlycontinuous island population (Sosa 2001). Ifintraarchipelago events are crucial to explainingpatterns of genetic variation within lineages, com-parisons between archipelagos or different oceanbasins may in some respects be confounded.

Finally, in this discussion we have assumed thatbottlenecks are essentially stochastic, but conceptu-ally, at least, we might recognize that they couldsometimes be regarded as deterministic. Whereasstochastic bottlenecks are survived by a randomsubset of the population, deterministic bottlenecksare survived by a specific genetically ‘pre-adapted’subset. A deterministic bottleneck will give rise to agreater subsequent degree of resistance to a newepisode of the same disturbance, whilst a stochasticone won’t. We can illustrate this idea with twohypothetical examples. First, in the event of theintroduction of the myxoma virus to an insular rab-bit population, a few individuals may survive toperpetuate the population, and these will tend to bethose least affected by the disease: subsequentlygenetic resistance will spread through the popula-tion. Second, imagine that a volcanic eruption pro-duces a lava flow that destroys entirely a plantcommunity, except for those plants that happen tohave germinated on small hillocks that protrudeover the level of the lava. The survivors will domi-nate the seed rain in the next generation, but havenot been selected for their resistance to being cov-ered with lava.

In conclusion, what have we established in thissection? First, the restriction of populations to anextremely small size, either on colonizing an island(the founding event), or subsequently in the his-tory of a lineage (other bottlenecks) may, in theory,produce a variety of effects:

● loss of genetic diversity (heterozygosity)● addition of novel genetic combinations● the introgression of genotypic variation fromrelated species (or varieties).

F O U N D E R E F F E C T S, G E N E T I C D R I F T, A N D B OT T L E N E C K S 171

Page 185: Whittaker. Island Biogeography. 2007

Thus, bottlenecks can variously diminish het-erozygosity (a general indicator of fitness), or incontrast can serve to ‘shake things up’, generating asort of mini-genetic revolution (sensu Carson andTempleton 1984), contributing to rapid phases ofevolutionary change in island populations.Secondly, although the number of studies is grow-ing, providing some support for the operation ofthese mechanisms, their relative contribution toisland evolutionary change remains uncertain(Grant 2002). Thirdly, notwithstanding recentdevelopments in molecular biology and genetics, itis hard to be sure how well the specific phenotypicproperties upon which the success of a lineagedepends are measured by tools such as allozymeelectrophoresis (see papers in Clarke and Grant1996). It is surely notable that in comparison torelated continental lineages, island endemic formsoften display far greater morphological differencesthan are reflected in estimates of average heterozy-gosity (e.g. Crawford et al. 1998), which implies thatsomething about these island environmentsencourages rapid lineage diversification throughthrowing specific switches in the genetic make-up.That ‘something’ likely reflects the general impov-erishment, disharmony, and environmental/bioticnovelty of remote islands. While founder eventsand other largely stochastic bottlenecks are part ofthe story, the emergent patterns encourage evolu-tionary scientists to invoke significant roles fordeterministic causes alongside or in preference tostochastic founding effect processes, as will be dis-cussed later.

7.2 After the founding event:ecological responses to empty nichespace

Do islands possess vacant niche space? The rela-tionship between isolation and impoverishmentmay not necessarily take a simple linear or log–linear form (Chapter 4), but it is accepted thatoceanic islands possess fewer species for their areathan equivalent areas of continent. Yet, if this repre-sents an equilibrium condition (whether dynamicor static), this could be interpreted as indicatingthat the islands involved are actually ‘full’ and that

the ‘missing’ ecological niches are lacking becausethere is no space for them. Although this argumentmight apply to older oceanic islands, surely theremust be plenty of vacant niche space in the speciespoor and disharmonic early stages of the lifespan ofan island? This is generally assumed to be the case,but the concept of empty niche space turns out tobe another slippery one to pin down and quantify.This is partly because the term ecological niche hasitself been variously defined. When first introducedit reflected the figurative usage of ‘a place in thecommunity’, but developed into the description ofthe types of opportunities that occur for species touse resources and avoid predators (Schoener 1989).Hutchinson’s (1957) usage was of a different form:the n-dimensional hypervolume of environmentalaxes within which a species occurs. A plant or ani-mal has a certain range of an environmentalresource, such as temperature, at which growth isoptimal. Outside this optimal range it may survivebut not thrive, until a point is reached, the limits ofits species tolerance, beyond which the species can-not persist. Each resource gradient forms a part ordimension of the hypervolume.

In the island context, the term empty niche refersto the lack of representation of a particular main-land ecological guild or niche, providing evolution-ary opportunities for colonizing taxa to exploit thevacancy: the idea being that a fundamental nicheexists and it is just a matter of filling it. As theassembly rules debate shows, it can be quite diffi-cult to determine if such a fixed form of architec-tural blueprint exists and so while the concept ofthe empty niche is of intuitive value, it can proveextremely difficult to apply.

The literature discussed in the following sectionsmostly refers to a form of niche theory that derivesfrom the resource gradient idea. Typically, it doesnot encompass a multidimensional characterizationof the niche, but is limited to a single or very lim-ited set of characters, such as mobility, body size, orjaw size, each of which provides a surrogate meas-ure for one dimension of the n-dimensional hyper-volume of the organism. The ambiguities of theniche concept are avoided in many such studies bydescribing the distributions of species populationsalong resource spectra. As Schoener (1989) points

172 A R R I VA L A N D C H A N G E

Page 186: Whittaker. Island Biogeography. 2007

out, much of this literature is, in essence, concernedwith competition, chasing the elusive question(Law and Watkinson 1989) of the extent to whichcompeting species can coexist. The answer remainselusive in the island evolutionary context because itis easier to detect functional changes (particularlymorphological changes) in island forms, than it is todetermine the role of interspecific competition inrelation to other potential causes of such change.The best chance here is to examine contemporaryprocesses, which may then be inferred to hold awider historical relevance. The greatest method-ological difficulties come when it is the ‘ghost ofcompetition past’ that is being sought, i.e. whenresearchers are setting out to deduce past eventsfrom current pattern (Law and Watkinson 1989).

Notwithstanding these cautions, if we are tounderstand how adaptive radiations of organisms,such as the Galápagos finches or Hawaiian honey-creepers have come about on oceanic islands, wehave to be able to build explanatory models thatstart with the arrival on a remote island of the firstseed eater, the first nectar feeder, or even the firstland-bird species. The founding population thuscan exploit hitherto largely untapped resources: inshort, empty niche space. As the island fills up witha mix of further colonists and of neoendemicspecies the availability of empty niche space mustfluctuate, eventually declining again as the islandages and subsides.

Ecological–evolutionary responses to the differ-ential occupancy on remote islands (as comparedwith continents) provide interesting insights, whichwe now attempt to review, starting with two gen-eral responses. First, the phenomenon of ecologicalrelease, i.e. of niche expansion by colonist species,and second, density compensation, the term givenfor the general pattern of increased average densityof island species that goes hand in glove with thelower richness of island ecosystems.

Ecological release

Ecological release occurs when a species in colo-nizing an island thus encounters a novel bioticenvironment in which particular competitors orother interacting organisms, such as predators, are

absent. The response can take perhaps two mainforms: (1) loss of ‘unnecessary’ features (e.g. defen-sive traits, bold patterning), and (2) increase in vari-ation in features such as beak morphology.

Lomolino (1984a) argues that ecological releaseattributable to the absence of predators may bemore widespread than generally recognized and,conversely, that competitive release may be overes-timated. He cites examples of Fijian fruit bats of thegenus Pteropus being more diurnal in the absence ofpredatory eagles than those on less isolated Pacificislands, and of the meadow vole Microtus pennsyl-vanicus being essentially indiscriminant of habitattype on islands without one of its major predators,Blarina brevicauda.

Sadly, the loss of defensive traits commonlyinvolves a lack of fear of, and ability to flee from,humans and terrestrial vertebrate predators:undoubtedly a contributing factor to the demise ofisland species such as the dodos and the manyspecies of flightless rails extinguished from islandsacross the Pacific (Steadman 1997a). The SolomonIslands rail, Gallirallus rovianae, is not only flightlessbut also exemplifies another tendency: the loss ofbold patterning (Diamond 1991a). The loss of elab-orate or distinctive morphology and a return tosimpler song types both appear common featuresof island endemics, especially where islands pos-sess only one representative of a genus (Otte 1989).The general interpretation offered for these changesis that organisms released from competition withclosely related species no longer require such accu-rate mating barriers, and so these features are grad-ually lost.

The second form of release in the absence of closecompetitors on an island is to free a colonist fromconstraining selective pressure, thereby allowing itto occupy not only different niches but also a widerarray of niches than the continental ancestral form(Lack 1969; Cox and Ricklefs 1977). This form ofecological release, the increase in niche breadth,forms a key part in the reasoning of the competitivespeciation model we consider later and is an impor-tant part of many scenarios for island evolution(e.g. adaptive radiation). Island finches provideclassic examples of evolutionary increases in nichebreadth within lineages. In illustration, granivorous

E C O L O G I C A L R E S P O N S E S TO E M P T Y N I C H E S PAC E 173

Page 187: Whittaker. Island Biogeography. 2007

finch species on remote islands such as Hawaii andthe Galápagos are highly divergent in terms of beaksizes (and seed sizes utilized) compared withfinches on continents and, plausibly, this can berelated to an absence of competitor lineages on theislands (Schluter 1988).

Increases in niche breadth within an island line-age are not always morphologically apparent. Theancestor of the Darwin’s finch of Cocos Island,Pinaroloxias inornata, colonized an island where‘empty’ niche space and the lack of interspecificcompetition provided conditions for ecologicalrelease. This endemic finch has diversified behav-iourally and, while showing little morphologicalvariation, exhibits a stunning array of stable indi-vidual feeding behaviours spanning the range nor-mally occupied by several families of birds (Wernerand Sherry 1987). This intraspecific variabilityappears instead to originate and be maintainedyear-round behaviourally, possibly via observa-tional learning.

Density compensation

Density compensation refers to higher than nor-mal (i.e. than mainland) densities of a species on anisland, linked to the lower overall richness of theisland assemblage, and often suggested to indicatea form of competitive release. It was recognized byCrowell (1962) in a comparative study of birds onBermuda with those of similar habitats on theNorth American mainland. This density adjust-ment may occur to an apparently excessive degree,in which case it is termed density overcompensa-tion (Wright 1980). The occurrence and degree ofdensity compensation is variable, and a singlespecies may exhibit higher, lower, or comparabledensities on different islands as compared withthe mainland. This is, on reflection, unremarkable,as particular species experience different condi-tions on different islands, e.g. in relation to pres-ence/absence of predators, the presence/absenceof competitors, or the suitability of habitats.

To examine the generality of density compensa-tion effects, we should consider entire guilds orentire assemblages rather than just subsets ofspecies. Remote islands are typically considered

species poor for their area, and one corollary of thisis that the species–abundance distribution isexpected to be attenuated, with fewer very rarespecies able to sustain a presence than would befound in an average equivalent area of mainland.Where colonizing species find ‘empty niche space’and undergo ecological release, this effect will be allthe greater. It follows that the average populationdensity of species occurring on islands should begreater than for a mainland assemblage, and thatthis effect should be more marked the lower therichness per unit area (Fig. 7.1).

As discussed in the island ecology section, varia-tion in island species richness can best be accountedfor by bringing additional variables, particularlyclimatic and habitat factors, into the regressionmodels. Given which, we ought to take account sta-tistically of the same variables in testing if overalldensity levels differ from those that would be antic-ipated as a function of the overall resource base ofan island and the accompanying richness of speciesthus supported. In short, density compensationacross an entire assemblage is expected as a simplecorollary of lower richness, and this ‘null model’should first be considered before reaching for otherfactors in explanation (Williamson 1981).

Even so, there is a considerable literature ondensity compensation, focusing on a variety ofexplanations for apparent cases of overcompensa-tion. Two prominent ideas are: (1) a species beingable to find more food per unit effort and thusattaining a higher abundance within the samehabitat without a niche shift; and (2) nicheenlargement, such that a species utilizes a widerrange of habitats, foraging strata, foraging tech-niques, or dietary components (MacArthur et al.1972). In turn, these mechanisms may be under-pinned by the compositional character of theisland assemblage, for example, the absence of toppredators, or of large-bodied competitors. In thelatter case, a similar bird biomass might bedistributed amongst more numerous smaller-bodied bird species.

Blondel and Aronson (1999) provide an example(Table 7.1) of increased density through both mech-anisms (1) and (2) among tits (Parus species) inCorsica. The island has three species, and the nearby

174 A R R I VA L A N D C H A N G E

Page 188: Whittaker. Island Biogeography. 2007

E C O L O G I C A L R E S P O N S E S TO E M P T Y N I C H E S PAC E 175

Speciesimpoverishment

Continentalresources

Species richness

Islandresources

Speciesabundance

Density compensation

Figure 7.1 Density compensation in islands. Island isolation results in species impoverishment (lower obtuse triangle), which promotes theacquisition of available resources by the species present on the islands, allowing an increase in their density (upper obtuse triangle). From Olesenet al. (2002), slightly modified.

Table 7.1 Niche enlargement and density inflation in tit species (great tit Parus major, blue tit P. caeruleus, coal tit P. ater and crested tit P. cristatus) in matching habitat gradients on Corsica compared with the nearby French mainland. Values are breeding pairs/10 ha. Habitatsfollow a gradient of increasing structure complexity. Each species occupies on average more habitats and has higher population sizes on Corsica than on the nearby mainland. The crested tit (P. cristatus) is not present on Corsica (from Blondel and Aronson 1999)

Species Low Medium High Low High Old matorral matorral matorral coppice coppice oakwood

MainlandGreat tit 2.2 3.1 3.2Blue tit 11.5Coal tit 0.2Crested tit 1.8

Total mean 2.2 3.1 16.7Corsica

Great tit 1.6 1.7 2.5 3.6 2.6 4.7Blue tit 0.5 0.2 3.3 7.9 14.2Coal tit 1.2 2.1 4.1

Total mean 1.6 2.2 2.7 8.1 12.6 23.0

mainland has four. Comparable habitats have moretit species in Corsica than in the mainland, and theCorsican population densities of all tit species aregreater.

Several other mechanisms have been suggestedthat may influence density patterns in addition tothose mentioned (e.g. see MacArthur et al. 1972,

Wright 1980). It is hard, therefore, to generalize onthe combination of factors involved: they are spe-cific to particular species on particular islands. It issufficient for present purposes to note that main-land and island species densities can vary substan-tially. The phenomena of ecological release anddensity inflation do not apply universally to all

Page 189: Whittaker. Island Biogeography. 2007

island populations, but there appears to be goodevidence for their operation for many island line-ages, providing selective environments thatencourage genetic adjustments in the island popu-lations. From first principles, we can anticipate thatthey will be particularly prevalent on remoteislands at early stages of faunal and floral coloniza-tion. We go on to consider other niche shift trendsin a later section.

7.3 Character displacement

Character displacement can, and frequently does,refer to size changes among island forms. What dis-tinguishes this form of niche shift is the causalinterpretation: that it is competition between twofairly similar varieties or species that brings aboutselection, in one or both, away from the region ofresource overlap (Brown and Wilson 1956).Diamond et al. (1989, p. 675) define ecological char-acter displacement as ‘the effect of competition incausing two initially allopatric species to divergefrom each other in some character upon attainingsympatry’. Darwin (1859) termed this ‘divergenceof character’, and it has also been termed ‘charactercoevolution’ or ‘coevolutionary divergence’,although there is actually not one, but a suite ofrelated theoretical ideas centred around this notion(Otte 1989).

Caribbean anoline lizards provide examples ofcharacter displacement, Schoener (1975) findingshifts in perch height and diameter consistent withcompetitive effects. Species of similar size werefound to affect one another more than dissimilarlysized species, while larger species affected smallerones more than the reverse (cf. Roughgarden andPacala 1989). Furthermore, according to a distribu-tional simulation using Monte Carlo analysis, thedistribution of lizard species over microhabitattypes on satellite islands of the Greater Antilles wasalso found to be consistent with competitive effects(Schoener 1988). However, the results of this analy-sis depended on the assumptions built into the nullmodel––a general problem with null models inecology and biogeography (Chapter 5).

There is also some evidence for competitivedisplacement in Hawaiian crickets. Otte (1989)

presents an array of observations, including dataon song variation in Laupala, a genus of swordtailcrickets. The songs were found to be hyperdis-persed, the divergence in signals being greaterbetween coexisting populations than betweenallopatric populations, suggesting that competitiveinteractions are important. Otte concluded thatcharacter displacement is probably extremely com-mon in the Hawaiian crickets and, by extension,elsewhere, although it is difficult to demonstrateunequivocally. While the most convincing demon-strations of character displacement tend to be fromsystems involving few interacting species (cf. theanoline lizard studies), such effects may also bedetectable, and certainly are relevant, where asomewhat larger network of interacting species isinvolved.

Given the difficulty in validating ideas of charac-ter displacement (Connell 1980), it would be ideal ifan experiment were to be devised that controlledfor other complicating factors. The principal prob-lem is that character changes with time are gener-ally not observed, but are merely inferred bycomparison of the characteristics of populations inallopatry with those in sympatry. Such studies pro-vide corroboration but not validation.

Diamond et al. (1989) have reported a naturalexperiment which comes close to solving this prob-lem, in demonstrating divergence within popula-tions in sympatry for less than three centuries. Inthe mid-seventeenth century, Long Island, off NewGuinea, erupted violently and destructively, initiat-ing a primary succession on both Long itself andthe nearby islands of Tolokiwa and Crown. Byanalogy with the recolonization of Krakatau, it isreasonable to assume a delay of a few decades forre-vegetation to proceed to the point at which a for-est bird assemblage would have developed. Thusthe bird populations of the three islands can betaken to have been founded less than 300 years ago.The avifauna contains two different-sized speciesof honeyeaters of the genus Myzomela—M. pamme-laena and M. sclateri. They are the only islands onwhich the two coexist. Allopatric populations onother islands can be identified as the probablesources for the founding populations of the Longgroup. Thus, the serendipitous ‘experiment’ has a

176 A R R I VA L A N D C H A N G E

Page 190: Whittaker. Island Biogeography. 2007

maximum duration of about three centuries, forwhich two otherwise allopatric species have co-occurred. Furthermore, within the Long group,the two species occur together in all habitats and atall altitudes, often being found in the same flower-ing tree. They are thus sympatric even at the finestintraisland scale. Diamond et al. (1989) found thatthe Long populations are significantly more diver-gent from each other than are the allopatric sourcepopulations. The larger species, M. pammelaena, isbigger still on Long than in the source populations,and the smaller M. sclateri is even smaller.Comparing samples of the two species fromallopatric populations, the weight ratio was foundto lie between 1.24 and 1.43, but for the sympatricLong island populations the value was 1.52.

The 300-year time span may be unremarkable byanalogy with similar niche shifts in other circum-stances. Conant (1988) found that Laysan finches(Telespyza cantans) introduced to Pearl and HermesReef underwent measurable shifts in bill shapewithin 18 years, in adjusting to their local food sup-ply. Similarly, significant change in diaspore size inLactuca muralis in Barkley Sound took place withinjust five plant generations (see below; Cody andOverton 1996).

7.4 Sex on islands

There are many ways in which plants and animalsreproduce, the simplest dichotomy being betweensexual and asexual modes. What works best in theisland laboratory? How does the possession of aparticular breeding system affect the chances of col-onization or speciation on islands? Does hybridiza-tion create new forms and thus engenderdiversification, or does it restrict the tendency toradiation by pulling back two diverging forms fromthe brink of irrevocable parting?

Dioecy and outcrossing

Plants exhibiting sexual reproduction may as a sim-ple generalization be classified as monoecious(male and female flowers on the same plant), her-maphrodite (bisexual flowers), or dioecious (plantsthat, generally, bear either male or female flowers

throughout their lifespan). At least theoretically, itshould be more difficult for dioecious species tocolonize a remote island, given the apparent needto have both sexes present for success (Ehrendorfer1979; Sohmer and Gustafson 1993). In practice,early comparison by Bawa (1980, 1982) suggestedthat islands might have relatively high proportionsof dioecious species. He noted that 28% of theHawaiian plant species are dioecious, compared toa figure of 9% for a well-specified tropical mainlandflora from Panama. Since then, the Hawaiian figurehas been revised downwards, to 15% for speciesthat are strictly dioecious, or to 21% if a broadergrouping of sexually dimorphic species is used.This is still the highest incidence of dioecy of anywell-specified flora (Sakai et al. 1995a). However, itis by no means clear that Hawaii is representative:indeed dioecious species are reportedly scarce inboth the Galápagos (McMullen 1987) and theCanaries, where only 2% of the endemic flora isdioecious (Francisco-Ortega et al. 2000).

The high proportion of dimorphic species in theHawaiian endemic flora is intriguing. Sakai et al.(1995a) find that dimorphy is frequent in partbecause many colonists were already dioecious, butthat autochthonous evolution of dioecy hasoccurred in at least 12 lineages, including severalspecies-rich lineages. Interestingly, one-third of cur-rently dioecious species derive from monomorphiccolonists. Further study of other island floras willbe needed to establish how representative thesepatterns are of other oceanic island floras.

Baker (1955) argued that hermaphroditic plantswith self-compatible reproduction systems, able toestablish (potentially) from a single founder indi-vidual, would have better chances of island colo-nization than dimorphic or obligated out-crossers.From the data available it appears that the floras ofHawaii (Carr et al. 1986), New Zealand (Webb andKelly 1993) and the Galápagos (McMullen 1987) dohave smaller proportions of self-incompatible taxathan nearby continents, consistent with Baker’srule. Moreover, in some species, such as Fragariachiloensis, dioecious in America but hermaphroditicin Hawaii, or Coprosma pumila, dioecious in NewZealand but monoecious in Macquarie (Ehrendorfer1979), there is evidence that ‘continental’ allogamy

S E X O N I S L A N D S 177

Page 191: Whittaker. Island Biogeography. 2007

(out-crossing) has been substituted by insular auto-gamy (self-compatability). We might anticipate thatany initial advantage in autogamy at the pointof colonization should be of limited duration inrelation to subsequent lineage development.Interestingly, 237 species of Canarian endemicplants (almost half the endemic flora) exhibit fea-tures (such as unisexual flowers, heteromor-phic styles or self-incompatibilty) promoting out-breeding (Francisco-Ortega et al. 2000).

Patterns of dioecy and self-crossing vs. outcross-ing in island floras currently remain poorly under-stood. As they became better quantified, suchproperties may shed interesting light on the evolu-tionary assemblages of island floras in terms ofwhat they reveal about the importance of pollinatormutualisms (below), and may ultimately help us tounderstand why some lineages have radiated onislands while others are absent or have failed toradiate (Sakai et al. 1995b; Barrett 1996).

Loss of flower attractiveness

Island plants often display a loss of flower showi-ness, such that they tend to have small, actinomor-phic (radially symmetrical), white or greennon-showy flowers with simple bowl-shaped corol-las, instead of the large, tubular, zygomorphic(bilaterally symmetrical), bright-coloured flowersfrequently displayed on the continents (Carlquist1974). This trend, is evidenced in a comparison ofAustralian and New Zealand floras, and probablyreflects the generalistic and promiscuous nature ofthe island pollinators (Barrett 1996), as well as theirsmaller sizes when compared to the continentalpollinators. Inoue and Kawahara (1996) were ableto demonstrate a direct correlation between theflower size of Campanula species and of their polli-nators. Flowers were larger on the mainland, wherethey were visited by larger pollinators (bumblebeesand megachilid bees) lacking from the islands.Pollination in the island populations was carriedout by smaller pollen-collecting halictid bees,which were shown experimentally to be indifferentto flower size, in contrast to the preference forlarger flowers shown by the nectar-feedingmegachilid and bumblebees on the mainlands.

Anemophily

Some island plants appear to have shifted fromancestral animal pollination to wind pollination(anemophily). Examples include Rhetinodendron inthe Juan Fernández islands, Plocama and Phyllis inthe Canaries, and Coprosma in New Zealand(Ehrendorfer 1979), and it has been suggestedas a general island trend (see e.g. Carlquist 1974,Ehrendorfer 1979). Possible benefits include: (1) gaining independence from impoverishedpollinator services (e.g. Goodwillie 1999;Traveset 2001), (2) windy conditions favouringanemophily, and (3) outcrossing fitness benefitsyielded by wind pollination (Barrett 1996).However, as a definitive analysis of the prevalenceof anemophily has yet to be carried out, it wouldbe hasty to conclude in favour of these mecha-nisms. Preliminary data for the incidence ofanemophily in the floras of well-studied archipel-agos yield values varying from 20% for Hawaii(Sakai et al. 1995a,b), to 29% for New Zealand and34% for the Juan Fernández islands (Ehrendorfer1979). In contrast, McMullen (1987) stressed thepaucity of wind pollination within the native floraof the Galápagos. Furthermore, anemophily maybe linked to other important features, such aswoodiness or dioecy, thus not of itself indicatingselective advantage.

Parthenogenesis

Reproductive mode may also be tied up with colo-nization probabilities in certain animal taxa.According to Hanley et al. (1994), parthenogenetic(asexually reproducing) lizard species are relativelycommon on islands. They suggest that this may bebecause of: an enhanced colonization ability; theopportunity to flourish in a biotically less diverseenvironment (in which the relative genetic inflexi-bility that might be associated with parthenogene-sis is not too disadvantageous); or the opportunityto escape from hybridization or competition withtheir sexual relatives. Their study examined thecoexistence of parthenogenetic and sexual forms ofthe gecko Lepidodactylus on Takapoto atoll in FrenchPolynesia. The two forms of Lepidodactylus,

178 A R R I VA L A N D C H A N G E

Page 192: Whittaker. Island Biogeography. 2007

although closely related and hybridizing to alimited extent (through the activities of males of thesexual form), are considered separate species. Theircoexistence appears to be stable in the short term atleast, as the proportions of their two populationswere similar in 1993 to a previous survey in 1986.Hanley et al. note that the asexuals are actuallyextremely heterozygous relative to the sexual pop-ulation. This may seem odd at first glance, butapparently all parthenogenetic vertebrates with aknown origin have arisen from a hybridizationevent, in large part accounting for their higher lev-els of heterozygosity. In this case, far from the asex-ual form suffering from competition with thesexual, it is the parthenogenetic form that has thewider distribution within the atoll (perhaps indica-tive of hybrid vigour), with different clones havingoverlapping but different habitat preferences, andwith the sexual form more or less restricted to thelagoon beach. The coexistence of the two formsappears to be facilitated by the specialization of theless successful sexual taxon to beach habitats.

Hybridization

According to figures cited by Grant and Grant(1994), speciation through hybridization mayaccount for as many as 40% of plant species (allallopolyploids arise in this way), but is consideredto be much rarer in animals. However, cross-breed-ing in animals is not so rare, and there have beennumerous studies of hybrid zones, whereinhybridization takes place between two recognizedspecies. When successful, it provides a form of geneflow between them (cf. Kaneshiro 1995, above).Linking the themes of pollination and hybridiza-tion, Parrish et al. (2003) report evidence ofhybridization of dioecious fig species on islands inthe Sunda Strait region of Indonesia (including theKrakatau islands), which indicates a breakdown inpollinator specificity on the islands (as hypothe-sized by Janzen 1979), possibly linked to temporaryshortages of the matching fig wasp pollinatorspecies. Hybridization is also considered to be com-mon within the Hawaiian flora. Indeed, even in thespecies-rich genus Cyrtandra, which was oncethought not to hybridize much (if at all), recent

studies suggest it to be quite common (Sohmer andGustafson 1993).

Grant and Grant (1994) studied the morphologi-cal consequences of hybridization in a group ofthree interbreeding species of Darwin’s finches onDaphne Major (Galápagos) between 1976 and 1992.Geospiza fortis bred with G. scandens and G. fuligi-nosa. The hybrids in turn backcrossed with one ofthe recognized species. Interbreeding was alwaysrare, occurring at an incidence of less than 5% of thepopulations, but nonetheless was sufficiently fre-quent to provide new additive genetic variance intothe finch populations two to three orders of magni-tude greater than that introduced by mutation.During the course of their studies Grant and Grant(1996b) noted a higher survival rate among hybridsfollowing the exceptionally severe El Niño event of1982–83. This event led to an enduring change inthe habitat and plant composition of the island,which appears to have opened up the food nichesexploited by the hybrids, and provided for most ofthe back-crossing observed during the study. Theirwork thus incidentally demonstrates that signifi-cant fluctuations in climate can have important eco-logical and evolutionary consequences.

7.5 Peculiarities of pollination anddispersal networks on islands

We have established that islands tend to be speciespoor and disharmonic, that island colonists gothrough founding events and subsequent expan-sion in numbers, and that evolutionary ecologicalmechanisms subsequently include density compen-sation, niche expansion, and niche shifts. To pro-vide another example of density compensation orover-compensation, densities of anoline lizards asgreat as two individuals per square metre havebeen measured on some Caribbean islands (Bennettand Gorman 1979). Such high densities must, it isreasoned, lead to a greater relative level of intraspe-cific competition for resources than is usual. In suchcircumstances, we can expect to see selection for theexploitation of new food resources, such as nectar,pollen, fruits, or seeds. Here we review evidence forthe distinctive ways in which island pollination anddispersal networks have developed.

P E C U L I A R I T I E S O F P O L L I N AT I O N A N D D I S P E R S A L N E T W O R K S O N I S L A N D S 179

Page 193: Whittaker. Island Biogeography. 2007

We highlight two features. First, because of theimpoverishment in species, island pollination anddispersal networks tend to feature endemic super-generalist species, both among the animals andplants (Olesen et al. 2002). Secondly, pollinationand/or dispersal relationships can appear bizarrein relation to those found on continents, becausethe functions are carried out, in the absence of morecommon continental pollinators or dispersers, bygroups of animals not frequently involved in suchfunctions (Traveset 2001).

The emergence of endemic super-generalists

Traditionally, pollination studies focused on singleplant species or, at most, on guilds of related plantspecies and the insects pollinating them (Travesetand Santamaría 2004). This reflects formerly domi-nant ideas that pollination mutualisms were theresult of highly co-evolved interspecific relation-ships. More recent community-level work hasshown that pollination processes are typically moregeneralized in nature than previously thought,with many insect and plant species interacting, andthus yielding weak pairwise relationships.However, the connectance level (the percentage ofall possible animal–plant interactions within a net-work that are actually established) has been foundto diminish in species-rich communities (Olesenand Jordano 2002). On islands, where species-poorcommunities prevail, the existence of pollinationnetworks with a high degree of connectance is thusto be expected. Studies within the Azores, theCanaries, and Mauritius show connectance levelsvarying between 20 and 30% (Olesen et al. 2002).The exception to this range was provided by theMacaronesian laurel forests, a community domi-nated by palaeoendemic tree species, where thevalue obtained was 9%, indicating a higher level ofspecialization.

Recent work suggests that island networks ofteninvolve a few super-generalized animal pollinatorsserving many plant species, and a few super-generalized plant species that are visited by manypollinators (Olesen and Jordano 2002). Bombuscanariensis, the endemic Canarian white-tailedbumblebee, provides us with one such example. It

occurs from the coast up to 2000 m and is known topollinate 88 plant species (Hohmann et al. 1993). Ina laurel forest studied on La Gomera, Bombuscanariensis was found to visit 48% of the plantspecies participating in the network. Similarly,the endemic Canarian bee Anthophora allouardii vis-its at least 46% of the plant species of a subdesertscrub community on Tenerife (Olesen et al. 2002).An even greater degree of generalism is displayedby an endemic Halictus bee in Flores (Azores)(60%), by the endemic day gecko Phelsuma ornatafrom Ile aux Aigrettes (Mauritius) (71%), and by theendemic bee Xylocopa darwini from the Galápagos(77%).

Among plants, the Macaronesian islands againprovide examples of endemic plant species withgeneralized pollinator relationships. Theseinclude Aeonium holochrysum, a member of thecoastal subdesert scrub, visited by 80% of theknown community of pollinators, and Azorinavidalii, an Azorean endemic shrub, visited by75% of the pollinator community. Conversely,Lavandula buchii, an endemic plant restricted tothe old Teno and Anaga massifs of Tenerife, ispollinated by only 10 insect species (Delgado2000), in comparison with its continental congenerLavandula latifolia from Southern Spain, which ispollinated by 80 species (Herrera 1989). This illus-trates that a more comprehensive comparativeanalysis is needed to establish if the apparenttendency towards generalist species is a generalfeature of remote islands.

It has been suggested that one reason for thesuccess of many introduced species on islands isthe formation of ‘invader complexes’, wherebyintroduced species preferentially establish interac-tions with other introduced species (D’Antonioand Dudley 1993). However, Olesen et al. (2002)find little support for the invader complex conceptin pollination networks analysed on the Azoresand Mauritius. Rather, endemic supergeneralistspecies tend to include newcomers in their net-works, either in the form of endemic insects polli-nating exotic plants, or by endemic plants beingpollinated by exotic insects. In these cases, theendemic super-generalists may be pivotal to thesuccessful establishment of alien species that

180 A R R I VA L A N D C H A N G E

Page 194: Whittaker. Island Biogeography. 2007

have left behind their continental networks(Olesen et al. 2002).

Unusual pollinators

The absence of many common continental pollina-tors (bees, wasps, humming-birds, etc.) or dispersers(some guilds of land-bird, non-volant mammals)from remote islands has provided opportunities foranimals that do not usually perform these functionsto acquire or extend these roles. The best illustrationof this island feature comes from recent work onlizards, which have adopted roles as pollinatorsalmost exclusively on islands, including NewZealand, Mauritius, the Balearic islands, andTasmania. Olesen and Valido (2003) report that of 37lizard species now known to act as pollinators, onlytwo species are from mainland areas, both in factpeninsulas (Baja California and Florida).

The first published observations revealinglizards as pollinators derive from Madeira in the1970s, when Teira dugesii was observed drinkingnectar and licking the stigmatic lobes for pollen onboth native and introduced ornamental plantspecies (Elvers 1977). Several species of Canarianendemic lizards (Gallotia spp.) are now known tofeed on nectar of Euphorbia, Echium, and Isoplexis.Similarly, on the Balearic islands, Podarcis lilfordi,an endemic lizard, is known to visit some 20 plantspecies, including the flowers of the nativeEuphorbia dendroides, for its highly concentratednectar (Traveset and Sáez 1997). Of special interestis the role of Hoplodactylus geckos in New Zealandand of Phelsuma day geckos as pollinators inIndian Ocean islands. These geckos are known tovisit many plant species (Whitaker 1987; Olesenet al. 2002), transporting pollen over considerabledistances (Nyhagen et al. 2001). The introductionof exotic predators (e.g. rats, mongooses, mon-keys) has the potential to disrupt these pollinatorservices, potentially leading to extinction of nativeplant species.

Unusual dispersal agents

Again it appears that lizards have a disposition toact as dispersal agents on islands, a role performed

less extensively by mainland lizard species. Ofsome 200 lizard species reported to have a role inseed dispersal, two thirds were island species(Olesen and Valido 2003). Examples include 18New Zealand lizard species, among them thegecko Hoplodactylus maculatus and the skinksOligosoma grande and Cyclodina alani (Whitaker1987). On the Balearic islands, Podarcis lilfordi andP. pityusensis are known to feed on the fruits ofmore than 25 species (Pérez-Mellado and Traveset1999). All the endemic Canarian lacertids (sevenGallotia species) and skinks (three Chalcides species)are known to feed on fruit. Gallotia lizards eat fruitsof at least 40 native plant species, which is slightlymore than the half of the Canarian plant speciesbearing fleshy fruits, and also feed on the fruits ofat least 11 introduced species (Valido and Nogales1994).

Another ‘unusual’ plant dispersal agent comesin the form of the giant tortoises (Geochelone spp.)of the Galápagos (Racine and Downhower 1974)and Aldabra atoll (Hnatiuk 1978). Today, theAldabran species (Geochelone gigantea) has beenintroduced on several islets off Mauritius wheregoats and rabbits have recently been eradicated,as a functional substitute for extinct Mauritiantortoises (Geochelone triserrata and G. inepta), whichare considered to have been the original dispersersof the endemic Ile aux Aigrettes ebony (Diospyrosegrettarum). After a pilot study in a fenced enclo-sure, viable fruits of the ebony were found dis-persed in tortoise faeces away from the parenttrees (Zavaleta et al. 2001).

7.6 Niche shifts and syndromes

Thus far in this chapter we have considered someof the filters that might prevent lineages success-fully establishing on islands and the implicationsof small population sizes and of the possession ofparticular life-cycle traits connected to breedingsystems. We have also begun, inevitably, to con-sider emergent patterns of changing biological andniche characteristics. Here, we continue thisprocess under the heading of some of the moreprominent island syndromes, starting with the lossof dispersal powers.

N I C H E S H I F T S A N D S Y N D R O M E S 181

Page 195: Whittaker. Island Biogeography. 2007

The loss of dispersal powers

Given that oceanic islands are so hard to reach, itmay appear paradoxical that so many flightlessforms occur on them. For a distant island poppingout of the sea for the first time, there will inevitablybe a strong bias towards groups that are highly dis-persive, as these will, by definition, have a greaterchance of colonizing. This was reflected in the bio-geographical patterns discussed in Chapter 3, of fil-ter effects, the gradual loss of taxa with increasingdistance from continental source areas, and ofdisharmony. The relevance of these effects is broad,and wide ranging, from the evolutionary opportu-nities thus presented on islands like Hawaii, to theecological structuring in less remote islands such asKrakatau (Chapter 5). The island biogeography ofremote islands thus bears the imprint of dispersalfiltering of faunas and floras as its first organiza-tional principle.

Once on an isolated island, however, there maybe a strong selective force against dispersal ability,particularly in insects and birds. Examples given byWilliamson (1981) of flightless forms include the 20endemic species of beetle on Tristan de Cuhna, allbut two of which have reduced wings. A number ofideas have emerged in explaining such tendencies.One put forward by Darwin (letter to J. D. Hooker,7 March 1855; Roff 1991) is that highly dispersivepropagules are more liable to be lost from the genepool, for instance by a highly dispersive plant orinsect being blown off to sea, so selecting for less-dispersive forms among the island population. Asecond idea, applicable to flightless birds such asthe famously extinct dodos, is that the lack of pred-ators removes the selective advantage of flight inground-feeding species. Indeed, flight might be dis-advantageous in that the energy used in maintain-ing and using flight muscles is effectively wasted.However, flightless species within generally volantgroups can also be found on continents, so beforeattempting to test ideas concerning loss of disper-sal, it might be prudent to confirm that the inci-dence of flightlessness on islands is higher thanexpected by chance alone.

Roff (1991, 1994) has undertaken just such analy-ses for insects and for birds. Darwin’s original

observations on insects were based, of course, onrelatively few data, and in his analysis of the muchlarger data sets now available Roff (1991) foundthat statistically there is no evidence of an associa-tion between oceanic islands and insect flightless-ness. He argues that the large size of most oceanicislands makes flightlessness of little positive selec-tive value: only a small fraction of the population islikely to be lost from such sizeable land masses (seealso Ashmole and Ashmole 1988). It should bestressed, however, that Roff’s analyses are based ona dichotomy between volant and non-volant forms.He does not take into account island forms thathave evolved a reduced flight capability or behav-iour without actually losing the ability altogether.

In his second article, on birds, Roff (1994) pointsout two important factors that complicate theanalysis. First, in the case of most bird groups, thesmall sample size limits the potential for formal sta-tistical analysis. Secondly, the phylogenetic con-straints on flightlessness must be considered. Thecondition may have evolved once, followed by sub-sequent speciation within a flightless lineage, or itmay have evolved several times, in extreme casesonce for each flightless species. For example,among important flightless bird groups, the 18known penguin species are considered to be mono-phyletic, having speciated from a common flight-less ancestor, whereas the 41 ratite species(ostriches, etc.) represent between one and five evo-lutionary transitions to flightless forms, dependingon whose phylogeny one accepts. In contrast,flightlessness in rails has most probably evolvedseparately for each of the 17 (or more) flightlessspecies, among the 122 (or more) recognized railspecies. Truly flightless rails are only found onislands and they represent about half of the islandrail species (Roff 1994). More ratites occur onislands than on continents, but the taxonomic dis-tribution does not indicate that flightlessness ismore likely to evolve on islands in this group.Examples of flightlessness from other groups ofbirds include the following island forms: the NewCaledonian kagu (Rhynochetos jubatus), the twoextinct dodos (Raphus cucullatus in Mauritius andPezophas solitaria in Rodrigues), New Zealand’skakapo (Strigops habroptilus, the only flightless

182 A R R I VA L A N D C H A N G E

Page 196: Whittaker. Island Biogeography. 2007

parrot), the Auckland Islands teal (Anas auck-landica), the Galápagos cormorant (Phalacrocoraxharrisi), the three Hawaiian ibises (Apteribis), andfour Hawaiian ducks (Anas); these 13 species repre-senting at least eight evolutionary transitions.Taking all these factors into account, Roff con-cluded that there is an insular tendency to flight-lessness, although this could only be formallyshown for the rail group. A separate analysis byMcCall (1998) records that flightlessness hasevolved independently in at least eleven extant birdfamilies, eight of which currently contain bothvolant and non-volant species. They noted that thevolant species in these families tend to be charac-terized by relatively short wings. As short-wingedspecies pay a relatively high energetic cost to fly,their findings support the hypothesis that the ener-getic costs of flight provides a key factor in select-ing for flightlessness in these families.

On the topic of flightless rails, Diamond (1991a)reported a newly discovered species, Gallirallusrovianae, which is believed to be either flightless orweak-flying. It was collected in 1977 from NewGeorgia, Solomon Islands. The ancestral form isvolant and more boldly patterned than the newspecies. Diamond regards G. rovianae and its rela-tives on other oceanic islands as cases of convergentevolution, whereby 11 groups of rails have inde-pendently evolved weak-flying or flightless formsand also, in several cases, reduction in boldly pat-terned plumage, on widely scattered oceanicislands. These weak- or non-flying rails do not farewell in contact with people, and as many as half ofthose extant at the time of European discovery havesubsequently become extinct. Moreover, subfossilremains of other extinct species have been found onmost of those Pacific islands that have beenexplored by palaeontologists, and it is likely thatthere are many more extinct forms awaiting discov-ery (Steadman 1997a). There is thus clearly a strongselection pressure for reduced flying ability in rails,and this appears to be related to the energetic andweight burden of unnecessary flight muscle onislands lacking terrestrial mammalian predators.

From the plant world, the genus Fitchia (in theAsteraceae), endemic to the Polynesian islands of the south Pacific, exemplifies reduction in

dispersability (Carlquist 1974; Williamson 1981). Itsclosest continental relatives are herbs with spikedfruits transported exozoically (i.e. by externalattachment to an animal), but in Fitchia the spikesare relict, the fruits are much larger, the plants aretrees, and their fruits drop passively to the forestfloor. Endemic Pacific island species of Bidens (alsoin the Asteraceae) illustrate various steps in the lossof barbs and hairs and other features favouring dis-persiveness, these changes being accompanied byecological shifts from coast to interior and, in somecases, wet upland forest habitats (Carlquist 1974;Ehrendorfer 1979). It might be argued that theseexamples suffer from a certain circularity of reason-ing, with traits being used to infer the reasons fortheir selection. To allay such concerns, a demon-stration is needed of such changes actually occur-ring. Once more, the Asteraceae provide theevidence.

Cody and Overton (1996) monitored populationsof Hypochaeris radicata and Lactuca muralis, bothwind-dispersed members of the Asteraceae, on 200near-shore islands in Barkley Sound (Canada). Theislands ranged from a few square metres in area upto about 1 km2. Significant shifts in diaspore mor-phology occurred within 8–10 years of populationestablishment, the equivalent in these largely bien-nial plants of no more than five generations. Thediaspore in these species consists of two parts, atiny seed with a covering (the achene), which is sur-rounded by or connected to a much larger ball offluff (the pappus). The clearest findings were forLactuca muralis, the species with the largest samplesize. Founding individuals had significantlysmaller seeds (by about 15%) than typical of main-land populations, illustrative of a form of a non-random founder effect termed immigrant selection(Brown and Lomolino 1998, p. 436). Seed sizes thenreturned to mainland values after about 8 years,but pappus volumes decreased below mainlandvalues within about 6 years. If the pappus isthought of as a parachute and the seed the payload,then the ratio between the two indicates the dis-persability of the diaspore (Diamond 1996).Figure 7.2 provides a schematic representation ofthe trends in dispersability. Cody and Overtonwere able to quantify these phenomena because

N I C H E S H I F T S A N D S Y N D R O M E S 183

Page 197: Whittaker. Island Biogeography. 2007

they were working on a micro-scale version of theoceanic island examples usually cited, i.e. theirpopulations were very small, occupying tiny areas,and their plants were short-lived—ideal for moni-toring short-term evolutionary change.

The development of woodiness in herbaceousplant lineages

Several plant families with predominantly herba-ceous species in the continents are characterized byhaving woody relatives on islands. Outstandingexamples include the silversword alliance andlobelioids in Hawaii, tree lettuces in JuanFernández and Macaronesia, tree sunflowers in

St Helena and Galápagos and buglosses and daisiesin Macaronesia (Givnish 1998).

Whether island woodiness is a derived or basalfeature has long been debated. Advocates for theformer view include both Darwin (1859) andCarlquist (1974, 1995). Those arguing that it is abasal, relictual characteristic include Bramwell(1972), Sunding (1979), and Cronk (1992). Thereare cases of woody island plants that areconsidered relictual, for instance the Atlanticisland laurel-forest tree species discussed inChapter 3, but these species are unexceptional inbeing woody forms. For predominantly herba-ceous taxa represented on islands by woodyforms, analyses of molecular phylogenies all pointto woodiness being a derived characteristic, hav-ing developed from originally herbaceouscolonists (Kim et al. 1996; Panero et al. 1999;Emerson 2002).

Hawaii provides good examples. Most colonistswere weedy, herbaceous, or shrubby, with onlya minority of tree-like plants: subsequently,woodiness has developed in the insular lineages(Carlquist 1995). The Amaranthaceae andChenopodiaceae provide classic cases, in whichthe endemic species are only woody (or in casessuffrutescent—woody only at the base of thestem), whereas in continental source areas they areprincipally herbaceous (Carlquist 1974; Sohmerand Gustafson 1993; Wagner and Funk 1995). Forthe Atlantic islands, analyses of nuclear rDNA ofwoody Macaronesian Sonchus and five relatedgenera have likewise shown them to have beenderived from continental herbaceous ancestors,most probably from a single founding species.During the adaptive radiation of the lineage, it hasundertaken a limited number of interisland trans-fers, has diversified in to differing habitats, andhas developed considerable morphological diver-sification, including the development of thewoody lineage (Kim et al. 1996). Similarly, theMacaronesian clade of Echium (Boraginaceae),consisting of 29 endemic species (2 on Madeira, 24on the Canaries, and 3 on the Cape Verde islands),of which 26 exhibit woodiness, derive from asingle herbaceous continental ancestor (Fig. 7.3;Böhle et al. 1996). The acquisition of woodiness

184 A R R I VA L A N D C H A N G E

Mainlandpopulation

Low Medium High

Dispersability

Oldpopulation

Intermediateage

Newpopulation

Time on

island

Dispersal to island

Figure 7.2 A schematic diagram of the evolution of reduceddispersability of wind-dispersed plants on islands, as suggested by thechanges recorded in Lactuca muralis (Asteraceae) on near-shoreislands in British Columbia, Canada. Maximal dispersabilitycharacterizes the youngest island populations, as the founding event(s)will be from the upper end of the dispersal range of the mainlandpopulation. Thereafter selection acts to decrease dispersability ofpropagules on islands as more dispersive diaspores are lost to sea.(Simplified from Cody and Overton 1996, Fig. 1.).

Page 198: Whittaker. Island Biogeography. 2007

from herbaceous ancestors has been repeatedseveral times in some lineages, as in theMacaronesian endemic genus Pericallis (Paneroet al. 1999).

Several reasons have been suggested for thedevelopment of lignification on islands. As reviewedby Givnish (1998), they include the following.

1 The competition hypothesis (Darwin 1859).Herbaceous colonists would obtain a competitiveadvantage by growing taller and developing intoshrubs/trees.2 The longevity hypothesis (Wallace 1878).Woodiness might allow extended lifespans,

increasing the chance of sexual reproductionwhere pollinators are scarce.3 The moderate insular climate hypothesis(Carlquist 1974). In comparison with the closestcontinents, islands typically have milder (andmoist) climates, promoting the development ofherbs to tree-like growth forms.4 The promotion of sexual out-crossing hypothe-sis (Böhle et al. 1996). Evolution of the woodyhabit accompanies increased longevity, which pro-vides greater likelihood of cross-pollination inenvironments such as islands where pollinationservices may be unreliable. This hypothesis is anextension of hypothesis (2).

N I C H E S H I F T S A N D S Y N D R O M E S 185

E. russicum

E. sabulicolaE. parviflorumE. plantagineumE. lusitanicumE. albicansE. italicumE. asperrimumE. vulgareE. creticumE. horridumE. tuberculatumE. rosulatum H

1. 2.

4.

5.

9.

6. 7. 8.

MadeiraArchipelago

CanaryIslands

Cape VerdeArchipelago

Continental Species

Island Species

3.

100100

100100

98100

Singleisland

invasionthrough

herbaceouscontinental

ancestorfollowed bysuccessive

islandcolonization

and rapidspeciationof woody

forms.

89100

9481

9796

Madeira

CapeVerde

8298

94100

100100

96-

94100

1.0%

HHHHHHHHHHH

E. hypertropicum WE. vulcanorum W

E. candicans WE. nervosum WE. webbii WE. aculeatum WE. leucophaeum WE. virescens WE. simplex WE. giganteum WE. hierrense WE. brevirame WE. bonnetii HE. strictum WE. callithyrsum WE. onosmifolium

Lobostemon

E. russicum

WE. handiense WE. decaisnai WE. pininana WE. wildpretii WE. auberianum W

(Eastern Europe)(South Africa)

HWL. fruticosus

Canary

Islands

Figure 7.3 Phylogeny and biogeography of Echium showing that the woody species of Macaronesia are part of a monophyletic archipelagicradiation (source: Böhle et al. 1996). Altogether there are 29 endemic species of Echium in Macaronesia. The numbers refer to Echium speciesthat exemplify a growth form or habit: (1) Echium plantagineum, (2) E. humile, (3) E. rosulatum, (4) E. parviflorum, (5) E. onosmifolium, (6) E.decaisnei, (7) E. simplex, (8) E. wildpretii, (9) Lobostemon regulareflorus. The technical details of the phylogeny are as follows. It was based onanalyses of 70 kb of non-coding DNA determined from both chloroplasts and nuclear genomes. Numbers above branches indicate the number oftimes the branch was found in 100 bootstrap neighbour-joining replicates, using the Kimura distance, whereas the numbers below branchesindicate the number of times the branch was found in 100 bootstrap parsimony replicates.

Page 199: Whittaker. Island Biogeography. 2007

5 The taxon cycling hypothesis (Givnish 1998).Successful island colonizers are usually weedyspecies that established themselves easily inopen/marginal habitats, where they can grow rap-idly. Increased specialization and selection in com-petitive environments, with scarce resources, mayhave precipitated a shift toward persistence throughwoodiness (Givnish 1998, Panero et al. 1999). Thishypothesis is an extension of hypothesis (1).

It is too early to conclude which of the above mightprovide the best explanation, if indeed there is asingle answer.

Size shifts in island species and the island rule

It presents another apparent paradox that twoconflicting traits are viewed as common in islandforms: gigantism, examples of which include theGalápagos and Indian Ocean tortoises (Darwin1845; Arnold 1979), and nanism (dwarf forms),illustrated by island subspecies of ducks, whichare commonly smaller than mainland species(Lack 1970). Of course, size variations within taxaare not restricted to islands, and so there is thedanger of inferring a general rule from a few strik-ing cases. For instance, the famous Aldabran andGalápagos giant tortoises (Geochelone spp.) reachup to 130 cm in length. However, giant tortoise arereadily eliminated by humans, and other giantGeochelone occur, or occurred, in South America,Madagascar, the Seychelles, the Mascarenes, andon the African mainland. It is not certain whetherlarge size evolved after reaching islands orwhether it merely bestowed the capabilities ofdispersing to islands as remote as the Galápagos(Arnold 1979).

Unusual size has been reported for many islandtaxa, including plants and insects, but it is amongstvertebrates that the most striking general patternsare found, for there appears to be a graded trendfrom gigantism in small species to dwarfism inlarger species. This has been termed the island rule(van Valen 1973, Lomolino 1985, 2005), and hasbeen found to hold for mammalian orders, non-volant mammals, bats, passerine birds, snakes and

turtles (Fig 7.4, Box 7.1). How well particular taxafit this island rule depends, as ever, on the assump-tions involved in the analysis (e.g. compareLomolino 2005 and Meiri et al. 2006), but the syn-thesis provided by Lomolino (2005) provides both acompelling body of evidence and a persuasiveinterpretation of apparent exceptions. Analyses ofdata for continental mammal species have sug-gested that there is an optimum body size forenergy acquisition at about 1 kg body weight(Damuth 1993). Linking this observation to theisland rule, it would seem that islands favour theevolution of a body size most advantageous forexploiting resources in particular dietary cate-gories, and that this must in turn be linked in someway to the constrained area of islands, and theabsence of the full array of competitors and preda-tors found within a continent.

The body size at which equivalence in body sizebetween island and continental populations isreached varies between taxa. Lomolino (2005) inter-prets these values as indicating the optimal size fora particular body architectural plan (bauplan) andecological resource exploitation strategy (Fig. 7.5).Whereas in species-rich mainland communities,interspecific interactions (especially predation andcompetition) facilitate evolutionary divergencefrom the optimal size, on islands selection favourschange towards the optimum body size. Forinstance, islands often lack terrestrial vertebratepredators, but do have birds of prey. This switch inpredation regime might favour larger rodents,which would then become dominant in the popula-tion through intraspecific competitive advantage(Reyment 1983, Adler and Levins 1994). At theother end of the scale, carnivores (Box 7.1) andartiodactyls (even-toed ruminants) typicallybecome smaller, suggesting small size to be advan-tageous in restricted insular environments of lim-ited food resources (Lomolino 1985).

There is considerable variability around the trendline (Fig 7.4), some of which may be explained byvariables such as island isolation and area. Forexample, body size of the tricoloured squirrel(Callosciurus prevosti) decreases with increasing iso-lation, and increases with island area (Heaney1978). Some size changes can only be understood

186 A R R I VA L A N D C H A N G E

Page 200: Whittaker. Island Biogeography. 2007

N I C H E S H I F T S A N D S Y N D R O M E S 187

0.00

0.50

1.00

1.50

2.00

2.50

1 10 100 1 000Body size of mainland population (g)

10 000 100 000 1 000 000

DidelphimorphiaInsectivoraLagomorphaRodentiaTerrestrial carnivoresCarnivores with aquatic preyArtiodactyla

S i

Figure 7.4 Demonstration of the island rule: body size trends for populations of insular mammals. Si � relative size of insular forms expressedas a proportion of body mass of their mainland relative. (Redrawn from Fig. 1 of Lomolino 2005).

Box 7.1 Dwarf crocodiles and hominids

The dwarf crocodiles of Tagant (Mauritania)Seventy years after the last reports of theirexistence and six years after the IUCN listed themas extirpated, relict populations of the Nilecrocodile (Crocodylus niloticus) have beenrediscovered in wetlands in southern Mauritania(Shine et al. 2001). These populations appear tohave been isolated by climatic change associatedwith the Pleistocene/Holocene transition, andhave persisted in association with seasonal ponds,known as gueltas. In summer, these gueltas canbe smaller than 1 ha. Today it is thought thatthere may be small populations of these dwarfcrocodiles in some 30 or so of these Mauritanianwetlands, feeding mainly on the guelta fishes, incompetition with human settlers. In these small‘islands’ the species has undergone dwarfism,resulting in adults of no more than 2–3 m bodylength, just half that of adult Nile crocodiles.

Dwarfism in human islanders?Considerable press interest followed the discoveryof the remains of six individuals belonging to

what has been claimed to be an extinct species ofhuman (Homo floresiensis) from Liang Bua Cavein the island of Flores, Indonesia (Brown et al.2004; Morwood et al. 2004). The adults wereonly 1 m tall, weighing just 25 kg and with brainssmaller than one third that of modern humans.They are thought to have reached Floresc.50 000 years BP, persisting alongside modernhumans until finally becoming extinct about18 000 years BP.

Where did they come from? Although Homoerectus were considered to have reached Java asearly as 1.5 Ma, they were until fairly recentlythought not to have crossed Wallace’s line (the25 km gap of open ocean dividing Bali fromLombok). Finds of 0.8 Ma stone tools in Floresmight indicate that they did reach Flores, althoughno bones have yet been found. Thus H. floresiensismay have evolved from ancient H. erectus. Humandwarfism due to insularity has not previously beenreported, but this change in body size is consistentwith the island rule, as discussed in the text.

Page 201: Whittaker. Island Biogeography. 2007

188 A R R I VA L A N D C H A N G E

Body size of mainland population (g)

Evolutionary shift from flying passerine to bau

plan and ecological strategies of

a large, ground-dwelling herbivore

a b

0.00

0.50

1.00S i

1.50

2.00

1 10 100 1 000 10 000 1 00 000 1 000 000

Passerine birds

Op Ou

Ungulates (Artiodactyla)

Figure 7.5 Lomolino’s (2005, Fig. 6) graphic of the island rule shows the rule for two different bauplans and a possible transition betweenthem. In essence, small forms tend to get larger (arrow at point a) and larger forms get smaller (arrow at point b) but following different lines,matching separate optima, dependent on the fundamental body plan and ecological role. Most size changes are gradual but occasionally majorevolutionary transformations occur, involving shifts in bauplan and ecological role in response to the distinctive island opportunities and selectiveforces, e.g. the flight loss and gigantism of the now extinct moa of New Zealand. An evolutionary transition from a passerine bird bauplan to thattypical of a large ungulate (large non-volant herbivore) would involve a transition from the left-hand line, with optimum Op (passerine) to theright-hand line, with optimum Ou (ungulate).

within Lomolino’s model as a result of a shift toanother bauplan, i.e. to another niche (implyingboth size and function change). As an example,New Zealand moas were giant flightless birds that,in the absence of terrestrial mammals, essentiallyacquired the bauplan of ungulates (Fig. 7.5;Lomolino 2005).

It is also possible that there might be a form ofnon-random founding effect (termed immigrantselection by Brown and Lomolino 1998) involved insome size shifts. For instance, in active dispersalevents, the founder group might comprise signifi-cantly larger individuals from within the mainlandpopulation, better able to survive a long-distancejourney to an island. Or, conversely, significantlysmaller plant propagules may be better predis-posed to arrive on a new island if a passive disper-sal event has to occur (e.g. Cody and Overton 1996,

above). However, it is hard to see how immigrantselection would assist in generating the size reduc-tions in larger mammals. Support instead for theoverwhelming role of selection comes from a recentstudy of anthropogenically fragmented forest inDenmark, where body size changes consistent withthe island rule were reported within just 175 years(Schmidt and Jensen 2003, cited in Lomolino 2005).

Fossil and subfossil remains provide furtherexemplification of the island rule. For instance, theMediterranean islands once featured an endemicfauna of dwarf hippopotami, elephants and deer,and giant rodents (the so called HEDR fauna),which survived from the Pliocene to the sub-Recentperiod (Reyment 1983). Schüle (1993) argued thatfor animals like elephants, which start large, islandsprovided insufficient resources to sustain suchlarge body weights. Even infrequent scarcity could

Page 202: Whittaker. Island Biogeography. 2007

provide strong selection pressure towards moreagile, smaller forms. Some of the best evidence fornanism seems to be for fossils of the Cretan form ofdwarf elephant, which show a wide size range,interpreted as representing stages in the develop-ment of the dwarf form. ‘The flatland, lumberingancestral form, with good swimming capabilities,had been irrevocably transformed into a cleverclimber of rocky slopes’ (Reyment 1983, p. 302). Theprize for the greatest degree of dwarfing goes to theMaltan Elephas falconeri, which was barely morethan a metre in height (Lister 1993) and only a quar-ter the size of the putative ancestor, E. namadicus.

These remarkable decreases in size may haveoccurred quite quickly. Vartanyan et al. (1993)reported remains of the woolly mammoth, whichappear to have persisted on Wrangel island, 200 kmoff the coast of Siberia, when the island becameseparated from the mainland c.12 000 BP. The mam-moth became extinct from its continental range byaround 9500 BP (either because of climate change, orhumans, or a combination of both), but it persistedon Wrangel until some point between 7000 and4000 BP. Not only did the population thus attainrelictual status, but between 12 000 and 7000 BP itappeared (on the basis of fossil teeth) to shrink inbody size by at least 30%. However, size changesneed not be so rapid, nor follow so soon after colo-nization. Sicilian elephants probably took between250 000 and 600 000 years to shrink to about halfthe dimensions of the ancestral Elephas antiquus andabout 1 million years to reach one quarter of theancestral size.

All Pleistocene ungulates, other mammals of anyconsiderable size, flightless birds, and giant tor-toises on the Mediterranean islands were almostcertainly wiped out by the first humans to arrive.This is, however, only recognizable in the archaeo-logical record on Cyprus and Sardinia. It appearsthat the extant wild ungulates on theMediterranean islands were all introduced byhumans.

Although the island rule now appears to beclearly established, there is evidence of Holocenesize reductions in lizards from small islands in theCaribbean (Pregill 1986). In particular cases, theevidence is strongly supportive of some form of

co-evolutionary competitive pressure broughtabout by formerly allopatric taxa coming into sym-patry (cf. Roughgarden and Pacala 1989). In othercases, the activities of humans, such as habitatalteration and the introduction of terrestrial preda-tors, may be responsible not only for the extinctionof larger lizards from within island assemblagesbut also for size reductions (Pregill 1986). In illus-tration, in the Virgin Islands, populations of the StCroix ground lizard (Amereiva polyps) have declinedsteeply over the past century, and body size ofmature adults has also decreased. The ecology ofthese islands has been greatly affected by humans,and in particular, by the introduction of the mon-goose (Pregill 1986). Associations between humancolonization and the extinctions of larger speciesfrom within insular lizard assemblages are alsoknown from the fossil records of the Canaries,Mascarenes, and Galápagos archipelagos. Indeed,Pregill (1986) concludes that insular lizards thatdraw our attention because of their exceptional sizemay have been rather ordinary in prehistory.Fossils of extinct large frogs and snakes are alsoknown from the Caribbean. Humans have alsoselectively eliminated larger vertebrates across theglobe during the late Quaternary: so this form of fil-tering is far from being just an island pattern.

Food chain links may be important in selectingfor both smaller and larger forms. An interestingexample is Forsman’s (1991) study of variation inhead size in adder Vipera berus populations onislands in the Baltic Sea. In accord with the generaltrend for smaller vertebrates noted above, it wasfound that relative head length of adders wassmallest in the mainland population. An interestingfurther feature of this study was that, within theisland set, relative head length increased on theislands with increasing body size of the main prey,the field vole, Microtus agrestis. This was inter-preted as the outcome of stabilizing selection forhead size within each population, i.e. an evolution-ary response to the variation in body size of themain prey species and the small number of alterna-tive prey species available on islands. Schwanerand Sarre (1988) also linked large body size in asnake species to food resources. In this case, theAustralian tiger snakes (Notechis ater serventyi) on

N I C H E S H I F T S A N D S Y N D R O M E S 189

Page 203: Whittaker. Island Biogeography. 2007

Chappell Island in the Bass Strait eat an abundant,high-quality resource, which requires little effort tolocate, but which is highly seasonal: chicks of mut-ton birds. Here, larger body size might enablegreater fat storage, thus enhancing survival of indi-viduals through long periods of fasting.

The size shifts and associated morphologicalchanges, like other ‘micro-evolutionary’ processes/patterns reviewed in the present chapter, some-times lead to island populations being recognizedas subspecies, or full species, but do not necessarilydo so. The island rule is, as Lomolino (2005) writes,an emergent pattern, and although numerous par-ticular mechanisms might be involved, varyingfrom taxon to taxon, the generality of the pattern(occurring in many taxa, all over the globe) sug-gests that a relatively powerful underlying selec-tive force must apply, and that this in turn mustrelate to island disharmony. Of the various factorsdiscussed, those Lomolino holds to be most impor-tant are summarized in Figure 7.6.

Changes in fecundity and behaviour

Vertebrates colonizing islands also frequently dis-play changes in behaviour (Sorensen 1977; Stampsand Buechner 1985). For birds we have already

discussed examples of changes in feeding niche,loss of flight, loss of defensive instincts on remoteislands, and a general tendency to follow the islandrule. To provide an example of a behaviouralchange, Lack observed that birds on the Orkneyisles select a wider range of nesting sites than on theBritish mainland (Table 7.2). Lack (1947b) alsonoted that island birds frequently have reducedfecundity, expressed as smaller clutch sizes thantheir continental relatives. The same trend has beenreported for island lizards and mammals (Stampsand Buechner 1985), often combined with the pro-duction of larger offspring. Further trends that havebeen claimed for insular animals include a diminu-tion of sexual dimorphism, a tendency towardsmelanism, and the relaxation of territoriality(Stamps and Buechner 1985; Wiggins et al. 1998).

Small clutch sizesBird clutch sizes are known (1) to increase withlatitude, and (2) to decrease on islands whencompared with continental areas at the same lati-tude (Lack 1947b; Ricklefs 1980; Blondel 1985).Furthermore, clutch sizes decrease with decreasingisland area and with increasing isolation within thesame archipelago (Higuchi 1976), and at least forsome related bird species, clutch size also decreases

190 A R R I VA L A N D C H A N G E

Dw

arfi

smG

igan

tism

Rel

ativ

e im

port

ance

of f

orce

s pr

omot

ing

Body size of populations on the mainland

Immigrant selection

Intensified intraspecific competition

Ecological release from large competitors and predators

Intensified intraspecific competition

Ecological release from predators

Resource limitation and specialization for insular niches

Figure 7.6 Factors hypothesized to be important in generating the island rule in island vertebrates and how they vary in relation to body size.(Redrawn from Fig. 9 of Lomolino 2005.)

Page 204: Whittaker. Island Biogeography. 2007

with increasing endemism level of the island focalspecies. For instance, the Canarian endemic bluechaffinch (Fringilla teydea) has a mean clutch size of2.0 eggs (Martín and Lorenzo 2001), the endemicsubspecies of chaffinch Fringilla coelebs canariensis amean clutch size of 2.6 (Martín and Lorenzo 2001),whereas the northwest African subspecies ofchaffinch (F. c. africana) shows a value of 3.9 (Crampand Perrins 1994). Canarian endemic laurel-forestpigeons (Columba bollii and C. junoniae) lay one egg,instead of the two eggs laid by the native but non-endemic rock dove (C. livia). The Canarian kestrel(Falco tinnunculus canariensis) has a mean clutch sizeof 4.41 eggs, whereas Moroccan kestrels, at thesame latitude, have an average clutch size of 4.80.

Several hypotheses, more or less interconnected,have been suggested in explanation of the decreasein fecundity of island birds.

● The energy reallocation (Cody) hypothesis. Thisis a refinement of David Lack’s hypothesis on theevolutionary significance of clutch size, and hasbeen especially advocated by Cody (e.g. 1971). Itposits that greater environmental predictability onislands will lower mortality, resulting in lower pop-ulation fluctuations than in mainland regions. Thismight result in selection for smaller clutch size,with reinvestment of the saved energy in othercomponents of fitness, such as better quality ofyoung or increased longevity for parents throughbetter foraging efficiency, predator avoidance orcompetitive ability (Blondel 1985).

● The resource predictability (Ashmole) hypothe-sis. First postulated by Ashmole (1963) and sup-ported by Ricklefs (1980), this hypothesis predicts

that clutch size is determined by the differences infood resources between the non-breeding and thebreeding season, increasing the clutch size in directproportion to seasonality. In stable island environ-ments, where seasonality is buffered by the ocean,there is less seasonal fluctuation in food supply,thus favouring a diminution of the clutch size.

● The density compensation hypothesis. Islandisolation precludes the arrival of many mainlandcompetitors, which drives density compensation inthe island species (MacArthur et al. 1972), resultingin higher population sizes and thus in higherintraspecific competitiveness for resources, favour-ing higher investment in fewer young (Krebs 1970).

● The reduced predation and parasitism hypothe-sis. Island isolation prevents many groups of conti-nental predators and parasites arriving, thus theinsular bird populations have higher survival rates,and competitiveness, which results in a diminutionin clutch sizes.

Relaxation in territorialityInsular lizards, birds, and mammals often exhibitreduced situation-specific aggression toward con-specifics. This relaxation in aggressive behaviourcan be expressed in the form of: (1) reduced terri-tory sizes, (2) increased territorial overlap withneighbours, (3) acceptance of subordinates on theterritory, (4) reduced aggressiveness, or (5) aban-donment of territorial defence (Stamps andBuechner 1985). These changes are often associatedwith unusually high densities, niche expansion,low fecundity and the production of few, competi-tive offspring. Two main non-exclusive hypotheses

N I C H E S H I F T S A N D S Y N D R O M E S 191

Table 7.2 Alterations in nesting sites in some Orkney birds first noted by David Lack (after Williamson 1981, Table 6.4).In all cases the normal mode is also found in Orkney

Species Normal mode Orkney mode

Fulmarus glacialis (fulmar) On cliffs On flat ground and sand dunesColumba palumbus (woodpigeon) In trees In heatherTurdus philomelos (song thrush) In bushes and trees In walls and ditchesTurdus merula (blackbird) Woods and bushy places Rocky and wet moorlandAnthus spinoletta (rock pipit) Sea cliffs Out of sight of seaCarduelis cannabina (linnet) Bushes and scrub Cultivated land without bushes; reedy marshes

Page 205: Whittaker. Island Biogeography. 2007

have been suggested to explain this relaxation interritoriality:

● Resource hypothesis. This hypothesis suggeststhat territorial behaviour is primarily adjusted toresource densities, and that resources on islandsare, because of the lack of competitors resultingfrom isolation, more abundant than in the main-land. More resources will result in reduced territorysizes, increased territorial overlap, and changes inthe vigour of territorial defence on islands (Stampsand Buechner 1985).

● Defence hypothesis. In addition to any effects ofresources, the buffer effect exerted by the ocean onisland weather will result in milder climates andthis, together with the absence of predators, willresult in the survival of more conspecific individualswithout territory. These ‘have-nots’ will intrude andcontend with those that own territories. This will ele-vate the costs of defence for owners of insular terri-tories, selecting for reduced territories, increasedterritorial overlap, and acceptance of subordinates. Ifdefence costs became exaggerated, animals mightbenefit by expending less energy in territory defenceand reallocating their resources into producingfewer but more competitive young. Thus, thishypothesis links the observed trend of small clutchsizes of insular birds to territory size via the mecha-nism of defence costs (Stamps and Buechner 1985).

As with many of the phenomena discussed inthis chapter, these ideas are interesting, but atpresent we cannot conclude which of them mighthold greater explanatory power across differentislands.

The island syndrome in rodents

Although the various evolutionary changes in eco-logical niche in island populations have mostlybeen considered separately so far, it will be appar-ent to the reader that many of the changes evolvetogether. A proper understanding of individualchanges requires models that incorporate severalsuch phenomena simultaneously. Adler and Levins(1994) set out to construct such a model for islandrodents, synthesizing results from a range of empir-ical studies, principally of mice and voles. Theynoted that island populations of rodents tend toevolve higher and more stable densities, better sur-vival, increased body mass and reduced aggres-siveness, reproductive output, and dispersal. It isstriking that island rodent populations of differentspecies and from disparate geographic areas, oftendemonstrate similar sets of patterns. Adler andLevins termed these collective island-mainland dif-ferences the island syndrome. They summarize themain traits and some of the more likely explana-tions for them in tabular form (Table 7.3).

192 A R R I VA L A N D C H A N G E

Table 7.3 Short-term and long-term changes in island rodents and proposed explanations (from Adler and Levins 1994, Table 2)

Island trait Proposed explanation

Reduced dispersal Immediate constraint (short-term response) and natural selection against dispersers (long-term response)

Reduced aggression Initially, reduced population turnover, greater familiarity with neighbours, and kin recognition. Long-term directional selection for reduced aggression

Crowding effect Isolation (‘fence effect’ resulting from reduced dispersal) and reduced number of mortality agents such as predation, both of which result in crowding of individuals and consequently higher population densities

Greater individual body size Initially, a reaction norm as a response to higher density.Long-term directional selection for increased body size in response to increased interspecific competition

Lower reproductive output per individual Initially, a reaction norm as a response to increased density.Long-term directional selection in response to decreased

mortalityGreater life expectancy (higher survival probabilities for individuals) Reduced number of mortality agents such as predation

Page 206: Whittaker. Island Biogeography. 2007

The higher densities of rodent populations indi-cated by the crowding effect are an important part ofthis analysis. As already highlighted, because islandstend to have fewer species on them than equivalentcontinental areas, population densities will differbetween islands and mainlands, in effect as a statis-tical artefact. Williamson (1981) therefore distin-guishes the notions of density compensation anddensity stasis. Density compensation is whereisland communities have the same total populationdensity but distributed over fewer species, i.e. withlarge population sizes per species (above). Densitystatis is where the overall population of the commu-nity on the island is less than that of the referencemainland system, such that population sizes perspecies are the same as the mainland. Distinguishingbetween the two conditions is tricky and necessitatesan appropriate experimental design.

According to Adler and Levins (1994) a number ofstudies of island rodents have avoided the pitfallsidentified by Williamson, demonstrating that higheraverage densities of rodent populations do indeedoccur in some cases on islands. They cite severalexamples of experimental studies using fencedenclosures, which have shown that rodent popula-tions in such circumstances can reach abnormallyhigh densities and often destroy the food supply(the ‘fence effect’ of Table 7.3.), thereby indicatingdispersal to be an important regulator of populationdensities under normal circumstances in whichthere is no (complete) fence. As rodent populationson small islands lack an adequate dispersal sink, the

fence effect may be invoked as a force providingselective influence on the population.

One of the mainstays of island ecological analysisin recent decades has been the study of the relativesignificance of island area and isolation (Chapter 4).Adler and Levins’ (1994) scheme extends to dealingwith how these factors may work in an island evo-lutionary context (see also Adler et al. 1986). Theyhypothesize that average population densitymight, in general, increase with increasing isolationand hence with reduced dispersal, but that thisfence effect might decline with an increase of islandarea to the point where it disappears altogether asislands more closely resemble a mainland. Thus theisland syndrome might be expected to occur only inislands which are: (1) sufficiently isolated, (2) nottoo large so as to resemble mainland, and (3) not sosmall that they cannot support persistent popula-tions. Figure 7.7 illustrates the differing effects thatarea and isolation are envisaged to have. Effectswhich may be expressed largely as reaction normsin the short term i.e. within the range of phenotypesof the founding population, may, under sustainedselection pressure, produce rapid evolutionarychange and locally adapted island populations thatdiffer from mainland populations. The effect of iso-lation is a direct one, on dispersal, whereas that ofarea is less direct. Larger islands and mainlandstypically have more predators, competitors, andhabitat types, and because of this (and especiallybecause of predation effects), densities aredepressed compared with those of small islands.

N I C H E S H I F T S A N D S Y N D R O M E S 193

REDUCEDDISPERSAL

GREATERNEIGHBOURFAMILIARITY

REDUCEDAGGRESSION

MORESTABLESOCIAL

STRUCTURE

GREATERPOPULATION

STABILITY

FOUNDERPOPULATION

BETTERSURVIVAL

REDUCEDPREDATION

HIGHERDENSITY

REDUCEDREPRODUCTIVE

EFFORT

GREATERBODYSIZE

Isolation

Reducedarea

Figure 7.7 Schematic diagram showing the initial effects of island isolation and area on rodent populations. Long-term effects of insularity aredirectional selection for increased body size, reduced reproductive output, and reduced aggression (Redrawn from Adler and Levins 1994, Fig. 2).

Page 207: Whittaker. Island Biogeography. 2007

Adler and Levins’ (1994) model is based on studiesof rodents, but there is no particular reason whyit could not be extended to and tested on othersystems.

7.7 Summary

This chapter has been concerned with what wemight think of as micro-evolutionary change; theoften small, incremental alterations in genotype,and in niche, which provide the building blocks ofthe more spectacular emergent patterns. We haveconsidered essentially random processes associatedwith founding events and genetic drift, and wehave discussed at greater length an array of islandspecies traits or syndromes that appear to point togeneral selective forces operating on islands. Wehave also set out some of the many hypotheses putforward in explanation of these changes.

The founder principle is that small bridgeheadpopulations bring with them a biased subset of thegenetic variation of the source population. Suchbottlenecks appear to be important in the rapidevolutionary divergence of mainland and islandlineages, although the scale and significance offounder effects remains a contested issue. Analysesof reproductive behaviour and the sexual systemsof colonists provide further insights into evolution-ary change on islands. Hybridization appears tohave been more important in island evolution thanonce realized.

Island forms have undergone a wide variety offorms of niche alteration, including shifts andexpansions in feeding niche, alterations in nesting

sites in birds, loss of defensive attributes, and lossof dispersal powers. One of the most striking pat-terns is the so-called island rule, of size increase insmall vertebrates, and size reduction in large verte-brates. Each of these trends requires careful evalua-tion before a significant island effect is accepted. Insome cases niche shifts can be due to character dis-placement caused by competitive interactions, or toecological release in the absence of the usual arrayof competing species. Particular syndromes of traitscan be identified which can be understood in rela-tion to the geographical context of the island sys-tems under examination. For example, islandpollination networks seem to be characterized bythe emergence of endemic super-generalists, whichcan be linked to the disharmonic nature of remoteisland ecosystems.

Although much of the discussion has been incon-clusive, raising as many questions as answers, wehope that we have set out some of the more impor-tant evolutionary–ecological mechanisms, the toolkit from which island evolution has fashioned thepeculiar features of island endemics. In the finalsection, we have begun to see how these mecha-nisms can be put together into more complex theo-ries that might explain emergent patterns of islandfaunas and floras, including the most spectacularradiations of species and genera from single ances-tral colonists. To understand these in more detailwe have now to consider the process of speciation,what it entails, and how we can organize ideasabout island speciation into alternative explanatoryand descriptive frameworks. This we do in the nextchapter.

194 A R R I VA L A N D C H A N G E

Page 208: Whittaker. Island Biogeography. 2007

. . . it is the circumstance, that several of the islands pos-sess their own species of the tortoise, mocking-thrush,finches, and numerous plants, these species having thesame general habitats, occupying analogous situations,and obviously filling the same place in the natural econ-omy of this archipelago, that strikes me with wonder. Itmay be suspected that some of these representativespecies, at least in the case of the tortoise and some of thebirds, may here-after prove to be only well-marked races;but this would be of equally great interest to the philo-sophical naturalist . . .

(Charles Darwin in 1845, writing about the Galápagosarchipelago. From Ridley 1994, p. 79)

Having covered the more important micro-evolutionary processes and outcomes in the previouschapter, the aim of this chapter is to provide thebasic frameworks by which we understand macro-evolutionary patterns of change on islands. Wethus start with the nature of the species unit, andthen we outline a set of alternative frameworkswithin which we can organize ideas aboutisland evolution: specifically these frameworksare distributional, locational, mechanistic, andphylogenetic. We reserve consideration of themost striking emergent patterns of island evolutionto Chapter 9.

Evolutionary change does not necessarily implyspeciation, and in the previous chapter we focusedon the changes in niche that characterize islandforms without paying much attention to whetherthey constitute ‘good’ species or merely varieties. Infact, the determination of the species unit, althoughof central importance in ecology and biogeography,is not always clear cut. As highlighted in the abovequotation, Darwin was uncertain as to the taxo-nomic status of the members of that most emblem-atic group of island evolution, the Galápagos

finches. When he collected them he viewed the dif-ferent kinds of finches as merely variants of a singlespecies (Browne and Neve 1989). Moreover, the fullsignificance of his Galápagos collection did notimmediately strike Darwin. Rather, his key insightsinto species transmutation came some time afterthe voyage, as he was attempting to make sense ofhis observations and of the taxonomic judgementsmade of his collections. Although he was right tostate that the Galápagos biota would prove ofabiding interest whatever the taxonomic status ofthe members of the radiating lineages, it is impor-tant for the ‘philosophical naturalist’ to appreciatethe nature of the species unit.

8.1 The species concept and its place in phylogeny

No one definition has as yet satisfied all naturalists; yetevery naturalist knows vaguely what he means when hespeaks of a species.

(Darwin 1859, p. 101)

. . . it will be seen that I look at the term species, as onearbitrarily given for the sake of convenience to a set ofindividuals closely resembling each other, and that it doesnot essentially differ from the term variety, which is givento less distinct and more fluctuating forms.

(Darwin 1859, p. 108)

If we are to study distributions of organisms, letalone speciation, it is a prerequisite that we have acurrency, i.e. units that are comparable. Thus the most fundamental units of biogeography arethe traditional taxonomic hierarchies by which theplant and animal kingdoms are rendered downto species level (and beyond). So, for example,

195

CHAPTER 8

Speciation and the island condition

Page 209: Whittaker. Island Biogeography. 2007

a particular form of the buttercup can be describedby the following nomenclature:

Kingdom PlantaeDivision SpermatophytaOrder RanunculalesFamily RanunculaceaeGenus RanunculusSpecies bulbosusSubspecies bulbifer

How is this taxonomy decided? Are the unitsactually comparable? The following is a fairly typi-cal textbook definition of the species unit: ‘groupsof actually or potentially interbreeding naturalpopulations which are reproductively isolated fromother such populations’ (Mayr 1942, p. 120). This isa form of the biological species concept—promotednotably by Ernst Mayr—which has been predomi-nant for the past 60 years (Mallet 1995). Yet thequestion of ‘reproductive isolation’ (Table 8.1) isoften an unknown quantity, as populations may begeographically separated (non-overlapping distri-butions are termed allopatric). In such circum-stances they may exhibit sufficient morphologicaldifferentiation to be considered subspecies or evenseparate species, yet actually remain capable ofinterbreeding if placed into a common enclosure.Then again, species may be capable of interbreeding

in the laboratory and have sympatric distributions(i.e. occur in the same geographic area) but remainreproductively isolated in the wild because ofbehavioural differences. Sometimes a rider isadded to restrict the definition to those individualsthat can successfully interbreed to produce viableoffspring. Yet this rule too can be broken, as specieswhich are recognized as ‘good’ may interbreedsuccessfully in hybrid zones (Ridley 1996), anexample being the native British oaks Quercuspetraea and Q. robur. Studies of the genetics ofBritish populations have shown that both maintaintheir integrity over extensive areas of sympatry, butform localized ‘hybrid swarms’ in disturbed ornewly colonized habitats (White 1981). Apparentlythis is a stable situation, as Miocene fossil leafimpressions suggest that these taxa became differ-entiated no less than 10 million years BP. Some pre-fer to treat such partially interbreeding populationsas ‘semispecies’ within ‘superspecies’ (White 1981).Hybridization between closely related islandspecies appears to be fairly common (Silvertownet al. 2005), especially early on in the developmentof an adaptive radiation (Grant 1994).

For many groups, moreover, the taxonomyremains poorly known and subject to revision. Forexample, molecular analyses of the Bufo margari-tifera complex of toads in Central and SouthAmerica suggest that toads recognized by thisname actually represent composites of morpholog-ically cryptic species that are genetically unique(Haas et al. 1995). If the criterion of being reproduc-tively isolated is a difficult one to employ on livingorganisms, it is even more so with many fossilorganisms. Often the best that can be done is todetermine whether the morphological gapsbetween specimens are comparable with those forliving species that are reproductively isolated.

Thus, even the species unit, although central toso many biogeography purposes, has blurred edges(see also Otte and Endler 1989; Ridley 1996). Theunits below the species are a matter of considerabledebate and strong opinions are held that they areeither important ecologically and therefore matteror that they are arbitrary and the whim of the tax-onomist (see discussion in Mallet 1995). The signif-icance of this issue can be shown by reference to the

196 S P E C I AT I O N A N D T H E I S L A N D C O N D I T I O N

Table 8.1 A classification of isolating mechanisms in animals (fromMayr 1963, slightly modified)

1. Mechanisms that prevent inter-specific crosses (pre-matingmechanisms)

a. Potential mates do not meet (seasonal and habitat (ecological)isolation)b. Potential mates meet but do not mate (ethological isolation)c. Copulation attempted but no transfer of sperm takes place(mechanical isolation)

2. Mechanisms that reduce full success of inter-specific crosses(post-mating mechanisms)

a. Sperm transfer takes place but egg is not fertilized (gameticmortality)b. Egg is fertilized but zygote dies (zygote mortality)c. Zygote produces an F1 hybrid of reduced viability (hybridinviability)d. F1 hybrid zygote is fully viable but sterile, or produces deficientF2 (hybrid sterility)

Page 210: Whittaker. Island Biogeography. 2007

Lepidoptera: butterflies are thought to compriseabout 17 500 full species, but the number of cur-rently recognized subspecies approaches 100 000(Groombridge 1992). These problems are general totaxonomy and systematics, but have particularforce in relation to islands on which there is oftenno adequate way of testing the potential reproduc-tive isolation of geographically isolated ‘species’,‘subspecies’, or ‘varieties’ on different islandswithin an archipelago or ocean basin. Even in well-studied lineages the integrity of the species unit canbe difficult to resolve. For example, although recentphylogenies based on mtDNA sequences (Sato et al.1999) recognize 14 species of Galápagos finches(13 on the Galápagos plus the Cocos Island finch,which is descended from a Galápagos ancestor),Zink (2002) argues that the species-level taxonomywithin the genera Geospiza (ground finches, 6 species) and Camarhynchus (tree finches, 4 species)should be considered unresolved. Zink raises thepossibility that Geospiza and Camarhynchus are ineffect each functioning as a single, highly variablespecies, which would mean reducing the number ofrecognized species from 14 to 6. This taxonomicuncertainty is a product of the degree of hybridiza-tion going on in this relatively youthful radiation(Grant and Grant 1996a). Elsewhere in this volumewe follow general practice in recognizing 14 species.

In a review of the species concept, Mallet (1995)notes the prevalence, even at textbook level, of asmany as seven or eight different notions of how todefine species. This may be bewildering to many,and for the present purposes a way out is needed.This is not the place for a lengthy review of theproblems of taxonomy and systematics; the intenthas instead been to ensure that the reader appreci-ates that difficulties exist with the species conceptand its application. The solution Mallet recommendsis to return to Darwin’s pragmatic usage of ‘species’and ‘varieties’, but to make use of new knowledgefrom genetics as well as morphology in determiningwhen the species label is appropriate. Darwin’s posi-tion is made clear by the quote given above, and inthe following passage from On the origin of species:

. . . ‘varieties have the same general characters as species,for they cannot be distinguished from species, except,

firstly, by the discovery of intermediate linking forms . . . ;and except, secondly, by a certain amount of difference fortwo forms, if differing very little, are generally ranked asvarieties . . . ‘

We now recognize that there is more to it thanthis, but even with modern genetic techniques itcan be difficult to agree on what constitutes a‘good’ species.

If species are in essence arbitrary units, or ‘well-marked varieties’, then it will be apparent thatdetermining the point at which speciation hasoccurred within a radiating lineage is also problem-atic. ‘That is like asking exactly when a childbecomes an adult. I am content to know that ini-tially there was one lineage and now there aremore.’ (Rosenzweig 1995, p. 87). The debates onthis crucial matter will continue, and althoughsome may not like it, the view taken here is a prag-matic one, following Darwin and Wallace: the pre-cise level at which species are defined is arbitrary.As with other currencies, exchange rates canchange as a function of the methods and assump-tions used in the calculation.

At higher levels, species are grouped into genera,then families, and so forth. In flowering plants suchgroupings of species are based principally on theevolutionary affinities of floral structure and so it ispossible to find many different growth forms andecologies within a single family. Thus theAsteraceae (Compositae—the daisy family) all havea composite flower structure, but some are herbsand others are trees. In general, there tends to beless variation in functional characteristics within agenus than between genera (within the same fam-ily) and there is presumed to be a closer evolution-ary relationship, with all members of a genusultimately descended from a common ancestor,different from that of other genera within thesame family. Some species are so distinct that theymay be the only members of a genus or family,whereas others belong to extremely species-richgenera, e.g. figs (Ficus; Moraceae). Insular examplesinclude, at the one extreme, Lactoris fernandeziana,the monotypic representative of the endemic familyLactoridaceae from the Juan Fernández islands(Davis et al. 1995); and at the other extreme, on the

T H E S P E C I E S C O N C E P T A N D I T S P L AC E I N P H Y L O G E N Y 197

Page 211: Whittaker. Island Biogeography. 2007

Hawaiian islands, over 100 species of Cyrtandra ofthe widespread family Gesneriaceae (Otte 1989). Ingeneral, the greater the taxonomic distinctionbetween organisms, the more ancient their differen-tiation, although calibrating the timescale is diffi-cult, particularly for fossil organisms.

In recent decades the methods of traditional evo-lutionary systematics have been challenged by thedevelopment of phylogenetic systematics, orcladistics, which presumes to supply a more objec-tive means of quantifying the relatedness of a taxo-nomic group. This involves scoring the degree towhich different organisms share the same uniquelyderived characteristics that other taxa outside thegroup do not possess. These characteristics are thenused to construct a cladogram, or phylogeny (phy-logenetic tree), which is a branching sequence set-ting out the most parsimonious (‘simplest’) modelfor the relationships between taxa, i.e. it provides ahypothesis for the evolutionary development of alineage. The data employed may be morphologicalor genetic (i.e. based on DNA sequences). Thepapers in Wagner and Funk (1995) provide exam-ples of both as applied to Hawaiian plants and ani-mals, as well as an excellent explanation of themethods of cladistics.

Historical biogeographers also make use of whatare termed area cladograms. An area cladogram isconstructed by first determining the phylogeny andthen replacing the species names with the geo-graphic location in which those species are found.This new tree in effect provides a series of hypothe-ses of the sequence of dispersal and vicariance(population subdivision by barrier formation) invol-ved in the evolutionary development of the line-age, emphasizing the role played by geological/environmental history in lineage development.

Although the proponents of phylogeneticsystematics are often forthright in praising theapproach and condemning traditional systematics,it should be recognized that the decisions as towhich characters to include in the phylogeny, andthe rules used in its construction, are the choice ofthe user. The phylogeny thus constructed is there-fore merely one approximation, albeit hopefully abasically reliable one, to the history of events thatproduced the lineage. Once such a phylogeny has

been constructed, it can then be used to explore thelikely sequence of interisland and intraisland speci-ation events within archipelagos, or betweenislands and mainlands.

The most exciting opportunities have beenopened up by advances in molecular biology andgenetics and these methods have been seized uponin island evolutionary studies. Such studies benefitfrom use of the ‘genetic clock’ which, on theassumption of a more or less constant rate ofaccumulation of mutations and thus of geneticdifferences between isolates, allows estimates of thedates of lineage divergence (Box 8.1). An excellentillustration is provided by the studies of Thorpeet al. (1994) on the colonization sequence of thewestern Canarian lizard, Gallotia galloti, in whichthe direction and timing of colonization as postu-lated by genetic clock analyses (nuclear and mito-chondrial DNA divergence) was shown to beentirely compatible with the independentlyderived geological data for the timing andsequence of island origins.

It should be noted that these molecular clocksmay not always run to time. A key problem isexemplified by Clarke et al.’s (1996) study of twospecies of land snails (Partula taeniata and P. sutu-ralis) on the island of Moorea (French Polynesia). Itwas found that although the two species exhibitsignificant morphological and ecological differ-ences, they are genetically close. Clarke et al. attrib-uted this closeness to ‘molecular leakage’ or‘introgression’: the convergence of genetic structurethrough occasional hybridization. Such introgres-sive hybridization is now considered to be quitecommon in particular island lineages (Clarke andGrant 1996). Even low rates (as low as 1 in 100 000)may be enough to upset the phylogenetic trees andmolecular clocks upon which scenarios of islandevolution discussed in Chapter 9 now rest (Clarkeet al. 1996; Zink 2002). For further insights into theseand other methodological problems of biogeogra-phy and systematics, see Myers and Giller (1988),Thorpe et al. (1994), Ridley (1996), the exchangebetween Herben et al. (2005) and Silvertown et al.(2005); and for some reassurance about the use ofmolecular clocks, see Bromham and Woolfit (2004)and Box 8.1.

198 S P E C I AT I O N A N D T H E I S L A N D C O N D I T I O N

Page 212: Whittaker. Island Biogeography. 2007

8.2 The geographical context ofspeciation events

Distributional context

In order to appreciate the full gamut of possibilitiesinvolved in speciation it is helpful, if slightlyartificial, to distinguish the geography of specia-tion from the mechanism (Table 8.2; Box 8.2). Thegeographical context can be viewed either indistributional or locational terms, i.e. distinguish-ing the degree of overlap of populations involvedin speciation events on the one hand, from the

issue of where the evolutionary change is takingplace.

The terms sympatry and allopatry denote,respectively, two populations (or species) whichoverlap in their distributions and two populations(or species) which have geographically separatedistributions. In geographical terms, a new species(or subspecies) may, in theory, arise in one of threecircumstances. If the new form arises within thesame geographical area as the original, then it maybe termed sympatric speciation, if it arises in azone of contact (hybrid zone) between two species

T H E G E O G R A P H I C A L C O N T E X T O F S P E C I AT I O N E V E N T S 199

Box 8.1 Molecular clocks

Dating events in evolutionary biogeographypresents considerable challenges because of theextended time periods involved, extending waybeyond many conventional dating techniquessuch as radiocarbon isotopes. Traditionally, fossilsare dated by means of the geological record andtheir place in the stratigraphic column, but thiscan be a very imprecise approach and doesn’tprovide a means of dating colonization andspeciation events in extant taxa. Molecularclocks provide a solution to this problem basedon the idea that proteins and DNA evolve at amore or less constant rate. The extent ofmolecular divergence is then used as a metric forthe timing of events within the development ofthe lineage.

Geneticists have developed an impressive arrayof rather complicated analytical procedures toensure that such molecular clocks are reasonable.Nonetheless, they are inevitably based on limitedamounts of information and involve a number ofimportant assumptions. They should be regardedas providing plausible hypotheses for thedevelopmental sequence and timing of particularmonophyletic lineages. Greater confidence can beplaced in such molecular clocks if they can beindependently calibrated.

In illustration, Carranza et al. (2006) havemade use of a molecular clock in their study of

the phylogeography of the lacertid lizardPsammodromos algirus in the Iberian peninsularand across the Strait of Gibraltar in North Africa.They sampled genomic DNA for three genesfrom 101 specimens of the subfamilyGallotiinae, mostly of the target species, butincluding specimens of Gallotia species from theCanaries. These specimens provide a means of‘rooting’ the phylogenetic tree, and also ameans of calibrating it. This is because it isreasonable to assume that G. caesaris caesaris,endemic to the island of El Hierro, commenceddivergence from its nearest relative, G. c.gomerae, endemic to the island of La Gomera,shortly after the formation of El Hierro(approximately 1 Ma). The clock used byCarranza et al. indicated that diversification inthe genus Gallotia began on the Canaries about13 Ma, with a rather earlier data of 25 Ma forthe first speciation events within theIberian/North African genus Psammodromus.

This case study is illustrative of a growingnumber of studies that place the development ofa monophyletic island lineage into contextalongside related mainland lineages. This body ofwork has produced some interesting surprises,for instance showing evidence of successfulback-colonization of island forms into mainlandregions (e.g. Nicholson et al. 2005).

Page 213: Whittaker. Island Biogeography. 2007

200 S P E C I AT I O N A N D T H E I S L A N D C O N D I T I O N

Table 8.2 A simplified framework for speciation patterning

Form Pattern

DistributionalSympatric OverlappingAllopatric SeparateParapatric Contiguous

Locational (and historical)Neo-endemic Change on islandPalaeo-endemic Island form relict

MechanisticAllopatric Drift and selectionPolyploidy Changes in chromosome numberCompetitive speciation Other sympatric mechanisms

Tree form (phylogeny)Anagenesis Replacement of originalAnacladogenesis Alongside originalCladogenesis Lines diverge and replace original

Box 8.2 On archipelago speciation, allopatry and sympatry—a clarification

We provide this simple summary as a quick look-up and clarification of the ideas involved.

● Allopatric speciation is speciation occurringbetween two (or more) populations that are geographicallyisolated from each other. All island neo-endemics exemplifyallopatry in the sense that they are isolated from the main-land source pool. However, within an archipelago, popula-tions may also be allopatric at the inter-island scale, oreven within islands, for example, isolated within forestislands surrounded by lava (kipukas, Hawaii).● Sympatric speciation is a term applied to caseswhere speciation occurs without geographical separationof the populations involved, i.e. the opposite of allopatricspeciation. In practice, the past roles of allopatry and sym-patry can be difficult to distinguish at the intra-islandscale, or even at the intra-archipelago scale where theislands are close to one another and the organismsinvolved have high dispersal abilities.● Parapatric speciation refers to situations interme-diate between the extreme cases of allopatry and sympa-try, where the geographical segregation is incomplete, andtwo separating forms are contiguous in space with ahybrid zone in the area of contact/overlap.

● Archipelago speciation is a term sometimes usedwhere speciation of island lineages has occurred mostlybetween islands rather than within them. This tends to implyan allopatric model, especially where inter-island distancesare considerable, but may in fact involve both allopatric andsympatric phases. The taxon cycle model, as applied to theWest Indian avifauna, is also an example of archipelago spe-ciation. Sometimes, environmental differences across anarchipelago such as the Canaries are great enough thatadaptive differences are evident, but sometimes verylittle niche segregation is evident, in which case the archipel-ago speciation might be considered non-adaptive.

● Are these terms operational? Given the hugeenvironmental changes involved in the life cycle of isolatedvolcanic islands, and archipelagos, it can be very difficult tobe sure that current distributions indicate the circum-stances in which speciation occurred. Indeed, taking theGalápagos finches as a classic case study of island evolu-tion, it appears likely that this adaptive radiation hasinvolved allopatric, sympatric and parapatric phases(Chapter 9). Similarly, the distinction between adaptive andnon-adaptive changes may not always be clear cut: evolu-tionary change commonly involves a mix of stochastic anddirectional mechanisms.

it may be termed parapatric speciation, and if itarises in a separate geographical area it maybe termed allopatric speciation. In reality the dis-tinction between these conditions may be blurredas: (1) the spatial scale we choose to describe thegeography of the populations may not be concor-dant with the scale at which the members of thepopulations interact; and (2) we may be whollyignorant of the past history of the distributionaloverlap of the populations such that, for instance, acurrently sympatric species pair may well havespeciated while geographically isolated from oneanother.

As evident from the debate concerning vicari-ance vs dispersalist explanations in biogeography(Chapter 3), many authors have assumed thatallopatric speciation is by far the dominant modeof speciation. The demonstrable occurrence ofsympatric forms of speciation exposes the weakness

Page 214: Whittaker. Island Biogeography. 2007

of making such assumptions (Sauer 1990). Theproduction of island neoendemics of course occursin isolation from the original continental sourcepopulation, and to that extent can be considered tohave occurred allopatrically. However, muchisland speciation has occurred in the condition ofisland-scale sympatry (i.e. differentiation of twoforms occurring on one island), but on the finerwithin-island scale, this still allows for differingdegrees of population overlap (Keast and Miller1996, p. 111).

Diehl and Bush (1989) suggest that populationsutilizing different, spatially segregated habitatsshould nonetheless be considered sympatric whenall individuals can move readily between habitatswithin the lifetime of an individual. This must bejudged separately for different taxa as, for example,a within-island barrier for land snails may not be abarrier to bird populations (Diamond 1977), whichin cases might move freely even between separateislands within an archipelago. Indeed, the distinc-tion between allopatry and sympatry is somethingthat varies substantially within higher taxa. In illus-tration, the Krakatau islands, isolated from themainlands by about 40 km, are beyond the appar-ent colonization limits of certain plant families(Whittaker et al. 1997), yet for some species ofplants and their insect pollinators, the island popu-lations can be considered as freely interbreeding(panmictic) with populations on the mainland(Parrish 2002). Within large oceanic archipelagos itis probably commonplace for large radiations ofmonophyletic lineages to have involved a mix ofsympatric and allopatric speciation events, as forexample, appears to apply to the Hawaiian fruitflies Drosophila (Tauber and Tauber 1989) andSarona (a genus of phytophagous insects) (Asquith1995).

Locational and historical context—island ormainland change?

We have previously introduced the distinctionbetween neoendemic forms (those that haveevolved in situ on the island) and palaeoendemicforms (which have become endemic because of theextinction of the same or closely related forms

from mainland populations). This general ideawas recognized by A. Engler as long ago as 1882and is discussed in two papers by Cronk (1987,1990). Cronk compared the two species of theendemic St Helenan genus Trochetiopsis (redwoodand ebony) with other species in the same family(Sterculiaceae), based on the morphology of pres-ent-day pollen and ancestral pollen in sediments.From this he concluded that the ancestors of thepresent species probably arrived by long-distancedispersal during the Tertiary period, from Africanor Madagascan stock (Fig. 8.1). The lineage hassince diverged into two forms, but the degree towhich they differ from their mainland relatives isonly partly explained by in situ change. Evolutionand extinction elsewhere among the branch towhich the ancestor belonged has meant that therest of the family has effectively evolved awayfrom the St Helena genus over the period sincecolonization.

The key point is that in attempting to understandevolution on islands, we cannot assume that overtimescales of millions of years, evolution on thecontinents that supplied the initial island colonistshas somehow stood still. We cannot therefore inter-pret all island/mainland differences, and in partic-ular all insular endemism, as necessarily a functionof evolutionary change on the island in question.‘There is no doubt that adaptive radiation occurs onislands, but this may be interpreted as the elabora-tion of a relict ground-plan, without requiring thatground-plan to have arisen by evolution on theisland’ Cronk (1992, p. 92).

Paulay (1994) comments that relict taxa are wellknown on large, old, continental fragment islands,the lemurs of Madagascar being classic examples.The primitive weevil family Aglycyderidae of bothPacific and Atlantic islands provides another case ofancient forms restricted to islands. Although short-lived oceanic islands may not be thought to be idealfor long-term survival of relict lineages, if thespecies involved have good dispersal abilities theymay survive by island hopping. Thus three of thesix families that comprise the most primitive orderof pulmonate land snails are restricted to islands,yet representatives may be found on even the mostremote Pacific islands, strongly suggesting their

T H E G E O G R A P H I C A L C O N T E X T O F S P E C I AT I O N E V E N T S 201

Page 215: Whittaker. Island Biogeography. 2007

distributional boundaries to be set by biologicalfactors rather than an inability to reach mainlandareas.

The labelling of particular species or traits as‘relictual’ and the relative importance of palaeo- andneoendemism on islands has proven controversial(Chapter 2; Bramwell 1979; Cronk 1992; Elisens1992; Carlquist 1995). One example discussed inChapter 7 is the question of whether woodinessof island forms of normally herbaceous taxacould be considered relictual (basal) or a derivedcharacteristic. The application of modern genetictechniques argues convincingly for the latter (Kimet al. 1996; Panero et al. 1999). In practice, as moremodern genetically derived phylogenies becomeavailable it seems likely that the dichotomy betweenpalaeo- and neoendemic forms will becomeincreasingly irrelevant. Instead, we will be able tocompare the complex, and often individualistic,patterns of lineage development across and betweenislands, with recent work even pointing to cases ofback-colonization of island forms to mainlands (e.g.Carine et al. 2004; Nicholson et al. 2005).

8.3 Mechanisms of speciation

When examining the mechanisms of speciationrather than focusing on the geographical circum-stances, it becomes clear that there are differentprocesses at work. Rosenzweig (1995) summarizesthese under the three headings: geographical orallopatric speciation, competitive speciation, andpolyploidy (involving an increase in the chromo-some number). Each form holds relevance to theisland context, but as insularity defines the islandcondition, the allopatric form of speciation will bedealt with first.

Allopatric or geographical speciation

The essence of this model has been summarized byRosenzweig (1995, p. 87):

● A geographical barrier restricts gene flow withina sexually reproducing population.● The isolated subpopulations evolve separatelyfor a time.

202 S P E C I AT I O N A N D T H E I S L A N D C O N D I T I O N

St. Helena

Trochetiopsis

Dombeya

Trochetia

Paradombeya

Pentapetes

Melhania

Harmsia

Cheirolaena

Astiria/Ruizia

Dendroleandria/Paramelhania

Figure 8.1 Distribution of genera in the subtribe Dombeyeae (Sterculiaceae). The closest relatives of the Trochetiopsis of St Helena are Trochetia(Mauritius) and Dombeya. The centre of generic diversity is in Madagascar and the Indian Ocean islands. According to Cronk, present-day oceancurrents (arrows) differ from those predominant at the time the ancestral Dombeyeae stock colonized St Helena during the late Tertiary, whenthere was a stronger flow from the Indian Ocean to the South Atlantic. (Redrawn from Cronk 1990, Fig. 3, with permission of the NewPhytologist Trust.)

Page 216: Whittaker. Island Biogeography. 2007

● They become unlike enough to be called differentspecies.

● Often the barrier breaks down and the isolatesoverlap but do not interbreed (or they interbreedwith reduced success).

The first step of this model suggests a vicariantevent (barrier formation). This scenario applies tocertain ancient continental fragment islands (e.g.Madagascar) and to former land-bridge islands. Insuch cases, the island population is assumed not tohave experienced a bottleneck founding event.For true oceanic islands, the starting point isdifferent: dispersal across a pre-existing barrier.The founding population may be quite small, andthus the founding event may involve a bottleneckeffect (Chapter 7). In either scenario, consideringa now remote island population, it develops ineffective isolation from the mainland source popu-lation. Subsequent environmentally-determinedbottlenecks, genetic drift effects and selectivepressures in the novel island context then providethe engine for further differentiation from the main-land form.

The failure of the isolating barrier may not be afrequent feature for oceanic islands, but can hap-pen, for instance, as a function of sea-level changeproviding ‘stepping stones’ between island andmainland. This can lead to double invasions,where a second invasion of the mainland formoccurs notwithstanding the continued existence ofan oceanic barrier. This was the hypothesis putforward to explain the occurrence of two speciesof chaffinches on the Canaries (Lack 1947a). Theendemic blue chaffinch (Fringilla teydea) was con-sidered to have evolved from an early colonizationevent, to a point where it was sufficiently distinctto survive alongside a later-arriving populationof mainland common chaffinches (F. coelebs).More recent genetic work based on mitochondrialDNA has supported this basic idea, with aninteresting twist, as it seems likely that the latercolonizing F. coelebs reached the Canaries via theAzores and Madeira (Marshall and Baker 1999).As often the case with such data, the resultsand interpretation are not straightforward, but thiswas regarded as the most likely sequence, with

strong support also for a back-colonization eventfrom the Canaries to Madeira. Hence, it seemslikely that the sequence of population movementsamongst the islands was rather more complex thanimplied by the double invasion model. Similarcomplexity of movements may be necessary toaccount for other Atlantic island phylogenies(e.g. Kvist et al. 2005).

Interestingly, available evidence shows that thegreatest degree of radiation within the Macaronesianflora can be attributable to monophyletic genera,with paraphyletic lineages paradoxically producingfewer endemic species. Silvertown (2004) reportedthat 20 monophyletic endemic genera account for269 endemic plant species, while 20 paraphyleticlineages account for just 38 species. Double ormultiple invasions occur within the monophyleticlineages, but the movements occur at the intra-Macaronesia scale rather than in the sense ofmainland–island colonization events. Silvertownattempts to explain this paradox in terms of a com-petitive mechanism involving niche pre-emption(see discussion in: Silvertown 2004; Saunders andGibson 2005; Silvertown et al. 2005).

The barriers that exist between islands within anarchipelago are undoubtedly important features ofremote oceanic island groups. As Rosenzweig(1995, pp. 88–9) puts it ‘Suppose propagules occa-sionally cross those barriers, but usually afterenough time for speciation has passed. Then theregion and its barriers act like a speciation machine,rapidly cranking out new species.’ The radiationsthat characterize many taxa on the Hawaiianislands reflect this, with interpretations for the rela-tionships among the 700 (or more) species ofDrosophila involving a number of phases of interis-land movements as a part of this most spectacularof radiations (e.g. Carson et al. 1970; Carson 1992).Radiation on archipelagos is examined further inthe next chapter.

An illustration of the contrast between the archi-pelago context and the single remote island is pro-vided by that most famous of examples of islandevolution, Darwin’s finches. The Geospizinae arerecognized as a distinct subfamily, found in theGalápagos archipelago, which comprises nineislands larger than 50 km2 (and a larger number of

M E C H A N I S M S O F S P E C I AT I O N 203

Page 217: Whittaker. Island Biogeography. 2007

smaller islets), with predominantly desert-like con-ditions. Here they have radiated into 13 differentspecies (Sato et al. 1999; but see Zink 2002). Thefamily is found also on Cocos Island, a lone,forested island of about 47 km2, which is, just, thenearest land to the Galápagos. Here there has beenno possibility of the island hopping that appears tocharacterize archipelago speciation, and the groupis represented by a single endemic speciesPinaroloxias inornata, one of just four species of landbird (Williamson 1981; Werner and Sherry 1987).Analyses of mtDNA data indicate that the Cocosfinch is probably a descendent of a Galápagos treefinch (Sato et al. 1999).

Yet, as Sauer (1990) cautions, within an archipel-ago, in addition to single species being present ondifferent islands, quite commonly several siblingspecies occur on a single island, where they aremainly segregated by habitat. In many instances,isolation may have been only indirectly relevant as,by screening out competitors, isolation offersopportunities for adaptive radiation to thosespecies that do arrive. Some, at least, of this radia-tion may take the form of the next class of mecha-nism, competitive speciation.

Competitive speciation

This is the term that Rosenzweig (1978, 1995) givesto cover a variety of related modes of sympatricspeciation. The idea here is that a species expandsits niche to occupy an unexploited ecologicalopportunity, followed by that species’ sympatricbreak-up into two daughters, one in essence occu-pying the original niche, and the other the newlyexploited one. The expansion happens because of alack of competition with other species, a featurecommon to remote island faunas and floras. Butthe break-up into separate species happensbecause of increased competitive pressure betweenthose portions of the population best able toexploit the two different niches. The operation ofthese processes must take place over the courseof many generations, and observation of the com-plete process in the field is thus not an option.Rosenzweig illustrates the idea by means of a

‘thought experiment’, the essence of which, placedin the island context, can be rendered as follows.

A colonizing species of bird will have a particularfeeding niche, determined morphologically andbehaviourally, and perhaps most clearly expressedby features such as bill size and shape (Lack 1947a;Grant and Grant 1989). According to theory, suchcharacters would be expected to have a unimodaldistribution of values. The environment of theisland contains empty niches, unexploited ecologi-cal opportunities, which for the purpose of ourexperiment are taken to be distributed bimodally,i.e. there are two resource peaks. The species,unconstrained by competition with the full range ofmainland forms, expands from its original modalposition somewhere along this resource space sothat it occupies each of these niches. Individualswith genotypes that match best to one or the otherof these niches become numerically dominantwithin the population, which has by this stagegrown close to its carrying capacity. At this point,disruptive selection kicks in, such that the mid-range phenotypes lose fitness relative to those thatare better suited to one or the other of the mainfeeding niches. Those individuals adapted prima-rily to a peak in the resource curve may also exploitthe valleys either side of that peak. If they do so atall successfully, no opportunity remains for valleyphenotypes in between the two peaks. They will befew in number in the population, and their off-spring will have fewer resources to tap. Althoughthey may also reach up to the peaks, they will notbe effective competitors in those portions of theresource continuum. Valley genes will thus have lit-tle success and will be bred out, as in time membersof each ‘peak’ population that can recognize othersof their type (and breed with them alone) will be ata selective advantage. Therefore, isolating behav-ioural mechanisms develop and at this point thelineage can be viewed as having split.

If neither resource peak matches closely the orig-inal niche of the species, the divergence may resultin two novel forms and the effective disappearanceof the original species (termed anagenesis: seebelow). There may also be more than two resourcepeaks, thus accommodating a larger radiation.

204 S P E C I AT I O N A N D T H E I S L A N D C O N D I T I O N

Page 218: Whittaker. Island Biogeography. 2007

This idea, of ‘adaptive landscapes’, was originallyintroduced by Sewall Wright (1932) and has beenincorporated into a number of different evolution-ary models (see e.g. Templeton 1981, Futuyma1986, Otte and Endler 1989).

Intentionally, the account given here is a simplerendition of ideas culled from a much larger bodyof literature. Based on laboratory evidence, compet-itive speciation may occur in as few as 10–100generations, not as fast as polyploidy (almostinstantaneous), but faster than geographical specia-tion, which seems to require thousands or evenhundreds of thousands of years (Rosenzweig 1995).

Grant and Grant (1989) recorded an early phasein the process of sympatric speciation in a study ofa temporary reproductive subdivision withinGeospiza conirostris, one of the Galápagos (Darwin’s)finches occurring on Isla Genovesa. The populationwas partly subdivided ecologically for a briefperiod, but this division then collapsed againthrough random mating. All of the known sym-patric species of ground finches differ by at least15% in at least one bill dimension, suggesting thatthis may indicate a threshold in niche separationneeded to sustain separate species. During thetemporary subdivision, the two groups of malesdiffered by a maximum of 6% in bill dimensions,and this, occurring as it did over but a brief period,was insufficient to foster any discrimination in amating context. One of the dry-season food nichessustaining the division declined catastrophicallyand the division collapsed. Nonetheless, this studyshows how sympatric division may be fostered intaxa in which polyploidy does not occur. Grant andGrant would have had to be extraordinarily luckyto come along just at the right time to record a sym-patric split that ‘stuck’. They therefore did notregard the failure of this event as evidence againstthe idea, which they concluded to have some rele-vance for vertebrates. Rather, they took it to indi-cate that in order to foster sympatric divergenceleading to speciation, the niches or habitats towhich different members of a population adaptmust be markedly different and display a long-term persistence. This is more likely to occur onlarger, higher islands.

How important is competitive speciation in theisland context? This is difficult to judge as it cannotbe decided merely by investigating the degree ofgeographical overlap between sibling species inecological time. This is because, given the proba-bilistic nature of dispersal and the magnitude ofpast environmental change, populations currentlyisolated from one another may once have beensympatric and vice versa. Indeed, it may commonlybe the case that archipelago speciation—diversifi-cation within archipelagos—involves a mix ofallopatric and sympatric phases of subtly varyingcombination from one branch of a lineage to thenext (cf. Clarke and Grant 1996). Over the lifespanof a large island, the same may also apply within asingle island. The relative roles of allopatric andcompetitive speciation within the island contextcannot therefore be quantified, but one thingremains clear; isolation is crucial, either directly orindirectly (respectively).

Polyploidy

One important class of sympatric speciation isthrough polyploidy, a condition comparativelycommon in plants and many invertebrates, but notin higher animals: it is unknown in mammals andbirds, for instance (Grant and Grant 1989;Rosenzweig 1995). Polyploid species are those thathave arisen by an increase in chromosome number.Two main forms of polyploid can be distinguished.If the new species has twice the chromosome com-plement of a single parent species it is termed anautopolyploid. If it has the chromosomes of both oftwo parent species (i.e. a crossing of lineages) it iscalled an allopolyploid. According to Rosenzweig(1995), autopolyploids are rare, but at a conserva-tive estimate 25% or more of plant species may beallopolyploids. Barrett (1989) states that currentestimates for the angiosperm subset are as high as70–80% of species being of polyploid origin.

Rosenzweig (1995) presents an intriguing analysisof the proportionate importance of polyploids indifferent floras, showing that there is a generaldecrease in the relative contribution of polyploids infloras with decreasing latitude. The best trend was

M E C H A N I S M S O F S P E C I AT I O N 205

Page 219: Whittaker. Island Biogeography. 2007

found for continental floras; for islands there wasmuch more scatter in the relationship. Rosenzweigdoes not distinguish the nature of the islands in hisanalysis, i.e. whether they were predominantlyoceanic or continental in origin, and this may be aconfounding factor in his analysis. It is necessary tolook a little further into the data to establishwhether polyploidy has provided a significant con-tribution to island evolution and whether it hasdeveloped in situ. The following studies are cited inorder of increasing proportion of polyploidy.

Stuessy et al. (1990) examined about one third ofthe 148 endemic plants of the Juan Fernándezislands (33�S), which are volcanic, oceanic islandsoff the coast of Chile. Remarkably, none of thespecies examined was found to be polyploid.Borgen (1979) reports 25.5% polyploidy among360 Macaronesian endemic plant species from theCanaries (28–29�N). This is a low proportion incomparison to most known floras, and is signifi-cantly lower than the 36.4% polyploids amongst the151 non-endemic Canarian species for which datawere available (Borgen 1979). It appears also thatseveral of the polyploid events were fairly ancientones, subsequent to which gradual speciationwithin certain polyploid lineages has occurred (anexample being the genus Isoplexis in theScrophulariaceae), but to a lesser degree thanwithin diploid lineages (Humphries 1979). It maybe concluded that polyploidy has not been a majorevolutionary pathway in Macaronesian lineages.The New Zealand (c.35–47�S) flora consists ofapproximately 1977 species, 40% of which areknown cytologically according to Hair (1966).Polyploidy is absent in the gymnosperms, but char-acterizes 417 (63%) of the 661 angiosperms investi-gated, a figure approaching the 70–81% polyploidysuggested for Oceanic–Subarctic and Arctic areas ofhigh northern latitudes. New Zealand is an ancientarchipelago, and its flora has been claimed to havebeen primarily of continental origin, relictual ofGondwanan break-up. However, this has been dis-puted by Pole (1994), who argues against the large-scale continuity of lineages in the New Zealandflora, claiming that palynological data support

derivation of the entire forest flora from long-distance dispersal (predominantly from Australia).This remains a hotly contested issue (see e.g. Heads2004, Cook and Crisp 2005, McGlone 2005) makingit difficult to put a time frame on the interpretationof the New Zealand data.

Few firm conclusions can be drawn from the datacurrently available. Polyploidy is more important insome island floras than others. The latitude and ageof the islands in question may each be of relevance,although often, as shown, age and origin of the floraare imperfectly known. If the Macaronesian datacan be taken as a good guide, it appears that themore spectacular radiations of island plant lineagesdo not tend to rely upon polyploidy.

8.4 Lineage structure

A number of different models and facets of islandevolution have been introduced in this chapter. Toomany frameworks may serve to confuse, but thereis one more that appears particularly helpful inmaking sense of island speciation patterns, andalthough the terms may not be instantly memo-rable, the ideas seem useful. In some cases, evolu-tionary change takes the form essentially of thecontinuation of a single lineage, whereas in othercases the splitting of lineages is involved. Stuessyet al. (1990) provide three terms that codify whatthey see as the chief outcomes (Fig. 8.2).

● Anagenesis is when the progenitor species/formbecomes extinct.● Anacladogenesis is when the progenitor sur-vives with little change alongside the derivedspecies. What is meant by alongside is open to dif-ferent interpretations—it could be within the sameisland, or less restrictively, within the same archi-pelago.● Finally, cladogenesis is where the progenitor ispartitioned into two lines and becomes extinct in itsoriginal form, a model suggestive of the classicexamples of adaptive radiation. This frameworksets aside cases where lineages converge or cross,via hybridization.

206 S P E C I AT I O N A N D T H E I S L A N D C O N D I T I O N

Page 220: Whittaker. Island Biogeography. 2007

8.5 Summary

In order to understand evolutionary change it isnecessary to have a basic familiarity with the cur-rency of evolution and knowledge of the problemsassociated with it. This chapter therefore begins bydiscussing some of the problems of the speciesconcept, the arguably arbitrary distinction betweenspecies and varieties and the higher-order phyloge-nies of relationships that can be constructed for -particular lineages. Recent developments inphylogeny, including the use of cladistic andmolecular biological techniques, have providedmany new insights into long-standing islandbiogeographical issues, although such techniquesshould be viewed with proper caution as merelyconstructing realistic hypotheses of the unknownreal genealogies.

A distinction is drawn between distributional,locational (and historical), mechanistic, and phylo-genetic frameworks for describing island evolution.It is shown that divergence can take place boththrough isolation (allopatry) and in conditions of

distributional overlap (sympatry), and indeed thatthese concepts can be applied on different scales:between island and mainland populations, withinan archipelago, and within an island. Evolutionarychange within a lineage may be accomplished by avariety of genetic mechanisms, including hybridi-zation and polyploidy, and by more subtle alter-ations following from founding events, geneticdrift, and natural and sexual selection in the novelbiotic conditions of oceanic islands. Sympatricspeciation may occur through more than one route.Perhaps the classic island examples are those thatfollow the competitive speciation model of nicheexpansion and break-up in to daughter populationsexploiting different resource peaks. Polyploidydoes not seem to have been a major route of speci-ation on islands, although it undoubtedly occurs.The final framework considered in the chapter isthat of phylogenetic trees, which can be used todenote whether lineages develop without branch-ing (anagenesis), with branching and extinction ofthe original form (cladogenesis), or by branchingand survival of the original line (anacladogenesis).

S U M M A RY 207

5%

24%71%ANAGENESIS

(progenitorbecomes

extinct)

CLADOGENESIS(progenitor ispartitioned intotwo lines andbecomes extinct)

ANACLADOGENESIS(progenitor surviveswith little change)

Figure 8.2 General patterns ofphylogeny in the endemic vascularflora of the Juan Fernándezislands. The figures give thepercentage of each form oflineage development evident in123 endemic plant species. (Fromdata in Stuessy et al. 1990.)

Page 221: Whittaker. Island Biogeography. 2007

Studies of island biotas are important because the rela-tionships among distribution, speciation, and adaptationare easier to see and comprehend.

(Brown and Gibson 1983, p. 11, citing one ofA. R. Wallace’s general biogeographic

principles)

Having considered the evidence for micro-evolutionary change on islands and the frameworksof speciation in the previous two chapters, we nowexamine how these building blocks fit together toproduce neoendemics on islands. Although othercategorizations are possible, we identify the follow-ing emergent patterns or models of evolutionarychange on islands. The first is anagenesis, whichrefers to speciation without much radiation of line-ages, and which may be most applicable to singleisolated islands of moderate size and limited habi-tat heterogeneity. The second and third models, thetaxon cycle and adaptive radiation, find their mostdramatic illustrations within archipelagos and con-stitute multiagent evolutionary–ecological models:both stress adaptive changes. In contrast, the notionof non-adaptive radiation refers to the diversifica-tion of monophyletic lineages without apparentniche alterations, and emphasizes geographic isola-tion within or between islands.

Until comparatively recently, these scenariosrelied heavily on distributional data (sometimeswith supporting ecological data), for species classi-fied using traditional taxonomic systematics. Overthe last two decades, numerous studies have appliedmodern phylogenetic techniques to the descriptionand analysis of island lineages, allowing the con-struction of scenarios for the historical patterns ofevolutionary and distributional change on a lineageby lineage basis, at scales ranging from intraislandto across whole ocean basins. These scenarios are

based around genetic relationships rather thanadaptive characteristics. We consider some exam-ples of these scenarios in the section on islandhopping radiations. We also draw attention to theimportance of changes to the platforms on whichisland evolution takes place and the implications ofthe life cycle of volcanic islands for patterns of phy-logenesis. The final part of the chapter attempts toplace the principal island evolutionary concepts andmodels into a simple framework of land area versusisland or archipelago isolation.

9.1 Anagenesis: speciation with littleor no radiation

The Juan Fernández archipelago contains two mainislands, Masatierra (4 Ma) and Masafuera (1–2 Ma),located respectively some 670 km and 850 km westof mainland Chile. Of the native vascular plants,18% of the genera and 67% of the 148 species areendemic to the archipelago. There is even onemonotypic endemic family, the Lactoridaceae, rep-resented by its only species Lactoris fernandezianaStuessy et al. (1990) estimate that the presence of the69 genera with endemic taxa on these islands maybe explained by 73 original colonization events: inmost cases only a single introduction per genus. Onthe basis of phylogeny and distribution of theendemic flora, Stuessy et al. (1990) have attributed71% of endemic species to the anagenesis model, inwhich the progenitor form becomes extinct; 24% tothe anacladogenesis model, where the progenitorsurvives little changed while a peripheral or iso-lated population diverges rapidly (perhaps whileentering new habitats); and 5% to the cladogenesismodel, in which the lineage divides into two linesand the original form fails to survive (Fig. 8.2).

208

CHAPTER 9

Emergent models of island evolution

Page 222: Whittaker. Island Biogeography. 2007

The significance of anagenesis as a model of islandevolutionary change may well have been over-looked given the obvious fascination provided byspectacularly branching lineages like the Hawaiianhoneycreeper-finches. In the absence of systematicmolecular data, a simple yet fairly conservative basisfor estimating the proportion of speciation eventsfitting the anagenesis model is to count the numberof cases where a genus is represented on an island/archipelago by a single endemic species. On thisbasis, Stuessy et al. (2006) estimate that for 2640endemic angiosperm species from 13 island systems,about one quarter of speciation events match theanagenesis model.

Anagenesis describes an emergent phylogeneticpattern and does not imply a tightly specifiedexplanatory model. For instance, a single endemicon an island might arise through a combination of apronounced founder effect followed by genetic driftof the newly isolated island population. On theother hand, the colonist populations may be subjectto strong directional selection pressure in the novelbiotic and abiotic environment provided by aremote island, and show apparently clear adaptivefeatures in response. There is insufficient geneticdata to reveal whether species conforming to thismodel typically show unusual levels of genetic dif-ference and thus whether chance or selection domi-nates. Where anagenesis is important on an island itindicates that intraisland vicariant events and com-petitive speciation have each been relatively infre-quent, such that the motor for speciation is thefurther arrival of new colonists combined with envi-ronmental novelty and possibly environmental (e.g.climatic) change through time. It seems likely thatanagenesis will be of greatest relative importance onislands of relatively limited environmental ampli-tude (Stuessy et al. 2006), and especially where suchislands are themselves isolated (as the case of theJuan Fernández archipelago) so that opportunitiesfor repeated phases of island hopping leading tomultiple speciation events are limited.

9.2 The taxon cycle

Until recently it was generally held that islands rep-resent a form of evolutionary blind alley, i.e. the flowof colonization events was considered essentially

unidirectional from continents to islands, implyinga generally lower fitness of island endemic forms.We now have evidence that sometimes island formssuccessfully colonize mainlands (e.g. Nicholsonet al. 2005), but the general trend is undoubtedlymainland to island, and the taxon cycle helpsexplain why this is so.

The term ‘taxon cycle’ was coined by Wilson(1961) in his studies of Pacific ants. The model hasbeen adapted to fit circumstances by differentauthors working on different taxa and regions, andhence it is not a discrete model allowing easy refu-tation. The shared features of different invocationsappear to be that the evolution takes place withinan island archipelago, in which immigrant speciesundergo niche shifts, which are in part driven bycompetitive interactions with later arrivals, suchthat the later arrivals ultimately may drive the ear-lier colonizing inhabitants towards extinction. Themethodology employed in all the early studies wasthat patterns of geographic distribution and tax-onomic differentiation provided the empirical basisfrom which were inferred cycles of expansion andcontraction in the geographical distribution, habitatdistribution, and population density of species inisland groups (Ricklefs and Cox 1978).

Melanesian ants

As developed by Wilson (1959, 1961), the taxoncycle described ‘the inferred cyclical evolution ofspecies [of Melanesian ants], from the ability to livein marginal habitats and disperse widely, to prefer-ence for more central, species-rich habitats with anassociated loss of dispersal ability, and back again’(MacArthur and Wilson 1967, glossary).

Wilson (1959, 1961) recognized differences in theranges of the ponerine ants as a function of theirhabitat affinities (Fig. 9.1). Marginal habitats (littoraland savanna habitats) were found to contain bothsmall absolute numbers of species and higher per-centages of widespread species. These data wereinterpreted as a function of changes in ecology anddispersal capability as ants moved from theOriental region, and particularly its rain forests,through the continental islands of Indonesia,through New Guinea, then out across the Bismarckand Solomon islands, on to Vanuatu, Fiji, and

T H E TA X O N C Y C L E 209

Page 223: Whittaker. Island Biogeography. 2007

Samoa. Wilson bundled the species into threestages. Stage 1 species are those that dominate in themarginal habitats: open lowland forest, grassland,and littoral habitats (Fig. 9.2). They have greaterecological amplitudes than the other species, asthey also occur in other habitats. They tend to betrail-making ants, nesting in the soil. Stage 1 speciestypically show a continuous distribution with notendency to break into locally distinct races. Stage 2of the process sees ants ‘returning’ to the dominantvegetation types, of interior and montane forest,within which they are more likely to be foundnesting in logs or similar habitats. If they succeed inadapting to the inner rain forests they eventuallydifferentiate to species level within Melanesia,forming superspecies or species groups. As theydifferentiate, they are liable to exhibit reduced geneflow across their populations. Stage 3 species occupysimilar habitats to those of stage 2, but evolution hasnow proceeded to the point where the species groupis centred on Melanesia and lacks close relatives inAsia. This may be, in part, a function of change inMelanesia, and in part of the contraction in thegroup remaining in Asia. Progress within the cycleis thus marked by a general pattern of restriction of

species to narrower ranges of environments withinthe island interiors. Meanwhile, these lineages havebeen replaced by a new wave of colonists occupy-ing the beach and disturbed habitats.

Pivotal to this inferred evolutionary system isthat the process is driven by the continuing (albeitinfrequent) arrival of new colonist species. Theircolonization initiates interactions that push earliercolonists from the open habitats. This is because theearlier colonists are likely to have become moregeneralized in the time since they themselves colo-nized, as they spread into additional habitats, thuslosing competitive ability in their original habitats(Brown and Gibson 1983). This model is persuasivein that it provides answers to some seeminglypuzzling phenomena. For example, it helps explainwhy species inhabiting interior forests of theoceanic islands are more closely related to speciesof disturbed habitats on New Guinea, than to thosespecies that have similar niches and which occur inmature forest on New Guinea. Although largely aunidirectional model, it was suggested that somelineages did find a way out of their alleyway,perhaps temporarily. Occasionally, a stage 3 speciesmay readapt to the marginal habitats, becoming

210 E M E R G E N T M O D E L S O F I S L A N D E VO L U T I O N

Per

cent

Increasing faunal size

Littoral Savanna Monsoonforest

Lowland rain forest(marginal)

Lowlandrain forest(interior)

Montanerain forest

100

0

50

Figure 9.1 Proportions of ponerine ant species in different habitats, as a function of their geographical distribution. The marginal habitats, tothe left, contain both smaller absolute numbers of species and higher percentages of widespread species. Dark columns, species widespread inMelanesia; stippled columns, species restricted to single archipelagos in Melanesia but belonging to groups centred in Asia or Australia; blankcolumns, species restricted to single archipelagos and belonging to Melanesia-centred species groups. (Redrawn from Wilson 1959.)

Page 224: Whittaker. Island Biogeography. 2007

a secondary stage 1 species, and expanding in itsdistribution (Fig. 9.2).

Caribbean birds

Ricklefs and Cox (1972) put forward a taxon cyclemodel to account for the biogeography of theavifauna of the Lesser Antilles. Their version dis-tinguishes an additional stage (Table 9.1, Fig. 9.3),but the basics are shared with the ant study.

Stage I The first stage is the invasion of an islandor archipelago by a mainland species. In theCaribbean, this means a species spreading acrossmany islands. It is thus taken to be a dispersiveform with interchange occurring between main-land and islands and among islands. The species islikely to be one of disturbed or coastal habitats, assuch species typically have good dispersive ability.There is therefore little differentiation initiallybetween mainland and island forms.

Stage II The colonist may then expand its niche,invading other habitats, and becoming moregeneralized in the use of resources. Species at thisstage have more spotty distributions as selection

against mobility reduces gene flow. They graduallyevolve local forms and become restricted to a sub-set of the islands. Species then become vulnerableto being out-competed in their original colonizingniche by further specialized colonists and maybecome restricted to interior forest habitats so thattheir niche breadth narrows again.

Stage III As they proceed through the stages ofthe cycle, species become highly differentiatedendemics that ultimately become extinct and arereplaced by new colonists from the mainland. StageIII species thus evidence a longer history of evolu-tion in isolation, being found as scattered endemicforms.

Stage IV The fourth stage of the cycle is when ahighly differentiated endemic species persists as arelict on a single island. From here the next step isextinction.

One reason for the success of colonists is that arecent arrival may have left behind predators, par-asites, and competitors on colonizing an island,enabling it to flourish despite the existence of localspecies with a longer period of evolutionary adjust-ment to the conditions on that island. An important

T H E TA X O N C Y C L E 211

Innerrain

forest

Marginalhabitats

South-eastern Asia Melanesian Islands

Time

AB

CD

E

F

Stage 1Stage 2

Stage 3

Figure 9.2 The inferred taxon cycle of ant species groups in Melanesia, in which is traced the hypothesized histories of groups derivedoriginally from Asia. The following sequence was postulated: stage 1—species adapt to marginal habitats (A) in the mainland source region(SE Asia) and then cross the ocean to colonize these habitats (B) in New Guinea. These colonizing populations may become extinct in time (C).Alternatively, the cycle enters stage 2—the ants invade the inner rain forests of New Guinea and/or surrounding islands (D). If successful in re-adapting to inner rain forest habitats they in due course diverge (E) to species level. Stage 3—diversification progresses within Melanesia,while the group remaining in Asia may, in cases, be contracting its range, such that the lineage becomes centred in Melanesia. A few members ofthese lineages, especially those on New Guinea, may re-adapt to the marginal habitats (F), and expand secondarily. (Redrawn from Wilson 1959.)

Page 225: Whittaker. Island Biogeography. 2007

part of the mechanism for the cycle as put forwardby Ricklefs and Cox (1972) was the evolutionaryreaction, or counter-adaptation, of the pre-existingisland biota to the newly arrived species, such thatover time they begin to exploit or compete with theimmigrant more effectively, thus lowering its com-petitive ability. The subsequent arrival of furtherimmigrant species will then tend to push the earliercolonists into progressively fewer habitats andreduce their population densities. By this reason-ing, range of habitat use and population densityshould each be diminished in species at anadvanced stage of the taxon cycle.

Ricklefs and Cox (1978) examined these proposi-tions from their earlier paper on the basis ofstandardized frequency counts taken in nine majorhabitats on three islands: Jamaica, St Lucia, andSt Kitts. Each species was first assigned to one oftheir four taxon cycle categories (Table 9.1a). Non-passerines did not show distinct trends in relationto stage of cycle, but the data for passerines weremore interesting. The most common stage I speciesdemonstrated habitat breadths and abundancesrarely attained by mainland species, indicating that

colonization of islands must involve some degreeof ecological release (cf. Cox and Ricklefs 1977),although the phenomena appeared to be confoundedby variation in the proportion of species in eachstage of the taxon cycle. Passerine species at ‘later’stages of the cycle tend to have more restrictedhabitat distributions and reduced population den-sities. Early-stage species tend to occupy open, low-land habitats, whereas late-stage species tend to berestricted to montane or mature forest habitats.Similar findings were reported for birds on theSolomon Islands by Greenslade (1968). Ricklefs andCox (1978) found these trends to be clear for Jamaica,with 35 passerines, but they are difficult to detecton St Kitts, with 13 species, none of which is endemic.

The Ricklefs and Cox model was subsequentlycriticized by Pregill and Olson (1981), who contestedthat the four stages are simply criteria that define aset of distributional patterns and that almost anyspecies in any archipelago would fall into one ofthese categories without necessarily following theprogression from stage I to IV in turn. To take a par-ticular island, many of the endemics of Jamaica,especially those endemic at the generic level, are

212 E M E R G E N T M O D E L S O F I S L A N D E VO L U T I O N

Table 9.1 The taxon cycle as applied to the avifauna of the Caribbean by Ricklefs and Cox (1972, 1978) (from Ricklefs1989)

(a) Characteristics of distribution of birds in the four stages of the taxon cycle

Stage of cycle Distribution among islands Differentiation between island populations

I Expanding or widespread Island populations similar to each otherII Widespread over many neighbouring Widespread differentiation of populations on

islands different islandsIII Range fragmented due to extinction Widespread differentiationIV Endemic to one island N/A

(b) Number of species of passerine bird on each of three islands

Jamaica St Lucia St Kitts

Stage I 5 8 6Stage II 10 7 6Stage III 8 9 2Stage IV 12 2 0Total 35 26 14

Area (km2) 11 526 603 168Elevation (m) 2257 950 1315

Page 226: Whittaker. Island Biogeography. 2007

widespread and abundant, contrary to the originalmodel. Ricklefs and Cox (1978) took this as illus-trating that species on Jamaica have been able todiverge sufficiently in ecological terms for thesegeneric-level endemics to avoid competition frommore recent colonists to the island. But is this a caseof special pleading? In addition, Pregill and Olsonargue that not all endemic species belong to groupsthat were once widespread. Fossil data indicate thatsome colonized an island from the mainland anddifferentiated at the species level without havingdispersed to other islands, an example being theJamaican becard (Platypsaris niger). Similarly, theyargued that very similar distributional patterns canarise from quite different evolutionary histories.

For example, the smooth-billed ani (Crotophaga ani)is absent from fossil deposits in the West Indies,whereas the white-crowned pigeon (Columba leuco-cephala), a species that is practically endemic to theWest Indies but which, nonetheless, has a stage Idistribution like that of the ani, occurs commonly inPleistocene deposits.

Generally, analyses of faunal turnover in islandstudies are based on the unstated premise thatenvironments are effectively stable, i.e. that bioticforcing factors, and not environmental factors, areresponsible for turnover. In contrast, Pregill andOlson’s (1981) analyses of the fossil record and ofrelict distributions, particularly amongst thexerophilic vertebrates of the Caribbean, indicated

T H E TA X O N C Y C L E 213

Lesser

Antilles

1949

1931

1924 1936

1900

Early Stage I

Shiny cowbird

Late Stage I

Grey kingbird

Stage III – IV

Adelaide warblerHouse wren

1

2

Stage III

1

2

4

3

Stage II

Antillean bullfinch

1

234

7

9

8

6

Figure 9.3 The distribution of members of the Lesser Antillean avifauna, illustrative of the proposed stages of the taxon cycle. Species are asfollows: shiny cowbird, Molothrus bonairiensis; grey kingbird, Tyrannus dominicensus; Antillean bullfinch, Loxigilla noctis; house wren, Troglodytesmusculus; and Adelaide warbler, Dendroica adelaidae. The small figures indicate differentiated populations (subspecies). The house wren becameextinct on the islands marked* within the 20th century. Dates refer to first colonization. (Redrawn from Ricklefs and Cox 1972, Fig. 1.)

Page 227: Whittaker. Island Biogeography. 2007

that conditions of climate and habitat wereappreciably different during the last glacial periodand indeed earlier in the Pleistocene (cf. Buskirk1985). Lowered sea level combined with climaticchanges to produce connections between islandspresently separated and also produced a generalincrease in the extent of xeric habitats, such as aridsavanna, grassland, and xeric scrub forest. They listseveral examples from the fossil record of speciescharacteristic of dry, open country, which once hada much wider distribution than they have had inlate Holocene times; these include the burrowingowl (Athene cunucularia), the Bahaman mockingbird(Mimus gundlachii), thick knees (Burhinus), falconsand caracaras (Polyborus and Milvago), curly-tailedlizards (Leicophalus), and rock iguanas (Cyclura).The significance of such losses of species on parti-cular islands is exemplified by the figures for NewProvidence Island, where 50% of the late Pleistoceneavifauna and 20% of the fossil herpetofauna nolonger occur, many of these species being xerophilic.

Pregill and Olson (1981) provide a number ofother examples of species distributions that theyregard as most parsimoniously explained by refer-ence to long-term environmental changes. Forinstance, they list a number of pairs of Hispaniolanbird species, such as the todies (Todus subulatus andT. angustirostris) and the palm-tanagers (Phaeni-cophilus palmarum and P. poliocephalus), which theyregard as most probably having differentiated attimes during the Quaternary when Hispaniola wasdivided into north and south islands. Hence, inopposition to the taxon cycle model, they offeredthe argument that the distributional patterns iden-tified might be primarily an evolutionary outcomeof climates becoming wetter (and warmer) since theend of the Pleistocene rather than being interpretablesimply in terms of interspecific interactions.

At the time of this debate the distributional andfossil data were inadequate for a clear resolution ofthe argument. However, recent genetic analyseshave provided a new angle on this controversy.Within the taxon cycle model it is assumed that awidespread species moves out in a single colon-ization phase from a mainland source, and spreadsout according to a stepping-stone process of colon-ization through the archipelago, producing a

monophyletic group within which birds on islandsclose to one another should be more closely relatedthan those distant from one another. Klein andBrown (1994) point out that the ornithologist JamesBond (after whom the fictional character wasnamed) proposed multiple colonizations frommainland sources for West Indian island popula-tions of widespread species. Multiple colonizationevents would mean that the representatives of aspecies on different islands would not form amonophyletic group relative to samples from thevarious mainland source pools. Modern techniquesallow Bond’s idea to be tested.

Klein and Brown (1994) studied mitochondrialDNA (mtDNA) from specimens of yellow warbler(Dendroica petechia) collected from North, South,and Central American sites, as well as from theWest Indies. The most parsimonious tree con-structed from their phylogenetic analyses indicatedcolonization of some islands in the Lesser Antillesby Venezuelan birds, and colonization of theGreater Antilles by Central American birds.Furthermore, they found evidence of multiple colo-nizations not only of the West Indies as a whole, butalso of individual islands. Such events can involvethe introgression of characters from one lineage toanother. Moreover, the phylogenetic data indicatedthat birds of adjacent islands are not always eachother’s closest relatives, against the assumptions ofa stepping-stone model for colonization. The multi-ple colonizations of the West Indies by yellow war-bler are at variance with the taxon cycle model asoutlined above. At least some of the differencesbetween populations are a consequence of whatcould be thought of as a series of founder effects,rather than exclusively because of in situ selectivepressures and drift. Klein and Brown (1994) notethat a number of recent studies of bats in theCaribbean provide similar lines of evidence formultiple colonizations of individual islands, and ofone group of related haplotypes being widespreadwhile another was confined to the Lesser Antilles.

Similar analyses based on mtDNA have nowbeen conducted on many other bird lineages in theWest Indies (papers reviewed in Ricklefs andBermingham 2002), allowing a comprehensivere-evaluation of the debate over the taxon cycle

214 E M E R G E N T M O D E L S O F I S L A N D E VO L U T I O N

Page 228: Whittaker. Island Biogeography. 2007

model. Multiple colonization, as exemplified by theyellow warbler, occurred in a few other cases, butoverall the results provide remarkably good sup-port for the original model as a fair representationof the temporal sequence followed by manyspecies. Some 20 lineages colonized the LesserAntilles between 7.5 and 10 Ma, with individualisland populations typically persisting for about4 million years, and with single island endemicstypically amongst the oldest species, as required bythe model. Some species have undergone phases ofsecondary expansion. Phases of expansion and con-traction do not appear to be correlated acrossspecies and contrary to Pregill and Olson (1981)many do not correlate with glacial climate transi-tions. Indeed, many of the cycles predate the mostintense glaciation episodes of the last 2 millionyears. However, some 17 lineages appear to havecolonized the Lesser Antilles between about 0.75and 0.55 Ma, which suggests that although taxoncycles occur independently of climatic fluctuations,the onset of the most extreme Pleistocene climaticfluctuations may indeed have influenced the com-position of the Lesser Antillean avifauna.

In conclusion, Ricklefs and Bermingham (2002)find evidence for expansion phases occurring atintervals on the order of 106 years, and that theseepisodes are not synchronous across independentlineages. They interpret these data as supportingthe idea that taxon cycles are related to changes inspecific aspects of population–environment rela-tionships, such as host-parasite interactions: asdemanded by the ‘counter-adaptation’ argument ofRicklefs and Cox (1972). Future analyses of the phy-logenies of the parasites may allow a more rigoroustest of this mechanism.

Caribbean anoles

The anoline lizards of the Caribbean comprise about300 species, or between 5 and 10% of the world’slizards (Roughgarden and Pacala 1989). OnCaribbean islands they are extremely abundant,replacing the ground-feeding insectivorous birds ofcontinental habitats. The distribution of species/sub-species relates best not to the present-day islands butto the banks on which the islands stand, separated

from other banks by deep water. The anoles havelong provided one of the classic illustrations ofcharacter displacement, whereby island banks haveeither one intermediate-sized species or two speciesof smaller and larger size respectively. This has beentaken to indicate that where two populations haveestablished, interspecific competition between themhas resulted in divergence of size of each away fromthe intermediate size range so that they occupydistinct niche space (Brown and Wilson 1956). Thisprogresses to the point at which the benefit of furtherreduction in competition is balanced by thedisadvantage of shifting further from the centre ofthe resource distribution (Schoener and Gorman1968; Williams 1972). However, Roughgarden andPacala (1989) contend that within the islands of theAntigua, St Kitts, and Anguilla banks in the northernLesser Antilles, a series of independently derivedfacts contradict this model, and instead support aform of taxon cycle (see also Rummel andRoughgarden 1985).

The anolid taxon cycle of the northern LesserAntilles begins with an island occupied by amedium-sized species, which is joined by a largerinvader from the Guadeloupe archipelago. Ratherthan the two species moving away from the middleground, as in the character displacement scenario,both species evolve a smaller size. The medium-sized species becomes smaller as it is displacedfrom the middle ground by its larger competitor.This opens up niche space in the centre of theresource axis, selecting for smaller size in theinvader, which thus approaches the medium(presumably optimal) size originally occupied bythe first species. The range of the original residentspecies therefore contracts, culminating in itsextinction. At this point, the invader has takenup occupancy of the niche space of the first speciesat the outset of the cycle. The islands remain in thesolitary-lizard state until (hypothetically) anotherlarge invader arrives.

The key points in the Roughgarden and Pacala(1989) analysis are as follows. Intensive scrutiny ofthe systematics of anoline lizards over a period offour decades resulted in a phylogenetic tree com-prising five main groups. The northern LesserAntilles are populated by the bimaculatus group,

T H E TA X O N C Y C L E 215

Page 229: Whittaker. Island Biogeography. 2007

which is subdivided into two series at the next levelof the hierarchy (Fig. 9.4). The wattsi series comprisesmall brown lizards that perch within a few feet ofthe ground. The bimaculatus series comprise large ormedium-sized green or grey-green lizards, whichare relatively arboreal. Islands (for which readisland banks) have either one species or two living inthe natural habitat, i.e. excluding small enclaves ofintroduced anoles living near houses. Anguilla,Antigua, and St Kitts are the island banks havingtwo species, in each case a larger bimaculatus and asmaller wattsi.

Earlier research by Williams (1972, cited byRoughgarden and Pacala 1989) had produced twoempirical ‘rules’ concerning the sizes and ratios ofthe anoles:

1 Species from islands in which only one anole ispresent are intermediate in size (snout–vent length)between the body sizes seen on islands in whichtwo anole species coexist. This rule is correct for 11of 12 island banks having a solitary species (theyrange in length from 65 to 80 mm) and it appliesthroughout the eastern Caribbean.

2 Species on a two-species island differ in bodylength by a factor of 1.5–2.0, such that the largerspecies exceeds 100 mm and the smaller is less thanor equal to 65 mm in length, the size of the smallestsolitary species. This rule is correct for four of the

five two-species island banks, and again appliesthroughout the eastern Caribbean.

These patterns appeared to be consistent with thecharacter displacement model. It was also thoughtthat the original residents were members of thebimaculatus series, and that the invading specieswere members of the wattsi series, with both invad-ing from Puerto Rico.

Although the character displacement mechanismappeared to explain much of the variation in ano-line lizards through the Lesser Antilles, there weresome exceptions and further research producedmore problems in respect to the northern islands(Roughgarden and Pacala 1989):

● On St Maarten there are two anoles of nearly thesame body size. By the character displacementhypothesis, one or both should be a new arrival.However, both are differentiated (visibly and bio-chemically) from other anoles in the region.

● Medium-sized lizards should become larger butPleistocene cave deposits demonstrated unequivo-cally that larger anoles have become smaller, andno fossil data show a medium-sized species becom-ing larger.

● The two-species community supposedly has itsorigins in two colonists of similar size, whichdiverge in size. However, experimental studies of

216 E M E R G E N T M O D E L S O F I S L A N D E VO L U T I O N

pogus

schwartziwattsi (2)bimaculatusleachisabanusgingivinusnubiluslividus

marmoratus (9)

terraealtae (2)ferreusoculatus (4)

St. MaartenSt. Kitts Bank

Antigua Bank

Antigua BankSt. Kitts Bank

Saba

St. Maarten BankRedonda

wattsi

seriesbim

aculatus series

Montserrat

Guadeloupe Bank

Illes de SaintesMarie GalanteDominica

Figure 9.4 Phylogenetic tree for Anolis of the bimaculatus group (containing the wattsi series and the bimaculatus series) in the northernLesser Antilles. Those end-points represented by dots are monotypic species, the others are geographic races or subspecies according to currentnomenclature. This tree is merely one trunk of a larger phylogenetic tree for eastern Caribbean anolids, constructed on the basis of detailedsystematic/phylogenetic data by Roughgarden and Pacala (1989). (From Roughgarden and Pacala 1989, Fig.1.)

Page 230: Whittaker. Island Biogeography. 2007

competition and accidental introductions of anolesserve to demonstrate that an invader very similar insize to an established resident does not easily estab-lish on an island. A second species arriving on anoccupied island needs to be larger than the occu-pant in order to become established.

Having dismissed the original model, Rough-garden and Pacala (1989) put forward 12 strands ofevidence in further support of their taxon cyclemodel, from which the following points are taken.The only known historical extinction of an anole inthe Lesser Antilles is of the smaller species from atwo-species island. North of Guadeloupe, only thewattsi series has sufficient geographic variationwithin an island bank to have led to a subspecificnomenclature, the Antigua bank having A. wattsiwattsi on Antigua proper and A. wattsi forresti onBarbuda. This may be interpreted as indicating alonger presence of the smaller-sized lineage, thewattsi series, than the larger bimaculatus series onthe northern Lesser Antilles. This is contrary to theoriginal model but in accord with the taxon cyclemodel. The source for the bimaculatus series appearsto be Guadeloupe, and this is geographically sensi-ble given that the prevailing currents run towardsthe northern islands. In short, data on the ecologyof colonization, the phylogenetic relationshipsamong anoles, their biogeographical distribution,and the fossil record, refute the character displace-ment explanation for the northern Lesser Antilles,but are consistent with the taxon cycle model.Yet, in advocating this model, Roughgarden andPacala (1989) caution that such a cycle does notappear to happen often, and there is no evidencefor its operation other than in the northern LesserAntilles.

Phylogenetic analyses shed further light on this,by showing that the Lesser Antilles are occupied bytwo distinct lineages of anoles. Dominica and islandsto the north contain species related to those onPuerto Rico, whereas islands to the south harbourspecies with South American affinities (Losos 1990,1996). The patterns found of character displace-ment in conditions of sympatry in the northernLesser Antilles contrast with the evidence forchange in body size in the southern Lesser Antilles.

In the latter case change appears to be unrelated towhether a species occurred in sympatry with con-geners, and appears instead to result from a processof ecological sorting, such that only dissimilar-sized species can colonize and coexist (Losos 1996).

Evaluation

The taxon cycle has, until recently, been difficult totest. This is in part because of the fluidity of the the-oretical framework, and in part because of the pos-sible competing influences of environmentalchange, and of humans in ‘messing up’ the biogeo-graphical signal. It has, for instance, been postu-lated that size reductions and extinctions of somelarger lizards in the Caribbean were linked not tointrinsic biological processes so much as to humancolonization of the islands (Pregill 1986)—anentirely believable proposition. However, themolecular phylogenies now available for LesserAntillean avifauna and anoles have served to reaf-firm the existence of taxon cycles. A history of taxoncycling has also been proposed for weevils of thegenus Galapaganus on the Galápagos (Sequeira et al.2000), suggesting that it might be worthwhile test-ing for taxon cycles in other island and indeedmainland settings.

9.3 Adaptive radiation

The word ‘adaptation’ gives an erroneous impression ofprediction, forethought or, at the very least, design.Organisms are not designed for, or adapted to, the presentor the future—they are consequences of, and thereforeadapted by, their past.

(Begon et al. 1986, p. 6)

Island evolutionary forces reach their most spectac-ular embodiment in the patterns termed adaptiveradiations. The Hawaiian honeycreepers (some-times ‘honeycreeper-finches’) and drosophilids andthe Galápagos finches are the most famous illustra-tions from the animal kingdom. Hawaii and theCanaries provide some of the better-known plantexamples. The term adaptive radiation is takenhere to refer to the evolutionary development ofdistinct species (or varieties) from a single ancestralform (i.e. the lineage is monophyletic), where the

A D A P T I V E R A D I AT I O N 217

Page 231: Whittaker. Island Biogeography. 2007

radiation is distinguished by niche differentiationamongst the members of the lineage. As we haveseen, selection and adaptation does not necessarilylead to radiation. Similarly, as we will discuss later,radiation within monophyletic lineages does nothave to be adaptive. The term non-adaptive radia-tion can be applied to cases where allopatric specia-tion via founder effect and genetic drift mechanismshas been dominant in lineage development,although it is best to think in terms of a gradientbetween entirely adaptive and non-adaptive endpoints, with most systems occupying a space some-where in the middle of such a conceptual gradient.As with the taxon cycle, adaptive radiations are theresult of a combination of driving forces and mech-anisms; and they are drawn from a common toolkit. Adaptive radiation scenarios differ from thetaxon cycle model in being concerned solely withthe expansion phase of a monophyletic lineage,whereas taxon cycle theories attempt to describethe expansion and subsequent contraction of par-ticular taxa within an entire evolutionary assem-blage (e.g. ponerine ants, or land birds).

The availability of vacant niche space is a key fea-ture in adaptive radiations, as it allows a form ofecological release, allowing the diversification thatsometimes leads to speciation within a lineage. Assuggested by its name, Metrosideros polymorpha isvery diverse and multiform. It occupies habitatsranging from bare lowlands to high bogs, andoccurs as a small shrub on young lava flows and asa good sized tree in the canopy of mature forest. Itis the principal tree of the Hawaiian wet forestsand, as an aside, at least some individuals can befound in flower at any time of the year, a patternsupported by enough of the nectar-producing florato have allowed the evolution of the nectar-feedinglineages among the honeycreepers. Despite itsradiation of forms, all members of this complexwere originally allocated to M. polymorpha, i.e.Meterosideros has radiated a lot but has speciated toa lesser degree. Some highly distinct populationshave now been recognized as segregate species, butactive hybrid swarms also exist (Carlquist 1995),indicating continuing gene flow.

We now consider in a little more detail what wemean by the term adaptive. First, according to the

above quote from Begon et al. (1986), adjustment tothe abiotic and biotic environment works by natu-ral selection. This is effected by the successful formscontributing more progeny to the following gener-ations, a form of filtering that tells us about the pastand which may be of quite varying fit to con-temporary circumstances (e.g. of altered climate,depleted native biotas, altered habitats and numer-ous exotic species). Gittenberger (1991) has arguedthat there should be no automatic labelling ofradiations as adaptive, but that a degree of special-ization of species in to different niches should beinvolved before the term is used.

Once these points are registered, there is little tothe core theory of adaptive radiation that we havenot already covered. Adaptive radiation invokesnot so much a particular process of evolutionarychange as the emergent pattern of the most spec-tacular cases, the crowning glories of island evolu-tion. For this very reason the best cases are liable toinvolve a wide range of the evolutionary processespreviously introduced.

The data in Table 9.2 provide recent estimates forthe degree of endemism and an idea of the degreeof radiation involved in the biota of Hawaii. Thesefigures are not cast in stone, as they depend on thetaxonomic resolution and assumptions involved intheir calculation. For example, Paulay (1994) citesan estimate that there may be as many as 10 000Hawaiian insect species, and that they may haveevolved from 350–400 successful colonists. Not-withstanding such uncertainties, Hawaii clearlyprovides spectacular examples of radiations in taxaas diverse as flowering plants, insects, molluscs,and birds. It is estimated that the 980–1000 flower-ing plant species arose from between just 270 and280 original colonists (Fosberg 1948, Wagner andFunk 1995), the pattern of radiation being indicatedby the following statistics: there are 88 families and211 genera, the 16 largest genera account for nearly50% of the native species, about 91% of which areendemic (Sohmer and Gustafson 1993).

The biogeographical circumstances in whichradiations take place are reasonably distinctive.Radiations are especially prevalent on large, high,and remote islands lying close to the edge of agroup’s dispersal range (Paulay 1994). MacArthur

218 E M E R G E N T M O D E L S O F I S L A N D E VO L U T I O N

Page 232: Whittaker. Island Biogeography. 2007

and Wilson (1963, 1967) termed such peripheralareas the radiation zone. Here the low diversity ofcolonists, and the disharmony evident in the lack ofa normal range of interacting taxa, facilitate in situdiversification (Diamond 1977). In illustration, antsdominate arthropod communities across most ofthe world, but in Hawaii and south-east Polynesiathey are absent (or were, prior to human interfer-ence). In their place, there have been great radia-tions of carabid beetles and spiders, and evencaterpillars have in a few cases evolved to occupypredatory niches (Paulay 1994).

For less dispersive taxa, their radiation zones maycoincide with less remote archipelagos, which havea greater degree of representation of interacting taxathan the most remote archipelagos. Hence, the cir-cumstances for radiation reach their synergisticpeak on the most remote islands, the epitome beingHawaii (below). Examples that fit this idea of maxi-mal radiation near the dispersal limit include birdson Hawaii and the Galápagos, frogs on theSeychelles, gekkonid lizards on New Caledonia,and ants on Fiji; exceptions—taxa that have notradiated much at remote outposts—include terres-trial mammals on the Solomons and snakes andlizards on Fiji (MacArthur and Wilson 1967). Equally,it is clear from even the most remote island archi-pelagos that not all lineages within a single taxonhave radiated to the same degree. Nearly 50% of thec.1000 native flowering plant species of Hawaii arederived from fewer than 12% of the c.280 successful

original colonists (Davis et al. 1995). Most of the restof the colonists are represented by single species.Such differences may reflect the length of time overwhich a lineage has been present and evolvingwithin an archipelago, but it is clear that some earlycolonists have nonetheless failed to radiate. In short,although these geographical circumstances may beconducive to radiations, they are not the only factorsof significance and they do not lead to radiationswithin all lineages.

Darwin’s finches and the Hawaiianhoneycreeper-finches

Although the Galápagos are renowned for otherendemic groups, notably the tortoises (Arnold1979) and plants (Porter 1979), the most famousgroup of endemics must be Darwin’s finches(Emberizinae; genera Geospiza, Certhidea, Platyspiza,Camarhynchus, and Cactospiza). The context withinwhich these birds and the other creatures haveevolved is as follows. The Galápagos are in theeast Pacific, 800–1100 km west of South America(Fig. 9.5a). Although equatorial, they are compara-tively cool and average rainfall in the lowlands isless than 75 cm/year (Porter 1979). There are some45 islands, islets and rocks, of which 9 are islands ofarea greater than 50 km2. Isabela, at 4700 km2, rep-resents over half of the land area and is four timesthe size of the next largest island, Santa Cruz.Isabela and Fernandina have peaks of some

A D A P T I V E R A D I AT I O N 219

Table 9.2 Number of presumed original colonists, derived native species, and endemic species for a selection of the Hawaiian biota (Sohmer and Gustafson 1993)

Animal or plant Estimated no. Estimated no. of % Endemic group of colonists native species species

Marine algae ? 420 13Pteridophytes 114 145 70Mosses 225 233 46Angiosperms 272 c. 1000 91Terrestrial molluscs 24–34 c. 1000 99Marine molluscs ? c. 1000 30–45Insects 230–255 5000 99Mammals 2 2 100Birds c. 25 c. 135 81

Page 233: Whittaker. Island Biogeography. 2007

1500 m ASL, but most of the islands are relativelylow. Volcanic in origin, they are true oceanic islands,never having been connected to mainland (Perry1984). They remain volcanically active, many lavaflows being recent and still unvegetated. Geologicalevidence suggests islands to have existed in theGalápagos for between 3.3 million and 5 millionyears, although the four westernmost islands(Pinta, Marchena, Isabela and Fernandina) are just0.7 to 1.5 million years old (Simkin 1984).

Including the 13 finch species, 28 species of landbirds breed on the Galápagos, of which 21 areendemic. There are four mockingbirds, which arerecognized as separate species on the basis ofmorphological differences, although as they haveallopatric distributions, it is not known if they couldinterbreed or not (Grant 1984). It is also unclearwhether their genus (Nesomimus) is sufficiently dis-tinct morphologically to warrant separation fromthe mainland genus (Mimus). Similar problems existwith the finches. As noted by Darwin (1845) in thequotation at the start of Chapter 8, such uncertain-ties do nothing to diminish their interest. Darwin’sfinches will be the focus of attention here as theyhave radiated to the greatest degree, 13 Galápagos

species being generally recognized (e.g. Sato et al.1999; but see Zink 2002), with a fourteenth(Pinaloroxias inornata) occurring on nearby CocosIsland (Fig 9.5b; Chapter 8). In fact it is extremelydifficult to identify all the Galápagos finches, as thelargest members of some species are almost indis-tinguishable from the smallest members of others(Grant 1984). Collectively, they feed on a remarkablediversity of foods: insects, spiders, seeds, fruits, nec-tar, pollen, cambium, leaves, buds, the pulp of cac-tus pads, the blood of seabirds and of sealionplacenta. It is principally through changes in beakstructure and associated changes in feeding skillsand feeding niches that the differentiation betweenthe finches has come about (Lack 1947a; Grant 1994).In illustration, the woodpecker finch (Camarhynchuspallidus) uses a twig, cactus spine or leaf petiole as atool, to pry insect larvae out of cavities. Small,medium and sharp-billed ground finches (Geospizafuliginosa, G. fortis, and G. difficilis) remove ticksfrom tortoises and iguanas, and perhaps mostbizarre of all, sharp-billed ground finches on thenorthern islands of Wolf and Darwin perch on boo-bies, peck around the base of the tail, and drink theblood from the wound they inflict..

220 E M E R G E N T M O D E L S O F I S L A N D E VO L U T I O N

Isabela

DarwinWolf

GALÁPAGOSISLANDS

Pinta

Santa CruzSan Cristóbal

Española

Cocos Island

FernandinaSOUTH

AMERICA

5°N

80°W85°W90°W

(a)

Figure 9.5 The Galápagos archipelago and Cocos Island and their finches. (a) Map, whereby the dashed line approximates the 1800 m depthcontour. Five degrees on the equator represent approximately 560 km. (Redrawn from Williamson 1981, Fig. 9.1.). See (b) on the next page.

Page 234: Whittaker. Island Biogeography. 2007

A D A P T I V E R A D I AT I O N 221

Scratchfor seedson theground

Feed onseeds on theground andthe flowers andpulp of PricklyPear (Opuntia)

Feed in treeson beetles

use spines heldin the bill toextract insectsfrom barkcrevices

Feed on leaves,buds, and seedsin the canopy oftrees

Warbler-like birdsfeeding on smallsoft insects

Ce. olicacea

Ce. fusca

P. crassirostris

C. pauper

G. fortis

G. fuliginosa14 g

20 g

34 g

21 g

28 g

18 g

21 g

34 g

13 g

10 g

(b)

8 g

20 g

20 g

13 g

G. magnirostris

G. scandens

G. conirostris

G. difficilis

C. parvulus

C. psittacula

C. pallida

Pi. inornata

Figure 9.5 (b) Darwin’s finches, the Emberizinae, are endemic to these islands. There are a number of different phylogenetic hypotheses of howthese taxa relate to each other. The genetic distance between each species is shown by the length of the horizontal lines. The maximum amountof black colouring in the male plumage and the average body mass are shown for each species. This version is taken from Townsend et al. (2003)after Petren et al. (1999).

A comprehensive theory for this radiation wasdeveloped by Lack (1947a) and has been updatedand summarized by Grant (1981, 1984; and seeVincek et al. 1997). The key points are as follows.One of the Galápagos islands was colonized fromthe mainland in a one-off founding event. Geneticanalyses suggest that the effective population size of

the founding flock was at least 30 individuals(Vincek et al. 1997). The founding populationexpanded quite rapidly, undergoing selectivechanges and/or drift. After some time, members ofthis population colonized another island in thearchipelago, where conditions were slightly different.Further changes occurred through a combination of

Page 235: Whittaker. Island Biogeography. 2007

random genetic changes, drift, and selection (e.g. fordifferent feeding niches). A degree of differentiationwould then have been evident between the separatepopulations. At some point, individuals of one ofthe derived populations flew to an island alreadyoccupied by a slightly differentiated population.The result of this might sometimes have been aninfusion of the newcomers into the establishedpopulation, but where the two populationsexhibited a greater degree of behavioural separa-tion, hybridization would have been limited.Selection would have favoured members of the twogroups that fed in different ways from each otherand so did not compete too severely for the sameresources.

Recent research has shown that female finchesappear able to distinguish between acceptable andunacceptable mates on the basis of beak size andshape and also on the different patterns of songs(Grant 1984). This may have been the means bywhich females were able to select the ‘right’ mates,thus breeding true, and producing progeny thatcorresponded with the ‘peaks’ rather than the ‘val-leys’ of the resource curves. Such characteristicswould have a selective advantage in the popula-tions over time, thus allowing two or more speciesto exist in sympatry.

To put these points together: the radiation of thelineage has taken place in the context of a remotearchipelago, presenting extensive ‘empty niche’space, in which the considerable (but not excessive)distances between the islands has led to phases ofinterisland exchange only occasionally. Differingenvironments have apparently selected for differ-ent feeding niches both between and within islands.Thereafter, behavioural differences between formsmaintain sufficient genetic distance between sym-patric populations to enable their persistence as (tovarying degrees) distinctive lineages.

As Grant (1984) cautioned, this is not the onlyway in which speciation can occur. Theoretically alineage may have split into two non-interbreedingpopulations on a single island, as suggested by themodel of competitive (sympatric) speciation(Chapter 8), provided that the environment pre-sented two or more sufficiently distinct resourcepeaks (i.e. was sufficiently heterogeneous). Further

progress has been made in understanding theinteractions and exchanges that may occur betweensympatric populations of Galápagos finches, high-lighting the importance of sympatric episodes inlineage development (Grant and Grant 1996b), butbefore discussing this work we will outline the caseof the Hawaiian equivalents to Darwin’s finches.

The Hawaiian islands have formed as a narrowchain from a hotspot, which appears to have beenoperational for over 70 million years, although theoldest high island of the present group, Kauai, isonly about 5.1 million years old (Table 2.3). It isbelieved that the hotspot has never been closer toNorth America than it is today and its positionrelative to Asia has probably also been stable. Mole-cular clock data suggest that few lineages exceed10 Ma, i.e. the pre-Kauai signal in the present biotaof the main islands is actually rather limited(Wagner and Funk 1995; Keast and Miller 1996).

At least 20 natural avian colonizations have beensuggested (Tarr and Fleischer 1995), and endemicscomprise approximately 81% of the native avi-fauna. There have been other avifaunal radiationson Hawaii. They include the elepaio (Chasiempissandwichensis), a small active flycatcher endemic toHawaii. Distinctive subspecies occur on Kauai andOahu, and a further three occur within theyoungest and largest island, Hawaii itself (Prattet al. 1987). The native thrushes of Hawaii areplaced in the same genus (Myadestes) as the soli-taires of North and South America. Five species arerecognized by Pratt et al. (1987), each occurring onits own island or island group. One was last seen in1820, and three of the remaining species areseverely endangered, with populations—if theysurvive—numbering fewer than 50 individuals(Ralph and Fancy 1994).

The Hawaiian honeycreepers (or honeycreeper-finches) have shown an even greater radiationthan Darwin’s finches. They are a monophyleticendemic group perhaps best considered a subfam-ily, the Drepanidinae (see Pratt 2005). For a longtime it was believed that this radiation had resultedin 23 species in 11 genera. From a single type ofancestral seed-eating finch, the group had radiatedto fill niches of seed-, insect-, and nectar-feedingspecies, with a great variety of specialized beaks

222 E M E R G E N T M O D E L S O F I S L A N D E VO L U T I O N

Page 236: Whittaker. Island Biogeography. 2007

and tongues; thus providing one of the most popu-lar illustrations of evolutionary radiation (Fig. 9.6;Carlquist 1974; Williamson 1981). It is now knownthat there were actually many more species in therecent past. Estimates of the number of speciesknown historically range from 29 to 33, withanother 14 having recently been described fromsubfossil remains: most extinctions occurringbetween the colonization of the islands by thePolynesians and European contact (Olson andJames 1982, 1991; James and Olson 1991; Tarr andFleischer 1995). As noted, different genera aregenerally recognized within this subfamily, includ-ing Psittirostra (the Hawaiian name of which is ou)which have short, conical beaks, for seed eating(essentially they are still finches), and Pseudonestor(Maui parrotbill), which uses its powerful beak to

tear apart twigs in order to reach wood-boring bee-tles. The main evolutionary line, however, is quitedifferent. Its members have longer, narrower bills,and they feed on insects (in the case of Paroreomyza,the Molokai creeper) and on nectar (Loxops (Akepa)and Hemignathus (e.g. Akialoa) ).

Tarr and Fleischer (1995) observe that genetic dif-ferentiation between species of drepanidines is lessthan would be expected on the basis of their mor-phological divergence, indicating that they are theproduct of a relatively recent arrival (in the order of3.5–8 million years), followed by rapid adaptiveradiation. They concluded that evidence for anallopatric model of speciation was fairly convincingfor some of the honeycreepers, but they were notable to rule out sympatric speciation (particularly inthe light of recent fossil finds). Himatione sanguinea

A D A P T I V E R A D I AT I O N 223

Insects, no nectar

Fruitandseeds

Insects andsome nectar

Nectar and some insects

Fruit

67 8 9 13

14

11

12

17

?

16

15

1

10

4

5

3

2

Figure 9.6 The inferred pattern of evolution of dietary adaptations as represented by 16 of the Hawaiian honeycreepers (family Fringillidae,Subfamily Drepanidinae). 1, Unknown finch-like colonist from Asia; 2, Psittirostra psittacea ; 3, Chloridops kona (Psittacirostra kona) (extinct);4, Loxioides bailleui (Psittacirostra bailleui) ; 5, Telespyza cantans (Psittacirostra cantans) ; 6, Pseudonestor xanthophyrs ; 7, Hemignathus munroi(H. wilsoni) ; 8, H. Iucidus ; 9, H. obscurus (H. procerus); 10, H. parvus (Loxops parva) ; 11, H. virens (Loxops virens) ; 12, Loxops coccineus; 13,Drepanis pacifica (extinct); 14, Vestiaria coccinea ; 15, Himatione sanguinea ; 16, Palmeria dolei ; 17, Ciridops anna (extinct). Source: Cox and Moore(1993, Fig. 6.11). The taxonomy preferred here follows Pratt et al. (1987), with that given in Cox and Moore (1993) in brackets where different.

Page 237: Whittaker. Island Biogeography. 2007

sanguinea (the apapane) and Vestiaria coccinea (thei’iwi), two of the more widespread taxa, show anintriguing lack of differentiation, and it has been sug-gested that this may reflect relatively recent rangeexpansion. This may have occurred in response tothe arrival and spread of the tree Metrosideros poly-morpha, or possibly because of the loss of competitorsdue to the extinction of other avian taxa. If theseinterpretations are correct, these two species mayrepresent the early stage of a new cycle of dispersaland differentiation (Tarr and Fleischer 1995).

Having set out the basics of the Hawaiian pic-ture, we can pick up the thread of the discussion ofthe interactions and exchanges between sympatricpopulations by reference to comparative studiesbetween the Hawaiian and Galápagos lineages.Studies of morphological variation within popula-tions of Darwin’s finches have revealed that there isconsiderable variation in body size and beak traitswithin populations of ground finches (Geospiza),and that the amount of variation itself differsamong populations (Grant 1994). In studies of six ofthese species, Grant and Grant (1994) foundevidence for the limited occurrence of introgressivehybridization, allowing gene flow between popula-tions of different species and contributing to theintraspecific variation observed.

Grant (1994) set out to establish from the study of524 museum specimens of the 7 species of Hawaiianhoneycreepers with finch-like bills (which might betermed honeycreeper-finches) whether the samehybridization process was at work in a lineage thathas diversified to a much greater extent. The meas-urements demonstrated that variation within pop-ulations of Hawaiian honeycreeper-finches was lessthan among the Geospiza. The single Hawaiianspecies with both sympatric and allopatric popula-tions did not show greater variation in thesympatric population, as would have been expe-cted if hybridization was effective. It was con-cluded that there was no evidence of hybridizationoccurring within the past 100 years. Grant reportedthat G. C. Munro had suggested in 1944 thatRhodacanthis flaviceps might be a hybrid producedfrom R. palmeri and R. psittacea, and indeed the onlymorphological hint of hybridization among theHawaiian species came from the close similarityfound between R. flaviceps and R. palmeri. If

R. flaviceps is indeed of hybrid origin, then hybridiza-tion must have been going on for a long time, as thespecies is recognizable in fossil material.

Grant (1994) put forward two hypotheses whichmight explain why the hybridization evident in theGalápagos finches was not apparent in the Hawaiianequivalents.

● Geological and electrophoretic studies indicatethat the honeycreeper-finches have been present inthe Hawaiian archipelago for a much longerperiod, perhaps three times longer than have thefinches of the Galápagos, and over this time havediversified further and have evolved prezygotic(behavioural) and/or postzygotic isolating mecha-nisms (Table 8.1).

● The Hawaiian honeycreeper-finches have evolvedgreater dietary specializations in the generally lessseasonal and floristically richer Hawaiian islands. Inan environment with more distinctive resourcepeaks, and following the line of reasoning outlinedin the section on competitive speciation, stabilizingselection for specialist feeding may have providedstrong selective pressure against an ecological nicheintermediate between the two parental species.

Grant put forward a model to capture the impor-tant distinctions in a hypothetical phylogeny for theradiation of sympatric taxa (Fig. 9.7). An early phaseof divergence is characterized by occasional geneticcontact through interbreeding, followed by a genet-ically independent phase. The duration of the phaseof introgression will depend on the ecological isola-tion attained, which will be related to environmen-tal heterogeneity. From genetic distance measures offinch species known to hybridize from theGalápagos and from North America, Grant (1994)estimates the apparent duration of the phase ofintrogression, i.e. of hybridization, as up to 5 millionyears. He speculates that it may have been occur-ring over the whole of the estimated 2.8 millionyears of the radiation of Darwin’s finches and quitepossibly for much of the 7.5 million years estimatedby some (but see Tarr and Fleischer 1995) for thediversification of the Hawaiian honeycreepers.

All such figures involve error margins and in thecase of the Hawaiian species the evidence for sucha model is very largely indirect. Hybridization has

224 E M E R G E N T M O D E L S O F I S L A N D E VO L U T I O N

Page 238: Whittaker. Island Biogeography. 2007

also been observed between the Galápagos landand marine iguanas which, although of mono-phyletic origin, are actually placed in separategenera. However, analysis of the DNA of iguanapopulations on Plaza Sur suggested that if geneexchange occurs it must be at a very much lowerrate than in the finches (Rassmann et al. 1997)..

The studies reviewed in this section allow us toconclude that both allopatric and sympatric epi-sodes can be involved in radiations of island archi-pelago birds. Although events must vary from onelineage to another, Grant and Grant (1996b) suggestthe following general scenario for sympatric con-geners, based largely on their work in the Galápagos:

● First, following colonization of different islands,there is an initial phase of differentiation ofallopatric populations.● Then, interisland movements re-establish sympa-try, following which further differentiation takesplace between the two populations.

● Over a period of several million years, occasionalintrogressive hybridization occurs, often inhibitingspeciation; but perhaps in cases the resultingrecombinational variation plays a creative role,facilitating further divergence.● Eventually, when the lineages have diverged farenough, this form of exchange effectively ceases toplay a role.

Hawaiian crickets and drosophilids

The Hawaiian endemic crickets are thought to havederived from as few as four original colonizingspecies, each being flightless species arriving in theform of eggs carried by floating vegetation. Theancestral forms were a tree cricket and a sword-tailcricket from the Americas, and two ground cricketsfrom the western Pacific region (Otte 1989). Threeof the successful colonists have radiated exten-sively, and Hawaii now has at least twice as manycricket species as the continental United States.Much later, a further eight species have colonized,but these are considered to have been introducedby humans and will not be discussed further.

The tree crickets (Oecanthinae) have been calcu-lated on phylogenetic grounds to have colonizedHawaii about 2.5 million years ago, radiating into 3genera and 54 species (43% of the world’s knownspecies), with the greatest diversification seenwithin the older islands, which were occupiedearliest (Otte 1989). They have radiated into habitatsnot occupied by their mainland relatives. Otte con-siders competitive displacement to have had a rolein the within-island evolution of the Hawaiiancrickets, and indeed distributional and cladistic datasuggest that virtually all speciation takes placewithin islands. However, much of it likely occurs inlocally isolated habitats, as the active geology andgeomorphology of lava tubes, lava islands, incisedvalleys, and mega-landslips promotes repeatedintraisland vicariance events (Carson et al. 1990;Carson 1992). This emphasis on within-island speci-ation concurs with data for the plant genus Cyrtandraand the land snail genus Achatinella, each of whichis represented by over 100 species on Oahu alone.

Drosophila and several closely related genera inits subfamily include about 2000 known species, ofwhich Hawaiian drosophilids (the closely related

A D A P T I V E R A D I AT I O N 225Ti

me

Ecological variation

Morphological variation

(a) (b)

Figure 9.7 A model illustrating the relative significance ofintrogressive hybridization during divergence, as a function ofresource space. In the early introgressive phase of divergence, newlyformed species remain in genetic contact through occasionalinterbreeding, but this is followed by a genetically independentphase. The length of time over which introgression occurs may varybecause of ecological factors and the heterogeneity of the resourcebase. Resource-use curves are shown at the top of the diagram.Species that have become resource specialists (model a) are lesslikely to hybridize than are generalists (model b). These hypotheticalrepresentations were offered by Grant (1994) for the Hawaiianhoneycreepers (model a) and Galápagos finches (model b),respectively. (Redrawn from Grant 1994, Fig. 5.)

Page 239: Whittaker. Island Biogeography. 2007

226 E M E R G E N T M O D E L S O F I S L A N D E VO L U T I O N

0 100 200 km

N

Oahu(4)Kauai

(13)

Mauicomplex

(12)

Hawaii(7)

Oahu(29)

Kauai(12)

Mauicomplex

(40)

Hawaii(26)

2

5

1

1

1

3

2

2

2

1

6

10

7

15

Figure 9.8 The dispersal events among the Hawaiian islandssuggested by the inter-relationships of the species of picture-wingedDrosophila flies (above) and of silverswords (tarweeds) (below).Arrow widths are proportional to the number of dispersal events, andthe number of species in each island is shown in parentheses.(Redrawn from Cox and Moore 1993, Fig. 6.9, after an original inCarr, G.D. et al. (1989). Adaptive radiation of the Hawaiiansilversword alliance (Compositae—Madiinae): a comparison withHawaiian picture-winged Drosophila. In Genetics, speciation and thefounder principle, (ed. L. Y. Giddings, K. Y. Kaneshiro, and W. W.Anderson), pp. 79–95. © Oxford University Press, New York).

genera Drosophila and Scaptomyza) account for some600–700 species (Brown and Gibson 1983, afterCarson et al. 1970)—although, as noted earlier, theeventual figure could be as great as 1000 species(Wagner and Funk 1995). The ancestors of theHawaiian drosophilids (a single species, or at mosttwo) probably arrived on one of the older, now sub-merged islands and have radiated within andbetween islands. Carson and colleagues have stud-ied the chromosome structure of members of thepicture-wing group, using the karyotypic phy-logeny as a means of reconstructing the radiation.

The sequence, as might be expected, appears tostem from the oldest island, Kauai (age 5.1 millionyears), and, in general, the older islands to the westcontain species ancestral to those of the youngerislands to the east, i.e. most, but not quite all, of theinterisland colonizations have been from older toyounger islands (this is termed the progressionrule). In undertaking this analysis, the islands ofthe Maui Nui complex (Maui, Molokai, Lanai, andKahoolawe) may be taken as a single unit, as theyhave fused and separated at least twice in theirshort history, as a result of sea-level changes,adding another element of complexity to the picture.Starting from Kauai, at least 22 interisland colo-nizations are required to account for the phylogenyof the picture-wing species found on the islands ofOahu, the Maui Nui complex, and the Big Island ofHawaii itself. The precise details of this sequence ofevents may vary according to the means and dataused to construct the phylogeny, but the broad pat-tern is clearly established (Fig. 9.8). Furthermore,the trend of colonization from old to young islandsfound for these drosophilids is the most frequentfound amongst the other lineages investigated todate (Wagner and Funk 1995). However, the major-ity of speciation events among the drosophilidshave occurred within islands and, for example,Carson (1983) notes that of 103 picture-wingDrosophila species, all but 3 are endemic to singleislands or island complexes. Brown and Gibson(1983, citing Carson et al. 1970) report that nearly 50species of Drosophila have been reared from theleaves of Cheirodendron (in which they occur as leafminers). If species can currently coexist in aCheirodendron forest, and even be found in the same

leaf, it is difficult to rule out the possibility that insome cases populations may have diverged whilein sympatry. However, as already indicated, incommon with other large oceanic islands (Box 9.1)these islands have provided tremendous opportu-nities for within-island isolation, and it is generallyconsidered that isolation of this form has beencrucial to lineage development..

One form of within-island isolation prevalent onHawaii is when lava flowing down the flanks

Page 240: Whittaker. Island Biogeography. 2007

A D A P T I V E R A D I AT I O N 227

Box 9.1 Canarian spiders, beetles, and snails

The Canarian invertebrates provide anoutstanding example of radiation on rather lessisolated islands than Hawaii and the Galápagos.There may be as many as 6000 native Canarianinvertebrate species, of which half are consideredto be endemic (Oromí and Báez 2001). Ingeneral, the Canarian invertebrate fauna ischaracterized by the absence of certain groups(e.g. of scorpions, light-worms, Cicadae,Solifugae, and dung-beetles) but this has beencompensated for through active speciation withincolonist lineages, producing a high species–genusratio. Additionally, 99 genera are consideredexclusive to the archipelago, 58 of them beingpresent in more than one island (multi-islandgenera) and 41 restricted to one island, of which25 are found on the island of Tenerife (Oromíand Báez 2001). Not all lineages have radiated extensively; for example, as many as 57 of theCanarian endemic genera are monotypic,indicating a lengthy period of insular isolationand change without radiation.

Radiation is particularly evident in land snails,spiders and beetles, with at least 24 differentgenera producing 15 or more endemic species (seeTable). The largest genera of beetles (Laparocerus),diplopods (Dolichoiulus), and spiders (Dysdera) arerepresented in all the major habitats, from thecoast up to 3000 m, in arid and wet zones, inforested and open areas (even in lava tubes), allover the archipelago, a pattern strongly supportiveof the label adaptive radiation. Several of thegenera listed in the table are also present on otherMacaronesian islands, although the degree towhich they have radiated is variable betweenarchipelagos (Oromí and Báez 2001). For instance,among diplopods, there are 2 species ofDolichoiulus on Madeira but 46 on the Canaries; incontrast, the figures for Cylindroiulus are 25species on Madeira and 2 on the Canaries. Thismay reflect the timing of the colonization of eacharchipelago by different genera, but thisexplanation requires testing via molecularphylogenies.

As on other geologically young volcanic islands(e.g. Hawaii), there are fine examples of lavatubes on the Canaries. These features provide verysimilar habitats that are also very different fromthe environment above ground. Moreover, theyare typically quite isolated within each island.Different tubes often feature their own sets ofalbino and blind Dysdera spiders and Lobopteracockroaches. Given the difficulty of such formsdispersing between lava tubes, they are mostparsimoniously explained as examples of parallelevolution from above-ground relatives.

Number of species of endemic invertebrates of the CanaryIslands, within Canarian endemic (bold) and non-endemic generarepresented on the islands by at least 15 endemic species(source: Oromí and Báez 2001)

Group Name Number of species

Snails Hemicycla 76Napaeus 45Obelus 21Plutonia 18

Spiders Dysdera 43Oecobius 35Spermophorides 22Pholcus 16

Diplopods Dolichoiulus 46Hemiptera Cyphopterum 24

Asianidia 17Issus 15

Beetles Laparocerus 68Attalus 51Cardiophorus 31Tarphius 30Acalles 27Calathus 24Hegeter 22Nesotes 20Oxypoda 16Pachydema 16Trechus 15

Isopods Porcellio 18

Page 241: Whittaker. Island Biogeography. 2007

leaves undamaged ‘islands’ of forest, termedkipukas, which can vary from a few square metres tomany hectares in area. Whether kipukas maintaintheir isolation long enough to allow for speciationmay vary between taxa, but in one particularlywell-studied picture-wing, Drosophila silvestris, ithas been found that populations in differentisolates show a remarkable degree of genetic differ-entiation. Carson et al. (1990) and Carson (1992)have suggested that the many phases of develop-ment of these shield volcanoes provide conditionsof continuing division and flux, acting as a generalcrucible for parallel active evolutionary change inthe associated organisms.

The forces that created the Hawaiian islands and theirlong-ago foundered predecessors not only formed thefirmaments upon which life could diversify but also mayhave played a heretofore unappreciated direct role in theacceleration of evolutionary processes as they operate inlocal populations (Carson et al. 1990, p. 7057).

Geological evidence suggests that the surfaceof Mauna Loa (the main volcano of the island ofHawaii), has been replaced at an average rate ofabout 40% per 1000 years during the Holocene, andaccordingly the populations of Drosophila and otherspecies of the volcano flanks must have beenrepeatedly ‘on the move’ (see also Wagner andFunk 1995). It is thus entirely reasonable to positthat species which are sympatric today were at leastlocally allopatric at the time their lineages diverged.At this sort of scale, however, the terms allopatryand sympatry are best regarded as hypothetical endpoints, potentially involving little or no interchangeon the one hand, and free interchange on the other.Given the dynamism of these environments in evo-lutionary time, and the fluxes in population sizesand connectivity which are implied, the reality maylie somewhere between these absolutes, with eachspecies existing in the form of metapopulations(Chapter 10), i.e. units experiencing a degree ofinterchange and of supply and re-supply as localisolates wax and wane (Carson 1992).

Adaptive radiation in plants

Figure 9.8 shows that the general trend of colon-ization from old to young Hawaiian islands inDrosophila (the progression rule), is matched by

similar trends in the radiation of the silverswordalliance (tarweeds), which is a group of 30 speciesin 3 genera (Dubautia, Argyroxiphium, and Wilkesia)in the Asteraceae. The alliance appears to havedescended from a colonization event by a bird-dispersed herbaceous colonist (itself a hybrid fromthe genera Madia and Raillardiopsis) about 5 Ma,matching the age of the oldest island, Kauai(reviewed by Emerson 2002). They have diversifiedinto a range of life forms, including monocarpicrosette plants, vines, mat-forming plants, and trees,and they occur in habitats ranging from desert-liketo rain forest.

For Macaronesia, Humphries (1979) estimated anative vascular flora of 3200 species, of which about680 (c.20%) are endemic. The Canaries are the largestin area, the highest, and the richest, with 570endemics, about 40% of the native flora. Forty-fourMacaronesian lineages are endemic at the genericlevel also, with exactly half of them restricted to theCanaries (Santos-Guerra 2001). Some lineages haveradiated to a significant degree, with Argyranthemum(26 species; Asteraceae), Monanthes (17 species; Cra-ssulaceae), and Aichryson (14 species; Crassulaceae)the most diverse endemic genera. Other non-endemic genera have radiated to even greaterextents, notably Aeonium (31 species; Crassulaceae),Sonchus (29 species; Asteraceae), and Echium (28species; Boraginaceae) (Table 9.3).

Support for the use of the term ‘adaptive radia-tion’ in several of the above Macaronesian generacomes from studies of features such as habit, leafmorphology, and habitat affinities (e.g. Table 9.4).Humphries (1979) argues that evidence of similarmorphology in species of the same and of differentgenera (termed parallel or convergent morphology)occurring in similar habitats, is supportive of amodel of selection having favoured particularadaptive outcomes.

Within the Macaronesian flora as a whole, mostgenera are represented by only one or two species,and most of those with over four occur in theCanaries. In general, radiation of lineages has beengreatest on the larger islands with the greatestdiversity of habitats. The island with the greatestnumber of endemic species (320 species) is Tenerifeand it also has the largest numbers of Aeonium(12 species), Sonchus (11 species), and Echium

228 E M E R G E N T M O D E L S O F I S L A N D E VO L U T I O N

Page 242: Whittaker. Island Biogeography. 2007

A D A P T I V E R A D I AT I O N 229

Table 9.3 Endemic (in bold) and non-endemic Macaronesian vascular plant genera represented by 10 ormore endemic species. (Sources: Santos-Guerra 2001 for Macaronesia; Borges et al. 2005 for the Azores;Sziemer 2000 for Maderia; Izquierdo et al. 2001 for the Canaries; Arechavaleta et al. 2005 for the Cape Verdeislands). When total species number does not match the sum, some species are shared among archipelagos

Plant genus Azores Madeira Canaries Cape Verde Total

Aeonium – 2 28 1 31Echium – 2 23 3 28Lotus 1 5 19 5 28Argyranthemum – 4 22 – 26Sideritis – 1 23 – 24Limonium – 2 18 5 23Sonchus – 3 18 1 22Monanthes – 1 16 – 17Cheirolophus – 1 15 – 16Micromeria – 1 15 1 16Pericallis 1 1 13 – 15Euphorbia 2 2 10 1 14Aichryson 1 3 11 – 14Crambe – 1 12 – 13Teline – 1 9 1 11Convolvulus – 1 10 – 11Tolpis 2 2 7 1 11

Table 9.4 Adaptive radiation in the endemic genus Argyranthemum (Asteraceae) in the Macaronesian islands (data fromHumphries 1979). Francisco-Ortega et al. (1996) recognize 26 monophyletic species in this genuss

Environment Adaptive features Islands Species

Broadleaved forests Type A: shrubby, unreduced Tenerife and Gomera A. broussonetiileaves Madeira A. adauctum ssp.

jacobaeifolium and A. pinnatifidum

South-facing, lowland arid Type B/C: reduced Tenerife A. frutescens ssp. gracilescens, andlignification, habit, A. gracilecapitulum size, and leaf area

Gran Canaria A. filifolium

Sub-alpine and high Type D: dies back after Tenerife A. tenerifae and A. adauctum ssp.montane flowering each year, leaves dugourii

very dissected, hairy

Exposed northern coastal Type E: reduced habit, large Canaries A. coronopifolium,areas capitula, increased A. frutescens ssp. canariae, ssp.

succulence succulentum and A. maderense

(9 species). Most endemic species have a restricteddistribution, 48% occurring on a single island, and afurther 15% on two. From this and other considera-tions, Humphries (1979) concluded that allopatricspeciation has been crucial in the flowering plants.

Furthermore, from studies of a sample of 350 of theendemic species, he concluded that 90% hadevolved by gradual divergence following the break-up of a population (including migration from onearea to another), with the remainder accounted for

Page 243: Whittaker. Island Biogeography. 2007

by abrupt speciation involving hybridization, poly-ploidy, and other forms of sudden chromosomalchange. However, it appears that those notably fewgroups which have radiated adaptively into a widerange of different conditions, have done so withoutmajor chromosomal change (Humphries 1979).

In general, it appears that the ‘species swarms’represented by the species-rich Macaronesiangenera are essentially interfertile, i.e. they readilyhybridize, particularly in disturbed areas. How-ever, stabilized hybrids are rare and so it seems thatthe ecological differentiation between species is suf-ficient to maintain species integrity in most groups.It follows that although gene flow through intro-gressive hybridization may occur between species,hybridization has led to few speciation ‘events’(Humphries 1979; Silvertown et al. 2005). Thus,while details differ, it appears that the essential fea-tures of evolution in the Macaronesian flora, incases involving spectacular radiations, are commonto other studies encountered in this section (see alsoKim etal. 1996). In terms of Rosenzweig’s (1995)three modes of speciation, the data support a rolefor each of the geographical, competitive, and poly-ploid models in declining order of significance.

9.4 From valley isolates to island-hopping radiations

Both the taxon cycle and adaptive radiation scenar-ios involve a focus on evolutionary–ecologicalmechanisms to account for the ways in which line-ages develop within islands and archipelagos. Inthis section we consider scenarios that are essen-tially biogeographical rather than mechanisticmodels of island evolution. Here the focus is onreconstructing the sequence of movements acrossarchipelagos that have been followed by particularlineages, and establishing if in sum there are emer-gent tendencies within biotas which may be relatedto the developmental history of the islands, or if thepatterns found could be accounted for by chancedispersal and isolation events within and betweenarchipelagos. We start with an illustration of howthe lack of dispersal can lead to a form of non-adap-tive (selectively neutral) radiation within a single

island and progress to tracing lineages that havehopped back and forth across entire ocean basins.

Non-adaptive radiation

Gittenberger (1991) discusses the radiation of landsnails of the genus Albinaria within the island ofCrete. They have diversified into a species-richgenus without much niche differentiation, to occupymore or less the same or only a narrow range of habi-tats, yet only rarely do more than two Albinariaspecies live in the same place. Gittenberger thusposits non-adaptive radiation as a logical alternativeto adaptive radiation, while recognizing that theprocess of radiation may involve a blend of the twoover time (cf. Lack 1947a). Cameron et al. (1996), intheir study of land snails from Porto Santo island(Madeira), also concluded that the majority of the 47endemic species recorded had arisen from radiationof an essentially non-adaptive character by repeatedisolation within this highly dissected volcanic island.Barrett (1996) suggests that the term non-adaptiveradiation may also be an appropriate way of describ-ing diversification in particular plant lineages, citingspecifically Aegean island populations of Erysimum.Macaronesian plant examples might include manyof the species of Limonium (23 species), Cheirolophus(16 species), or Helianthemum (9 species) that exploitsimilar habitats in different islands.

Non-adaptive radiation as described above isattributable to stochastic mechanisms, such as gen-etic drift acting on small, isolated populations, pro-ducing an array of sibling species within an islandor archipelago. Within animals, another possiblyimportant mechanism is sexual selection, which hasbeen invoked in some studies of cichlid fishes ascontributing a selective yet non-adaptive compo-nent within a spectacular radiation (see Box 9.2).

Speciation within an archipelago

We may differentiate endemic species within anarchipelago on simple distributional grounds intosingle- and multiple-island endemics. Single-island endemics (SIEs) are those species restrictedto a single island. They may be the result of recentin situ speciation or may represent the relictual

230 E M E R G E N T M O D E L S O F I S L A N D E VO L U T I O N

Page 244: Whittaker. Island Biogeography. 2007

F R O M VA L L E Y I S O L AT E S TO I S L A N D - H O P P I N G R A D I AT I O N S 231

Box 9.2 The explosive radiation of cichlid fishes in the African Great Lakes

When considering freshwater organisms, lakescan be viewed as islands, and the African GreatLakes (Tanganyika, Malawi, Victoria, etc.),clustered in the Rift Valley, can be considered asan archipelago. They possess one of the mostimpressive examples of radiation in the form ofabout 1000 species of cichlid fishes (Cichlidae),with some 200 in Lake Tanganyika, 300 in LakeVictoria, and over 400 in Lake Malawi. AlthoughLake Tanganyika’s cichlids very likely derive frommultiple riverine ancestors, those in Lake Victoriaapparently derive from just two separate ancestralspecies and those in Lake Malawi from a singlefounder population (Danley and Kocher 2001).

Lake Tanganyika developed as a shallowdepression linked to the Congo River systemabout 20 Ma, and achieved its isolation about9–12 Ma. Molecular clock estimates indicate thatthe Tanganyika lineages diverged from theirancestors at least 5 Ma. Lake Malawi is at most1–2 million years old, and the cichlid radiation isestimated to date back only approximately0.7 Ma. Finally, Lake Victoria is considered to dateto between 0.2–0.7 Ma, but may have beencompletely dry as recently as 12 400 years ago,suggesting a truly explosive rate of radiationsubsequently (Johnson et al. 1996).

Many models have been proposed inexplanation of these spectacular cases of

radiation. One of the more straightforward is thethree-stage radiation model (Streelman andDanley 2003):

1 Major habitat diversification. In the firstadaptive radiation stage, an initial divergence ofthe lineages takes place, resulting in the appear-ance of different clades related to the majoravailable habitats: the rock-dwellers, the sandy-substrate dwellers, and the pelagic species.

2 Trophic diversification. This second adaptivestage is well studied in the rock-dwellers of LakeMalawi, and involves the rapid directionalselection of the feeding apparatus, attainingresource partitioning between species. Thisprocess has yielded 10–12 recognized endemicgenera, featuring morphological differences infeatures such as jaw shape and tooth structure.

3 Sexual selection. The most intriguing step ofthe model is the third phase, a non-adaptive radi-ation process, involving sexual selection throughmate choice by females. This mechanism relies oncolour differences in males and is thought tooccur only in transparent waters, where thecolours are evident. As cormorants are thought tobe able to predate the colourful males more easilyin transparent waters, it is assumed that brightcolouration must come at a cost to fitness andthus can be considered essentially non-adaptive.

distribution of a formerly more widely distributedspecies, which may or may not have originated onthe island to which it is currently restricted. Forinstance, Saxicola dacotiae (the Canary island chat) istoday restricted to Fuerteventura, having beenextirpated from the nearby islet of Alegranza a cen-tury ago. Although not endemic, another similarcase is Pyrrhocorax pyrrhocorax (red-billed chough),now restricted within the Canaries to La Palma butknown from fossil evidence to have occurred onTenerife, La Gomera, and El Hierro (Martín andLorenzo 2001). Such losses may be the product ofnatural causes as envisioned within the taxon cyclebut in this case are attributable to humans.

In contrast, multiple-island endemics (MIEs) areattributable to three different scenarios.

● First, they may have originated on one island,most typically the oldest island, and have subse-quently colonized a younger island or islands,where as yet there has been insufficient time (or toomuch interisland gene flow) for speciation to haveoccurred.● Second, they may be palaeoendemics that haveevolved little since colonizing an archipelago, buthave been lost from their continental source areas.● Third, in a small number of cases they may beexamples of parallel microevolution, in which sim-ilar environmental pressures on separate islands

Page 245: Whittaker. Island Biogeography. 2007

have favoured the same mutations, resulting in tworelated taxa of separate (i.e. paraphyletic) originbeing incorrectly classified as one species. The first scenario is considered by far the mostfrequent (e.g. Funk and Wagner 1995).

Examples of parallel microevolution have beendocumented within Canarian Nesotes beetles. Onmorphological criteria, N. fusculus was considereda MIE, occupying the xeric coastal zones of GranCanaria, La Gomera and Tenerife. However, phy-logenetic analysis showed this ‘species’ to be para-phyletic, with the ‘fusculus’ phenotype havingevolved independently on each of the threeislands (Rees et al. 2001). A similar scenario appliesto N. conformis, a laurel forest dweller, found onGran Canaria, Tenerife, La Palma, and El Hierro.Phylogenetic analyses revealed that the laurelforest phenotype evolved twice within GranCanaria and once more on the western Canaries(Emerson 2002). This last example appears toshow that paraphylly can pertain even for anapparent SIE.

Just over half (57%) of the endemic Canarianflora (bryophytes �vascular plant species) areSIEs, with two thirds of these species being foundeither on Tenerife or on Gran Canaria (Table 9.5).Nevertheless, each island also harbours more MIEsthan SIE species. For instance, Tenerife harbours151 MIEs and 107 SIEs and Gran Canaria harbours

100 MIEs and 75 SIEs. Intriguingly, the number ofendemic species found on two or more islandsdoes not decline smoothly with increasing numberof islands as might be expected, with some 45species being shared by the five high islands ofGran Canaria, Tenerife, La Palma, La Gomera, andEl Hierro (Table 9.5). This pattern is repeated foranimals, of which 57.5% (1679 species) are SIEs and42.5% (1239 species) are MIEs. Tenerife (644species) and Gran Canaria (351 species) againaccount for the majority of SIEs (data fromIzquierdo et al. 2001).

How proportions of SIEs and MIEs vary bet-ween taxa and archipelagos remains to be estab-lished. We provide a preliminary compilation offloristic data for three further volcanic archipela-gos: the Azores (Borges et al. 2005), the Cape Verdeislands (Arechavaleta et al. 2005) and the Galápagos(Lawesson et al. 1987) (Table 9.6). As in theCanaries, the SIEs of the Cape Verde and Galápagosare the largest single category, but they account foronly 22% of endemics, in contrast to 57% SIEs inthe Canarian endemic flora. In the Azorean flora,the commonest pattern is of MIEs found on all nineislands, again constituting 22% of endemics. Thesesimple analyses serve to show that the proportionsof SIEs and MIEs vary considerably between archi-pelagos. We can only speculate as to the reasons forthese differences. It may be that the Canaries are insome respects more effectively isolated from one

232 E M E R G E N T M O D E L S O F I S L A N D E VO L U T I O N

Table 9.5 Distribution patterns amongst the endemic plant species of the Canaries (Source: Izquierdo et al. 2001, reworked)

Islands occupied 1 2 3 4 5 6 7 Total

Species no. 289 75 44 30 45 9 19 511% 56.6 14.7 8.6 5.9 8.8 1.8 3.7 100More frequent distributions in T 107 TC 18 PGT 9 PGTC 8 HPGTC 42 HPGTCF 5order of importance

C 75 PT 17 GTC 8 HPGT 7 GTCFL 2 PGTCFL 3

P 36 FL 17 HPT 5 HGTC 7G 35 GT 11 HPG 5 TCFL 3H 13 HP 4 CFL 5 HPTC 2F 11 PG 3 HGT 4L 12 HG 2

T, Tenerife; C, Gran Canaria; P, La Palma; G, La Gomera; H, El Hierro; F, Fuerteventura; L, Lanzarote.

Page 246: Whittaker. Island Biogeography. 2007

another than is the case for the other archipelagos,so that the spread of newly evolved endemicsacross the archipelago is hampered. For instance,several islands of the Galápagos archipelago (SantaCruz, Santiago, Baltra, Pinzón, Rábida, Santa Fe,Isabela, and Fernandina) were merged into a singleentity during the last glaciation event, as also hap-pened in the Cape Verde islands. However, this wasnot the case for the archipelago showing the great-est tendency to multiple-island occupancy, theAzores, where only the Faial–Pico fusion occurred(F. García-Talavera, personal communication).Alternatively, and perhaps of greater relevance, thehigh proportion of SIEs on the Canaries may reflectgreater opportunities for within-island speciationand radiation provided by the high topographiccomplexity and habitat diversity of the westernislands of the group, especially within Tenerife, withits steeply dissected Tertiary age Anaga and Tenopeninsulars, and the central Quaternary age massif,which is capped by the 3718 m peak of El Teide.

Island-hopping allopatric radiations: do cladesrespond to islands or to habitats?

Oceanic island archipelagos have been describedas ‘speciation machines’ (Rosenzweig 1995). But,as we have seen, the proportions of single-and multi-ple-island endemics may be highly variable betweenarchipelagos, indicating that these ‘machines’

can function in rather different ways. As highlightedin Box 8.2, the application of the label ‘archipelagospeciation’ does not tell us whether speciation isadaptive or non-adaptive. Advances in the applica-tion of molecular analyses in recent decades allow usto approach these phenomena in rather differentways, and to ask different questions (Emerson 2002;Silvertown et al. 2005), such as: (1) do clades respondto islands or to habitats? (2) do they move acrossoceans by routes determined by the shortest dis-tances? (3) do they move back and forth rather thanunidirectionally?

Within an archipelago we may posit two hypo-thetical extreme scenarios for the speciationmachine that may be distinguishable in the phylo-geographical structure of lineages (clades). First, aparticular taxon may respond to islands more thanto habitats, i.e. species inhabiting an island aregenetically closer to other species on the sameisland than to species exploiting the same ecosys-tem on different islands. In this case, a singlefounder event per island is involved in the process,followed by an adaptive radiation driven bynatural selection, which can be more or less com-plex according to the island’s habitat diversity.Second, we can think of clades responding to habi-tats instead of to islands if the species in a certainhabitat are taxonomically closer to other speciesoccupying the same habitat all over the archipel-ago, than they are to their island congenerics. This

F R O M VA L L E Y I S O L AT E S TO I S L A N D - H O P P I N G R A D I AT I O N S 233

Table 9.6 Endemic plant species distribution in the Azores, the Cape Verde and the Galápagos archipelagos (Azores and Cape Verde datainclude both bryophytes and vascular plants, whilst Galápagos data only include vascular plants). The data shown are the number and percentageof plant species in each distributional class. The largest class is highlighted in bold for each archipelago (Sources: Borges et al. 2005 for Azores;Arechavaleta et al. 2005 for Cape Verde; Lawesson et al. 1987 for Galápagos)

Archipelago 1 2 3 4 5 6 7 8 9 or Totalisland islands islands islands islands islands islands islands more endemic

islands species

Azores 4 5 2 5 8 8 6 13 17 77(%) 5.2 6.5 2.6 6.5 10.4 10.4 7.8 17.0 22.1 (100)Cape Verde 16 6 12 2 8 10 8 1 9 72(%) 22.2 8.3 16.7 2.8 11.1 14.0 11.1 1.4 12.5 (100)Galápagos 32 16 14 17 9 10 13 8 29 148(%) 21.6 10.8 9.5 11.5 6.1 6.8 8.8 5.4 19.6 (100)

Page 247: Whittaker. Island Biogeography. 2007

scenario needs many more founder events, at leastone for each occupied habitat on each island, andwould likely involve colonization events back andforth within an archipelago. We have cast theseextreme scenarios at the species level, but theymight also be considered at the subspecies orpopulation levels.

What do phylogeographical analyses show? Letus begin with evidence that the two extremescenarios proposed are realistic. For the first sce-nario (responding to islands), an interesting animalexample of intraisland radiation is provided byanolid lizards in the Greater Antilles, where asmany as 140 different species occur. Cuba andHispaniola are each inhabited by six different eco-morphs (species specialized to use a particularstructural microhabitat: crown–giant, grass–bush,trunk, trunk–crown, trunk–ground, and twig),whereas five ecomorphs live on Puerto Rico andfour on Jamaica. The presence of the same set ofecomorphs on each island might be accounted forin two ways: (1) each evolved once and then colo-nized other islands, or (2) each ecomorph evolvedindependently on each island (Emerson 2002).The phylogeny of anoline lizards depicted by Lososet aal. (1998) for the Great Antilles suggests that, withtwo exceptions, the members of the same ecomorphclass on different islands are not closely related.Rather, the species assemblages on each island seemto be the result of independent events of adaptiveradiation. For the second scenario (responding tohabitats) Francisco-Ortega et al. (1996) provide ananalysis of the endemic Macaronesian plant genusArgyranthemum, which contains 26 monophyleticspecies. Between them, they have colonized almostall the habitats existing in the Canaries (see Table9.4). The phylogeny supports the notion thatArgyranthemum respond largely to habitats, whichmeans that during their evolution within this islandregion they have often dispersed successfullybetween islands, but also that they have generallyremained within the habitat type in which they orig-inated. Similar vagility of species among the islandsof the Canaries has been revealed in the phylogeog-raphy of the plant genera Aeonium, Crambe, Sideritis,and Sonchus, although these clades show more fre-quent evolutionary shifts between habitats than

have occurred in Argyranthemum (see e.g. Kim et al.1996; Silvertown et al. 2005; Trusty et al. 2005).

Funk and Wagner (1995) analysed phylogeo-graphical data from genetic analyses of more than20 different Hawaiian taxa (among them crickets,Drosophila, spiders, honeycreepers, the silverswordalliance, and Clermontia bellflowers). They inferredmore than 100 speciation events associated withinterisland dispersal, whereas intraisland radiationaccounted for over 200 speciation events, thus yield-ing an approximate ratio of 1:2. This distinction is,however, oversimplistic as most of these eventsmerely form one part of more complex evolutionaryscenarios. They recognize nine of them: four basicpatterns (progression rule, intraisland radiation,stochastic, and back-dispersal), four combined pat-terns (progressive clades and grades, terminal reso-lution, recent colonization, and extinction), andfinally the possibility of an unresolved pattern; towhich we have added two further scenarios derivedfrom studies in the Canaries: repeated colonization,and fusion of palaeoislands (Table 9.7). The follow-ing exemplification is drawn from Funk andWagner (1995) except where indicated.

● Progression rule. Taxa exemplifying this patternshow dispersion from the older to the youngerislands of the archipelago, with or without in situradiation. In the Hawaiian context, this implies ini-tial colonization of Kauai, the oldest extant highisland, when it was still young, either from the con-tinent, or from an older now-submerged or erodedisland (see Chapter 2). As each new volcanic islandbecame available for colonization, dispersaloccurred from the older to the younger island, withspeciation following. This pattern is displayed byDrosophila flies, Tetragnatha spiders, and the plantgenera Hesperomannia, Remya (Asteraceae), and Kokia(Malvaceae). In some archipelagos, island age corre-lates with distance from the principal source regionthus potentially confounding interpretation. Forexample, the age of the Canarian islands decreaseswith distance from Africa (i.e. from east to west).This colonization path has been taken by many taxa,e.g. the wild olive tree Olea, Gallotia lizards, Hegeterbeetles, and Gonopteryx butterflies (Fig. 9.9; Thorpeet al. 1994; Marrero and Francisco-Ortega 2001a,b).

234 E M E R G E N T M O D E L S O F I S L A N D E VO L U T I O N

Page 248: Whittaker. Island Biogeography. 2007

However, the Canaries also provide exampleswhere phylogeographical analysis has shown a pat-tern of colonization not predictable by distance tothe nearest continent and in opposition to the pro-gression rule. For instance, finches of the genusFringilla (chaffinches), originally from the Iberianpeninsular, colonized the western Canaries via theAzores and Madeira (Marshall and Baker 1999),while Pericallis (Asteraceae) may have colonizedthe Canaries from America (Panero et al. 1999).

● Intraisland radiation. Within-island radiationfollowing colonization by an ancestral species hasalready been discussed at length, and may involveboth adaptive radiation and non-adaptive radia-tion. The latter may be related to periods of intrais-land isolation and the complex geological andenvironmental history of an island. Hawaiianexamples of intraisland radiation include Geranium,silverswords (Argyroxiphium and Wilkesia), andPlatydesma (Rutaceae). This pattern is frequentlyfound as part of a larger archipelagic radiation.

● Stochastic. Apparent stochasticity in distributionin relation to the developmental history of thearchipelago may result from high vagility (as seems

to be the case for the Hawaiian honeycreepers(Drepanidinae) in so far as we can understand theirnow fragmented distribution), or from a recentestablishment on the archipelago, colonizing adja-cent islands in a random fashion (e.g. Tetramolopium,in the Asteraceae).

● Back-dispersal. This refers to lieages that gener-ally may follow the progression rule, but in whichone or more cases of dispersal from a younger to anolder island occur, with subsequent speciation occur-ring from the back-dispersed colonist. This phenom-enon can only be detected with certainty bymolecular analysis. Examples include Progna-thogryllus crickets, and Kokia (Malvaceae) on Hawaii(Funk and Wagner 1995) and Pimelia beetles in theCanaries (Juan et al. 2000). Recent studies havedemonstrated that the back-dispersal pattern caneven reach the continent from which the ancestralspecies of the lineage originated. Examples of plantlineages back-colonizing Africa from the Canariesinclude species in the genera Sonchus (S. bourgaeui, S.pinnatifidus), Aeonium (A. arboreum, A. korneliuslemsii)(Santos-Guerra 1999) and Convolvulus (C. fernandesii)(Carine et al. 2004).

F R O M VA L L E Y I S O L AT E S TO I S L A N D - H O P P I N G R A D I AT I O N S 235

Table 9.7 Biogeographical-evolutionary patterns exhibited by selected Hawaiian (Funk and Wagner 1995) andMacaronesian taxa (sources given in text). Some taxa provide examples of several patterns

Pattern typology Hawaiian taxa Macaronesian taxa

Progression rule Drosophila, Hesperomannia, Olea, Gallotia, Hegeter, GonopteryxHibiscadelphus, Kokia, Remya,Tetragnatha

Radiation Clermontia, Cyanea, Geranium, Echium, Limonium, PericallisPrognathogryllus, Platydesma

Stochastic Drepanidinae, Scaveola, Echium, CrassulaceaeBack-dispersal Clermontia, Geranium, Kokia, Pimelia

Laupala, SaronaProgressive clades and grades Drosophila, Laupala, DubautiaTerminal resolution Cyanea, Drosophila, SaronaRecent colonization Tetramolopium, Clermontia Parus caeruleusExtinction Clermontia, Geranium, ArgyroxyphiumRepeated colonization Unlikely to have occurred Calathus, Tarentola, Ilex, Lavatera,

HederaFusion of palaeo-islands Gallotia, Steganecaurus, Pimelia,

Eutrichopus, DysderaUnresolved Clermontia, Sarona Chalcides

Page 249: Whittaker. Island Biogeography. 2007

● Progressive clades and grades. A clade is anentire group descended from one common ances-tor (i.e. it is monophyletic). Grade refers instead tothe level of adaptation, such that two organismscan be classed as of equivalent grade but not have acommon ancestor (e.g. Old World and New Worldfruit bats). Progressive clades and grades are joinedtogether because both result from radiations thatdiffer only in which taxon of the group is involvedin the dispersion to another island. This complexcombined pattern presents a common feature inwhich species from a particular island are groupedtogether (either in a clade or in a grade) and theoverall pattern in the area cladogram is consistentwith the progression rule. Hawaiian examples are

shown by Tetragnatha spiders, Prognathogryillus andLaupala crickets, as well as Hybiscadelphus(Malvaceae) and Dubautia (Asteraceae) plant genera.● Terminal resolution. Taxa showing this patternhave either an unresolved base or a base that hasexperienced a radiation on an older island, withsubpatterns involving several well-defined clades.The latter can be either progressions or radiations,in each case indicating repeated introductions froman older, now largely or entirely submerged island.Hawaiian examples can be found in the plant gen-era Schiedea and Alisinidendron (Caryophyllaceae)and Cyanea (Campanulaceae), as well as in Sarona(Miridae) plant bugs.

236 E M E R G E N T M O D E L S O F I S L A N D E VO L U T I O N

Canary Islands

L

F

C

TG

H

P AFRICA

TPHH G C F L Isl.

?

?

Gallotiagen.

(*)

(*) Psammodromus

Gallotiinae

LacertinaeEurasia-Africa

Africa

(a)

Figure 9.9 Phylogenies of (a) Gallotia lizards (based on mtDNA) and (b) Olea europeaea subspecies cerasiformis(based on internal transcribed spacer 1 (ITS-1) sequences, randomly amplified polymorphic DNAs (RAPD), andintersimple sequence repeats (ISSR) analyses) on the Canary Islands (after Marrero and Francisco-Ortega 2001 figures15.5 and 15.6; in turn based on papers by González et al. 1996, Fu (2000) and Hess et al. (2000)). L = Lanzarote, L.Isl. = islets close to Lanzarote, F= Fuerteventura, C = Gran Canaria, T = Tenerife, G = La Gomera, H = El Hierro,P = La Palma.

Canary Islands(b)

L

F

C

TG

H

PAFRICA

TP GP C F

subsp. cerasiformisOlea europaea

G T CCC CC FCvar. sylvestrisWest Mediterranean

subsp. europaeaWest Mediterranean

Madeira

subsp. laperrineiMorocco–Hoggar

O. cuspidataEast and South Africa

Page 250: Whittaker. Island Biogeography. 2007

● Recent colonization. Where a comparativelyrecent colonization event has taken place, landfallmay have occurred on any of the extant islands, suchthat some taxa first colonize a young island, withsubsequent dispersal from younger to older islands,while others may colonize an intermediate-agedisland, subsequently dispersing to both younger andolder islands. This pattern seems to be followed byTetramolopium (Asteraceae) and Geranium(Geraniaceae) in the Hawaiian flora. The blue tit(Parus caeruleus teneriffae group) provides a Canarianexample, with mitochondrial DNA data indicatingfirst colonization on Tenerife and subsequent spreadboth to older and younger islands (Kvist et al. 2005).

● Extinction. If a taxon is specific to habitats suchas high summits, found only on young islands,then the loss of such habitats from older islandsmay result in extinction from all but the youngestislands. This pattern can be distinguished fromrecent colonization in cases where the molecularclock age estimate indicates a much longer periodof lineage development than could be provided bythe islands on which the taxon currently occurs. InHawaii this pattern is exemplified by Clermontia(Campanulaceae), Geranium (Geraniaceae), andArgyroxiphium (Asteraceae) plant genera.

● Unresolved. Some cladograms fail to provide aclear scenario either because of the inherent com-plexity of the evolutionary sequences involved orthrough lack of sufficient data.

● Repeated colonization from outside thearchipelago. This pattern would not be expected insuch a remote situation as Hawaii (although seeTerminal resolution, above), but appears common inthe Canaries (just 60 km from Africa during low sea-level stands). Repeat colonization produces para-phyly (separate clades) within a taxonomic group:for example, at least three different colonizationevents have been detected in Calathus beetles andTarentola geckos, and two in the plant genera Ilex(Aquifoliaceae), Hedera (Araliaceae), and Lavatera(Malvaceae) (Juan et al. 2000, Emerson 2002).However, within the flora, the greatest degree ofradiation occurs in genera that appear to be mono-phyletic (debated in Herben et al. 2005; Saundersand Gibson 2005; Silvertown et al. 2005).

● Fusion of palaeoislands. A repeated phylogeo-graphic pattern found within Tenerife, is theoccurrence of allopatric sister taxa in the Anaga(north-east) and Teno (north-west) peninsulas.These areas were separate islands until the con-struction of the Las Cañadas massif during thePleistocene fused them together into a single, muchlarger island (Chapter 2). Gallotia lizards, Chalcidesskinks, Pimelia, Calathus, and Eutrichopus beetles,Steganecarus mites, Dysdera spiders, and Lobopteracockroaches among animals (Juan et al. 2000) andAeonium, Cheirolophus, Limonium, Lotus, Sideritis,and Teline among plant genera (Báez et al. 2001),have present-day distributions related to this histor-ical sequence. For instance, Pimelia beetles showlineages associated with the west and the east ofTenerife, which have spread over the central regionsduring the Quaternary (Juan et al. 2000).

Island-hopping on the grand scale

As will already have become evident, molecularphylogenetics now allows biogeographers toreconstruct sequences of movement on the grandscale, across entire ocean basins. We have alreadygiven examples of how the Canaries have receivedcolonists not only from Africa, the nearest conti-nental landmass, but also from Europe via theAzores and Madeira, from the Cape Verde islandsto the south, and even from America. In short, theCanaries have received colonists from all points ofthe compass, and have also supplied back-coloniststo the African mainland.

The Indian Ocean also provides examples ofcomplex scenarios of island-hopping. For example,coastal lizards of the genus Cyrptoblepharus appearto have colonized the western Indian Oceanislands by transoceanic dispersal from Australia orIndonesia. The genus appears then to have diversi-fied in Madagascar from where it separately colo-nized the East African coast, the Comoro islands(twice) and Mauritius (Rocha et al. 2006). Galleyand Linder (2006), in their review of the affinities ofthe southern African Cape flora, find that in addi-tion to some ancient disjunctions across the IndianOcean, there is also strong evidence for some veryrecent disjunctions, which can only be explained by

F R O M VA L L E Y I S O L AT E S TO I S L A N D - H O P P I N G R A D I AT I O N S 237

Page 251: Whittaker. Island Biogeography. 2007

dispersal events. Examples include Pelargonium,Wurmbea, and Pentaschistis, each of which hasAustralian members that are recently derived fromCape groups. For a further illustration, Pentaschistisinsularis is one recently derived species, sister to anAfrican species, which is present on AmsterdamIsland, halfway between Cape Town and Australia.Galley and Linder (2006) therefore argue that theformation of biogeographic disjunctions across theIndian Ocean should be seen as ongoing and notjust a relictual pattern from break-up of theGondwanan supercontinent.

In biogeographical terms, New Zealand providesone of the most contested insular systems, withsome panbiogeographers arguing that the vast bulkof New Zealand’s biogeography is attributable toGondwanan break-up and rafting of the fragmentsto their present-day positions. Here again, molecularphylogenies have made telling contributions to thestory, showing that many New Zealand lineageshave diversified only relatively recently, followingthe onset of late Tertiary mountain building (5–2 Ma)and Pleistocene climatic change (Winkworth et al.2002). Moreover, as Winkworth et al. (2002) show intheir review, molecular phylogenies are revealingcomplex patterns of transoceanic dispersal across thesouthern Pacific since the late Tertiary, in many casesin opposition to present day current systems.Metrosideros subgenus Metrosideros, one of the mostwidely distributed woody plant groups in thePacific, provides a powerful illustration of this.Analyses based on nuclear ribosomal DNA indicatethat one monophyletic clade underwent a dramaticrange expansion across Polynesia (includingHawaii) from a Pleistocene dispersal event out ofNew Zealand. Wright et al. (2000) interpret this interms of changes in wind-flow patterns in the south-ern hemisphere related to glacial periods.

Some have questioned whether the molecularclocks can be relied upon, arguing that they might ineffect ‘run fast’ in explosive radiations. Bromham andWoolfit (2004) test this proposition using 19 inde-pendent cases of explosive radiations in islandendemic taxa, but find no evidence of a consistentincrease in rates compared to mainland relatives. Theevidence of such recent island-hopping radiationsdoes not rule out the existence of ancient historical

elements in the biogeography of the southern oceans,but does conclusively demonstrate that long-distancedispersal has played a key role in shaping the con-temporary biogeography of the region, especially forplants (Sanmartín and Ronquist 2004; Cook andCrisp 2005; McGlone 2005). Thus, island-hopping onthe grand scale goes on, and has presumably done sofor a long time and by diverse vectors. Not only hasit been the means of colonization of all true oceanicislands, but it has also been important in shaping thebiogeographical affinities of continental fragmentislands. Not only that, but it is becoming increasinglyapparent that island forms do successfully colonizemainlands (Santos-Guerra 2001; Carine et al. 2004;Nicholson et al. 2005; Rocha et al. 2006), providingexplanations for at least a minority of the disjunctionsin continental biotas on opposite sides of oceans(Galley and Linder 2006).

9.5 Observations on the forcing factorsof island evolution

Darwin was right in 1845 to point to oceanic islandsas being of great interest to the ‘philosophicalnaturalist’. Evolutionary change on islands hasbeen shown in this and previous chapters to comein many forms, from the microevolutionary alter-ations of morphology in near-shore mice popula-tions to the macro-scale radiation of lineages. Manygeographical, ecological, and genetic mechanismshave been invoked in explanation of the evolution-ary changes recorded. These mechanisms providethe mode of delivery of change, but what can wesay about the underlying forcing factors of evolu-tionary change on islands, such as competition,facilitation, impoverishment, disharmony andenvironmental change?

Competition has often been invoked in discus-sions of island evolutionary change, but as detailedin the island ecology section of this book such invo-cations are hard to bring to a definite conclusion.Even where micro-evolutionary changes appearattributable to competition, it is difficult to extendsuch reasoning to past evolutionary change withany degree of certainty as the extent of past compe-tition between two currently sympatric species isunknowable (Connell’s (1980) ‘ghost of competition

238 E M E R G E N T M O D E L S O F I S L A N D E VO L U T I O N

Page 252: Whittaker. Island Biogeography. 2007

O B S E R VAT I O N S O N T H E F O R C I N G FAC TO R S O F I S L A N D E VO L U T I O N 239

past’). In illustration, Silvertown (2004) has arguedthat genetic data for Canarian endemic plants indi-cate that niche pre-emption by earlier colonist line-ages may have inhibited the success of later ones.Emerson and Kolm (2005a), on the other hand,argue for a sort of facilitation effect, claiming thatdiversification rate of a taxon is an increasing func-tion of higher species richness of that taxon (diver-sity begets diversity), using Canarian plant data asone of four illustrative data sets. In practice,Emerson and Kolm’s analysis is not based on a rate(i.e. per unit time) estimate but on the proportion ofspecies that are endemic to particular islands, andall that they really show is that richer Canarian flo-ras also have high proportions of endemic species.Our own unpublished analyses of the patterns ofsingle-island endemism suggest that the opportuni-ties for speciation show a roughly parabolic rela-tionship to island age, with speciation rates andproportions of SIEs peaking on islands of young tointermediate age (cf. Silvertown 2004), and thendeclining, consistent with what we term the islandimmaturity–speciation pulse model (Box 9.3). Itfollows that the correlation between plant speciesrichness and single-island endemism reported byEmerson and Kolm (2005a) fails to demonstratethat greater richness of a taxon leads to greaterrates of (i.e. faster) speciation in that taxon. BothSilvertown’s (2004) and Emerson and Kolm’s(2005a) studies have been criticized for the logicinvolved in the interpretation (see Cadena et al.2005; Saunders and Gibson 2005; and see replies byEmerson and Kolm 2005b; Silvertown et al. 2005),supporting our suggestion earlier in this book thatthe only certainty about claims regarding competi-tive effects is that they will be disputed.

Despite the problems in testing such ideas,ecological interactions including competition andinteractions across trophic levels (mutualisms,parasitism, disease, predation, etc.) are clearlyimportant within island ecology and evolution.This is demonstrated by the extent to whichremote, species-poor, disharmonic islands haveproduced radiations of lineages now occupyingecological niche space that they occupy nowhereelse, by the increased tendency towards generalistpollinator mutualisms on particular islands, by the

ground-nesting behaviour of many oceanic islandbirds (many now driven to extinction), and bymany other phenomena reviewed in these pages.Alongside these biotic mechanisms and forcingfactors, we must also consider how the role of abi-otic forcing factors influences rates and patterns ofevolution. This point is amply illustrated in studiesreviewed already in this chapter.

How the two sets of forces (biotic and abiotic)interact is hard to determine with any precision.But we might set out the following hypotheticalscenario (developed further in Box 9.3). In the earli-est stages of the life cycle of a remote volcanicisland, opportunities for evolution are severely lim-ited by the barren and volatile nature of the system.As the island builds in area, elevation, and topo-graphic complexity through time, it acquirescolonists and complex food webs of interactingorganisms. Opportunities for adaptive and non-adaptive radiation (and for mechanisms such ascounter-adaptation to operate) increase accord-ingly. We might therefore expect the highest rate ofevolution in this stage. As the island ages itbecomes less active volcanically and erosion, mega-landslips, and subsidence combine to reduce itsarea, topographic complexity, and elevationalrange. Opportunities for speciation into vacantniche space and into isolated habitat pockets arereduced along with the overall carrying capacity ofthe island. Eventually the island slips back into thesea, in the tropics perhaps forming an atoll and sosharing an increasingly cosmopolitan strandlinefauna and flora with other similar systems. Theunique forms found on the island have by this pointeither colonized other islands or failed altogether.

Such a scenario might well apply to theHawaiian islands, which have been likened to aconveyor belt, in which the islands undergo a pat-tern of birth, growth, maturity, and decline. Themore recent lava flows within the youngest islandare thus, today, the sites in which novel species andadaptations are most apparent. Kaneshiro et al.(1995, p. 71) in their analysis of species groupswithin the picture-wing Drosophila of Hawaii makethe following observation:

Most of these species, like many other extant terrestrialendemic fauna, show a very strong but by no means

Page 253: Whittaker. Island Biogeography. 2007

240 E M E R G E N T M O D E L S O F I S L A N D E VO L U T I O N

exclusive tendency to single-island endemism. Mostspecies thus appear to evolve on an island early in its his-tory and thereafter remain confined to that island.Colonists arriving at newer emerging islands tend to formnew species, a finding that has led to the serious consid-eration that speciation may be somehow related tofounder events.

Studies from the Juan Fernández islands byCrawford et al. (1992) also support the idea of highinitial rates of radiation. Electrophoretic data for theendemic genus Robinsonia (Asteraceae) suggest thatthe founding population arrived early in the4.0 million year history of Masatierra island, radiat-ing and speciating rapidly after colonization(Crawford et al. 1992).

Evolutionary change can be very slow paced butit can also occur with great speed, so the notion thatspeciation rates may vary during the life history ofan island is not unreasonable. In illustration ofrapid change, Losos et al. (1997) were able todemonstrate significant adaptive differentiationover a 10–14 year period in populations of Anolissagrei introduced experimentally into small islandsin response to varying vegetation of the recipientislands. Diamond’s study of character displace-ment in the myzomelid honeyeaters is anotherexample of the sort of context in which a fairly pre-cise time frame, a maximum of c. 300 years, couldbe put to a given evolutionary change (Chapter 7;Diamond et al. 1989). The limitation of such studiesis that they are concerned with microevolutionarychanges rather than full speciation: for the latter wedepend upon geological and biological dating tech-niques. Repeated speciation is demonstrated by the70 species of Mecyclothorax beetles endemic to 1 mil-lion-year-old Tahiti, and the picture-wing Drosophila,in which 25 of the 26 species on the island of Hawaiiare restricted to that approximately 0.6–0.7 million-year-old island (Paulay 1994), while the cichlidfishes of Lake Victoria provide the most spectacularexample of explosive radiation (see Box 9.2).

Although such large islands as Hawaii were onceassumed to be fairly well buffered from climaticvariations, the large-scale global climatic fluctua-tions of the Pleistocene are now known to haveaffected them (Chapter 2; Nunn 1994) and musthave some bearing on the biogeography of such

island systems (above). On ecological timescales,present-day weather conditions can, after all, behighly significant determinants of breeding successin bird species, including, it has been found, in oneof the Hawaiian honeycreepers, the Laysan finch(Morin 1992). Some of the most interesting insightsin this field come from the remarkable studies ofhybridization in Darwin’s finches by Grant andGrant (1996a), showing how influential the climaticanomaly of the 1982–83 El Niño was in affectinggene flow through hybridization. Thus, althoughspeciation rates may indeed be higher at earlystages in the life of an oceanic island, it remains tobe established if and how such patterns have beeninfluenced by global climate change during thePleistocene (cf. Kim et al. 1996).

9.6 Variation in insular endemismbetween taxa

The majority of island lineages have not radiatedspectacularly at the species level, but some types oforganism have done so repeatedly, in widelyseparated archipelagos. Examples include generawithin the Asteraceae in islands as remote from eachother as Hawaii, Galápagos, and Macaronesia, andthe finch lineages of Hawaii and the Galápagos.What traits might determine the extent to whichparticular taxa undergo speciation on islands?

The most obvious trait is dispersal ability. Thistrait varies hugely between different taxa, andlargely determines which fail to colonize, whichcolonize very occasionally, and which may colonizerepeatedly. The first group obviously cannot formisland endemics, whilst very frequent gene flowbetween islands and source regions may also limitspeciation. We can illustrate this by looking atendemism in ferns and seed plants on two archi-pelagos. As already mentioned, the Juan Fernándezarchipelago contains two main islands, about150 km apart. Within the angiosperm genera con-taining endemic species, 31% have at least onespecies on both islands; the corresponding figurefor pteridophytes (ferns) is 71% (Stuessy et al. 1990).Endemism within the Galápagos flora is distrib-uted across major taxa as follows: 8 of the 107pteridophyte species are endemic, 18 of the 85

Page 254: Whittaker. Island Biogeography. 2007

VA R I AT I O N I N I N S U L A R E N D E M I S M B E T W E E N TA X A 241

monocots are endemic, and 205 of the 351 dicots areendemic (Porter 1979, 1984). The low proportion ofendemic ferns may be explained as a function oftheir better dispersability. This is indicated by thedata in Fig. 9.10 which illustrate not only the tinysize of fern spores, but that seed sizes of grassesand sedges (monocots) are intermediate in sizecompared to the classic island endemics in thefamily Asteraceae. The proportion of endemism inthe Hawaiian flora, is considerably higher than forthe Galápagos flora, but again is greater in angio-sperms (91% endemic) than in ferns and their allies(70% endemism) (Table 9.2).

Dispersal traits thus provide one key element ofthe evolutionary syndromes characterizing remoteisland biotas. This is again illustrated by theHawaiian flora, for which Carlquist (1974) hasattempted to determine the most likely means ofcolonization of each of the ancestral colonists of thecurrent native flora (Table 9.8) (see also Keast andMiller 1996). Species dispersed by terrestrialmammals, for example, are not part of the game.

Porter (1984) has calculated that the Galápagosflora could be accounted for by 413 naturalcolonists. On this basis and assuming the age of thearchipelago to lie between 3.3 and 5 million years(Simkin 1984), the present native vascular plantflora of the islands could be accounted for by onesuccessful introduction every 7990–12 107 years.The error margins on such estimates may be large,

but they serve to indicate that successful dispersalevents to such a remote location are very infre-quent, judged on ecological timescales. Studies ofless remote islands (e.g. Krakatau) find that theimmigration rates vary through time, declining asthe most dispersive species colonize and the rich-ness of the flora builds up. It is hard to know if thesame variation in rates occurs in very remote situ-ations such as the Galápagos, but it seems likely(Box 9.3). In any event, the above data indicate thatremote islands receive new colonists sufficientlyinfrequently as to provide long periods for thosespecies that do colonize an archipelago early on to

100

80

60

40

20

00.01 0.1 1 10

Diaspore magnitude (mm)

Perc

ent e

ndem

ics

Ferns and allies

Grasses and sedges

Composites

MascarenesGalapagos

Key

Figure 9.10 Percentage of endemism forselected plant taxa of the Galápagos and theMascarenes, against diaspore magnitude (therough length of the dispersal element). (Adaptedfrom Adsersen 1995, Fig. 2.2.)

Table 9.8 Most probable means by which the original floweringplant colonists dispersed to Hawaii (data from Carlquist 1974)

Dispersal mode Percentage of colonists

BirdsMechanically attached 12.8Eaten and carried internally 38.9Embedded in mud on feet 12.8Attached to feathers by viscid 10.3

substanceOceanic drift

Frequent (able to float for prolonged 14.3periods)

Rare (most likely to have arrived) 8.5by rafting

Air flotation 1.4

Page 255: Whittaker. Island Biogeography. 2007

242 E M E R G E N T M O D E L S O F I S L A N D E VO L U T I O N

Box 9.3 The island immaturity–speciation pulse model of island evolution

Emerson and Kolm (2005a) recently undertookanalyses of the relationship between theproportion of single-island endemics (SIEs) and anumber of other island properties for arthropodsand plants of the Canaries and of Hawaii. Theyfound that the strongest statistical explanation forthe number of SIEs per island for each taxon wasthe species richness of that same taxon. Thisanalysis is interesting, but inconclusive. It mighttell us, as inferred by Emerson and Kolm(2005a,b), that high species richness creates theconditions for high rates of speciation (e.g.through competitive interactions), but it could

also be that the relationship is a by-product ofcircumstances whereby remote islands of highpotential carrying capacity cannot be filled byimmigration alone and so provide greateropportunities for speciation.

Here we set out the line of argument thatrather than ‘diversity begetting diversity’, it isopportunity that is key to understandingvariations in rates of speciation on remote islands,and that opportunity has a broadly predictablerelationship to the life cycle of an oceanic island.We term the resulting model the islandimmaturity–speciation pulse model. In setting

The island immaturity–speciation pulse model of island evolution, showing the form of variation expected in species richness (R), theintrinsic carrying capacity of the island (K ), immigration rate (I ) and speciation rate (S ). Islands such as the Canaries experience acomparatively lengthy old-age of declining elevation, with speciation rate declining in tandem: hence we envisage time as some form oflog function, as indicated on the figure. The greatest opportunities for speciation (i.e. where the gap between R and K is greatest) arewhen an intermediate number of the eventual colonist lineages have arrived, and where environmental complexity is great. This is likely tooccur not in extreme youth, but during the youth of the island nonetheless, with opportunities for adaptive radiation greatest in thisphase. At the very earliest stage of island colonization, increasing R is driven by I (mostly from older islands within the archipelago), butthe rate of I declines (as per MacArthur and Wilson 1963, 1967) as R increases and the proportion of the archipelagic pool already on theisland increases. Speciation takes time, and so the rate of speciation lags slightly behind the increased ‘opportunity’ (the latter is indicatedby the gap between K and R ), increasing to a peak early on in the islands’ history and then slowly declining. The amplitude of I and Scurves may vary as a function of effective isolation (compare with Heaney 2000, Fig. 5). Later speciation events, during the old age of anisland, are more likely to be consistent with the anagenesis model than with anacladogenesis. For a discussion of extinction rate (E), seebox text.

Page 256: Whittaker. Island Biogeography. 2007

VA R I AT I O N I N I N S U L A R E N D E M I S M B E T W E E N TA X A 243

out this temporal model, we describe theobservations on which it is based, and thenexamine what we expect in terms of critical rates,illustrating the model in a simple graphic, ofsimilar format and terms to MacArthur andWilson’s (1967) EMIB graphic (Fig. 4.1).

Analyses of species richness variation reviewedin Chapter 4 provide broad support for the ideaof an environmentally-determined carryingcapacity (K) for richness. Yet, many oceanic islandsfail to attain their hypothetical carrying capacity asthey are too young or have had too little timesince some past disturbance event(s) in which torebuild their diversity (Heaney 2000).

Remote islands receive colonists so rarely thatimmigration and speciation occur on similartimescales, and thus increasing proportions of thebiota of remote islands are made up of endemicspecies. Within archipelagos such as Hawaii andthe Canaries, young islands should present thegreatest opportunities in terms of the physicalattributes of size and topographic range, and thebiological attribute of ‘vacant’ niche space (seefigure). This implies opportunities for radiationthrough sympatric, parapatric, and allopatricmechanisms (see text for details).

Following attainment of their maximumaltitude, such islands are prone to suffercatastrophic losses due to mega-landslides,subsidence, and erosion (Chapter 2). In time, theireruptive activity slows, and although erosion thendominates, once their elevational range declinesbelow about 1000 m these catastrophic losses inarea become less likely (Hürlimann et al. 2004), sothat the attrition of the island slows as the islandsattain old age. Their eventual fate is to slip backinto the sea, or, in tropical seas, to persist asatolls.

Opportunities for phylogenesis (maximal valuesof K�R), and especially for adaptive radiation, canthus be expected to be greatest at a fairly earlystage in the process of biotic accumulation on aremote island, when there are sufficientlycomplete ecological systems to provide adaptiveopportunities, but still plenty of ‘unoccupied’niches. As opportunity allows, so speciation isinitiated, and the rate of production of SIEsincreases to an early peak. As the island ages,the complex eruptive and erosive history of the

stage may provide additional opportunities fornon-adaptive radiation by within-island isolation(vicariance or peripheral isolation) (cf. Carsonet al. 1990).

From this point in the island’s history, we canconsider two scenarios (using the notation in thefigure, and with E being the extinction rate):

1. As R gradually approaches the hypotheticalvalue of K, S might be expected to decline evenas R continues to increase.2 If the notion of a fixed K is considered an illu-sory concept (because increasing diversity allowsmore diversity to evolve, as MacArthur and Wilson1967, Emerson and Kolm 2005a), then S must stilleventually fall with the decline and submergenceof the island itself. At this stage however fast evolution may be working, K is diminishing (e.g.see Stuessy et al. 1998), and so E� (I�S) and theinevitable slide back into the ocean will necessarilydraw R (and with it S) down to zero (see figure).

Note that in practice, the hypothetical carryingcapacity of a volcanic island over the course of itsexistence can thus be predicted to describesomething like a parabolic function, but skewedto present a peak capacity early on, followed by afar more extended period of gradual decline. Thissmooth curve will be interrupted by theoccasional catastrophe (eruptions, mega-landslides; Chapter 2).

Within a single oceanic archipelago such as theCanaries or Hawaii, it is therefore to be expectedthat the islands with the greatest K and R valueswill tend to be those of greatest size, elevationand topographic complexity, and will thus also berelatively young. Hence, within the Canaries, asshown in this chapter, Tenerife, much of which isof �2 Ma and the oldest parts of which are�8 Ma, has greatest diversity. During ‘late youth’,islands have been in existence long enough tohave sequentially built their substantial size, longenough to have accrued colonists, and longenough for speciation events to have occurred ina wide range of colonists. Hence, highproportions of SIEs are to be expected within suchan island.

Hence, we expect to find that the relativecontribution of I and S to the richness of a remoteisland will vary through time as shown in the

Page 257: Whittaker. Island Biogeography. 2007

244 E M E R G E N T M O D E L S O F I S L A N D E VO L U T I O N

figure. Further, although both S and I may eachdecline during the phase of islandsubsidence/destruction, the relative importance ofS may be greater than in the very earliest stages(thousands of years) of the island’s history,during which (as argued by Heaney 2000) mostof the increase in richness comes fromimmigration.

To complete the model, we must consider theloss of SIEs. This can occur through twomechanisms: either colonizing another island, orsimply going extinct.

● Colonizing another island. When islands existfor lengthy periods, opportunities exist for someof their new endemics to island hop to other,younger islands, thus becoming (at least for awhile) an MIE and no longer an SIE. Many morelineages follow such a progression rule (see text)than back-colonize.● Extinction. We have not attempted to sketch anextinction-rate curve in the figure, but we anti-cipate that in terms of biotic drivers of extinction,E should increase as R approaches K (see figure),although remaining very low in absolute terms.However, these biotic drivers of extinction arelikely to be dwarfed by abiotic drivers; at earlystages by mega-disturbance phenomena (damag-ing eruptions, mega-landslides), in the later stagesof the island life cycle by the gradual attrition oftopographic relief and area (Stuessy et al. 1998),and over the last 10 000 years, by ‘relaxation’associated with the loss of area due to postglacialsea-level rise.

Over the life-cycle of an island as sketched inthe figure, we would therefore anticipate that inextreme youth I �S�E, from youth towards islandmaturity S � I �E, interrupted by phases of highE(catastrophe), and that in advanced old age,E �(S � I). If I and S do come into approximatebalance on an ageing island, then the attrition ofSIE status within the biota through theprogression rule should result in a decline in boththe number and the proportion of SIEs on the

oldest islands. This means that for archipelagoswith a sufficiently complete range of island ages,we expect to see a humped relationship betweenspeciation rate and island age, and a humpedrelationship between the proportion of SIEs andisland age.

From our analyses (not illustrated) of Canarianarthropod and plant data taken from Izquierdo etal. (2004) we found that the proportion of SIEs onthe islands is a humped function of island age.This is consistent with the logic that volcanicoceanic islands undergo a burst of speciationrelatively early on in their evolution, reflecting theopportunities provided by the vacant niche space,and topographic complexity.

If we accept the logic that the inherent carryingcapacity of volcanic islands describes a humpedcurve over time, it follows that older islands(which have longest to reach their K value) arelikely to have greater competitive pressuresoperating over longer periods of time thanyounger islands. Hence, if the ‘diversity begetsdiversity’ logic holds, these islands (e.g. Lanzarote)should not have lower proportions of SIEs thanislands of younger age (e.g. Tenerife), yet theyshow precisely this pattern. We thereforeconclude that the ‘diversity begets diversity’model (Emerson and Kolm 2005a,b) is at best anincomplete representation of the data, because itignores the non-linearity of relationships betweenisland life history and the key biologicalparameters and rates.

Unfortunately, the empirical evaluation of theseideas depends on simple exploratory regressionmodels, based on several simplifying assumptionsand tested on a necessarily small number ofislands. This makes for a blunt instrument giventhe complexities of the environmental histories ofarchipelagos like the Canaries. The mostprofitable route to testing the model proposedherein would therefore seem to be via analyses ofphylogenies of multiple taxa.

Acknowledgement. This box is a shortenedversion of an as yet unpublished manuscript,written with our colleagues R. J. Ladle, M. B.Araújo, J. D. Delgado and J. R. Arévalo.

Page 258: Whittaker. Island Biogeography. 2007

exploit the opportunities available, through combi-nations of adapative, non-adaptive and archipel-agic radiation.

Yet, still some taxa appear to radiate more thanothers, with some lineages persisting for millionsof years on islands without radiating. This sug-gests that some taxa have a greater inherent degreeof genetic plasticity than others (e.g. Werner andSherry 1987) or that they possess traits related totheir breeding systems that favour rapid evolu-tionary change on islands (see discussion inChapter 7).

9.7 Biogeographical hierarchies andisland evolutionary models

Williamson (1981) has suggested redefining islandtypes as oceanic where evolution is ‘faster’ thanimmigration, and as continental if immigration isfaster than evolution. He also offered the observa-tion that it is useful for any particular island contextto distinguish between those groups for which theisland is continental, readily reached by dispersal,and those for which the island is oceanic, leading tolittle dispersal. For instance, the Azores could beregarded as oceanic for beetles, intermediate forbirds, and almost continental for ferns. The gradualdiminution in the range and numbers of taxa thathave been able to cross ever-increasing breadths ofocean means that, in general, the more remote theisland the more disharmonic the assemblage,and the greater the amount of vacant niche space.Thus, the greatest opportunities for adaptive radia-tion are towards the edge of the range of particulartaxa, i.e. in the radiation zone of MacArthur andWilson (1967), reaching their synergistic peak in themost remote archipelagos. Thus, in the absence ofants (amongst other things), there have been greatradiations of carabid beetles and spiders in Hawaiiand south-east Polynesia (Paulay 1994).

Although diversification in certain lineages hasundoubtedly been enabled by the disharmony ofvery remote islands, the absence of particular inter-acting organisms (e.g. pollinators for particularplant species) may have restricted the colonization,spread, and evolution of other lineages. That is tosay, hierarchical relationships within the island

biota, partly predictable and partly the chanceeffects of landfall (e.g. the highly improbablearrival of the first drosophilids on Hawaii at a par-ticular point in time), must have a strong influenceon the patterns that have unfolded.

Once on an archipelago, the relative importanceof interisland and intraisland speciation may vary(Paulay 1994). Only a handful of isolated archipela-gos show large endemic radiations of birds orplants. The occurrence of a number of islands inproximity to one another appears to be particularlyimportant to birds, suggesting that interislandeffects must be important (cf. Grant and Grant1996b). Intraisland speciation is restricted to islandslarge enough to allow effective segregation of pop-ulations within the island. This is taxon dependent,so that for land snails and flightless insects anisland of a few square kilometres is sufficient.

As several authors emphasize, different islandgroups have their own special circumstances andhistories, and few theories can span them all. Acorollary of this is that each theory may have itsown constituency. The solution, as elsewhere, mustbe to set up multiple working hypotheses and seekevidence to distinguish between them. Yet, freque-ntly, a particular lineage reflects not the operationof a single process, but of several. Therefore it maynot be possible to explain the overall biogeographyof an island lineage by means of a single model,such as character displacement, a double invasion,or the taxon cycle. Moreover, alongside the evolu-tionary hypotheses we should also be on the lookout for anthropogenic effects. The historical impactof humans in eliminating species from manyislands, and thus selectively pruning phylogenies,turning multiple-island endemics into single-islandendemics, and altering distributions of extantspecies (e.g. Steadman et al. 1999, 2002), requirescareful consideration in evaluating island evolu-tionary models and phylogenies (e.g. comparePregill and Olson 1981; Pregill 1986 with Ricklefsand Bermingham 2002).

Having established the relevance of factors suchas the regional biogeographic setting, environmen-tal change, dispersal differences between taxa,island area, island habitat diversity, and island iso-lation and configuration, we should consider how

B I O G E O G R A P H I C A L H I E R A R C H I E S A N D I S L A N D E VO L U T I O N A RY M O D E L S 245

Page 259: Whittaker. Island Biogeography. 2007

they fit together (as, e.g. Fig. 7.7; Box 9.3). Figure9.11 is one attempt to place island evolutionaryideas and models into a simple island area–isolationcontext. The aim is to highlight the geographicalcircumstances in which particular evolutionaryphenomena are most prominent. This is not toimply that, for instance, founder events and drift areinsignificant in large and isolated oceanic islands, asthey do form elements of the macro-evolutionarymodels. Rather, it is that they emerge as the mainevolutionary storyline principally for small, near-mainland islands.

Working up the isolation axis, small islands oflimited topographic range and degree of dissectionare liable to be poorly buffered from pronouncedenvironmental fluctuations, such as are associatedwith El Niño events or cyclonic storms. Thosespecies or varieties that become endemic on suchislands are likely to have relatively broad niches, at

least in so far as tolerating disruptions of normalfood supplies is concerned, as illustrated by theSamoan fruit bat, Pteropus samoensis (see Piersonet al. 1996).

Large, near-continent islands are likely to havevery low levels of endemism. Mainland–islandfounder effects and drift are also unlikely to beprominent in their biota. Islands such as (mainland)Britain fit this category, having essentially a subset ofthe adjacent continental biota, most species havingcolonized before the sea-level rise in the earlyHolocene that returned the area to its island condi-tion. Large islands, with a somewhat greater degreeof isolation and considerable antiquity, are the siteswhere relictual elements are most frequently claimedin the biota (but see Pole 1994; Kim et al. 1996).

As frequently noted, the greatest degree of evolu-tionary change occurs on remote, high islands.Where these islands are found singly, or in very

246 E M E R G E N T M O D E L S O F I S L A N D E VO L U T I O N

Niche space

Isol

atio

n

Dis

harm

ony

Anagenesis

Adaptiveradiation

Increasedniche breadth

in island forms

Gigantismand nanism

Founder effectand genetic drift

Relictualelements

Taxon cycleand competitivedisplacement

Mac

ro-sc

ale

Area

Micr

o-scale

Impoverished strand-linesystems, few endemics

High island ArchipelagoLow island

Key

Loss ofunnecessary

traits (e.g. defence, bold

colouring)

Evolutio

n

Continentalsub-set

Figure 9.11 Conceptual model of island evolution. This diagram attempts to highlight the geographical circumstances in which particularevolutionary phenomena are most prominent. The scalars will vary between taxa.

Page 260: Whittaker. Island Biogeography. 2007

widely spread archipelagos, speciation is frequent,but often without the greatest radiation of lineages,fitting the model of anagenesis. Non-adaptive radia-tion features principally in the literature on landsnails, for high and remote islands, where the fine-scale space usage of snails and their propensity tobecome isolated within topographically complexislands allows for genetic drift of allopatric popula-tions. The adaptive radiation of lineages is best seenon large, diverse, remote archipelagos in which inter-island movements within the archipelago allow amix of allopatric and sympatric speciation processesas a lineage expands into an array of habitats.

Most cases proposed for the taxon cycle are fromarchipelagos which are strung out in stepping-stone fashion from a continental land mass (theAntilles) or equivalent (the smaller series of islandsoff the large island of New Guinea). Here the degreeof disharmony is less than for systems such asHawaii and the Galápagos, and, crucially, there arelikely to be repeat colonization events by organismsquite closely related, taxonomically or ecologically,to the original colonists.

Of course, Fig. 9.11 is a simplification, and theplacing of islands into this framework is likely tovary depending on the taxon under consideration,because of the different spatial scale at which dif-ferent organisms (snails, beetles, birds, etc) interactwith their environment, both in terms of spaceoccupied by individuals, and dispersal abilities.Therefore, the scalars for area and isolation for Fig. 9.11 should vary between taxa, such that, for

instance, the radiation zones for different taxa maycoincide with different archipelagos (Fig. 9.12).Such radiation is, however, best accomplished onhigher islands, and archipelagos of such islands,than it is on low islands or single high islands.

These simple graphical portrayals of islandevolutionary outcomes pay no attention to the tem-poral development of island biotas, an aspect ofisland biogeography that warrants further atten-tion in relation to the life cycle of particular islands(Box 9.3), the role of long term environmentalchange, and in terms of how particular lineagesdevelop. Questions that we might address include,for example: (1) how typical is the rapid expansionof Metrosideros across Polynesia documented byWright et al. (2000)? (2) how well do such episodesmatch to periods of environmental change, alteredcurrent systems, or island configurations? and (3)to what degree did the colonization of this particu-lar woody plant lineage on Hawaii transform theevolutionary theatre? Our ability to track themovements of such proficient island hoppersacross whole ocean-basins should allow us toaddress many such questions in the near future. Itis an exciting time to be working in or followingisland evolutionary biogeography.

9.8 Summary

This chapter outlines several emergent patterns bywhich island evolution may be understood. First,we note that some island speciation occurs with

S U M M A RY 247

Less dispersivetaxa

Dispersive taxa

Isolation

Source pool

Figure 9.12 Dispersive taxa radiate best at or near to their effective range limits (the radiation zone), but only moderately, or not at all, onislands near to their mainland source pools. Less dispersive taxa show a similar pattern, but their range limits are reached on much less isolatedislands. The increased disharmony of the most distant islands further enhances the likelihood of radiation for those taxa whose radiation zonehappens to coincide with the availability of high island archipelagos.

Page 261: Whittaker. Island Biogeography. 2007

little or no radiation (anagenesis), but instead bychange in the island form away from the coloniz-ing phenotype along a singular pathway. Changemay also occur in the mainland source region, suchthat these island forms are in part relictual. Nextwe consider the taxon cycle, wherein evolutiontakes place within the context of an island archi-pelago, driven by a series of colonization events oftaxonomically and/or ecologically related forms.Immigrant species undergo niche shifts, driven byinteractions with both existing island species andlater arrivals. Through time, the earlier colonistslose mobility, decrease in distributional range, andultimately may be driven to the brink of extinction.Evidence for the taxon cycle in the early literaturewas largely distributional in nature, but the exis-tence of taxon cycles is now increasingly wellsupported by molecular phylogenetic data.

The most impressive evolutionary patterns, theradiation of lineages, are to be found on the mostisolated of large, oceanic islands. Many suchradiations involve clear alterations of niche and maybe termed adaptive. Examples are drawn fromplant, bird, and insect data from Hawaii, Galápagos,and Macaronesia. The most spectacular casesinvolve multiple islands and some even spreadacross different archipelagos. They typically involvea mix of sympatric, parapatric and allopatric phasesand ‘events’, although distinguishing the imprint ofthese different forms of speciation within the envi-ronmentally dynamic situation of oceanic archipela-gos is typically difficult. Some radiations have beenpostulated to be essentially non-adaptive, as a prod-uct of genetic drift in within-island isolates ratherthan clear niche changes. The clearest examplescome from studies of land snails on topographicallycomplex islands.

The recent explosion in phylogenetic analyses ofisland lineages has not only allowed the testing andrefinement of the above evolutionary-ecologicalmodels, but has also allowed different sorts ofmodels to be developed and evaluated, providingdynamic biogeographical interpretations of lineagedevelopment within archipelagos and even acrossocean basins. Patterns documented include the

progression rule, whereby early colonizing taxadisperse from old to young islands, speciating oneach in turn; back-colonization, whereby a minorityof movements occur from young to old islands,or even to the continent from which the lineageoriginated; and the repeated colonization pattern,whereby two or more colonization events arerequired to explain the island phylogeny.

It appears on theoretical and empirical groundsthat the fastest rates of radiation occur among earlycolonists at relatively early stages of the existence ofan island, declining as the island itself ages anddeclines: we term this the island immaturity–speci-ation pulse model. Radiation of lineages may alsobe enhanced by the physical dynamism and envi-ronmental variation of large oceanic islands, whichcan drive repeated phases of within-island isola-tion. The importance of environmental change,including Pleistocene climatic change, has beengradually gaining more attention, but it is oftenvery difficult to separate abiotic from biotic forcingfactors. Differential dispersal abilities, and evolu-tionary changes in dispersal powers, provideanother key element to understanding the dishar-mony of island ecosystems and the context inwhich evolutionary change takes place in those lin-eages that do reach remote islands.

The final section of this chapter attempts to placethe principal island evolutionary concepts andmodels into a simple framework of area versusisland—or archipelago—isolation. It is suggestedthat different ideas come to prominence in differentregions of this natural experimental factor space.Thus, for example, the taxon cycle may be anappropriate model to test in a not-too-remote archi-pelago, while adaptive radiation is more typical ofthe most isolated archipelagos of high islands. Aseffective dispersal range varies between (andwithin) taxa, different evolutionary patterns mayemerge within the same archipelago by comparisonof different types of organism. These geographicaland taxonomic contexts provide a frameworkwithin which many of the ideas of island biogeo-graphy can be seen to be complementary ratherthan opposing theories.

248 E M E R G E N T M O D E L S O F I S L A N D E VO L U T I O N

Page 262: Whittaker. Island Biogeography. 2007

PART IV

Islands and Conservation

This view across the cemetery of Old San Juan, PuertoRico is illustrative of the density of settlement that canbe found in the coastal zones of oceanic islands.Habitat loss is a key driver of biodiversity loss on manyoceanic islands, and on Puerto Rico has been a keyfactor in the decline of the emblematic Puerto Ricanparrot (Amazona vittata), only a few dozen of whichpersist in the wild (see text).

Page 263: Whittaker. Island Biogeography. 2007

This page intentionally left blank

Page 264: Whittaker. Island Biogeography. 2007

The classical paradigm in ecology, with its emphasis on thestable state, its suggestion of natural systems as closed andself-regulating, and its resonance with the nonscientificidea of the balance of nature, can no longer serve as anadequate foundation for conservation.

(Pickett et al. 1992, p. 84)

The fragmented world of biological communities in thefuture will be so different from that of the past, that wemust reformulate preservation strategies to forms that gobeyond thinking only of preserving microcosms of theoriginal community types.

(Kellman 1996, p. 115)

10.1 Islands and conservation

Our species has had a profound influence on theecology and biodiversity of the planet, altering thecomposition and functioning of ecosystems acrossthe globe. Our actions in some respects representmerely the latest driver of change, concentratedlargely within one interglacial of a period, theQuaternary (the past c.2 Ma), which has seenrepeated dramatic fluctuations in climate. Duringthis time, individual species have experiencedalternating episodes of expansion and of contrac-tion and fragmentation of ranges and have conse-quently been re-sorted into combinations that haveoften differed radically from those found today(e.g. Bush 1994). There is nothing new about rangealterations. Now, however, it is humans who aredriving range shifts and extinctions. We have donethis for a long period of our history, and we aredoing so in an accelerating fashion and on a globalscale (Saunders et al. 1991; Bush 1996). Whether wesee particular changes in ecosystem properties asbeneficial or detrimental can depend as much on

our cultural perspective as on the nature of thechanges themselves. However, there is no doubtingthe profound importance of the human agent inbiogeographical terms: not only in extinguishingspecies but, as significantly, by bombarding areaswith sets of species gathered up from far-distantregions of the globe, breaking bounds of isolationthat have lasted for many millions of years, and inthe case of oceanic islands for their entire existence.

The accelerated extinction of species on a plane-tary scale is something that concerns all societiesand it should therefore be a high priority for natu-ral scientists to study the processes involved, todocument the changes involved accurately, todevelop predictive models, and to provide scien-tific guidance for the conservation of biodiversity.This field of study is termed conservation biology,and it is an area of research that has expandedgreatly in the last 20 years: yet, there remain manyuncertainties concerning the big picture. In illustra-tion, we are currently unable to say to within anorder of magnitude how many species we share theplanet with (Gaston 1991; Groombridge 1992). Asecond illustration is that although a large propor-tion of land plants and animals are believed to berain forest species, we cannot be sure how large aproportion, nor how much tropical forest exists, norwith much precision, how fast these ecosystems arebeing degraded of their biodiversity on a globalscale (Whitmore and Sayer 1992; Grainger 1993).We know there to be a crisis, but we cannot be sureof its precise magnitude.

Nature conservation as a social movement is notmotivated solely by a concern to reduce globalspecies extinction, but is also imbued with othervalues and goals. However, a full consideration of

251

CHAPTER 10

Island theory and conservation

Page 265: Whittaker. Island Biogeography. 2007

the theoretical and operational principles of natureconservation and the long history of the conserva-tion movement are beyond our present remit. Ourpurpose in this final section of the book is to con-sider two broad themes. First, in the present chap-ter, we examine the application of ideas derivedfrom island ecological biogeography to conserva-tion problems. This research programme is focusedlargely on the loss of species resulting from thereduction in area and fragmentation of ‘natural’habitats, and is based on the premise that islandecological rules can be applied to ‘islands’ of par-ticular habitat types within mainland landmasses.This work forms part of the emerging field ofconservation biogeography (Whittaker et al. 2005).Second, we return in the final two chapters to afocus on real islands, examining the particularconservation problems of oceanic islands and someof the management solutions that have been devel-oped. The rationale for focusing on such islands inan island biogeography book is obvious, but thisfocus is also justified by the wider significance ofislands as repositories of biodiversity and by theextent to which island endemics feature in the listsof recently extinct and currently endangeredspecies.

10.2 Habitats as islands

. . . the abundance of publications on the theory of islandbiogeography and applications [of it] has misled many sci-entists and conservationists into believing we now have aready-to-wear guide to natural area preservation . . .

(Shafer 1990, p. 102)

It is generally accepted amongst conservation bio-logists, that the ongoing fragmentation and reduc-tion in area of natural habitats is a key driver ofterrestrial species extinctions at local, regional, andglobal levels (Whitmore and Sayer 1992). Theremaining areas of more-or-less natural habitats areincreasingly becoming mere pockets within a sea ofaltered habitat; in short, they are habitat islands. Itwas therefore to island biogeography that manyscientists turned in the search for predictive modelsof the implications of fragmentation and for guid-ance as to how best to safeguard diversity.

The implications of increasing insularity forconservation were recognized in the seminal worksof Preston (1962) and MacArthur and Wilson (1963,1967), with Preston astutely observing that in thelong run species would be lost from wildlife pre-serves, for the reason that they constitute reducedareas in isolation. MacArthur and Wilson’s dynamic(EMIB) model provided the theoretical basis todevelop this argument, allowing conservationscientists to explore the outcomes of differing con-figurations of protected areas, assuming them to actas virtual islands in a ‘sea’ of radically altered andthus essentially empty habitats. One early manifes-tation of this research programme was the so-calledSLOSS debate, which posed the question ‘given theopportunity to put a fixed percentage of land intoconservation use is it better to opt for a Single LargeOr Several Small reserves?’ Here the focus is on theentire regional assemblage of species and how tomaximize the carrying capacity of the protectedarea (i.e. habitat island) system. We will come backto this debate a little later, as we have organized thisdiscussion to begin with the smallest units: popula-tions of particular species.

The most basic question is: ‘how many individu-als are enough to ensure the survival of a single iso-lated (perhaps the last remaining) population? Theresulting figure is termed the minimum viablepopulation. We will consider this first. Manyspecies of conservation concern are not, however,restricted to a single locality—or at least, not yet.Rather, their survival in a region might be depend-ent on a network of habitat islands. Most of thisbook has been concerned with real islands in thesea. Is it realistic to expect habitat islands withincontinents to behave according to the same princi-ples as real islands? Instead of the barrier of saltwater, a forest habitat island might be separatedfrom another patch of forest by a landscape ofmeadows, hedgerows, and arable land. The impli-cations for species movements between patchesmight be radically different. As we have seen, how-ever, this is probably too simple a distinction andtoo simple a question. Scale of isolation is crucial, aswell as the nature of the intervening landscape fil-ter. Mountain habitat islands within continents, butsurrounded by extensive arid areas, may be much

252 I S L A N D T H E O RY A N D C O N S E R VAT I O N

Page 266: Whittaker. Island Biogeography. 2007

more effectively isolated than real islands that hap-pen to be located within a few hundred metres of amainland. Many isolates are actually sufficientlyclose to one another that their populations are ineffect linked: they form metapopulations, thesecond theoretical element that we will consider.When we have added these tools to our armoury,we can then consider the whole-system implica-tions of fragmentation and how ‘island’ approacheshave contributed to their understanding.

Over the past few decades there has been a shiftin ecology from what Pickett et al. (1992) term thebalance of nature paradigm, which sees the worldas a self-regulating, balanced system, towards theflux of nature paradigm, which Pickett et al. (1992)believe provides a more realistic basis for conserva-tion planning and management. This frameworkdoes not deny the possibility of equilibrium, butrecognizes equilibrium as merely one special state.The flux of nature viewpoint stresses process andcontext, the role of episodic events, and the open-ness of ecological systems.

Conventionally, the ‘island theory’ input to con-servation debates has been based on the generalassumption that a ‘natural’ area is in a state of bal-ance, or equilibrium. Humans then intervene toremove a large proportion of the area from its natu-ral state, leaving behind newly isolated fragments,which are thus cast out of equilibrium. Subsequently,these fragments must shed species as the systemmoves to a new equilibrium, determined by the areaand isolation of each fragment. This way of thinkingis entirely consonant with the balance of natureparadigm.

By contrast, in the present volume we havestressed the importance of considering physicalenvironmental factors in island biogeography, andhave shown that ecological responses to environ-mental forcing factors are often played out over toolong a time frame for a tightly specified dynamicequilibrium to be reached. Moreover, relatively fewareas of the planet, if any, are pristine, untouchedby human hand, so most areas that we may be frag-menting already contain the imprint of past landuses (e.g. cultivation and abandonment, hunting)(e.g. Willis et al. 2004). Hence, if the startingassumptions concerning the prior equilibrial state

of the system are flawed, some of the theoreticalisland ecological effects may, in practice, be fairlyweak. These points are addressed in the presentchapter in the applied, conservation setting. Hencewe will consider the physical effects of fragmenta-tion as well as the biotic, and we will examine whatsorts of factors in practice cause species to be lostfrom fragments.

10.3 Minimum viable populations andminimum viable areas

How many individuals are needed?

What is the minimum viable population (MVP)?By this we mean the minimum size that will ensurethe survival of that population unit, not just in theshort term, but in the long term. It is often definedmore formally in terms such as the population sizethat provides 95% probability of persistence for 100or for 1000 years. Attempts to calculate the viabilityof single populations, i.e. whether the populationexceeds the minimum requirement, are termedpopulation viability analyses (PVA). Importantelements that have to be considered in PVA includedemographic stochasticity, species traits, geneticerosion in small populations, and extreme eventssuch as hurricanes or fires.

Demographic stochasticity of initially small pop-ulations can lead to losses from a series of isolateswithout a need to invoke any common factor suchas predation or loss of fitness. However, wheresmall populations persist for a reasonable length oftime, they may also lose genetic variability as theypass through bottlenecks. They may then lack thegenetic flexibility to cope with either the normalfluctuations of environment or an altered environ-ment, and they may also accumulate deleteriousgenes. In short they lose fitness (Box 10.1). It isgenerally held as axiomatic that an increase ininbreeding in small populations reduces fitness inanimals, although unambiguous demonstrations ofthe effect in natural populations are relativelyscarce (Madsen et al. 1996). The basic rule of con-servation genetics is that the maximum tolerablerate of inbreeding is 1% per generation. This trans-lates into an effective population size of 50 to

M I N I M U M V I A B L E P O P U L AT I O N S A N D M I N I M U M V I A B L E A R E A S 253

Page 267: Whittaker. Island Biogeography. 2007

254 I S L A N D T H E O RY A N D C O N S E R VAT I O N

Box 10.1 Fluctuating asymmetry in a fragmented forest system

Fluctuating asymmetry (FA) refers to differencesbetween the right and left sides in animalanatomical characteristics that are usuallybilaterally symmetrical, and is a form ofdevelopmental instability that may have geneticand/or environmental causes. Whatever thenature of the control, a high incidence of FA isregarded as a sign of a population underenvironmental stress. Recent work has suggestedthat small, isolated populations in fragmentedhabitats may exhibit a higher incidence of FA.Anciães and Marini (2000) test this hypothesis forwing and tarsus FA on 1236 mist-nettedpasserine birds of 100 species captured from 7fragments and 7 control areas in the Atlantic rainforest area of Brazil. They recorded a highermean level of FA in both features in the forestfragments than in the control areas, suggestingthat passerine birds in these forests exhibitmorphological alterations as a result of forestfragmentation and habitat alteration. Possibleproximal causes might include increased pressuresfrom predators, competitors, parasites and

disease, and/or decreased food supplies, shelteror nest site availability.

In general, terrestrial and understoreyinsectivores showed greater relative FA changesthan other foraging guilds, in line with the generalsensitivity of this guild to habitat fragmentation. Anexample is Dysithamnus mentalis, which forages inthe medium strata and may also follow army ants.It has previously been found to be an area-sensitivespecies, and indeed was observed in only two ofthe fragments in the study, where it showed higherlevels of FA in the tarsus than in the control forests.By contrast, the most abundant frugivores sampledin the study (Chiroxiphia caudate, Manacusmanacus, and Ilicura militaris) were commonlyfound in altered forest patches and seemedunaffected by fragmentation in their incidence andin terms of FA values.

This form of developmental instability mayprovide a form of early warning system, indicatingstresses on vertebrate species in isolates prior tomore profound changes, such as rapid populationdeclines and species extinctions.

ensure short-term fitness according to the calcula-tions of I. A. Franklin in 1980 (reported in Shafer1990). However, there will still be a loss of geneticvariation at such a population size, and Franklinrecommended 500 individuals in order to balancethe loss of variation by gains through mutation.These commonly cited values appear to deriverespectively from data from animal breeders andbristle number in Drosophila, and so should beregarded as merely first approximations (Fiedlerand Jain 1992). Lacy (1992) noted that crudelyestimated MVPs for mammals based on data forinbreeding effects from captive populations variedover several orders of magnitude, and that it wouldbe premature to generalize on MVPs in the wild onthis basis.

Typically, only a proportion of the adult popula-tion participates in breeding; and it is these animalsthat form the effective population size, which is

often substantially smaller than the total popula-tion size (Shafer 1990). A study of grizzly bears inthe Yellowstone National Park showed that to pre-vent inbreeding rates exceeding 1% required anoverall population size of at least 220 rather than 50animals (Shafer 1990).

It is intuitively reasonable that the loss of a broadgenetic base is likely to increase the probability ofextinction. However, we learnt in an earlier chapterthat putting a population through a bottleneck ofjust a few individuals may not be as damaging tothe genetic base as one might anticipate, providingthe population is allowed to expand again fairlyquickly. In illustration, the great Indian rhino wasreduced within the Chitwan National Park inNepal (one of its two main havens) to an effectivepopulation of only 21–28 animals in the 1960s. Inthis case the bottleneck was short. Rhinos have along generation time, and it has since recovered to

Page 268: Whittaker. Island Biogeography. 2007

about 400 individuals. Genetic variation within thepopulation is still remarkably high (Tudge 1991).Nonetheless, the Drosophila studies reviewed inChapter 7 showed that sustained or repeated bot-tlenecks do lead to loss of heterozygosity. Hence,although you may be able to rescue some, perhapsmany, species from a bottleneck of short duration,you cannot assume that this will apply to a sus-tained or repeated bottleneck. Just because a suiteof threatened species are hanging on in theirreserves now, it doesn’t mean that the same level ofbiodiversity can be sustained for a long time in afragmented landscape.

A further complication in assessing geneticeffects is that where a species is split into numerousseparate populations in fragmented habitats, theremay be multiple bottlenecks involved. This mayresult in reduced variation within each populationbut increased genetic differentiation between popu-lations (see Leberg 1991). This may be of signifi-cance to lengthening persistence in metapopulationscenarios (below).

Pimm et al. (1988) have undertaken an analysis ofisland turnover that illustrates how some speciestraits may be important to determining minimumpopulation sizes. They analysed 355 populationsbelonging to 100 species of British land birds on 16islands. They found that the risk of extinctiondecreases sharply with increasing average popula-tion size. They examined extinction risk for large-bodied species, which tend to have long lifetimesbut low rates of increase, and for small-bodiedspecies. They found that at population sizes ofseven pairs or below, smaller-bodied species aremore liable to extinction than larger-bodied species,but that at larger population sizes, the reverse istrue. This can be explained as follows. Imagine thatboth a large-bodied and a small-bodied species arerepresented on an island by a single individual.Both will die, but the larger-bodied species is liableto live longer and so, on average, such species willhave lower extinction rates per unit time. On theother hand, if the starting population is large, but itis then subject to heavy losses, the small-bodiedspecies may climb more rapidly back to highernumbers, whereas the larger-bodied species mightremain longer at low population size, and thus be

vulnerable to a follow-up event. They also foundthat migrant species are at slightly greater risk ofextinction than resident species. Numerous factorsmight be involved in migrant losses, such as eventstaking place during their migration or in their alter-native seasons’ range (e.g. Russell et al. 1994).Further studies will be necessary to establish thegenerality of this particular finding (see discussionsbetween Pimm et al. 1988; Tracy and George 1992;Diamond and Pimm 1993; Haila and Hanski 1993;and Rosenzweig and Clark 1994).

Isolates subject to significant environmentalchange or disturbance may need to have muchlarger populations than is otherwise the case toensure survival. A number of studies have recog-nized the potential of environmental catastrophes inthis context but, typically, they have not attemptedto evaluate the general significance of such catas-trophes (see Pimm et al. 1988; Williamson 1989b;Menges 1992; Korn 1994). Of course, it is inherentlydifficult to incorporate extreme events into suchanalyses. As Bibby (1995) remarks ‘how do you esti-mate the probability of a cat being landed on a par-ticular small island in the next five years?’ Yet, suchan event may have a crucial impact on an endan-gered prey species.

Mangel and Tier (1994) argue that environmen-tal catastrophes may often be more important indetermining persistence times of small popula-tions than any other factor usually considered, andtherefore should be incorporated explicitly whenformulating conservation measures. They go on toprovide computational methods for modelling per-sistence times which do take account of catastro-phes and which allow for quite complicatedpopulation dynamics. Unsurprisingly, they findthat the resulting MVPs are larger than thosederived from variants of the MacArthur andWilson model, or indeed other analyses of popula-tion viability which ignore catastrophes (Fig. 10.1)(see also Ludwig 1996).

An empirical demonstration is provided byHurricane Hugo, which in 1989 caused extensivedamage to the forests of the Luquillo mountains,the home of the last remaining wild populationof the Puerto Rican parrot (Amazona vittata). In 1975the parrot population had reached a low of 13 wild

M I N I M U M V I A B L E P O P U L AT I O N S A N D M I N I M U M V I A B L E A R E A S 255

Page 269: Whittaker. Island Biogeography. 2007

birds, as a result of a combination of habitatdestruction, capture for pets, and slaughter (forfood and to protect crops). Given protection, thepopulation built up from this low point, to pre-hurricane levels of 45–47, from which it wasreduced by the hurricane to about 25 birds (Wilsonet al. 1994). Intriguingly, post-hurricane conditionsmay have suited the parrots, as 6 pairs bred in 1991,the highest total since the 1950s, and 11 youngfledged the following year. The post-breedingpopulation thus swiftly climbed to between 34 and37 parrots. However, the number of birds is clearlyfar too small for comfort, and the species couldeasily be pushed to extinction by repeated cata-strophic hurricanes.

As endemic species tend not to occur on thesmallest of islands, extinctions in the global sensecaused primarily by natural disturbances are likelyto be rare events on ecological timescales, unlessthe species concerned has already been reduced inrange (Hilton et al. 2003). One possible example of aspecies finished off in this way, by hurricanes in1899, is the St Kitts subspecies of bullfinch Loxigillaportoricensis grandis, although the last record of thisbird was actually several years earlier, in 1880(Williamson 1989b). A more recent example of aspecies reduced to the status ‘critically threatened’

is the Montserrat oriole (Icterus oberi), whichalthough confined to only 30 km2 of forest withinthe island of Montserrat, was not considered glob-ally threatened before the eruption of the SoufriereHills volcano in 1995. The eruptive activity (as oftenthe case) spanned several years (indeed it had notceased as of the end of 2005), devastating much ofthe forest habitat of the oriole. The dramatic popu-lation decline extended even to the small areas ofintact forest, suggesting knock-on consequencesof the eruption such as reduced food supplies orincreasing nest predation. It proved difficult to esti-mate the oriole population but it is thought that ithad declined from several thousand individualsbefore the eruptions to perhaps 100–400 pairs in theyear 2000 (Hilton et al. 2003). The populations arecurrently being closely monitored (�www.rspb.org.uk/science/diaries/montserrat/index.asp�).

In order to make MVAs more realistic, effortshave now been made to include data on age struc-ture, catastrophes, demographic and environmentalstochasticity, and inbreeding depression. Reed et al.(2003) use this approach to derive MVP estimatesfor 102 vertebrate species, using a working defini-tion of MVP as ‘one with a 99% probability of per-sistence for 40 generations’. Their chosen speciesincluded two amphibians, 28 birds, one fish, 53

256 I S L A N D T H E O RY A N D C O N S E R VAT I O N

Time (t)

Popu

lati

on s

ize

at ti

me

t, N

(t)

0 100Time (t)

0 100

4050

(b)(a)

Figure 10.1 Simulation runs of population trajectories. (a) In the absence of catastrophes all five trajectories persist for the 100 time unitsshown. (b) In the presence of catastrophes two populations become extinct and two are at low population sizes after 100 time units. (FromMangel and Tier 1994, Fig. 5; original figure supplied by M. Mangel.)

Page 270: Whittaker. Island Biogeography. 2007

mammals, and 18 reptiles. Across this data set,mean and median estimates of MVP were 7316 and5816 adults, respectively. The values estimated didnot differ systematically between major taxa, orwith trophic level or latitude, but unsurprisinglywere negatively correlated with population growthrate. Interestingly, they also found that MVPs basedon shorter studies systematically underestimatedextinction risk, which appeared to be due to thegreater temporal variation in population sizecaptured in studies of longer duration. AlthoughReed et al. (2003) stress that the values theyobtained do not provide a ‘magic number’, they doserve to indicate the rough size of vertebrate popu-lations that may be needed for successful conserva-tion, i.e. about 7000 adults, which is higher thanmost previous estimates. In many cases, the arearequired for a large vertebrate population of thissize to be contained in a single reserve will be unob-tainable, requiring conservation strategies thateither allow population exchange to occur betweenmajor reserves, or that are not dependent onreserve systems.

How big an area?

How does a figure for a MVP translate into aminimum viable area (MVA)? With some speciesand communities, e.g. butterfly populations, afairly small area may suffice to maintain the requi-site number of individuals. In general, the higherup the trophic chain, the larger the area needed.Simply considering the home range of a few verte-brates is illustrative. It has been calculated that asingle pair of ivory-billed woodpeckers (Campephilusprincipalis) may require 6.5–7.6 km2 of appropriateforest habitat; that the European goshawk (Accipitergentilis) has a home range of about 30–50 km2; andthat male mountain lions (Felis concolor) in the west-ern United States may have home ranges in excessof 400 km2 (Wilcove et al. 1986). Even some plantand insect species may need surprisingly largeareas, as illustrated by Mawdsley et al.’s (1998)analyses of strangler fig species, which indicate thatthe MVA for some species may be as large as200 000 ha because of their low population densi-ties. Thus, for many species, reserves must be really

rather large if their purpose is to maintain a MVPentirely within their bounds.

There is of course, one very large provisoattached to the calculation of a minimum criticalarea. The approach assumes that the area con-cerned acts rather like a large enclosed paddock ina zoo, with freedom of association within it, but noexchange with any other paddocks or zoos. Unlessthe reserve does indeed contain the only popula-tion of the species, or they are completely immobilecreatures, there is always the possibility of immi-gration of other individuals from outside thereserve. Where a significant degree of exchangetakes place, this may buffer populations fromgenetic erosion and extinction, such that the con-cept of a fixed critical minimum area loses its mean-ing. In practice, this must be assessed on a taxon bytaxon, and probably species by species, basis.

Applications of incidence functions

Area requirements of particular species can beexamined by means of incidence functions estimat-ing the probability of a species occurring for habitatislands of given combinations of area and isolation(Box 5.2; Wilcove et al. 1986; Watson et al. 2005). Theinformation returned from such incidence func-tions is not equivalent to estimating MVAs. This isbecause the incidence functions constitute theproduct of population exchange within the net-work of suitable habitats within a region. Theytherefore tell you the properties of ‘islands’ onwhich a target species currently occurs, not thoseon which it may persist in the long term, or in analtered ecological context.

Simberloff and Levin (1985) examined the inci-dence of indigenous forest-dwelling birds of aseries of New Zealand islands, and of passerines ofthe Cyclades archipelago (Aegean Sea). In bothsystems they found that most species occur remark-ably predictably, with each species occupying allthose and only those islands larger than somespecies-specific minimum area. A minority ofspecies in each avifauna did not conform to thispattern, possibly because of habitat differencesamong islands and because of anthropogenicextinctions. Although stochastic turnover might be

M I N I M U M V I A B L E P O P U L AT I O N S A N D M I N I M U M V I A B L E A R E A S 257

Page 271: Whittaker. Island Biogeography. 2007

occurring, the structure in the data indicates thatthe habitat and human influences dominate. Thesefindings are consistent with the position expressedby Lack (1976), that most absences of birds fromislands within their geographic range can beexplained by unfulfilled habitat requirements.

Hinsley et al. (1994) quantified the incidencefunctions of 31 bird species in 151 woods of0.02–30 ha in a lowland arable landscape in easternEngland over three consecutive years (Fig. 10.2).For many woodland species the probability ofbreeding was positively related to woodland area,although some breeding occurred even in thesmallest of the woods. Only the marsh tit (Paruspalustris), nightingale (Luscinia megarhynchos), andchiffchaff (Phylloscopus collybita) failed to breed inwoods of area less than 0.5 ha in any year of thestudy, and each was uncommon in the area. Smallwoods appear to be poor habitat for specialistwoodland species, but are preferred by others.Their work showed that harsh winters could

temporarily but significantly alter the incidencefunctions of particular species. Specialist woodlandspecies were more likely to disappear from smallwoods after severe winter weather than weregeneralists, and could take more than a yearto recolonize. Variables describing the landscapearound the woods were important in relation towoodland use by both woodland and open countryspecies. For instance, long-tailed tits (Aegithaloscaudatus) were found to favour sites with lots ofhedgerows around them, whereas yellowhammers(Emberiza citrinella) preferred more open habitats,small woods, and scrub.

To establish the broader reliability of incidencefunctions as indicators of area requirements, theyneed to be shown to be reproducible across therange of a species. The results of Watson et al. (2005)reported in Box 5.2 for birds in the Canberra area,Australia, are not encouraging. They calculatedincidence functions for woodland habitat islands inthree different landscapes, with similar ranges in

258 I S L A N D T H E O RY A N D C O N S E R VAT I O N

Prob

abili

ty (

%)

WrenDunnockGreat TitTreecreeperG-s WoodpeckerMarsh Tit

Woodland area (ha)0 2 6 8 10 12 14

Prob

abili

ty (

%)

199019911992

Woodland area (ha)

0 10

100

80

60

40

20

0

100

80

60

40

20

0

20 3015 255

(a)

4

(b)

Figure 10.2 Incidence functions for the probability of breedingas a function of woodland area, based on 151 woods of0.02–30 ha area, in a lowland arable landscape in easternEngland. (a) Data representative of widespread and commonwoodland species (wren, dunnock, great tit) compared with thatfor more specialist woodland species (treecreeper, great spotted (G-s) woodpecker, marsh tit). All relationships are for 1990, exceptthat for the marsh tit, which is for 1991. (b) Interannual variation inthe incidence functions for the great spotted woodpecker, reflectinga period of severe weather in February 1991. Small woods werereoccupied in 1992. Vertical bars represent 1 SE; those below 10%are not shown. (Redrawn from Hinsley et al. 1994.)

Page 272: Whittaker. Island Biogeography. 2007

woodland area and isolation. Overall, they found amix of area-sensitivity, isolation-sensitivity, andcompensatory area-isolation effects. But they alsoreported significant variation in the form of theincidence functions, which they interpreted as mostlikely being the consequence of differences betweenthe three landscapes in terms of the habitat matrixin which the woodlands were embedded (cf.Ricketts 2001). Watson et al.’s study illustrates thatarea–incidence functions cannot be assumed to beconsistent features within the range of a species.

In conclusion, incidence functions can be usefultools, particularly if repeat surveys are available toestablish temporal variability in their form, butthey do not of themselves provide sufficient dataon which to base conservation policies. Notions ofthreshold population sizes and area requirements,as ever in island ecology, must be offset against dis-persal efficacy, habitat requirements, and specificfactors influencing mortality of particular species.Watson et al.’s (2005) findings also suggest that the‘island–sea’ analogy is at fault, and that we shouldbe paying greater attention to the ecology of thematrix (below).

10.4 Metapopulation dynamics

Most species are patchily distributed. Manyoccupy geographically separated patches that areinterconnected by occasional movements of indi-viduals and gametes, constituting a metapopula-tion. The first metapopulation models wereconstructed by Richard Levins in papers publishedin 1969 and 1970 (Gotelli 1991). The basic idea canbe understood as follows. Imagine that you have acollection of populations, each existing on patchesof suitable habitat. Each patch is separated fromother nearby habitat patches by unsuitable terrain.Although these separate populations each havetheir own fairly independent dynamics, as soon asone crashes to a low level, or indeed disappears,that patch will provide relatively uncontestedspace for individuals from one of the nearbypatches, which will soon colonize. Thus, accordingto Harrison et al. (1988), within a metapopulation,member populations may change in size independ-ently but their probabilities of existing at a given

time are not independent of one another, beinglinked by mutual recolonization following periodicextinctions, on time scales of the order of 10–100generations.

Studies of the checkerspot butterfly (Euphydryaseditha bayensis) in the Jasper Ridge Preserve (USA)provide one of the better empirical illustrations ofmetapopulation dynamics (Harrison et al. 1988;Ehrlich and Hanski 2004). The checkerspot butter-fly is dependent on food plants found in serpenti-nite grasslands. The study area is of 15 30 km andincludes one large patch (2000 ha) and 60 smallpatches of suitable habitat. The large ‘mainland’patch supports hundreds of thousands of adultsand is effectively a permanent population. Thesmaller populations are subject to extinctions, prin-cipally due to fluctuations in weather, but patchoccupancy may also be influenced by habitat qual-ity. A severe drought in 1975–1977 is known to havecaused extinctions from three of the patches,including the second largest patch, and so wasassumed to have eliminated all but the largest pop-ulation. Unfortunately, only partial survey datawere available from the period and this assumptioncannot be tested. By 1987, eight patches had beenrecolonized. Small patches over 4.5 km from the‘mainland’ patch were found to be unoccupied.Harrison et al. (1988) showed that the distributionof populations described an apparent ‘threshold’relationship both to habitat quality and distance.Patches had to be both good enough habitat andnear enough to the ‘mainland’ in order to beinhabited at that time. A similar relationship, butfrom a separate study, is shown in Fig. 10.3 (and seeEhrlich and Hanski 2004).

From models of the dispersal behaviour of thepopulations, assuming a post-1977 recolonization ofthe patches, it was predicted that more patcheswould become occupied in time. How manydepends on the assumptions of the model. Theyoffered two differing scenarios. The first assumedthat colonization continues until reversed by thenext severe drought, an event with an approximately50-year periodicity in the study area. The secondassumed that some extinction events occur betweensuch severe events, thereby requiring a continuousextinction model and producing an equilibrium

M E TA P O P U L AT I O N DY N A M I C S 259

Page 273: Whittaker. Island Biogeography. 2007

number of populations, with some turnover.The authors concluded that there was evidence forboth modes of extinction and also for variations indispersal rates over time. We may note that the fitwith equilibrium theory is thus uncertain. On theother hand, their data show a degree of independ-ence of population dynamics between the isolates,combined with a demonstration of recolonization,thereby fulfilling requirements of metapopulationtheory.

How does this view of extinction and recoloniza-tion differ from that of the EMIB? In the latter, theemphasis is on species number as a dynamic func-tion of area and isolation. Whereas in the check-erspot butterfly story the focus is on the status ofpopulations of a single species, with extinction seenprincipally as a function of (temporary) alterationsin the carrying capacity of the system. It is consis-tent with the metapopulation approach—althoughnot necessarily with all metapopulation models—that rates of movement and of population gainsand crashes may vary greatly over time. As Haila(1990) suggested, within a metapopulation system,the dynamics of the islands are to some extent inter-dependent. If the patches were much more isolated,their population fluctuations would be entirely

independent, and one of the alternative theoreticalmodels would be required. If, on the other hand, amore dispersive species was being considered, theentire system might constitute a single functionalpopulation, albeit one spread across a fragmentedhabitat. Thus, metapopulation models form abridge between the study of population ecologyand island theories of the EMIB form (Gotelli 1991).

Relaxing the single population assumption of theMVP model to allow for such ebb and flow betweenlocal populations will lengthen projected persist-ence times. Hanski et al. (1996) coin the term mini-mum viable metapopulation (MVM) for the notionthat long-term persistence may require a minimumnumber of interacting local populations, which inturn depend on there being a minimum amount ofsuitable habitat available.

Conservationists have suggested the creation ofmetapopulation configurations for endangeredspecies as a means of maintaining populationsacross areas of increasingly fragmented habitat. Themost celebrated example of this strategy is that ofthe Interagency Spotted Owl Scientific Committee.They proposed the creation and maintenance oflarge areas of suitable forest habitat in proximity, sothat losses from forest remnants will be infrequentenough and dispersal between them likely enoughthat numbers of the northern spotted owl (Strixoccidentalis caurina) will be stable on the entirearchipelago of forest fragments. Much of the workon metapopulations up to this point had been theo-retical (Hess 1996), and Doak and Mills (1994)expressed some concern that despite intensivestudy of the owl, the metapopulation models haveto assume a great deal about owl biology. Thesafety of the strategy cannot be demonstrated sim-ply by running the models and so, in effect, theexperiment is being conducted on the endangeredspecies. Criticisms of the strategy led to a US dis-trict judge ruling against it on the grounds that theplan carried unacknowledged risks to the owl(Harrison 1994).

Similar concerns were expressed by Wilson et al.(1994) concerning plans to subdivide the small captive flock of the Puerto Rican parrot into three groups, following the argument that a

260 I S L A N D T H E O RY A N D C O N S E R VAT I O N

0.01 0.1 1 10 100

Are

a (h

a)

100

10

1

0.1

0.01

Distance from source (km)

Figure 10.3 Occupancy of suitable habitat by the silver-studdedblue butterfly (Plebejus argus) in North Wales in 1990. Most patcheslarger than 0.1 ha were occupied (filled circle), provided that theywere within about 600 m of another occupied patch. Beyond thisdistance, no patches were occupied (open circle), regardless of patchsize. (Redrawn from Thomas and Harrison 1992.)

Page 274: Whittaker. Island Biogeography. 2007

metapopulation structure might bestow betterprospects of survival. The captive flock, of about 58birds, is maintained within the same forest area asthe wild population. In this case, the suggestion wasthat on subdividing the captive flock, one of thesepopulations should be transferred to a multispeciesfacility in the continental United States, with indi-viduals subsequently exchanged between the popu-lations. Wilson and colleagues pointed out thedanger of the accidental spread of disease, pickedup in this multispecies context, back to Puerto Rico.They also pointed out that mate selection is idiosyn-cratic in this, as in some other K-selected species,and that subdividing the population while it wasstill at a small size might be counter-productive forthis reason also.

As with the other models considered in thischapter, it is important to understand that meta-population models represent simplified abst-ractions, allowing ‘what-if’ scenarios to be explored(Harrison 1994). It is therefore dangerous to assignprimacy to any single model in determining con-servation policy.

The core–sink model variant

The classic metapopulation models assume equiva-lence in size and quality of patches, with all patchessusceptible to extinction and all occupied patchesable to act as re-supply centres (Fig. 10.4). Dawson(1994) noted two common scenarios that are inade-quately dealt with in such simplified models. First,if one patch is big, and the others small, as at JasperRidge, then they may have a core–satellite orsource–sink relationship, in which the core iseffectively immortal and the smaller patches areessentially sinks (Table 10.1). In a study of orbspiders on Bahamian islands, larger populationswere found to persist, whereas small populationsrepeatedly went extinct and re-immigrated fromthe larger areas and then become extinct again(Schoener and Spiller 1987), thus contributing littleor nothing to the chances of persistence of thewhole metapopulation (see also Kindvall andAhlen 1992; Harrison 1994). Dawson (1994) concluded that it is usually survival on the large

habitat fragments that ensures species survival, notan equilibrium turnover on many smaller patches(see Thomas et al. 1996 for a counter-example).Secondly, even seemingly isolated populations maybe saved from extinction through continued sup-plementation by individuals migrating from thecore population. In other words, the ‘rescue effect’may operate. Gotelli (1991) has shown that buildingthis effect into metapopulation scenarios has ahighly significant impact on the outcome.

M E TA P O P U L AT I O N DY N A M I C S 261

(a) Classic metapopulation

(b) Source-sink

Figure 10.4 Metapopulation scenarios. (a) In classic models,habitat patches are viewed as similar in size and nature: theoccupied patches (hatched) will re-supply patches that have lost their populations (unshaded), as indicated by the arrows, and at alater point in time will, in turn, lose their population and be re-supplied. (b) Much more commonly, large, effectively immortalpatches (source or ‘mainland’ habitat islands) (dense shading) re-supply smaller satellite or sink-habitat patches after density-independent population crashes (e.g. related to adverse weather):close and larger patches may be anticipated to be swiftly re-occupied(hatched) but more distant patches take longer to be re-occupied,especially if small (unshaded).

Page 275: Whittaker. Island Biogeography. 2007

It is easy to underestimate the power of naturaldispersal to re-supply isolated sites because theprocess is so difficult to observe. We can illustratethis by stepping outside the metapopulation litera-ture and considering the regular supply of migrantmoth species to the remote Norfolk Island, 676 kmfrom the nearest land mass (Holloway 1996). Asurvey over a 12 year period suggested that the res-ident Macrolepidoptera total 56 species, andmigrants as many as 38 species. In addition, someof the resident species are also reinforced bythe occasional arrival of extrinsic individuals. Therapid recolonization of the Krakatau islands, some40 km from the mainland, by so many species ofplants and animals, provides another example ofthe ability of many species to cross substantialdistances. There are many such demonstrations tobe found in the biogeographical literature, but theNorfolk Island data provide a powerful illustrationbecause quantifying the extreme, the tail of thedistribution, is generally difficult.

In contrast, there are many species of relativelylimited powers of dispersal, which like thecheckerspot butterfly may show distance limitationto patch recolonization on a scale of a few kilo-metres. For them to survive in increasinglyfragmented habitats may require active interven-tion by people, to enable dispersal of viablepropagules between isolates (Primack and Miao1992; Whittaker and Jones 1994a,b). Most metapop-ulation simulation models assume no distanceeffects, and thus may perform poorly in simulatingspecies that exhibit a strong gradient in arrival rateswithin a particular set of isolates (Dawson 1994;Fahrig and Merriam 1994).

The importance of continual immigration intoso-called ‘sink habitats’ is such that in some popu-lations the majority of individuals occur in sinkhabitats (Pulliam 1996). Furthermore, a habitat maycontain a fairly high density of individuals but beincapable of sustaining the population in theabsence of immigration. Density can, therefore, be apoor indicator of habitat quality. Thus, while somesuitable habitats will be unoccupied because oftheir isolation from source populations, other‘unsuitable’ habitats will be occupied because oftheir proximity to source populations. As pointedout earlier, if these effects operate, snapshot inci-dence functions may be misleading.

Deterministic extinction and colonization within metapopulations

Extinction models tend to be based on stochasticvariation, in some cases emphasizing demographicstochasticity, sometimes taking genetic stochasticityand feedback into account, and more recently incor-porating environmental stochasticity. However,Thomas (1994) has argued that, in practice, mostextinctions are deterministic, and that they canbe attributed directly to hunting by humans,introductions of species, and loss of habitat. Inmany cases documented in the literature, virtuallythe entire habitat was lost or modified, causing100% mortality. Stochastic extinction from surviv-ing habitat fragments is minor by comparison.According to this view, stochastic events are super-imposed as decoration on an underlying determin-istic trend. To put this in a metapopulation context,if the reason for patches ‘winking out’ is habitatchanges rather than—for sake of argument—aperiod of unfavourable weather, then the localhabitat is likely to remain unsuitable after extinc-tion and so will be unavailable for recolonization.In cases, fairly subtle changes in habitat may pro-duce a species extinction that might be mistakenlyclassified as ‘stochastic’. Thomas cites examplesfrom work on British butterflies, whereby evenchanges from short grass to slightly longer grasscan result in changes in the butterfly species foundin the habitat (cf. Harrison 1994).

262 I S L A N D T H E O RY A N D C O N S E R VAT I O N

Table 10.1 Problems that have been identified with classicmetapopulation models

1. Occurrence of core–satellite relations2. The rescue effect is not dealt with adequately3. Distance effects and varying dispersal abilities4. The occurrence of ‘sink’ habitats5. Metapopulations are very difficult to replicate, tests are

problematic6. Much extinction is deterministic

Page 276: Whittaker. Island Biogeography. 2007

Thomas (1994) usefully reminds us that there aremany reasons for species colonization of a patch,and they are not all island effects. He suggests sixcircumstances in which natural colonizations bybutterflies are observed:

● succession following disturbance● where new ‘permanent’ habitat is created close toexisting populations● introductions of species outside former rangemargins● regional increases in range● turnover-prone peripheral patches (which act inessence as ‘sinks’ and are therefore unimportant tooverall persistence)● seasonal spread.

Thomas regards butterflies as persisting inregions where they are able to track the environ-ment, and becoming extinct if they fail to keep upwith the shifting habitat mosaic, or if the shift inhabitat regionally is against them. Metapopulationmodels therefore need to be superimposed on anenvironmental mosaic, which in many cases willitself be spatially dynamic. ‘An environmentalmosaic perspective shifts the emphasis on to tran-sient dynamics and away from the equilibrium(balance) concept of metapopulation dynamics, forwhich there is little evidence in nature’ (Thomas1994, p. 376). The possibility of incorporatingpatchy disturbance phenomena into metapopula-tion models has since been developed by Hess(1996).

Value of the metapopulation concept

As Gotelli (1991) concludes, the empirical supportfor classic metapopulation models is fairly thin onthe ground and tests are liable to be difficult.Metapopulations are difficult to replicate, and thetimescale of their dynamics (many generations ofthe organisms involved) may be of the order ofdecades. In addition, it is readily appreciated thatpopulations are subdivided on many scales and sothe delimitation of the local population is often sub-jective. Hence, to achieve greater realism, meta-population simulations have to be extended fromthe original, simplistic models, to allow for the

differing degrees of population connectivity, anddiffering forms of interpatch relationship, to befound in real systems (Harrison 1994; Hanski 1996;Ehrlich and Hanski 2004). Fahrig and Merriam(1994) identify the following factors as potentiallysignificant:

● differences among the patch populations in termsof habitat area and quality● spatial relationships among landscape elements● dispersal characteristics of the organism of interest● temporal changes in the landscape structure.

These factors can be difficult to capture and mayrequire rather different variables to be assessed fordifferent species.

In modelling exercises there is typically just sucha trade-off between greater realism and improvedmodel fit on the one hand and generality on theother. In this respect metapopulation models areno better or worse than most other ecologicalmodels. The value of the metapopulation conceptis that it has led conservation biologists to givegreater attention to the spatial structure andtemporal interdependency of networks of localpopulations, which is a clear step forward fromconsidering them as essentially independententities.

10.5 Reserve configuration—the ‘SingleLarge or Several Small’ (SLOSS) debate

Given a finite total area that can be set aside for con-servation as a natural landscape is being convertedto other uses, what configuration of reserves shouldconservationists advocate? At one extreme is thecreation of a single large reserve; the alternative isto opt for several smaller reserves amounting to thesame area, but scattered across the landscape. Thisquestion, reduced to the acronym SLOSS, was castin terms of the assumptions and predictions of theEMIB (e.g. Diamond and May 1981). As one of themain aims of conservation is to maximize diversitywithin a fragmented landscape, might not islandecological theory, which focuses on species num-bers, provide the answer as to the optimal configu-ration of fragments?

T H E ‘ S I N G L E L A R G E O F S E V E R A L S M A L L’ ( S L O S S ) D E B AT E 263

Page 277: Whittaker. Island Biogeography. 2007

Following the island analogy, the first factors tobe included are isolation and area. Increased iso-lation of reserves reduces migration into them fromother reserves. If a reserve is created by clearance ofsurrounding habitat, then it follows that on initialisolation the immigration curve should bedepressed. The contiguous area of habitat is alsoreduced and thus extinction rate should increase.At the point of creation, therefore, the habitat islandcontains too many species (it may even gain fugi-tive displaced populations), and the result is that itbecomes supersaturated. It follows that it should intime undergo ‘relaxation’ to a lower species num-ber, a new equilibrium point (Fig. 10.5). Givenknowledge of isolation and area of patches, itshould be possible to estimate the number main-tained at equilibrium in a variety of configurationsof habitat patches. On these grounds, Diamond and

May (1981) favoured larger rather than smallerreserves, short rather than long interreserve dis-tances, circular rather than elongated reserves(minimizing edge effects), and the use of corridorsconnecting larger reserves where possible (Fig. 10.6).These widely cited suggestions spawned a largelytheoretical debate in which much appeared to hangon the validity and interpretation of the EMIB.Having considered the EMIB, MVP, and metapopu-lation ideas ahead of it, some of the limitations ofthe SLOSS approach will be immediately apparentand so can be dealt with quickly.

The simplest criticism of this approach followsthe line that the EMIB is a flawed model. Hence itprovides no firm foundation for the developmentof conservation policy. If policy-makers adopt suchtheories as though providing formal rules, as somedid in this case, this criticism is justified (Shafer

264 I S L A N D T H E O RY A N D C O N S E R VAT I O N

Nearpatch I

“Mainland”sample l Small

patch E

Farpatch I

Largepatch E

Biotic co

llapse

00

Imm

igra

tion

rat

e (I

)

d c b a

Ext

inct

ion

rat

e (E

)

Species number

Figure 10.5 According to the assumptions of the EMIB, reducing area causes supersaturation as immigration rate declines and extinction raterises. This causes loss of species, i.e. ‘relaxation’ to a lower equilibrium species number from a → b → c → d. In the extreme scenario of ‘bioticcollapse’, the immigration rate is so low that the equilibrium species number is near zero. (Redrawn from Dawson 1994, Fig. 4.) In practice thisabstract theory may lack predictive power as to either numbers lost or rate of loss, being founded on assumptions of the system passing fromequilibrium (pre-fragmentation), to non-equilibrium (at fragmentation), to a new equilibrium condition (when I and E balance again):circumstances which may not apply in complex real-world situations. This is not to deny species losses, they do occur, but very often they are nota random selection, but instead are drawn from predictable subsets of the fauna or flora, structured in relation to: particular ecologicalcharacteristics of the species in question; habitat or successional changes in the isolate; and the nature and dynamics of the matrix around it.Equally, isolates may acquire increased numbers of some species, or gain new species, sometimes of an ‘undesirable’ nature judged in terms oftheir impact on the original species.

Page 278: Whittaker. Island Biogeography. 2007

1990). The position adopted in this book, however,is that the EMIB has a place in a larger frameworkof island ecological theory. It is not that the theoryis either true or false, but that the effects it modelsmay be either strong or weak. If the effects are veryweak for the habitat island system being consid-ered, as often is the case, then the answer to theSLOSS question will not be supplied by the equili-brium theory.

The next obvious criticism of the attempts tosolve SLOSS using the EMIB is that the theory isinsufficiently precise on the amount of composi-tional overlap that will occur across a series ofisolates of varying area. Deployment of the EMIBrationale often comes down simply to the use ofspecies–area equations. Such an approach may givea rough idea of numbers of species on habitatislands, but not which habitats contribute most torichness, nor which species are most likely to be lostfrom the remnants (Saunders et al. 1991; Simberloff1992; Worthen 1996). At its simplest, a strongly non-nested series of small reserves may be anticipatedto hold more species than a single large reserve, butwhere the system is perfectly nested, the largestreserve will hold most species (discussed morefully below).

Interpretations of species–area relations based onequilibrium theory have it that the regression lineprovides the equilibrium number of species. The

statistical interpretation of the scatter is simply thatthe regression line represents the average numberof species for a given area and a point above itrepresents a positive error (Boecklen and Gotelli1984; Boecklen 1986). If a newly created habitatisland lies above the line it doesn’t necessarily fol-low that it is supersaturated and destined to losespecies. It could just be that habitat quality or het-erogeneity allows it to maintain more than theaverage number of species. In his analyses of alarge US bird census data set, Boecklen (1986)found that habitat heterogeneity is a significantpredictor of species number even after area hasbeen factored out. The significance of habitat effectsin the data set points to the possibility that inparticular cases several small reserves can incorpo-rate a wider (or better) array of habitat types andthus support more species than the single largeoption.

The discussion of SLOSS issues that follows willmostly be limited to biological and physical param-eters. However, it should be recognized that thereare often overriding practical considerations. Forinstance, the costs involved in establishing andmaintaining conservation and game parks in devel-oping countries depend on questions such as thefollowing (Ayers et al. 1991). Is perimeter fencingrequired? Is it necessary to patrol to prevent poachingor encroachment into the reserve? How many

T H E ‘ S I N G L E L A R G E O F S E V E R A L S M A L L’ ( S L O S S ) D E B AT E 265

D

E

F

A

B

C

Better Worse Better Worse

Figure 10.6 The suggested geometric principles for the design of nature reserves which were supposedly derived from island biogeographicstudies, and which were at the centre of the so-called SLOSS, or Single Large Or Several Small debate (redrawn after Diamond 1975b). These‘principles’ have been challenged on both theoretical and practical grounds (see text).

Page 279: Whittaker. Island Biogeography. 2007

wardens do you need? Do you have to equip andmanage park headquarters? In many situations,fewer larger reserves make more sense from thesepractical, financial, and inevitably political perspec-tives. Such considerations often have overridingimportance to decision makers, and many recentcommentators regard the theoretical debates ashaving contributed little of direct practical value toconservation (e.g. Saunders et al. 1991). However,the implications of different reserve configurationsstill demand attention, irrespective of the fate of thetheoretical frameworks or practical realities of landownership and opportunity that spawned them.

Dealing with the leftovers

In most cases, natural scientists are unable to exertgreat influence on the basic ground-plan of frag-mentation, and even where they do have a role toplay, it is never a free hand. There is, moreover, arange of biogeographical, economic and politicalconsiderations which can be crucial to the strategyadopted. This has been recognized even by theadvocates of the SLOSS ‘principles’. For instance,the proposed reserve system for Irian Jaya, onwhich Jared Diamond acted as an adviser, rightlypaid more attention to pre-existing reserve designa-tions, distributions of major habitat types, and cen-tres of endemism, and to economic andland-ownership issues, than it did to the SLOSSprinciples (see Diamond 1986).

Most often, conservationists today are in theposition of dealing with and managing a system ofhabitat fragments, and perhaps modifying its con-figuration slightly. Fragmentation may have beencarried out in a very selective fashion. Thus, on thewhole, farmers take the best land and leave orabandon the least useful. In much of southernEngland, this has meant that most remaining nativewoodland is on very heavy, clay-rich soils, with thelighter soils in agricultural use. The message is thatthe species mix in your fragments may not be rep-resentative of what was in the original landscape.To give just one example, the short-leaved lime(Tilia cordata) does not favour the heavier soils, andit is now known as a species of hedgerows. In thepast it was probably an important forest tree

(Godwin 1975). Equally, the juxtaposition of land-scape units may be critical. For instance, someAmazonian forest frogs require both terrestrial andaquatic environments to be present within a reservein order to provide for all stages of the life cycle(Zimmerman and Bierregaard 1986).

It must also be recognized that humans mayinterfere directly in events within reserve systems.Bodmer et al. (1997) point out that although muchwork has focused on extinction rates caused bydeforestation, many of the recorded extinctions inthe past few hundred years have been the result ofover-hunting. They collected data on the relativeabundance of large-bodied mammals in the north-eastern Peruvian Amazon in areas with, and with-out, persistent hunting pressure. They found thatspecies with long-lived individuals, low rates ofincrease, and long generation times are most vul-nerable to extinction. In one reserve in north-eastern Peru, it was found that nearly half the meatharvested by hunters in the buffer zones of thereserve originated from mammals that are catego-rized as vulnerable to over-hunting. Similarly,Peres (2001) uses game-harvest data to argue thatthere may be synergistic effects of hunting andhabitat fragmentation, such that persistent over-hunting in fragmented landscapes drives largervertebrate populations to local extinction, espe-cially from smaller habitat patches. Such results canbe used to help formulate appropriate policytowards such activities.

Trophic level, scale, and system extent

The optimal configuration of areas is liable to varydepending on the type of organism being consid-ered, and on the spatial scale of the system. If forcedto generalize with regard to trophic level, largerreserves are more appropriate for large animalspecies needing a large area per individual, pair, orbreeding group, and/or requiring ‘undisturbed’conditions, e.g. the Javan rhino (Rhinoceros sondaicus)(see Hommel 1990). Species requiring large areas areoften the ones most threatened by us and in need ofprotection, e.g. top carnivores, primates, and rhi-noceros. Whereas top carnivores, such as big cats,may require large territories, it might be possible to

266 I S L A N D T H E O RY A N D C O N S E R VAT I O N

Page 280: Whittaker. Island Biogeography. 2007

keep the highest diversity of butterfly species bymeans of a number of small reserves, each targetedto provide particular key habitats (Ewers andDidham 2005). They should not, however, be toosmall, lest they consist principally of ‘sink’ popula-tions. However, trophic level and body size are notthe only considerations. We have also to thinkabout how species use their habitat space. Forinstance, birds of prey may have large territoriesincorporating both good and poor habitat.Conservation of large birds of prey in Scotland,such as the golden eagle (Aquila chrysaetos) and theosprey (Pandion haliaetus), has required a range ofmeasures taken to discourage and prevent killing ofthe birds within a mixed-use landscape. It has alsorequired intensive but highly localized efforts toprotect nesting sites from egg collectors. The‘reserves’ required are tiny compared to the rangesof the individuals.

Although, in general, most studies advocaterelatively large reserves at the expense of largernumbers, in cases it may be better to have a set ofseveral small reserves. First, they may incorporatemore different habitats, i.e. they may capture betadiversity across a landscape. Secondly, competitionmay, in theory, lead to the exclusion of species ofsimilar niches from any given reserve and so it maybe good to have several reserves so that differentsets of species may ‘win’ in different reserves.Thirdly, there is an epidemiological risk inherent inhaving ‘all your eggs in one basket’, and, fourthly,particular species may rely upon small islands. Forinstance, small estuarine islands in the Florida Keysarea appear to provide important breeding sites forseveral species of waterbirds (Erwin et al. 1995).

In organizing our thinking about the implicationsof the SLOSS debate, we would do well to considertwo aspects of spatial scale: study system grain andextent (cf. Whittaker et al. 2001). Grain refers to thesize of the sampling unit in spatial analyses. In thiscontext, grain size is variable: we refer in fact to therange of areas of the habitat islands. Log–log plotsof species richness and area (e.g. Fig. 4.2) oftenindicate simple linear trends of converging richnessbetween fragmented and unfragmented areas assample area increases. However, recent theoreticaldiscussion has reopened the question of whether

this is the right approach to fitting trend lines tosuch data (e.g. Lomolino 2000c; Tjørve 2003; Turnerand Tjørve 2005), and in particular whether morecomplex forms of relationship should be tested forat the ‘small island’ end of the relationship.Empirically, it appears that superior model fits canbe derived for many ‘island’ data sets using a formof break-point regression, consistent with the exis-tence of a so-called small-island effect. In practice,this means that there is no systematic increase inrichness until a critical area threshold is crossed,producing a two-phase relationship shifting fromflat to increasing (Fig. 4.6; Lomolino and Weiser2001; Triantis et al. 2006). The implication of thisfinding is that the slope and intercept of ISARS fit-ted using log–log models may be more sensitive tothe range in area of isolates sampled than generallyrealized. This in turn may impact on our assess-ment of the relative richness gain from havinglarger-sized reserves: although it should again benoted that simply being able to predict the richnessof each isolate does not reveal the overall richnessof the system.

Turning to the study system extent, if we imaginetwo networks of a dozen habitat islands, each net-work identical in terms of the starting richness ofeach island, the conservation implications of increas-ing distance separating each of the habitat islandswork in perhaps two key ways. First, the furtherapart they are, the lower the probability of theirspecies populations mutually reinforcing oneanother, thereby preventing extinction from the net-work. Second, the further apart they are, the morelikely they are to sample different habitats, and ifthey are really far apart they may even sample dif-ferent biomes or different biogeographical sourcepools. Thus a widely scattered system of reservesmay capture more differentiation diversity (sensuWhittaker 1977): but whether it can retain it in thelong run may depend on the extent to which eachreserve can sustain its initial species complement inthe long term. Hence, the answer to whether it is bet-ter to have one large or several small reserves may besensitive to the grain and extent of the system. Ofcourse, the SLOSS ‘principles’ were advocated moststrongly for circumstances in which the biogeogra-phy of an area is poorly known. When working at a

T H E ‘ S I N G L E L A R G E O F S E V E R A L S M A L L’ ( S L O S S ) D E B AT E 267

Page 281: Whittaker. Island Biogeography. 2007

fine scale of analysis, the ‘island’ assumption is atleast a rational starting point for thinking about theSLOSS question. However, at coarse scales of analy-sis, the relevance of island theory is less clear, and soconservation biogeographers have developed otherapproaches, designed to make use of the improvedavailability of distributional data in digital form (atleast for some taxa). For instance, computer algo-rithms have been developed to achieve goals such asfinding the 10% of sites/grid cells that maximizerepresentation of rare or endemic species (e.g. seeAraújo et al. 2005a; O’Dea et al. 2006; and papers citedtherein). Such analyses provide a far more directapproach to conservation planning than the originalSLOSS guidelines and effectively supercede them atcoarse scales of analysis.

To sum up, on theoretical grounds, the answer tothe SLOSS debate is equivocal (e.g. Boecklen andGotelli 1984; Shafer 1990; Worthen 1996). The answerdepends, as we have argued here, on the ecology ofthe species (or species assemblage), on system grainand on extent. It also depends on a range of othersystem properties, which we will now explore.

10.6 Physical changes and thehyperdynamism of fragment systems

Large environmental changes can be involved inecosystem fragmentation, particularly whereforests are fragmented (Lovejoy et al. 1986;Saunders et al. 1991). The immediate and subse-quent differences in fluxes of radiation, wind,water, and nutrients across the landscape can havesignificant consequences for the ecology of theremnants, and have been summed up by Laurance(2002, p. 595) in the idea of hyperdynamism, whichhe defines as ‘an increase in the frequency and/oramplitude of population, community, and land-scape dynamics in fragmented habitats’. Thefollowing examples are illustrative.

● Radiation fluxes. In south-western Australia,elevated temperatures in fragmented landscapesreduced the foraging time available to adultCarnaby’s cockatoos (Calyptorhyncus funereuslatirostrus) and contributed to their local extinction.

● Wind. When air flows from one vegetation typeto another it is influenced by the change in height

and roughness. At the edge of a newly fragmentedwoodland patch, increased desiccation, wind-damage, and tree-throw can occur (e.g. Kapos1989). Increased wind turbulence can affect thebreeding success of birds by creating difficulties inlanding due to wind shear and vigorous canopymovement. Wind-throw of dominant trees canresult in changes in the vegetation structure, andallow recruitment of earlier successional species.

● Water and nutrient fluxes. Removal of nativevegetation changes the rates of interception andevapotranspiration, and hence changes soil mois-ture levels (Kapos 1989). In parts of the wheat beltof Western Australia, the new agricultural systemscause rises in the water table, which can bringstored salts to the surface, and this secondary salin-ity has caused problems both to agriculture and theremnant patches. In the fenlands of easternEngland, drainage for agriculture has led to peatshrinkage and a drop of 4 m in land level in130 years. Remnant areas of ‘natural’ wetland nowrequire pumping systems to maintain adequatewater levels (see also Runhaar et al. 1996).

● Fire regimes. Modified landscapes can be a majorsource of surface fires, for example from burning ofadjoining pastures. Penetration of such fires intofragment interiors can increase plant mortality, dis-turb the fragment boundaries and in time cause an‘implosion’ of forest fragments (Laurance 2002).

The physical effects of edges thus have importantknock-on effects for the biota, especially soon afterfragmentation (Table 10.2). In time, the systemadjusts to the new physical conditions, and thewoodland edge fills and becomes more stable. Yet,just as there is a wide range of disturbance regimesacross real islands, so must there be for continentalhabitats, and as we create new habitat islands, weinevitably alter disturbance regimes (Kapos 1989;Turner 1996; Ross et al. 2002). Furthermore, land-use of the matrix in which the habitat islands areembedded typically continues to change (Laurance2002). It might therefore be anticipated that manyhabitat islands will be characteristic of the dynamicnon-equilibrial position identified in Fig. 6.1 (cf.Hobbs and Huenneke 1992). Indeed, it has beensuggested that smaller habitat islands may be sub-ject to greater disturbance impacts than larger

268 I S L A N D T H E O RY A N D C O N S E R VAT I O N

Page 282: Whittaker. Island Biogeography. 2007

islands, thereby making such sites less suitable for,or removing altogether, particular species(McGuinness 1984; Simberloff and Levin 1985;Dunstan and Fox 1996).

10.7 Relaxation and turnover—theevidence

If habitat islands behave according to the expecta-tions of the EMIB, they should be supersaturatedwith species immediately after system fragmenta-tion. Subsequently, species numbers in the isolatesshould ‘relax’ to a lower level (Fig. 10.5). Immi-gration and extinction should continue to occur,both during the relaxation period and subse-quently, when the island has found its new, lowerequilibrium richness level. The time taken for relax-ation to occur is called the ‘lag time’ and the antici-pated eventual species loss is termed the extinctiondebt (Ewers and Didham 2005). If these anticipatedeffects are strong, even high-priority species—thevery species that you wish to conserve—may intime disappear from a reserve. As we have noted,however, the empirical evidence from a wide rangeof real and habitat islands is that turnover tends to

be heterogeneous (i.e. structured). Some speciestend to be highly stable in their distribution acrosshabitat islands. Most turnover involves ‘ephemerals’,species marginal to the habitat, or successionalchange (Chapters 4–6). The metapopulation studiesexamined above provide further insights intoturnover. They suggest a tendency for largepopulations to persist, while smaller satellite popu-lations may come and go, without jeopardizing themetapopulation. Nonetheless, species do disappearfrom particular habitat islands, and from entirelandscapes, and in cases the loss is a global one.Earlier, we noted that many species losses areattributable to deterministic causes, meaning hunt-ing, habitat changes, and the like. The relaxationeffect, in contrast, is based on stochastic processes.How strong is this island relaxation effect?

One of the best-known examples of relaxation onecological timescales is that of birds lost from BarroColorado island, in Panama. It is not an ideal studyin the present context in that, unlike most habitatislands, it is actually now surrounded by water. Theisland was formerly a hilltop in an area of continu-ous terrestrial habitat, but it became a 15.7 km2

island of lowland forest when the central section ofthe Panama Canal Zone was flooded to make Lake

R E L A X AT I O N A N D T U R N OV E R — T H E E V I D E N C E 269

Table 10.2 Classes of edge-related changes triggered by the process of forest fragmentation, as informed by theMinimum Critical Size of Ecosystem project. The first-order effects may lead to second-order and, in turn, third-orderknock-on effects (modified from: Lovejoy et al. 1986)

Class Description of change

Abiotic TemperatureRelative humidityPenetration of lightExposure to wind

BiologicalFirst-order Elevated tree mortality (standing dead trees)

Treefalls on windward marginLeaf-fallIncreased plant growth near marginsDepressed bird populations near marginsCrowding effects on refugee birds

Second-order Increased insect populations (e.g. light-loving butterflies)Third-order Disturbance of forest interior butterflies, but increased population of light-loving species

Enhanced survival of insectivorous species at increased densities (e.g. tamarins)*

*Does not apply to birds (Stouffer and Bierregaard 1995).

Page 283: Whittaker. Island Biogeography. 2007

Gatun in 1914. Of about 208 bird species estimated tohave been breeding on Barro Colorado islandimmediately following isolation in the 1920s and1930s, 45 had gone by 1970 (Wilson and Willis1975). Of these, some 13 have been attributed to‘relaxation’.

The other losses could be attributed to particularforms of ecological change. At the time of the analy-sis, much of the forest was less than a century old,following abandonment of farming activity before1914. Many of the birds lost were typical of secondgrowth or forest-edge, suggesting a probablesuccessional mechanism as forest regenerationreduced the availability of these more open habi-tats. Some were ground nesters, which may havebeen eliminated by their terrestrial mammalianpredators. The latter became abundant because ofthe disappearance of top carnivores (such as thepuma) with large area requirements (Diamond1984). This effect, of increasing numbers of smalleromnivores and predators in the absence of largeones, has been termed mesopredator release (Souléet al. 1988), and it has now been documented inseveral other systems (Fig. 10.7; Crooks and Soulé1999; Laurance 2002). Some of the birds lost weremembers of the guild of ant followers, which have

been found to be vulnerable to fragmentationelsewhere (below). In a simple sense, Barro Coloradoisland provides evidence of relaxation, but thestochastic signal is seemingly much smaller thanthat produced by changing habitat and food-webrelationships, and it might therefore be asked‘which is the signal and which the noise in thesystem?’ (cf. Sauer 1969; Lynch and Johnson 1974;Simberloff 1976).

That species numbers and composition maychange as a consequence of fragmentation is not indispute. However, it is remarkably difficult to findgood quantitative data for the island relaxationeffect for systems of habitat islands (Simberloff andLevin 1985; and see review in Shafer 1990). Moststudies, including that for Barro Colorado island,lack precise information about the species composi-tion before fragmentation (e.g. Soulé et al. 1988).The Barro Colorado island study is also fairly typi-cal in the losses recorded being mostly ‘determinis-tic’ in nature. The predictions of species lossthrough ‘relaxation’ derive from the EMIB, but the-oretically relaxation may also affect metapopula-tions. Hanski et al. (1996) speculate that the pace offragmentation in many regions has been so rapidthat ‘scores of rare and endangered species mayalready be “living dead”, committed to extinctionbecause extinction is the equilibrium toward whichtheir metapopulations are moving in the presentfragmented landscapes’ (p. 527). However, theirfindings are heavily dependent on models of thedata of a single species of butterfly, the Glanvillefritillary (Melitaea cinxia). The effect identifiedappeared to be far from instantaneous, and mightturn out in practice to be a relatively weak effect inrelation to some of the other processes discussed inthis chapter.

A classic illustration both of genuine reasons forconcern and of the uncertainty involved in projec-tions is provided by the tropical moist forests of theAtlantic seaboard of Brazil, which have beenreduced over the past few centuries to only about7% of their estimated former cover (Ribon et al.2003). Based on the ‘rules’ derived from islandtheory, some 50% of species are expected to goextinct. To date, no extinctions have been docu-mented with certainty (Whitmore and Sayer 1992;

270 I S L A N D T H E O RY A N D C O N S E R VAT I O N

Fragmentarea

Fragment age(time since isolation)

+

+ ––

––

Figure 10.7 Model of the combined effects of trophic cascades andisland biogeographical processes on top predators (for example,coyote), mesopredators (domestic cat) and prey (scrub-breeding birds)in a fragmented system. Direction of the interaction is indicated witha plus or minus. (From Crooks and Soulé 1999.)

Page 284: Whittaker. Island Biogeography. 2007

Ribon et al. 2003; Grelle et al. 2005), although manyspecies are Red-listed as vulnerable, endangered orcritically endangered based on range, or populationestimates (Ribon et al. 2003). Brooks and Balmford(1996) compare losses of birds projected using aspecies–area model with those listed by the IUCNas ‘threatened’ and find congruence, i.e. they arguethat the predictions are basically correct, but thatthere is a substantial lag between the habitatloss/fragmentation process and global extinctionof the species. There is, moreover, evidence oflocal extirpation within the existing range forbird species. Thus, within the Viçosa region(a 120 km2 area in southeastern Brazil) over thelast 70 years, it appears that at least 28 bird specieshave become locally extinct, with 43 ‘criticallyendangered’, and 25 ‘vulnerable’: in total 61% ofthe original avifauna has been significantly reducedin incidence (Ribon et al. 2003). Nectarivorousspecies were least at risk, next came omnivores andcarnivores, with frugivores and insectivores ingreatest jeopardy.

Assuming relaxation to be underway in thesefragmented systems, the next problem is to esti-mate the lag time between fragmentation and even-tual species losses. Brooks et al. (1999) attempt thisfor tropical forest birds in an area of fragmentedupland forest in Kenya. They base their analyses onisland theory, assuming z values (slope of the SAR)pre-isolation of 0.15, and post-isolation of 0.25 (i.e.for the new ISAR), and that the rate of decline inspecies is approximately exponential. This postu-lates that fragments will lose approximately halfthe species that will eventually be lost every xyears: this they termed the ‘half-life’. The difficultyis finding systems with ‘before’ and ‘after’ data.The system they chose for the test is the Kakamegaforest, where forest fragmentation occurred over60 years ago (but where the time frame is not per-fectly specified). Their analyses of five fragmentssuggests a half-life varying in relation to fragmentarea from 23 (100 ha) to 80 (8600 ha) years, and ofapproximately 50 years for a 1000 ha fragment. Asmight be anticipated from island theory, relaxationtimes were suggested to be scale dependent in rela-tion to both range in area (‘grain size’) and degreeof isolation (cf. Box 10.2).

One important caveat to this study is, however,that the historical data for the Kakamega area werederived from general collecting/observations fromthe Kakamega area, with values assigned to thefragments on the basis of the assumption ofz � 0.15 (slope of SAR) for their pre-isolation states.Hence, the figures derived for relaxation timesshould be seen as still of the nature of first approx-imations. Moreover, the relaxation times per frag-ment do not tell us overall species losses from thesystem. In practice, the data reviewed by Brooks et al.(1999) suggest that some loses have occurred, with10 of an estimated 73 Kakamega species not havingbeen recorded from these forests in recent years. Ofthose species counted as lost at least one is thoughtto have been trapped to extinction for the caged-birdtrade; three others were at their eastern range limitswithin the study area, so their loss from this areamight be related to other processes and may notbe indicative of their risk of global extinction. Inpractice, a single study system, of five fragments,based on relatively poorly specified empirical data,is a rather limited basis for building predictive mod-els, and there are surprisingly few other such studieswith which to compare the values obtained.

Box 10.2 appraises the use of species–area modelsin predicting regional species extinctions as a func-tion of regional habitat loss, arguing that althoughthere is a crude relationship between habitat lossand species loss, there is no logical connectionbetween such models and island theory. It shouldalso be recognized that even where species–areaanalyses are linked phenomenologically to islandtheory (i.e. they are being applied to systems offragments using parameters derived from othersuch systems, which also vary over similar rangesof area and isolation), they provide only a coarse,stochastic simulation of the impacts of fragmenta-tion. The actual drivers of extinction may bestructured, and in some instances avoidable byappropriate management. For instance, huntingand collection of birds and mammals may each beimportant drivers of the decline of particularspecies (Peres 2001; Ribon et al. 2003). With respectto the Atlantic forest system, conservation actionhas likely already been and will continue to becrucial to staving off the final extinction of highly

R E L A X AT I O N A N D T U R N OV E R — T H E E V I D E N C E 271

Page 285: Whittaker. Island Biogeography. 2007

272 I S L A N D T H E O RY A N D C O N S E R VAT I O N

Box 10.2 Does island theory provide a basis for the use of species–area regressions topredict extinction threat?

In the text, we discuss a number of studies thathave used species–area models to predict theeventual species losses following fragmentation(the so-called ‘extinction debt’). A growing bodyof literature has applied this approach at varying,sometimes very coarse, scales (e.g. Brooks et al.2002), to forecast species losses in the tropics as afunction of forest loss. Recent work has attem-pted both to assess extinction threat and validatethe application of species–area models bycomparing at a regional scale the forecasts ofspecies–area models to lists of threatened species.Where congruence is found, the analyses areregarded as mutually supporting: island effects areworking, systems are ‘relaxing’ to lower speciesnumber, and given time (the ‘lag effect’), will re-equilibrate to a lower species richness. Theapplication of species–area models is generallytaken to be supported theoretically by work onisland species–area relationships, from which theslope parameter z is derived.

Returning to the Atlantic forest system discussedin the text, the study by Grelle et al. (2005)assesses the sensitivity of species–area basedpredictions of extinction to the z parameter (slopeof the relationship) selected. Their paper providesan analysis of data for Rio de Janeiro State, inwhich present and historical coverage of forest areestimated at 19% and 100%, respectively.Following other recent studies, they first predictedextinctions using a species–area model with a zvalue of 0.25, and compared the predicted specieslosses with the number of species identified by anexpert workshop as threatened with extinction.Their analysis suggested that the species–areamodel overestimated extinction threat for each ofmammals, birds, reptiles and amphibians. Whenthey repeated the analysis for endemic speciesonly, they found that the species–area modelsuccessfully predicted threatened endemicmammals when using a z of 0.25, and threatenedendemic birds when using a z ranging between0.1–0.19.

We have detailed this study as it is one ofseveral in the literature (e.g. Brooks and Balmford

1996; Brooks et al. 2002; Thomas et al. 2004)that use species–area regression as a means offorecasting species threatened by, or committedto, extinction. However, the way in which thespecies–area models are used in this and otherstudies is conceptually decoupled from the islandtheory from which it seemingly derives. InChapter 4 we cite MacArthur and Wilson (1967)for the observation that ISAR z values may varybetween 0.2 and 0.35, and note that the range invalues may indeed be much greater (Williamson1988). Therefore, first, a z of 0.25 is a largelyarbitrary ‘middle’ value to take. Second, and morecrucially, this z value has been derived fromanalyses of true isolates: it tells us approximatelyhow many species are held in each of a series ofisolates/islands of different size. Note that thetotal number of species held across the series ofisolates is not derivable from the z parameter ofthe isolates: as the overall richness of the systemdepends on the degree of compositional overlap(nestedness) among the isolates. Yet, in recentstudies such as these, the z value is applied not toseparate fragments, but to the entire region. Thatis, the authors assume that the whole system (inthis case Rio de Janeiro State) is acting as a singlefragment, which has lost 81% of its area andwhich is therefore travelling along a trajectorytowards its lower equilibrium point. However, inpractice, we are talking about an area in whichforest habitat persists in very large numbers ofpatches, scattered across that region in varyingconfigurations. This loss of habitat is certainlybound to vastly reduce the number of individualsof forest species across that region, but whetherthat results in particular species falling below theirminimum viable population in all isolates is notknowable simply from the regional figure forhabitat loss (see e.g. Loehle and Li 1996).

There is thus no link of scientific logic that cancarry us from empirical demonstrations of z valuesfor islands and isolates, via MacArthur andWilson’s EMIB, and the related theoretical idea ofspecies relaxation (Diamond 1975b), through tothe assumption of a particular shape and slope of

Page 286: Whittaker. Island Biogeography. 2007

R E L AT I O N A N D T U R N OV E R — T H E E V I D E N C E 273

relationship between the percentage of forest losson a regional basis and regional species losses. Inshort, the use of species–area regressions in thisway appears to be currently without a theoreticaljustification, and thus represents merely anexercise in curve-fitting.

Taken as such, what do we learn from suchstudies? The Grelle et al. (2005) study illustratesthat, in practice, there is no reason to favour a zvalue of 0.25 over other slope parameters inmaking what amounts to a back-of-the-envelopesketch of how regional habitat loss may relate toloss of species regionally. In practice, we wouldbe better to use species-level assessments(whether underpinned by rigorous survey data orexpert assessments) than to rely on suchspecies–area fits. Such assessments enable

attention to be given to species judged most atrisk, and necessarily involve an assessment of whatfactors threaten particular species. This sort ofinformation would seem to be of inherent value toconservation agencies. Moreover, the argumentthat we should use species–area modelling in theabsence of adequate data for other regions orother taxa, arguably breaks down with therealization that the success of the study in fittingspecies–area models to independent data onspecies threats varied between taxa, withextinction risks of reptiles and amphibians notsuccessfully predicted in any of their models. Thisis one area within conservation biogeographywhere current theory appears inadequate and inurgent need of attention (as Whittaker et al.2005).

threatened species (Birdlife EBA factsheet 75,�www.birdlife.org/datazone/�, visited July 2004).Given these varying forms of human intervention, itmay be impossible to resolve the question of whetherthe Atlantic forests system provides support for, or arefutation of, the theory of species relaxation.

The assumption that fragmentation will necessarilylead to ongoing species losses through relaxationhas been challenged by Kellman (1996). He arguesthat reference to past environmental change andspecies responses to it shows that many species aremore flexible in their ecological requirements thangenerally recognized. Kellman contends that manyplant communities are actually undersaturated andthat systems of fragments may be able to sustainincreased species densities for significant periods oftime (although not necessarily indefinitely). Hebases his case in large part on studies fromNeotropical gallery forests, which constitute nar-row peninsulars of forest extending through amatrix of savanna ecosystems. According to equi-librium theory, the tips of the peninsulas should beimpoverished because of their relative isolation.This extension of island theory is termed the penin-sula effect. It has been tested at a coarse scale forFlorida and for Baja California, and has generallybeen found to be a fairly weak effect (Brown 1987;

Means and Simberloff 1987; Brown and Opler1990). In the Neotropical gallery forests, Kellman(1996) reports that there is no evidence for floralimpoverishment with distance from the peninsu-lar base, possibly because the frugivores thatdisperse the seeds of the plants are capable ofadjusting their ranging behaviour to accommo-date the long, thin peninsular configuration, orindeed to move through the sparsely woodedsavanna matrix surrounding the peninsulas. Thesepeninsulas have a long history, and thus appear toprovide evidence for adequate gene flow of plantpopulations to the peninsular tips over long peri-ods, i.e. there has been plenty of time for ‘lageffects’ to unfold, but at least for plants, there is noevidence of species impoverishment. However, it isunclear how good an analogue these peninsulasprovide for the rapid conversion of forested areas tofragmented systems.

There is a long-term fragmentation project whichsheds some light on these questions, and in whichdata were collected before fragmentation: theMinimum Critical Size of Ecosystem project. It islocated near Manaus, in the Brazilian Amazon, andsites were turned into habitat islands, isolated fromintact forests by just 70 to 650 m of pasture, between1980 and 1984. While some patches remained

Page 287: Whittaker. Island Biogeography. 2007

isolated, the land surrounding others wasabandoned during the study and became infilledby forest regrowth dominated by Cecropia.

Stouffer and Bierregaard (1996) examined the useof these forest fragments by understorey humming-birds. The three species found before fragmentationpersisted at similar or higher numbers in the frag-ments through the 9 years of the study. Use of thefragments did not differ between 1 and 10 ha frag-ments. Furthermore, it mattered little whether thesurrounding matrix included cattle pasture, aban-doned pasture, or Cecropia-dominated secondgrowth. This particular guild of birds thus appearsto be able to persist in a matrix of fragments, sec-ondary growth, and large forest patches.

By contrast, the guild of insectivores, whichdominate the understorey bird community, wasmuch more responsive to fragmentation: both abun-dance and richness declined dramatically (Stoufferand Bierregaard 1995). Three species of obligatearmy-ant followers disappeared within 2 years.Mixed-species flocks drawn from 13 species disag-gregated over a similar timescale, although three ofthe species persisted in the fragments. The 10 hafragments were less affected than the 1 ha frag-ments, and in time the species flocks reassembled inthose 10 ha fragments that were surrounded byCecropia regrowth. The communities in these recon-necting fragments converged compositionally onpre-isolation communities, while communities inthe smaller or more isolated fragments continued todiverge. Particular terrestrial insectivores (e.g.Sclerurus leafscrapers and various antbirds) failedto return to any of the fragments during the study.To sum up: (1) the impact of fragmentation on birdsis strongly ecologically structured, (2) different‘guilds’ respond in divergent fashion, and (3) thetypes of species most vulnerable to fragmentationare fairly predictable, with insectivores that partici-pate in mixed-species flocks being the most vulner-able guild (e.g. Lovejoy et al. 1986; Stouffer andBierregaard 1995, 1996; Christiansen and Pitter 1997;Ribon et al. 2003).

We may thus conclude that habitat fragmentationis associated with compositional change andspecies losses. Theory predicts significant losseswhere fragmentation is extreme, but over an uncer-

tain time frame. However, it has yet to be satisfac-torily demonstrated that the stochastic ‘island’relaxation effect is generally a strong one, and itseems likely that there will be a degree of scaledependency in relation both to area and isolation:factors that warrant further theoretical and empiri-cal attention. Recent work suggests that there maybe taxon dependent thresholds of habitat loss,above which few species losses occur, but beyondwhich losses accelerate (Ewers and Didham 2005).Given the significance of the assumptions of relax-ation to the debate on species extinction rates, it issurprising that more attention has not been given todemonstrations of these effects (see Simberloff andLevin 1985; Simberloff and Martin 1991; Whitmoreand Sayer 1992; Brooks et al. 1999; Laurance 2002).

10.8 Succession in fragmentedlandscapes

The processes of land-use change which create frag-mented systems typically initiate successionalchanges in the habitat island remnants. This wasapparent in the Barro Colorado island study.Another example is provided by Weaver andKellman’s (1981) study of newly created woodlotsin southern Ontario, Canada. The species losses thatoccurred did not take the form of stochasticturnover, with different species ‘winning’ in differ-ent woodlots. Instead, the ‘relaxation’ involved thesuccessional loss of a particular subset of species.With appropriate management, the observed lossescould be avoided. If occurring at all in the absenceof disturbance and succession, turnover appeared tobe very slow-paced among the vascular flora. In thissystem, the EMIB effects were clearly weak, andsubordinate to other ‘normal’ ecological processes.

In practice, nature-reserve management oftenpays considerable attention to ecological succession,particularly in small reserves. Failure to do so oftenleads to the loss of desirable habitats. This applies tomany of the lowland heath reserves of southernEngland, which have long been anthropogenicallymaintained. Without appropriate management,most areas suffer woodland encroachment. Theconcepts of minimum dynamic areas and patchdynamics (Pickett and Thompson 1978) are thus

274 I S L A N D T H E O RY A N D C O N S E R VAT I O N

Page 288: Whittaker. Island Biogeography. 2007

relevant to management plans. Simplified repre-sentations of woodland dynamics view a stand oftrees as going through phases of youth, building,maturity, and senescence, each of which may sup-port differing suites of interacting species. Reservesshould therefore be large enough (or managed) toensure that they contain enough habitat patches atdifferent stages of patch life-cycles to support a fullarray of niches. A related and often contentiousissue is how fire regimes are managed. In fire-proneregions there is often a cyclical pattern of post-burnsuccession and fuel accumulation, leading to anincreased likelihood of fire, repeating the cycle. Themaintenance of a particular fire regime and patchmosaic structure may be of crucial relevance tospecies diversity in a reserve system, but is oftenpoorly understood (Short and Turner 1994; Milbergand Lamont 1995). Fire is also politically contentiousbecause of the threat it poses to people and property.

Much active management of nature reserves isthus about keeping a mosaic of different succes-sional stages, and not allowing the whole of thereserve to march through the same successionalstage simultaneously. The relevance of such succes-sional dynamics to species changes in reserve sys-tems appears to have received much less attentionin the theoretical conservation science literaturethan its significance warrants (but see Pickett et al.1992). Moreover, continuing changes in the habitatsof the matrix can also be extremely important to thefate of species populations in the reserves them-selves (Stouffer and Bierregaard 1995, 1996; Gasconet al. 1999; Daily et al. 2003).

10.9 The implications of nestedness

As we established in Chapter 5, many island andhabitat island data sets exhibit nestedness. That is,when organized into a series of increasing speciesrichness, there is a significant tendency for thespecies present in species-poor (small) patches to befound also in successively richer (larger) patches.Although nestedness can be driven by selective col-onization or by habitat nestedness, it appears thatdifferential extinction plays a major role in produc-ing nested structure in many habitat island datasets (Wright et al. 1998). A nestedness analysis can

contribute a simple answer to the SLOSS question,as a strong degree of nestedness implies that mostspecies could be represented by conserving therichest (largest) patch. A low degree of nestedness,on the other hand, would mean that particularhabitat patches sample distinct species sets and thatan array of reserves of differing size and internalrichness may be required to maximize regionaldiversity (cf. Kellman 1996). Knowledge of nestedsubset structure might therefore provide a basis forpredicting the ultimate community composition ofa fragmented landscape, particularly if it is possibleto attribute patterns to particular causes (Worthen1996, Fischer and Lindenmayer 2005).

Blake’s (1991) study of bird communities in iso-lated woodlots in east-central Illinois shows hownestedness calculations can provide useful insights.He demonstrated a significant degree of nested-ness, particularly among birds requiring forest inte-rior habitat for breeding and among specieswintering in the tropics. In contrast, species breed-ing in forest-edge habitat showed more variabledistribution patterns. Blake’s findings resonatewith those reported from São Paulo by Patterson(1990), who found significant nestedness amongstsedentary bird species, but that the full data set (i.e.also including transient species) was non-nested.These results indicate that some species, often thoseof most conservation concern, will be lacking fromany number of small patches, but can be found inthe larger, richer patches.

Tellería and Santos (1995) have demonstratedthe apparent importance of nestedness of habitat.They studied the winter use of 31 forest patches(0.1–350 ha) in central Spain by the guild of pari-forms (Parus and Aegithalos, Regulus, Sitta, andCerthia—tits, goldcrests, nuthatches, and treecreep-ers). They found that birds with similar habitatpreferences tend to disappear simultaneously withreduction in forest size, thereby producing a nestedpattern of species distribution.

Simberloff and Martin (1991) have argued thatestablishing nestedness across a whole system is nolonger particularly exciting, but that some benefitcan come from examining discrepancies from nest-edness. Cutler (1991) developed an index, U, whichwas designed to account for both unexpected

T H E I M P L I C AT I O N S O F N E S T E D N E S S 275

Page 289: Whittaker. Island Biogeography. 2007

presences and unexpected absences. He used it toidentify what he termed holes and outliers. Holesare where widespread species are absent fromotherwise rich faunas, and outliers are whereuncommon species occur in depauperate faunas.He applied his index to the data for borealmammals and birds of the North American GreatBasin.He found that for mammals most of thedepartures from perfect nestedness were due toholes, whereas for birds most departures were dueto outliers. Cutler speculates that this differencecould be a function of the superior dispersability ofbirds, allowing them to generate greater numbersof outliers by recolonization events. ‘Supertramp’species (Chapter 5) might also appear as outliers inanalyses of this sort. The preponderance of holes inthe mammal data supports the importance ofextinction as a driving force in this system, with theholes representing departures from the generallypredictable sequence of species losses that haveoccurred during the Holocene.

Such analyses might appear to provide impor-tant information for conservation. However,Simberloff and Martin (1991) rightly caution that, toa large degree, extinction across these mountain-top habitat islands has been inferred rather thandemonstrated to have happened. The simple statis-tics can, moreover, mask more complex and impor-tant underlying details. In illustration, they showthat the same nestedness score can be generated forspecies with widely differing underlying distribu-tions. In the Maddalena archipelago (Sardinia), twospecies have a nestedness score of zero (by theWilcoxon statistic). One, the rock dove (Columbialivia), is found across the entire range of island sizes,and is absent from only 2 of 16 islands. The second,the little ringed plover (Charadrius dubius), has beenobserved only once in the archipelago, breedingonly in one year. As they point out, there may be dif-ferent explanations for accordance or deviationfrom nestedness in particular species. The explana-tion of such patterns requires basic information onspecies’ presence–absence, abundances, minimumarea requirements, habitat use at different sites, andtemporal regularity in occupying differing types ofpatch. In their view, given such data, it becomespossible to offer sensible input into conservation

decision-making—input in which the contributionof the nestedness statistics is fairly limited.

Moreover, there are many different indices forassessing nestedness, and much debate overwhich is the best to use for particular purposes(Rodríguez-Gironés and Santamaría 2006). Cook(1995) undertook a comparative study of 38 insularsystems using 6 different indices of nestedness,ordered both by area and richness. The analysesordered by richness returned higher nestednessscores than those ordered by area in between 35and 37 cases (depending on the metric used).Cook’s analyses show that the broad patterns ofnestedness are robust, but that there is a dangerof over-interpreting the details of a nestednessanalysis.

Formal consideration of compositional structurewas, for too long, sidelined in the SLOSS debateand related literature. The attention now beinggiven to species compositional patterns across sys-tems of habitat islands is thus welcome (Fischerand Lindenmayer 2005). A nestedness index canprovide one compositional descriptor and can per-haps aid identification of risk-prone species; butshould not be given primacy in conservation plan-ning. The discovery of nestedness at a particularpoint in time does not necessarily provide clearinsights as to the probability of maintaining thesame sets of species (or any particular species) overtime (Simberloff and Martin 1991). The isolates maybe subject to turnover and/or species attrition innew ways dictated by the changing biogeographiccircumstances of the landscape in which the frag-ments occur. As Worthen (1996, p. 419) put it, nest-edness is not a ‘magic bullet’, ‘ . . . no single indexshould be expected to distil the informational con-tent of an entire community, let alone predict how itwill react to habitat reduction or fragmentation’.

10.10 Edge effects

The boundary zone, or ecotone, between two habi-tats, being occupied by a mix of the two sets ofspecies, is often richer in species per unit area thaneither of the abutting core habitat types. In thereserve context, attention has focused on thereserve edge, typically where a woodland reserve is

276 I S L A N D T H E O RY A N D C O N S E R VAT I O N

Page 290: Whittaker. Island Biogeography. 2007

surrounded by a non-woodland matrix. For somespecies groups, possession of a large expanse ofsuch edge habitat can increase the overall speciesnumber found on a reserve, an apparently benefi-cial effect (Kellman 1996). But, the argument goes, ifthe additional species supported by the edge habi-tat are those of the matrix, then they do not dependon the reserve and should be discounted in evalu-ating the merits of the reserve system.

In illustration, Humphreys and Kitchener (1982)classified vertebrate species into those dependenton reserves and those which persist in disturbedareas in the intervening agricultural landscape ofthe Western Australian wheat belt (Fig. 10.8). Theyfound that the small habitat islands, which weremostly edge habitat, were disproportionately richin ubiquitous species. Those species restricted towoodlots typically required a larger area (cf.Hinsley et al. 1994). This pattern was found sepa-rately for lizards, passerine birds, and mammals.By analysing all species within a taxon together,small reserves came out in a better light than theywarranted. Similar results are reported by Watsonet al. (2004a,b) in a study of birds in littoral forestsand surrounding habitats in south-eastern Madag-ascar. Core forest locations were found to be richerthan edge or matrix habitats, with some 68% of theforest dependent species found to be edge-sensitive.Frugivorous species and canopy insectivores(including six endemic vanga species) were gener-ally edge-sensitive, while in contrast, sallyinginsectivores were edge-preferring. At least part ofthe edge-sensitivity recorded was attributable tochanges in vegetation structure at the remnant edges.Consistent with this edge sensitivity, forestdependent species were generally lacking fromfragments of less than 10 ha, thereby demonstrat-ing a small-island effect (Chapter 4), masked inanalyses of all bird species richness (Fig. 10.9).Hansson (1998) used a rather different approach inhis studies of edge effects in ancient woodlands inSweden, but again found differences in bird com-munity composition that appeared to relate to habi-tat structure at the edges. He analysed the degree ofnestedness, finding that whereas census data forbirds from the centres of the patches were nested,the samples from the edge habitat were non-nested.

In particular circumstances, ecotones may havenegative implications for core woodland species(Wilcove et al. 1986), in that although they may sup-port additional species or larger populations ofnon-woodland core species, these populations mayhave negative interactions with deep-woodlandspecies. Studies in North America have suggestedthat the nesting success of songbirds is lower near

E D G E E F F E C T S 277

Passerinebirds

0

1.0

0.8

0.6

0.4

0.2

0 500 1000 1500 2000

Edge species

Edge species

Edge species

Core species

Core species

Edge-core species

Edge-core species

Edge-core species

0

1.0

0.8

0.6

0.4

0.2

0

1.0

0.8

0.6

0.4

0.2

0 500 1000 1500 2000

0 500 1000 1500 2000

Mammals

Lizards

Pro

por

tion

Pro

por

tion

Pro

por

tion

(a)

(b)

(c)

Area (ha)

Corespecies

Figure 10.8 The proportion of the core species (those dependenton the reserves) and the edge species (those which can also surviveoutside the reserves in disturbed habitat) which were present in aseries of 21 nature reserves in the Western Australian wheatbelt inrelation to the number of each present in a reserve of 2000 ha. Thecurves with negative slopes represent the difference between thecurves for edge and core species, and represent the relative excess ofedge species over core species compared with their distribution at2000 ha. Data for (a) mammals, (b) passerine birds, and (c) lizards.(Redrawn from Humphreys and Kitchener 1982, Fig. 1.)

Page 291: Whittaker. Island Biogeography. 2007

the forest edges than in the interior because ofhigher densities of nest predators—e.g. blue jay(Cyanocitta cristata), weasel (Mustela erminea), andracoon (Procyon lotor)—around forest edges.Wilcove et al. (1986) estimated that as a result,reserves of less than 100 ha will not support viablepopulations of forest songbirds. A few studies ofbirds and invertebrates have suggested that somespecies prefer edge habitats for breeding eventhough mortality rates at edges may be higher, aphenomenon termed an ecological trap (Ewers andDidham 2005). Paton (1994) has undertaken acritical review of studies of nest predation in edgehabitats. He concluded that, in general, there doesappear to be evidence for a depression of breedingsuccess, due either to enhanced predation orthrough parasitism. However, these forms of edgeeffect usually operate only within about 50 m of anedge, a narrower belt than some have claimed (e.g.Terborgh 1992). Nonetheless, such an effect impliesthat the effective reserve area may be less than thatdescribed by the perimeter of the isolate.

Edge effects may be taxon or even systemspecific. For instance, Burkey (1993) undertook anexperimental study of egg and seed predation withdistance from the edge of a patch of Belizian rainforest. Egg predation rates were higher in a 100 medge zone, but, conversely, seed predation rates

were found to be higher 500 m into the forest than30 m and 100 m from the edge. In contrast toBurkey’s egg predation data, Delgado et al. (2005)report higher predation pressure under closedcanopy than within edge habitats in their study ofartificial nest predation by ship rat (Rattus rattus) inlaurel forest on Tenerife. This illustrates that therelationship between a reserve and its surroundingmatrix is not subject to easy generalization. Thereare species that share both zones, and just as thereare matrix species which may impact negativelyupon core reserve species, there may also be reservespecies (dependent on it for breeding and cover)which exploit resources in the matrix. The conceptsof source and sink populations may again be usefulin this context, as the maintenance of maximaldiversity across a landscape depends on sufficientsource or core habitat for species of each majorhabitat. The identification of edge effects provides apart of such an analysis.

10.11 Landscape effects, isolation, andcorridors

The benefits of wildlife corridors

The premise of much of the literature discussed inthis chapter is that habitat connectivity is beneficialto long-term survival, as it enables gene flowwithin populations and metapopulations. Habitatconnectivity might be achieved by having steppingstones, or corridors, of suitable habitat linkinglarger reserves together. In practice, habitat corri-dors act as differential filters, enabling the move-ment of some species but being of little value toothers. As Spellerberg and Gaywood (1993) pointout, we are not merely concerned with forest penin-sulas or hedgerows. All sorts of linear landscapefeatures, such as rivers, roads, and railways, mayact as conduits for the movement of particularspecies. Equally, they may represent barriers orhazards for others (Reijnen et al. 1996). Harris (1984)points out that landscapes with great topographicrelief channel their kinetic energy via dendritic trib-utaries, main channels, and distributaries, and thatthese in turn influence the landscape template incharacteristic ways. The stream-order concept of

278 I S L A N D T H E O RY A N D C O N S E R VAT I O N

Area (Log, ha)0.1 1 10 100 1000

Spec

ies

Ric

hnes

s

0

10

20

30

40

50

60

70

R2 = 0.88

R2 = 0.92

R2 = 0.78

Figure 10.9 Bird species richness–area relationships for littoralforests, southeastern Madagascar. Closed circles (and bold, solid line):all bird species; open circles: forest dependent bird species, fitted bylinear regression (solid line), and by break-point regression (brokenline), demonstrating a small-island effect. (From Watson et al. 2004b.)

Page 292: Whittaker. Island Biogeography. 2007

the fluvial geomorphologist may be a means ofunderstanding species distributions within thelandscape (Fig. 10.10), and of understanding move-ments of species between disjunct habitat patches(Fig. 10.11). Studies of features such as hedgerowshave not always found them to be as effective inconnecting up woodlands as was hoped, with somespecies moving as well through surrounding fieldsand others simply not moving well along thehedges. However, habitat corridors may be impor-tant in particular cases.

A useful illustration of how corridors can be ben-eficial comes from the study by Saunders andHobbs (1989) of Carnaby’s cockatoo (Calyptorhyncusfunereus latirostrus) from the Western Australianwheat belt, an area of 140 000 km2 in the south-westof the state, 90% of which has been cleared foragriculture. The Carnaby’s cockatoo is one ofAustralia’s largest parrots and was once the mostwidely distributed cockatoo in the region. It breedsin hollows in eucalypt trees and eats seeds andinsect larvae from plants in the sand-plain heath. Itcongregates in flocks, both to nest and forage.Individuals may live 17 years and breeding status isnot attained until at least 4 years of age, each pairrearing a single offspring at a time. The widespreadclearance of the land has removed extensive areas ofthe vegetation type in which they feed, replacing itwith annual crops that are useless to the birds. Inrecent developments, wide verges of native vegeta-tion have been left uncleared along the roads. Theseact to channel the cockatoos to other areas wherefood is available.

Cockatoos have not survived in areas of earlierclearances, carried out without these connectingstrips, as once they have run out of a patch ofacceptable habitat it takes a long time for the flockto find another patch of native vegetation. Themore patchy the vegetation, the less successful thebirds are at supplying adequate food to rear theirnestlings. Furthermore, the narrow road verges ofthese early clearances result in a higher incidence ofroad deaths. The example illustrates that the degreeand nature of connectivity of different landscapeelements—in this case of the breeding and feedinghabitats, and of the vehicular hazards—are criticalto the survival of this species in these newly

fragmented landscapes. Given the relatively longlife cycle of these birds, and their flocking behav-iour, failure can have a sudden and wholesaleexpression in the bird’s disappearance. The bigreduction in area is fairly recent, and the cockatoo isnot yet in equilibrium with the new regime. It isstill on a downward path and is viewed as a speciesin danger.

The benefits of isolation

Although much of the literature is concerned withmaintaining population movements, there are alsocontexts in which isolation of populations of a targetspecies might be desirable (Simberloff et al. 1992),e.g. where there is a threat from a disease organism.Young (1994) evaluated 96 studies of naturaldie-offs of large mammal populations, defined ascases where numbers crashed by 25% of theindividuals or more. He noted that populations

L A N D S C A P E E F F E C T S, I S O L AT I O N, A N D C O R R I D O R S 279

Figure 10.10 Association of different-sized carnivorous mammalspecies with stream order and typical food particle size in accordancewith the stream-continuum concept. (Redrawn from Harris 1984,Fig. 9.8.).

Page 293: Whittaker. Island Biogeography. 2007

subject to large-scale phenomena such as droughtand severe winters may not be protected from suchdie-offs by population subdivision. But, where thecrashes are produced by disease epidemics, subdi-vision may be beneficial, and the creation of linkingcorridors, and translocation efforts, may be harmful.Young noted in his review that most herbivore die-offs were due to starvation, but carnivore die-offswere more often attributed to disease. An epidemicneed not eliminate all the individuals in a popula-tion for it to have a crucial role in causing the even-tual extinction of a population from a reserve.Combining these considerations with meta-population modelling, Hess (1996) argues thatthere should theoretically be an optimal degree of

movement within a metapopulation to enable themovement of propagules of the target species butnot of disease. But much will depend on the natureof the intervening landscape matrix and the extentto which it filters target and ‘pest’ species (Ricketts2001; Ewers and Didham 2005).

The spread of exotic competitors and predatorsand of fire into reserves can also be problematic(Spellerberg and Gaywood 1993; Lockwood andMoulton 1994). It has been suggested that alienplant invasion can be reduced by isolating reservesand surrounding them with simplified croplandoceans (Brothers and Spingarn 1992). Short andTurner (1994) found that the decline and extinctionof certain species of native Australian mammals

280 I S L A N D T H E O RY A N D C O N S E R VAT I O N

0 8

km

N

Cas

ca

de

sR

ang

e

Peak

Wilderness

Diamond

Figure 10.11 A possible spatial and size-frequency distribution of old-growth habitat islands arranged along riparian strips at progressivelygreater distances from a current wilderness area in the Willamette National Forest, USA, designed to provide the optimal degree of connectivitybetween patches in a system structured by the dendritic pattern of drainage. In this hypothetical system, the most distant islands are generallylarger in order to counter presumed lower dispersal. (Redrawn from Harris 1984, Fig. 10.2.)

Page 294: Whittaker. Island Biogeography. 2007

were most parsimoniously explained by referenceto the roles of introduced predators (foxes and cats)and herbivores (rabbits and stock). This explainedthe continued persistence of the native species onoffshore islands from which the exotic species wereabsent. In these cases, isolation is beneficial(as Janzen 1983) and many reserves set in ‘seas ofaltered habitat’ within mainland Australia haveproved to be insufficiently isolated to prevent theirinvasion by exotic pests and pathogens.

Corridors or isolation?

Is it possible to offer guidance about the situationswhen corridors may be beneficial and when not?First, as Dawson (1994) notes, all-purpose corridorsdo not exist, as each species has its own require-ments for habitat, its own ability to move, and itsown behaviour. Corridors, like other landscapecomponents, therefore act as filters. He points outthat many rare and threatened species are unlikelyto benefit from corridors, because the corridorwould have to contain their presumably rare habi-tat, i.e. rare species may require odd corridors.

Some corridors are essential in providing linksbetween preferred habitats for animals that under-take regular seasonal migrations, e.g. some fish,amphibians, reptiles, and for large mammals inboth the seasonal tropics and the Arctic. On theother hand, for some populations, corridors mayact as ‘sinks’, drawing out individuals from themain habitat area but not returning individuals tosupplement it, in which case they may do moreharm than good. In other cases, they may be fairlyneutral in their ecological cost–benefit, but perhapsbe quite expensive to purchase and set up if notalready existing in a landscape.

Simberloff et al. (1992) point out that numerouscorridor projects have been planned in the USA,potentially costing millions of dollars, despite thelack of data on which species might use the corri-dors and to what effect. Once again, the cost–-benefit equation for corridors may depend onproperties of the system involved, including scaleconsiderations and assessment of the major driversof species loss. In the longer term, it has beenargued that climate change is likely to drive sub-

stantial shifts in the distribution of species, and thatthe resulting species migrations will be impeded bythe human sequestration of land to agriculture andother purposes. Hence, on these grounds, it wouldseem to be prudent to plan more or less continuoushabitat corridors that straddle major climatic/elevational gradients where this is feasible (e.g.Bush 1996, 2002).

Reserve systems in the landscape

As we have seen, the debate about habitat corridorshas broadened from fairly simple theoretical begin-nings to a consideration of how a whole range ofdiffering natural, semi-natural, and artificialfeatures are configured within a landscape. Forinstance, countless thousands of birds are killedeach year through collision with motor vehiclesand with overhead power cables (Bevanger 1996;Reijnen et al. 1996). Research has shown that deathsthrough collisions with cables can be greatlyreduced by consideration of important flight pathsduring construction, by design features of thegantries, and by attaching a variety of objects to thecables to enable birds to sight them (Alonso et al.1994). Paved roads and tracks can play an impor-tant role as corridors favouring the introduction ofalien plants (Arévalo et al. 2005) or animals(Delgado et al. 2001) into otherwise well-preservedecosystems, but also creating corridors of suitablehabitat that can in cases allow endemic species (e.g.lizards) to disperse through inhospitable forestmatrix, enabling genetic exchange between other-wise isolated populations (Delgado et al. in press).On the other hand, wide roads and, especially,fenced highways, may act to impede the move-ments of large terrestrial animals, thus fragmentingpopulations and interfering with migration. Suchinformation needs to be integrated into improvedmanagement of whole landscapes (Spellerberg andGaywood 1993). In short, a landscape ecologicalframework is needed.

Landscape components include both habitatpatches and the matrix in which the patches areembedded. Modelling exercises typically favourclumped configurations of reserves, treating thematrix as a uniform ‘sea’. In practice, the optimal

L A N D S C A P E E F F E C T S, I S O L AT I O N, A N D C O R R I D O R S 281

Page 295: Whittaker. Island Biogeography. 2007

configuration must take account of the details ofthe landscape involved (e.g. Fahrig and Merriam1994). For instance, there may be greater opportu-nity for dispersal between two distant reserveslinked by a river and its adjacent riparian corridor,than between two similar but closer reserves sepa-rated by a mountain barrier of differing habitattype (Fig. 10.11). For butterflies, differences in landuse as subtle as switches from one woodland typeto another can significantly influence the passage ofbutterflies from one favoured habitat patch toanother (Ricketts 2001). An early attempt to placeisland theories into realistic landscape contexts wasprovided by Harris (1984). Recognizing that therewere limits to the amount of land that would be putover to reserve use, Harris advocated a series ofreserves placed within the landscape as dictated bygeographical features such as river and mountains,to encourage population flows between reserves.Most reserves would remain in forestry manage-ment. Within such reserves, there should be anundisturbed core of forest that is never cut, pro-viding habitat for species requiring undisturbed,old-growth conditions, around which commercialoperations might continue, following a pattern ofrotational partial felling. This would provide amosaic of patches of differing successional stage,maximizing habitat diversity, while maintainingincome (Fig. 10.12). His strategy aimed to satisfythe requirements of island theories, patch dynamicmodels, and economic realities. The implementa-tion of such a policy requires concerted action froma wide variety of agencies and is thus easier tosketch out than to bring to a realization.

Species that don’t stay put

In the Northern Territory of Australia, as in manyother systems across the globe, mobility and mas-sive population fluctuations in response to a highlyvariable climate are features of the wildlife.Woinarski et al. (1992) cite as a particular examplethe magpie goose (Anseranas semipalmata). Theynote that a conventional reserve network consistingof discrete national parks (such as they have devel-oped for plants in the region) will not cater for the

need of the magpie geese to relocate in response topatchiness of rainfall. They identify four possiblestrategies for conservation in the Northern Territory:

● The status quo, with improvements to ensure allvegetation types are included in the network. Thiswould not solve the problem, as it is not justvegetation types but the year-to-year variation incarrying capacity across the region that matters.

● Seeking inclusion of known habitat of importantspecies into the reserve system. For the magpiegeese, this might mean inclusion of all wetlands.However, it would be difficult to decide whichspecies to concentrate on and it would be impracti-cal to gazette all the land in question because ofcompeting claims for use.

● Developing very large reserves which spanextensive slices of the environmental gradient, justas does the present Kakadu National Park, a20 000 km2 reserve. However, this would requirevast reserves in the semi-arid and arid region, inwhich, at the present time, the largest reserve is of1325 km2.

● Supplementing the representative reserve net-work with measures that protect wildlife habitat on

282 I S L A N D T H E O RY A N D C O N S E R VAT I O N

Figure 10.12 Harris’s (1984) proposed system for the managementof forest habitat islands on a long-rotation system, such that differentsectors are cut in a programmed sequence, but the core old-growthpatch is left uncut. This recognizes the importance of patch dynamics tohabitat diversity and how such management may serve both economicand conservation goals. (Redrawn from Harris 1984, Fig. 9.5.).

Page 296: Whittaker. Island Biogeography. 2007

large areas of unreserved land. They see this as theonly viable answer in the long term, requiring abroad educational and political strategy to get there(Woinarski et al. 1992; Price et al. 1995).

So, wildlife conservation must in part be con-cerned with the practicalities of conserving outsidereserves. The Northern Territory example mayseem an extreme case when viewed from manyparts of the world, yet the argument surely appliesto some degree everywhere.

10.12 Does conservation biology needisland theory?

Zimmerman and Bierregaard (1986) undertook ananalysis of Amazonian frogs which demonstratedthat knowledge of the autecological requirementsof the species would be critical to planning areserve system (not that a system would necessar-ily be designed just for frogs). They undertook aprediction by a species–area approach and showedthat it lacked relevance in the light of the availableautecological data. Frog species differ in the habitatrequired for breeding: whereas some are strictlystreamside breeders, others require landlocked per-manently or periodically flooded terra firma pools.The authors concluded that if such data are avail-able for enough species in the system being consid-ered, then there need be no recourse to abstractprinciples. If they are not available, it is still prefer-able to conduct a rapid survey and collect biogeo-graphic and ecological data. This is just one ofmany studies that have championed empirical overtheoretical ‘island’ approaches. Does conservationbiology need island theories?

A non-equilibrium world?

Worthen (1996) bemoans the fact that the bias towardsspecies–area relationships over compositional–arearelationships continues to be a feature of the litera-ture, including some of the most widely used intro-ductory ecology texts. This bias continues despitethe demonstration, in numerous studies, of nested-ness, community similarity of habitat fragments,the relevance of successional patterns, and a variety

of other forms of compositional structure. Thesenon-random patterns in how communities assem-ble, or alternatively how they change in response tofragmentation, demand that we reappraise the con-tribution of ‘island ecology’ within conservationbiology.

Much of the earlier applied island literatureseemed to assume that matrix contexts wererelatively insignificant and that stochastic decaytowards equilibrium would restore a new, depau-perate balance. These were arguably useful simpli-fying assumptions, but the behaviour of habitatislands is perhaps more realistically capturedwithin the notions of flux of nature described byPickett et al. (1992). This framework stresses physi-cal as well as biotic processes (e.g. succession), therole of episodic events (e.g. fires or hurricanes), andthe differing degrees of connectivity of populationsacross ‘filtering’ landscapes. We have also seen thatthe nature of the matrix can be crucial to what goeson in the habitat islands within the landscape. Howmuch of this is ‘island biogeography’ is a mootpoint, but insularity remains a key property andissue within these other topics.

The application of equilibrium (EMIB) assump-tions to reserve systems has been neatly critiquedby Saunders et al. (1991, p. 23):

It is commonly assumed that at some stage the remnantwill re-equilibrate with the surrounding landscape. It ishowever, questionable whether a new stable equilibriumwill be reached since the equilibration process is liable tobe disrupted by changing fluxes from the surroundingmatrix, disturbances, and influx of new invasive species.The final equilibrium can be likened to an idealized endpoint that is never likely to be reached, in much the samefashion as the climatic climax is now conceptualized insuccession theory. Management of remnant areas willthus be an adaptive process directed at minimizing poten-tial future species losses.

The general premise of much of the appliedisland biogeography literature is that (1) the alarm-ing fragmentation of many ‘natural’ systems willcontinue and therefore (2) if much wildlife is to sur-vive, it will have to do so in reserve systems, andthat (3) we have very little time in which to act.These concerns explain the initial attractiveness of

D O E S C O N S E R VAT I O N B I O L O G Y N E E D I S L A N D T H E O RY ? 283

Page 297: Whittaker. Island Biogeography. 2007

deriving general principles for reserve systemsfrom island theory. Increasingly, the resulting linesof reasoning have been attacked on the grounds:first, that many valid generalizations are self-obvi-ous platitudes; secondly, that models are often bothtoo simplistic and effectively untestable; and,thirdly, that many theoretical generalizations haveachieved the status of dogma (Doak and Mills1994). Yet, the general principles derived fromisland theory have influenced conservation policyand they remain pervasive. Because of this, it isimportant to understand both the strengths and,particularly, the weakness of the ‘island’ theoriesdiscussed in this and the previous chapters.

If our focus is on global losses of species diver-sity, then clearly the loss of—and fragmentationof—species rich habitats is a key concern. But, thestochastic consequences of reduced populationsizes are not necessarily simple to understand andmodel. Some species can persist well in fragmentedlandscapes providing that the areas of remaininghabitat are not too small and isolated. Much maydepend on the previous history of the region(extent of initial departure from equilibriumassumptions), the taxa considered, and the scalesensitivity of the systems, and how the habitatmatrix (the land outside the fragments) is managed.Simple species–area models, particularly as appliedon regional scales, may not in the end provide reli-able estimates of how many species really are ‘com-mitted to extinction’, i.e. still around but headingfor an inevitable nemesis. Some would regard thisassessment as complacent and over-optimistic. Forinstance, Pimm and Askins (1995) argue that con-cerns about the precise causal mechanisms ofextinctions, as expressed here, miss the point andthat, in practice, simple predictions of species lossesbased on species–area relationships are actuallyfairly reliable. From their analyses of North Americanforest birds they suggest that if such estimates err,they tend to be conservative, i.e. predicting fewerextinctions than actually occur. We are not con-vinced that their claim can be generalized, as somuch seems to depend on issues such as systemconfiguration and the actual drivers of speciesthreat and extinction (see discussion of the Atlanticforest analyses, above).

The problem of biodiversity loss (includingpopulation-level decline, genetic erosion and specieslosses) is a depressingly real one and the pace ofspecies extinctions is undoubtedly accelerating(Whitmore and Sayer 1992; Turner 1996). Yet, asSimberloff et al. (1992) observe, ideas derived fromisland theory, and the so-called SLOSS principles(specifically the benefits of corridors), becameembedded in policies of bodies such as the IUCNand World Wildlife Fund without having been val-idated. Thus, policy-makers and legislators have incases acted upon such uncertain scientific ‘princi-ples’. Once released from the bounds of scientificjournals, scientific information is often poorlyunderstood. It is therefore important not to offer upgeneral principles and rules where the evidence forthem is equivocal and, in cases, contradictory(Shrader-Frechette and McCoy 1993).

Nonetheless, we should not simply discard thetheoretical frameworks of island theory, metapop-ulation theory, etc. They have their relevance toparticular systems and a wider and important rolein helping us to understand the sorts of effects thatmight be operative in real-world contexts, and thattherefore should be tested for or monitored. AsDoak and Mills (1994) have put it, we should faceup to the problems rather than use them as anexcuse for ignoring the theoretical debates: ‘it isincumbent on us to teach such complexity tomanagers and nonbiologists, rather than attem-pting to snow them with undefendable over-generalizations’.

The notion that species disassembly (as well asassembly) is fairly structured and predictable, atleast in terms of functional ecological types orguilds, is one that has reappeared in several guisesin this chapter (e.g. from studies of relaxation,experimental fragmentation, incidence functions,nestedness, and edge effects). In the case of theManaus habitat fragmentation study, similar struc-turing was evident both in the initial losses (disas-sembly) and, in particular circumstances, in therecovery of bird communities that followed. Thisreminds us, first, that habitat change does not haveto be a one-way process, and, secondly, that itmight be possible to target conservation effortsfairly directly on to particular sets of vulnerable

284 I S L A N D T H E O RY A N D C O N S E R VAT I O N

Page 298: Whittaker. Island Biogeography. 2007

species. In this sense, ‘island’ approaches have pro-duced some dividends. It is now a question oflearning what they are, and differentiating betweenthose effects that are weak and those that are strong(the world is neither wholly non-equilibrial norwholly equilibrial). It seems that we can work outwhich sorts of species are most threatened andsome of the major reasons why they are threatened,and it should therefore be possible to offer somemanagement solutions. Some of these insights haveindeed been derived from ‘island’ studies.

For many species, reserves are essential. Veryoften this is because of specific threats posed bypeople (or our commensals) and the protection thatcan be afforded within the reserve. Simberloff(1992) comes down in general on the side of large,continuous blocks of habitat. He argues that noexisting theory adequately describes the jointeffects of loss of area and fragmentation, but thatthere are empirical observations, such as those todo with predation discussed earlier, to suggest thatfor many species the combined impact is indeedstrongly deleterious: ‘Probably any species that hasevolved in large, relatively continuous habitat hastraits that are maladaptive in small, isolated frag-ments’ (p. 85). Beyond this generalization, themeans, rate, and extent to which extinction is deliv-ered, are likely to be idiosyncratic. The theoreticalbasis for calculating likely regional and globalextinctions as a result of the rapid phase of habitatloss currently under way remains highly uncertain.For a careful review of the evidence and basis forcalculations see Whitmore and Sayer (1992).

Ecological hierarchies and fragmentedlandscapes

As with other branches of island theory, approachesto the debates on reserve configuration that takeno account of hierarchical interactions betweenorganisms of differing trophic levels are incomplete(Terborgh 1992). For many animal groups, plants(collectively or individually) determine the abilityof particular animal species to occupy a habitatisland. Successional changes in plant communitiesmay often explain turnover in bird or butterflycommunities (Chapter 5; Thomas 1994). Equally

important may be the activities of animals asdispersers and pollinators of plants (and as preda-tors), both in contexts of forest maintenance and ofrecolonization of disturbed areas. Different types ofanimals disperse different, albeit overlapping,suites of plant species. Studies of cleared areas inthe Neotropics have suggested that bats may bemore significant in seeding open habitat than areday-flying birds. The efficacy of birds, in particular,may also be positively influenced by the availabil-ity of some woody cover in an otherwise open area(Estrada et al. 1993; Gorchov et al. 1993; Guevaraand Laborde 1993). Thus, for instance, cheap plant-ings of small clumps of plants may encouragespeedy return of forest around the inocula.

The Krakatau recolonization data demonstratethe significance of birds and bats in transportingseeds between sites, and thus also supports theirsignificance for population interchange betweenforest patches in the tropics. Maintenance andrestoration of forest might thus necessitate protec-tion for vertebrate species, if not in their own right,for their interactions with plants. It is not ‘merely’biodiversity which is at stake, as in cases there maybe huge economic costs to the loss of pollinatorspecies on which economically useful plantsdepend (Pannell 1989; Fujita and Tuttle 1991; Coxet al. 1992; Ricketts 2004). At the same time, theKrakatau study demonstrated that particulargroups of plants, those dispersed by non-volantfrugivores and the large-seeded, wind-spreadspecies, are likely to be greatly hampered by patchisolation (Whittaker and Jones 1994a,b; Whittakeret al. 1997). These topics have recently seen a resur-gence of research interest (Ewers and Didham2005), with studies in both Africa and theNeotropics demonstrating the negative impacts onfrugivore-mediated plant dispersal of forest frag-mentation (e.g. Cordeiro and Howe 2001; Martínez-Garza and Howe 2003).

It seems likely that some area effects involvethreshold responses and that these operate via hier-archical links within ecosystems (Soulé et al. 1988;Terborgh 1992). For instance, below a certain size,an isolated habitat island may be incapable of sup-porting populations of large predatory vertebrates,with the consequence that higher densities of their

D O E S C O N S E R VAT I O N B I O L O G Y N E E D I S L A N D T H E O RY ? 285

Page 299: Whittaker. Island Biogeography. 2007

prey species may be maintained, with potentiallysignificant consequences further down the foodchain. Thus, the densities of medium-sized terres-trial mammals are between 8 and 20 times greateron Barro Colorado Island, which lacks top preda-tors, than in a comparable ‘mainland’ site in whichthey occur. Terborgh (1992) suggests that this maybe having important selective impacts on plantregeneration. On some smaller islets nearby in LakeGatun, not only are the (true) predators absent, butso too are those smaller mammals that act as seed-predators, the squirrels, peccaries, agoutis, andpacas which have such high densities on BarroColorado island. These small islets became domi-nated in the 70 or so years since their isolation bylarge-seeded tree species. This is plausibly becausethese plants no longer suffer high rates of attrition atthe point of germination and establishment, andhave therefore been able to out-compete the smaller-seeded species to a greater extent than usual.

Predators undoubtedly have a crucial role to playin fragmented landscapes, not just in terms ofpredation within the habitat island, but also withinthe matrix. Simberloff (1992) notes that in theManaus fragmentation study the bat falcon (Falcorufigularis) has repeatedly been seen to chase birdsflying over cleared areas from fragment to fragment.It may therefore be unsurprising that most under-storey birds will not willingly cross gaps of even80 m. Similarly, in the USA, the northern spottedowl (Strix occidentalis caurina), of metapopulationtheory fame, previously was a species which rarelyleft closed forest. In the newly fragmented land-scapes, as many as 80% of yearling males die,apparently from predation by great horned owls(Bubo virginianus) and goshawks (Accipiter gentilis)as they disperse over cleared areas. This was oneconsideration that persuaded the interagencyfederal committee responsible for the managementplan to shift their strategy towards larger, more con-tinuous blocks. This brief consideration of predationhelps us recall the point that the particular mecha-nisms responsible for failures to sustain populationsor to recolonize habitat islands may thus be poorlypredicted by ‘species–area’ type-approaches, or forthat matter by metapopulation models.

There may also be important dynamic features,such as the spatial and temporal oscillations thatcharacterize elephant populations in the TsavoNational Park in Kenya, on a cycle of roughly50 years. Even with very large reserves, migratoryanimals may ignore park boundaries and, espe-cially in lean years, may move into areas outside,where they are most unwelcome and perhaps alsounsafe. These patterns can provide big manage-ment problems, even when a very large area isenclosed as a reserve (Woinarski et al. 1992). Theelephants illustrate another idea: that of thekeystone species, meaning a species that is socritical to the functional character of an ecosystemthat its removal would cause a chain of alterations(e.g. Soulé 1986; Pimm 1991). Elephants have a keyrole as grazers. Without them successional changesoccur in the vegetation, altering the carrying capacity of the system for other species. Highdensities of elephant, on the other hand, can leadto overgrazing and widespread destruction oftrees. In such cases, conservation managementnaturally focuses around managing the keystonespecies.

Climate change and reserve systems

Range shifts driven by significant climatic changehave been a feature of the Quaternary period. Thereis no reason to consider these climatic fluctuationsto be at an end, and there is also mounting concernthat humans have initiated a phase of rapid climaticwarming. Researchers are now focusing on theabilities of plant and animal species to tracktheir required climatic envelope across the nowfragmented landscapes (Huntley et al. 1995). Keyfactors are the speed of climatic change, thedispersal abilities of the target species, andthe characteristics of the landscapes across whichspecies may have to move, which may range fromthe optimal to the benign to the hostile. Grapplingwith such scenarios requires a variety of researchtools (Bush 1996, 2002; Pearson and Dawson 2003;Araújo et al. 2005b), amongst which are the tools ofisland biogeography reviewed in these pages(Boggs and Murphy 1997).

286 I S L A N D T H E O RY A N D C O N S E R VAT I O N

Page 300: Whittaker. Island Biogeography. 2007

10.13 Concluding remarks: fromisland biogeography to countrysidebiogeography?

As made clear at the outset of this chapter, thisreview was not intended as a summary of all impor-tant branches of wildlife conservation. However, wehave stressed along the way the importance of aute-cological, distributional, and other forms of data.The student of conservation biology will also needto have a knowledge of such themes and issues asex situ conservation efforts in zoos, and other dedi-cated breeding facilities; germplasm (seed) banks;species translocation schemes; reserve managementpractices; CITES and other wildlife legislation;poaching and trade in endangered species; strategicconservation policy; the philosophy, politics, andeconomics of conservation and other fields besides(e.g. Primack 1993; Shrader-Frechette and McCoy1993; Whittaker et al. 2005).

It must also be recognized that there can be manyaims and purposes for shaping conservationmanagement, ranging from the aesthetic, throughthe scientific, to the economic. We may wish to con-serve systems or species which are ‘representative’,‘typical’, rare, speciose, nice to look at, of recre-ational value, or provide economic return (Ratcliffe1977; Shrader-Frechette and McCoy 1993). Suchmultiplicities of purpose require requisite tools; theisland theories have their place in the tool kit, butthey should not always be the first to be reached for.

Allied to the habitat remnant/reserve strategy,wherever possible, conservationists should arguefor greater priority to extensive rather than intensiveconservation, i.e. for environmental managementpolicies to encourage the survival (or passage) ofmany species outside of the more closely protectedreserves systems (cf. Rosenzweig 2003; Daily et al.2005). In short, to work to prevent habitat islandsand reserves becoming more and more like realislands—except in those biogeographical contextswhere insuralization is actually beneficial tosurvival prospects. We have seen in this chapterthat issues such as size, shape, and configurationwithin a landscape are important to reserve suc-cess, not just in terms of how many species will be

held within a reserve, but also which sets of species.The number of species held in a reserve (or reservesystem) is actually less important than to conservethose species which cannot survive outside theremnants (e.g. Newmark 1991).

Some recent efforts have been made to movebeyond an exclusive focus on (forest) fragmentstowards understanding the role of such habitatislands within mixed-use landscapes. This switch inemphasis comes under varying headers. Forexample, Watson et al. (2005) show that the incidencefunctions of woodland bird species in three differentlandscapes in the Canberra area (Australia) differsignificantly, seemingly as a function of differencesin properties of the landscape matrix within whichthe woodlands are embedded. Hence, Watson et al.(2005; and see Box 5.2) join others (e.g. Gasconet al. 1999; Cook et al. 2002, Ewers and Didham 2005)in calling for greater attention to ‘matrix effects’.Hughes et al. (2002) adopted a slightly differentapproach to their study in southern Costa Rica,focusing on the extent to which native forest speciesmake use of the surrounding countryside: theyfound that some 46% of bird species foraged oftenkilometres away from extensive areas of nativeforest. Although they stress that not all species canbe so readily accommodated outside large tracts ofnative forests, their work supports the importanceof developing ‘countryside’ landscapes that arebiodiversity-friendly and penetrable by nativefauna (as Harris 1994). Daily and colleagues (e.g.Daily et al. 2001, 2005; Hughes et al. 2002) coin theterm ‘countryside biogeography’ for this switch inattention from remnants per se, to the way in whichremnants function within whole landscapes. Thisswitch in emphasis is similar to that promoted byRosenzweig (2003) under the heading ‘reconciliationecology’. Whether we label this shift in emphasis‘matrix effects’, ‘countryside biogeography’, or‘reconciliation ecology’, the common element is arealization that effective conservation must includeconsideration of what happens outside reserves, asthe way we shape the countryside, whether we farmintensively or extensively, whether we retainhedgerows and trees within mixed landscapes, canhave profound implications for regional diversity

I S L A N D B I O G E O G R A P H Y TO C O U N T RYS I D E B I O G E O G R A P H Y ? 287

Page 301: Whittaker. Island Biogeography. 2007

and for abundances of wildlife (e.g. see Gascon et al.1999; Gates and Donald 2000).

Conservation requires pragmatic decision-making. As we continue to fragment landscapes,island effects may inform such decision-making,but should not be oversimplified. There is no singlemessage, and no single island effect; indeed,insularity may, in at least a minority of cases, bringpositive as well as negative effects (Lockwood andMoulton 1994). Island effects may be weak or strong.The implications of insularity vary, depending onsuch factors as the type(s) of organism involved, thetype of landscapes involved, the nature of theenvironmental dynamics, the biogeographicalsetting, and the nature of human use and involve-ment in the system being fragmented. It is unfortu-nate that the term ‘island biogeography theory’ hasbecome largely synonymous in a conservationsetting with a limited conception of island ecology,stressing the inevitability of stochastically driventrends to equilibrium. Both pure and applied islandbiogeography are richer than this. This richness ofideas and information needs to be understood andintegrated into teaching for both pure and appliedpurposes. Moreover, there is also a need for arenewed research effort within this area of conser-vation biogeography, in the search for improvedscientific guidance.

10.14 Summary

The island analogy can be extended to patches ofhabitats within continents. The conversion of more-or-less wild habitats to other uses is fragmenting,isolating, and reducing ‘wild’ areas across much ofthe globe. The implications of this insularization areexamined in this chapter from the scale of individ-ual populations up to whole landscapes.

The minimum viable population (MVP) is thesmallest number of individuals required to ensurelong-term population persistence. We do not knowhow big MVPs should be. We do know that MVPsvary from species to species, and that the effectivepopulation size is generally smaller than the actualpopulation size, i.e. not all individuals, or all adults,are involved in breeding. Population loss may bedue to stochastic demographic and/or genetic

effects, or to environmental disturbance and change.Stochastic demographic effects may have been over-estimated. Genetic effects are potentially complex,and vary between taxa. The significance of environ-mental change or catastrophe is probably crucial topopulation viability, but can be difficult to model.MVPs therefore need to be established separately fordifferent types of species, and management regimesmust be responsive to changing circumstances.

The area required by a MVP is termed the mini-mum viable area, and may be estimated fromknowledge of home range size, although thisapproach assumes no population exchange withother isolates. Another approach to area (or habitat,or isolation) requirements is to use incidence func-tions, although the patterns revealed may be con-founded by multicausality, and may fail to predictchanges that follow within fragmented landscapes.Moreover, recent work suggests that area require-ments for species presence in fragments may varysignificantly even within the same region.

Where geographically separated groups areinterconnected by patterns of extinction and recolo-nization they constitute a population of popula-tions, or a metapopulation. Here, the idea is thatparticular patches have their own internal popula-tion dynamics, but when they crash to local extinc-tion they are repopulated from another patchwithin the metapopulation. Such a scenario altersthe projections of population viability considerably.In many cases, however, habitat patches appear todescribe a source–sink rather than a mutual sup-port system. In such circumstances, a large patchacts as an effectively permanent population, withsmaller sink habitats around it being re-suppliedfrom the source population. The sink habitats mayhave little relevance to overall persistence-time ofthe metapopulation.

A heated debate developed around the SLOSS‘principles’, which represented an attempt toanswer the question of whether it was better toadvocate Single Large or Several Small reserves ofthe same overall area, based on the assumptions ofthe EMIB. Unfortunately, island theory provides noresolution to this debate. In practice, the optimalreserve configuration for one type of organism,landscape, or scale of study system, may not be so

288 I S L A N D T H E O RY A N D C O N S E R VAT I O N

Page 302: Whittaker. Island Biogeography. 2007

for another. Physical environmental change in bothfragment and ‘matrix’ habitats has not generallybeen given enough attention in the theoretical liter-ature. Successional change and altered disturbanceregimes may have long-term implications forspecies persistence in altered landscapes. The evi-dence for ‘relaxation’, i.e. stochastically drivendecline in species number, is surprisingly thin.Species composition and richness are liable tochange in fragmented systems: species are indeedlost. However, in relation both to SLOSS andmetapopulation scenarios, it is apparent that muchturnover may well be deterministic in nature andexplicable in relation to specific ecological processessuch as succession or meso-predator release.Various lines of data suggest strong structure to thedisassembly as well as to the assembly of systems.For instance, many systems and species arestrongly nested in their distributions, and this can,it seems, be produced in either colonization- orextinction-dominated systems.

It appears possible to predict the types of speciesthat will be in greatest difficulty in fragmentedlandscapes, and it is these species that requireprimacy in relation to reserve configuration, popu-lation monitoring, and management planning. Theconsequences of fragmentation can be severe, butthey are poorly predicted by classical islandtheories, which assume stochastic demographiceffects to dominate in a system of fragments search-ing for a new equilibrium in a sea of altered habi-tats. The nature of the connectivity of habitatpatches may be crucial. In some cases, corridorsand short dispersal distances may be beneficial,whereas in other circumstances isolated reservesmay actually provide the best chance of survival fora threatened species. Some species thrive in edgehabitats, others are disadvantaged by increasededge effects. The ‘filtering’ properties of the matrixcan be crucial and require careful assessment inconservation planning. Landscape effects may becomplex, with the same linear feature acting ascorridor, barrier, and lethal hazard for differentorganisms. Moreover, some species are mobile, sea-sonally or aseasonally, and their protection may

require a more extensive approach to conservationthan provided for by any reserve system on its own.

These considerations demand a re-evaluation ofthe place of island biogeography within conserva-tion biology. The equilibrium island theory turnsout to provide rather equivocal guidance, andappears to have often been invoked as a basis forpredictions of species loss where there is a logicaldisconnection between theory and prediction.Some of the effects traditionally associated withisland approaches may actually be fairly weak ingeneral, but other features of insularity may pro-vide powerful and fairly immediate influences onspecies persistence and system change. The role ofhierarchical ecological relationships, e.g. betweenplants and their dispersers, between vegetationchange and altering habitat space for butterflies,between understorey insectivores and theirpredators, and so forth, may be crucial to theconsequences of fragmentation. Such successionaland food-web controls may be the most powerfulforms of expression of islandness in the alteredlandscapes, and they typically involve interactionsand exchanges between the fragments and thesurrounding matrix. The attention being paid to theeffects of the matrix on persistence in remnants, andto the use of matrix habitats by native species,signals an on-going switch in emphasis fromisland thinking to what some term ‘countrysidebiogeography’, i.e. to the importance of incorporat-ing conservation concerns in managing whole land-scapes, not just isolated reserves.

A line must be drawn somewhere in any text, andin this chapter discussion stops short on alternativeapproaches to conservation problems, and onthemes such as the implication of large-scale cli-mate change under global warming scenarios.Neither is any attempt made to draw out guidingprinciples for conservation from island studies. Thegrand hopes for simple unifying principles haveproven to be as illusory as the end of the rainbow.It is nonetheless important to appreciate how theseideas have influenced conservation theory, just as itis important to tackle the ecological complexities ofincreasing insularity within real landscapes.

S U M M A RY 289

Page 303: Whittaker. Island Biogeography. 2007

We cannot discuss the ecology of islands without makinga few disparaging comments on goats. These creaturesmust be the true embodiment of the devil for a plant lover.

(Koopowitz and Kaye 1990, p. 72)

The biodiversity crisis is nowhere more apparent and inneed of urgent attention than on islands. Approximately90% of all bird extinctions during historic times haveoccurred on islands . . .

(Paulay 1994, p. 139)

11.1 Current extinctions in context

Throughout our discussions of island ecology andevolution, we have mostly set to one side the sig-nificance of humans as shapers of island biogeogra-phy. As this chapter will show, this is a questionablepractice, as human societies have had a profoundinfluence on island biodiversity in both recent his-torical and prehistorical time (Morgan and Woods1986; Johnson and Stattersfield 1990; Groombridge1992; Pimm et al. 1995). This influence has beenmanifested in many ways, but the most profoundmust surely be the extinction of numerous islandraces and species.

Of all the species that have ever lived on theEarth, only a small fraction are alive today. Naturalcatastrophes, such as volcanic eruptions, meteoriteimpacts, climate changes, marine transgressionsand regressions, in combination with biotic forcing,have been key agents of species extinction in thepast. Extinction rates have varied through time,with five phases recognized as particularly intenseand widespread events in the fossil record, the so

called mass-extinctions events. The most recent ofthese was the end-Cretaceous event, which resultedfrom a bolide impact approximately 65 Ma(Groombridge and Jenkins 2002). Whether the com-parison is entirely apposite is a moot point, butmany argue that a sixth mass extinction event isnow under way, which uniquely is being driven bythe activities of a single species.

Primack and Ros (2002) suggest that more than99% of recent species extinctions can be attributedto human activities. Extinctions of vertebratespecies over the last 400 years average 20–25 percentury, a rate between 20 to 200 times the back-ground extinction rate (i.e. the natural rate exclud-ing mass extinction events) (Groombridge andJenkins 2002). This anthropogenic episode begansome 100 000 years ago with the expansion of mod-ern humans from Africa throughout the world, andhas been particularly intense in Australia andNorth and South America. However, many of theless recent continental extinctions occurred concur-rent with episodes of pronounced climate change,and the extent to which humans caused many ofthe past losses has been much debated (a well-balanced evaluation is provided by Lomolino et al.2005). During this anthropogenic period, some ofthe most dramatic and convincing illustrations ofextinctions due to human action come from islandsworldwide (Diamond 2005).

Estimating the extent of the biodiversity crisis isa hard task, because we still have a remarkablypoor knowledge of the total diversity of the planet(the so-called Linnean shortfall), and of how thatdiversity is distributed (the Wallacean shortfall).

290

CHAPTER 11

Anthropogenic losses and threats to island ecosystems

Page 304: Whittaker. Island Biogeography. 2007

S TO C H A S T I C V E R S U S D E T E R M I N I S T I C E X T I N C T I O N S 291

Although approximately 1.7 million species havebeen described, typical estimates of the total diver-sity of the planet range from 5 million to 30 million,with figures as high as 100 million occasionallybeing cited. Many of the supposed undescribedspecies are thought to occur amongst invertebratetaxa in tropical forests, and ocean floors, althoughhuge uncertainty remains as to how diverse thesesystems really are (see, e.g. Lambshead andBoucher 2003).

Islands, too, are part of this uncertainty, undoubt-edly holding many species currently scientificallyunknown that are waiting to be discovered anddescribed. For instance, the cataloguing of theCanarian biota has been under way since theFrench priest Louis Feuillee came to the archipel-ago in 1724. Despite almost three centuries of scien-tific attention, species new to science continue to bediscovered, in recent decades at a rate approximat-ing one species every 6 days (Fig. 11.1), includingtwo large lizards (Gallotia intermedia and G. gomer-ana) and two trees (Myrica rivas-martinezii andDracaena tamaranae) (Izquierdo et al. 2004). Manynew finds have escaped detection until recently

because they persist only in small populations.Hence, no sooner are they discovered than manyare categorized as in danger of extinction.

11.2 Stochastic versus deterministicextinctions

Two main types of extinction events operating onislands can be differentiated (Table 11.1): largelystochastic extinctions, inherent to the naturaldynamics of island environments, and determinis-tic extinctions, related directly or indirectly tohuman activity. In the first group we include natu-ral disasters, such as volcanic eruptions, but alsothe consequences of the taxon cycle dynamics,which involve species moving down an evolution-ary cul-de-sac in becoming highly restricted localendemics, increasingly prone to natural disasters.The second group includes habitat-loss, degrada-tion or fragmentation, the introduction of alienspecies and predation by humans, whether for thepot or for other purposes.

When one species goes extinct, any evolutionaryinteractions involving that species must also cease.

1838 Brullé(59 species ofvarious insect

orders)

1839Brullé (14

lepidoptera),Mcquart

(16 diptera)

1852Shuttleworth(27 molluscs)

1862, Wollaston(93 coleoptera)

1864, Wollaston(224 coleoptera)

1865, Wollaston(59 coleoptera)

1865, Mousson(33 molluscs)

1936, Frey(23 diptera)

1908, Walshingham(42 lepidoptera) 1954, Lindberg

(65 hemiptera)

1987, Wunderlich(90 araneids)

1992, Wunderlich(94 araneids)

1833 Webb & Berthelot

(16 molluscs)

1792 1810 1820 1830 1840 1850 1860 1870 1880 1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 20000

500

1,000

1,500

2,000

2,500

3,500

3,000

250

225

200

175

150

125

100

75

50

25

0

Years

Spec

ies

Cum

ulat

ive

Spec

ies

Num

ber

Figure 11.1 The historical pattern of description of endemic invertebrates of the Canaries. The bars represent the number of species describedeach year, and the line the cumulative total. (After Izquierdo et al. 2001.)

Page 305: Whittaker. Island Biogeography. 2007

292 A N T H R O P O G E N I C L O S S E S A N D T H R E AT S TO I S L A N D E C O S YS T E M S

If this involves very tight mutualisms (e.g. as canbe involved in pollination services), then the sur-viving co-evolutionary partner, or ‘widow’ species(sensu Olesen and Valido 2004) may in time followthe same fate, unless it is able to find similar serv-ices from another locally available species. Suchlosses may then multiply, as other mutualisms areinterrupted, in a fashion analogous to the meso-predator release mechanism discussed in the pre-vious chapter. Such trophic cascades may followfrom both deterministic and stochastic extinctions.This term implies a cascading effect across trophicboundaries as an ecosystem unravels. There aresome spectacular examples driven by ‘ecosystemtransformer’ species (e.g. Rodda et al. 1999;O’Dowd et al. 2003), but it is by no means clearhow often and how far such cascading effects oper-ate. Many ecosystems in practice have a degree ofecological redundancy (Olesen and Jordano 2002),such that a cascading ‘ecological meltdown’ maybe relatively uncommon.

11.3 The scale of island losses globally

Species populations go extinct locally all the time.Of itself, a population extinction event is not neces-sarily of conservation concern. It becomes a matterof concern when it marks a pattern of range col-lapse, and especially when it represents the loss ofan endemic subspecies or species. In discussingisland species losses, we can distinguish differentlevels of extinction by the terms local species extinc-tion or island extirpation, as distinct from the finalextinction of the last population of a species, i.e.when an island endemic form is entirely eliminated.We focus on the latter unless otherwise denoted.

Those collating records of species extinctionscommonly take the historic period of island explo-ration and scientific record to have begun c.AD 1600.In the period so defined, significantly more speciesof plants and animals are known to have becomeextinct from islands than from continents: for thetaxa listed in Table 11.2, about 60% have been islandspecies. However, we know so little about thelosses of invertebrates that it would probably besafer to discount them from the analysis. For thegroups with the best data, the proportion of extinc-tions from islands varies from 60% for mammals(which are generally lacking from remote islands inthe first place), to 79% for molluscs, 81% for birds,and as high as 95% for reptiles. Moreover, of 121continental species losses cited by Groombridge(1992), about 66% can be classified as aquaticspecies, many of them inhabiting lakes, which canbe regarded as ‘negative islands’ (Box 11.1).

The values in Table 11.2 reflect a highly cautiousapproach to counting extinctions, and these num-bers are predicted to change radically over the com-ing decades as the various ‘extinction debts’(Chapter 10) are called in the world over. To date,therefore, relatively small numbers of recent terres-trial animal extinctions have been recorded frommainland tropical forest regions such as Amazoniaand the Atlantic forests of Brazil, although of coursethese are the areas from which large numbers ofextinctions are estimated to be occurring or about tooccur. Yet, when we place the figures for islandextinctions and threat into the context of how manyspecies originate from islands, it is clear that theyhave suffered disproportionate losses in recent timeand also that they remain priority sites for conserva-tion. In the case of birds, and allowing for the

Table 11.1 Largely stochastic, natural causes of extinction, versus deterministic causes arising from human activities on islands

Stochastic causes (inherent to islands) Deterministic causes (related to human activities)

Natural disasters (volcanic eruptions, landslides, hurricanes, Habitat loss, degradation, or fragmentationtsunamis)

Taxon cycle dynamics (including demographic collapse Alien species introduction (including competitors, predators,and inbreeding depression) hybridizing congeners, parasites, disease vectors, or diseases)

Pleistocene climatic fluctuations (including sea-level rise) Direct predation (including hunting, fishing, or specimen collection)—— Trophic cascade ——

Page 306: Whittaker. Island Biogeography. 2007

T H E S C A L E O F I S L A N D L O S S E S G L O B A L LY 293

numbers of island versus continental species, islandforms have had about a 40 times greater probabilityof extinction over the period since AD 1600 than havecontinental species (Johnson and Stattersfield 1990).

Recent palaeontological and archaeological stud-ies have made it abundantly clear that island

endemics have been extinguished in significantnumbers from islands and archipelagos all over theworld, before the explorations of modern collectorsbrought their biotas to the attention of Westernscience (Table 11.3). In the case of island birds, atleast, the established prehistoric losses greatly

Box 11.1 The decline of Lake Victoria’s cichlid fishes

Lakes are ‘negative islands’, that is, they aremore or less isolated freshwater areassurrounded by a hostile land matrix, usuallyconnected to other lakes through rivers. Lakesthus behave as islands in many biogeographicaland ecological respects. Long-lasting lakes (e.g.Lake Baikal in Siberia, the Great Lakes of theAfrican Rift Valley, and Lake Titicaca in SouthAmerica), have provided similar evolutionaryopportunities to oceanic islands, and in the caseof the African Rift Valley lakes (Tanganyika,Victoria, and Malawi) outstanding illustrations ofradiations of fish species (Box 9.2).

Lake Victoria, shared by Kenya, Tanzania, andUganda, is the largest tropical lake and thesecond largest freshwater lake in the world,comprising some 3000 km3 of water. It is theunique habitat of c. 300 endemic cichlid fishes,belonging to the genus Haplochromis, which arethought to have diversified since the lakerecharged after 12 400 BP (Box 9.2).Unfortunately, much of this endemic richness has

been depleted following the introduction of theNile perch (Lates niloticus) in 1954 in order toincrease the size of the fishing catch (Mackay2002). Today it is considered that at least half ofthe lake’s native fish species have gone extinct orare so severely depleted that too few individualsexist for the species to be harvested or recordedby scientists (Groombridge and Jenkins 2002).

Although predation by the Nile perch isbelieved to be the major cause of decline,important additional factors may includeincreasing pollution and sediment load, excessfishing pressure, and possible competition fromintroduced tilapiine cichlids (Groombridge andJenkins 2002). The lake itself has now becomedepleted of oxygen, as a consequence of themassive alterations in trophic organization withinthe lake. A shrimp tolerant to oxygen-poor waterstoday provides the major source food for the Nileperch, which itself forms the basis of a fisheryincreasingly focused on the export trade ratherthan providing a local protein source.

Table 11.2 Summaries of known extinctions from islands, continents, and oceans since c. AD 1600 (data fromGroombridge 1992; Steadman 1997a; Primack and Ros 2002). The values for several taxa differ from those given inthe first edition of this book. Figures given by different authorities vary depending on the criteria adopted, so thevalues given here should be taken as of largely indicative value

Group Islands Continents Oceans Total % insular

Mammals 51 30 4 85 60Birds 92 21 0 113 81Reptiles 20 1 0 21 95Molluscs 151 40 0 191 79Insects 51 10 0 61 84Plants 139 245 0 384 36

Total 504 347 4 855 59

Page 307: Whittaker. Island Biogeography. 2007

outnumber the historical losses. Table 11.3 providesa figure of 158 extinctions of island endemic birdsin prehistoric times, but a slightly higher figure, ofjust over 200 species (recorded as subfossils, andwhose disappearance is probably due to prehistoricman), is given by Milberg and Tyrberg (1993). Thereis no doubt that further archaeological and taxo-nomic work will add to the number of extinctionsfor the prehistoric period. It should therefore beunderstood that the year AD 1600 did not mark akey point in island history globally, and that its useby those compiling the statistics is essentially amatter of convenience. The primary place at whichwe ought to divide the analysis is the point atwhich islands have been colonized by humans.

Steadman (1997a) estimates that rates of natural,pre-human extinction in island birds may be atleast two orders of magnitude lower than post-human rates. The world’s avifauna would havebeen about 20% richer in species had the islands ofthe Pacific remained unoccupied by humans. As

many as 2000 bird species may have becomeextinct following human colonization but beforeEuropean contact; the list being dominated byflightless rails, but including moas, petrels, prions,pelicans, ibises, herons, swans, geese, ducks,hawks, eagles, megapodes, kagus, aptornthids,sandpipers, gulls, pigeons, doves, parrots, owls,owlet-nightjars, and many types of passerines. Theemphasis on rails stems from observations that vir-tually every Pacific island that has been examinedpalaeontologically has been found to have had oneor more rails of such reduced flight capabilitiesthat they must have been endemic species. Byextrapolation, it is conceivable that more than 800Pacific islands were once graced with one or morerail species in one of the most dramatic examplesof island radiation to have persisted into theHolocene (cf. Trewick 1997). Only a handful ofthese species remain today.

Island species continue to be disproportionatelyat risk. For instance, approximately one in six plant

294 A N T H R O P O G E N I C L O S S E S A N D T H R E AT S TO I S L A N D E C O S YS T E M S

Table 11.3 Estimates of actual and threatened losses of island endemic birds (data from Johnson and Stattersfield 1990; Steadman1997a). The total of 97 extinctions since AD 1600 is slightly at variance with the estimate of 92 species losses given in Table 11.2,and contrasts with the value of 21 species given in that table for the number lost from continental areas over this period

Region Number of extinctions Endemism

Prehistorica AD 1600–1899 AD 1900–94 Approximate % endemics number of threatenedendemics

Pacific Ocean 90 28 23 290 38Indonesia and Borneo – 0 2 390 22Indian Ocean 11 30 1 200 33Philippines – 0 1 180 19Caribbean Sea 34 2 1 140 22New Guinea and Melanesia 10 2 3 500 10Atlantic Ocean 3 3 1 50 50Mediterranean Sea 10 0 0 – –

Total 158 65 32 1750 23

a Data are lacking for prehistoric losses from Indonesia and the Philippines. The prehistory category consists of species fromprehistoric cultural contexts, and includes only species already described. Sources of possible overestimation include: (1) the so-calledLazarus effect, i.e. where species thought to be extinct are eventually rediscovered, (2) identification of species that did not exist, e.g.there were once thought to have been up to 11 species of flightless giant Moa on New Zealand, but recent genetic analyses ofbones indicates that there were only 2 species which, however, exhibited exceptional morphological dimorphism (Bunce et al. 2003).However, overall, there are doubtless many more extinct species yet to be discovered by the archaeologists; we can thus regard theseestimates as minimum figures.

Page 308: Whittaker. Island Biogeography. 2007

T H E AG E N C I E S O F D E S T R U C T I O N 295

species grows on oceanic islands, but one in three ofall known threatened plant species are islandendemics. The degree to which the remaining islandendemic birds are believed to be threatened is quan-tified in Table 11.3. Although IUCN data show thatthe threat of extinction is switching significantlytowards the continents (Groombridge 1992), statis-tics such as these underline the continuing impor-tance of focusing conservation efforts on islands.

11.4 The agencies of destruction

There are four major reasons why island species arereduced by human action: (1) direct predation; (2)the introduction of non-native species; (3) thespread of disease; and (4) habitat degradation orloss. In a global context, the loss of habitat is com-monly seen as the greatest problem for biodiversity(Lawton and May 1995), but on particular islandsother factors from this list may be more pressing.For instance, on the Galápagos, introduced animalsand plants have been described as: ‘ . . . absolutelythe most insidious, silent, and dangerous scourgethat exists in the archipelago’ (Galápagos Newsletter1996). Commonly, native endemic species arecaught in a pincer action, reduced by habitat loss toa rump, attacked by disease or predation, and fac-ing competition with exotic species. The syner-gisms between these forces result in the loss ofspecies which conceivably could cope with andadjust to one force alone. Similarly, there may sub-sequent trophic cascade effects (above), such thatthe loss of prey species might prejudice the survivalof native predators. For example, the very largeextinct New Zealand eagle, Harpagornis moorei, wasprobably dependent on moas and other large andnow extinct bird species (Milberg and Tyrberg1993).

Predation by humans

Humans hunt island species to eat and also forother reasons, such as the societal values attachedto brightly coloured feathers. In the case of theextinct Hawaiian mamo (Drepanis pacifica), it hasbeen estimated that the famous royal cloak worn byKamehameha I would have required some 80 000

birds to be killed (Johnson and Stattersfield 1990).Trade in such biological resources has a long his-tory, and was an important feature of the period inwhich the Polynesians spread across and held swayover the Pacific. Present-day trade in fruit bats (fly-ing foxes) provides an example of how hunting, notprimarily for local consumption, but for export, canreduce oceanic island populations to crisis point.The critical role that these bats play as pollinatorsand as dispersers of plants may mean that their losswill have significant knock-on impacts, i.e. they actin effect as ‘keystone’ species (Rainey et al. 1995;Vitousek et al. 1995). Hunting for collections contin-ues to be problematic. Today, the trade is drivenmostly by individual collectors, but in the pasthunting for museum collections was common andis, for example, believed to have dealt the finalblow to the spectacular New Zealand bird, the huia(Heteralocha acutirostris) (Johnson and Stattersfield1990).

Introduced species

As humans have spread across the islands of theworld we have taken with us, either purposefullyor inadvertently, a remarkable array of plant andanimal species (see Box 11.2). These anthropogenicintroductions are variously called exotic, alien ornon-native species. Some species persist simply asdomesticated (animal) or cultivated (plant) species,while others become naturalized, i.e. they form self-sustaining populations within modified habitats(Henderson et al. 2006). Of these, some become feralor invasive, expanding into intact or semi-intacthabitats. Of this subset, a few species cause seriousecological impacts. These species are termedecosystem transformers (Henderson et al. 2006).Based on a review of numerous case studies,Williamson (1996) has suggested that only about10% of introduced species become established, within turn about 10% of these species achieving peststatus.

The impact of non-native species can takenumerous forms. These include predation andgrazing impacts that are often strikingly apparent,while the changes can also be more cryptic, butnonetheless profound, such as alterations to

Page 309: Whittaker. Island Biogeography. 2007

296 A N T H R O P O G E N I C L O S S E S A N D T H R E AT S TO I S L A N D E C O S YS T E M S

Box 11.2 Some of the worst invasive alien species on islands

The examples given here demonstrate thatproblematic aliens that have invaded islands acrossthe world are drawn from many higher taxa, andoriginate from all parts of the globe. Although

most of these problematic invasives are continentalin origin, some are drawn from other islands.

Sources: Cronk and Fuller (1995); IUCN/SSCInvasive Species Specialist Group (2004)

Invasive alien species on islands

Common name Scientific name Origin Islands invaded

InsectsBlack twig borer Xylosandrus compactus Asia Madagascar, Mauritius, Seychelles, Java,

Sumatra, FijiCrazy ant Anoplolepis gracilipes West Africa Hawaii, Seychelles, Zanzibar, Christmas IslandCommon malaria Anopheles unknown Hawaii

mosquito quadrimaculatusFormosan subterranean Coptotermes formosanus China Hawaii

termiteTropical fire ant Solenopsis geminata Central America HawaiiArgentinian ant Linephitema humilis South America WorldwideLittle fire ant Wasmannia auropunctata Tropical America Galápagos, New Caledonia, Solomon Islands,

Hawaii, VanuatuWormsTriclad flatworm Platydemus manokwari New Guinea Hawaii, Guam, MarianasSnailsRosy wolfsnail Euglandina rosea SW United States French Polynesia, Hawaii, Bermuda, BahamasGiant Africa snail Achatina fulica NE Africa Hawaii, Samoa, Philippines, Sri Lanka, TahitiGolden apple snail Pomacea canaliculata South America Japan, Philippines, Borneo, New GuineaAmphibiansGiant toad Bufo marinus Central America Hawaii, Fiji, Samoa, Guam, Marianas, Caroline and

Solomon Is.Caribbean tree frog Eleutherodactylus coqui Puerto Rico Hawaii, Virgin IslandsReptilesBrown tree snake Boiga irregularis Australasia GuamBirdsIndian myna bird Acridotheres trisitis India New Zealand, New Caledonia, Fiji, Samoa, Hawaii, CookRed-vented bulbul Pycnonotus cafer South Asia Fiji, Samoa, Tonga, HawaiiMammalsEuropean hedgehog Erinaceus europaeus Western Europe New ZealandBrushtail possum Trichosurus vulpecula Australia New ZealandGrey squirrel Sciurus carolinensis North America Great Britain, IrelandStoat Mustela erminea Holarctic New ZealandLong-tailed macaque Macaca fascicularis South East Asia MauritiusHouse mouse Mus musculus India WorldwideFeral pig Sus scrofa unknown New Zealand,

Hawaii, GalápagosSmall Indian mongoose Herpestes javanicus South Asia Mauritius, Fiji, West Indies, HawaiiDomestic cat Felis catus Unknown? WorldwideFeral goat Capra hircus Asia WorldwideMaori rat Rattus exulans South Asia Marianas, New Zealand, Hawaii, Polinesia

Page 310: Whittaker. Island Biogeography. 2007

T H E AG E N C I E S O F D E S T R U C T I O N 297

(continued)

Common name Scientific name Origin Islands invaded

Brown rat Rattus norvegicus NE China WorldwideShip rat Rattus rattus India WorldwideRabbit Oryctolagus cuniculus Iberian Peninsula WorldwidePlant climbersOld man’s beard Clematis vitalba Europe New ZealandHiptage Hiptage benghalensis India Réunion, MauritiusMerremia Merremia peltata Africa, South Asia, Fiji, French Polynesia, Micronesia, Guam, Samoa,

Australasia Solomon Is., Tonga, VanuatuMile-a-minute weed Mikania micrantha Central America Philippines, Solomon, Sri Lanka, Mauritius, RarotongaBanana passion fruit Passiflora mollisima South America HawaiiHerbsWakeke Abelmoschus moschatus South Asia Samoa, Tonga, Micronesia, Marianas, Hawaii, PolinesiaCrofton weed Ageratina adenophora Mexico New Zealand, Hawaii, CanariesBroom sedge Andropogon virginicus SE USA HawaiiSweet vernal grass Anthoxanthum odoratum Eurasia HawaiiArundo grass Arundo donax India Micronesia , Guam, Palau, Fiji, Hawaii, Nauru,

New Caledonia, Norfolk, Christmas IslandKahili ginger Hedychium gardnerianum Himalaya Micronesia, French Polynesia, Hawaii, New Zealand,

Réunion, Jamaica, Azores, MadeiraIndian balsam Impatiens glandulifera Himalaya Great BritainCurly water weed Lagarosiphon major South Africa New Zealand, MascarenesElephant grass Pennisetum sp. Africa Hawaii, New Zealand, Sri Lanka, Guam, Galápagos,

CanariesWater fern Salvinia molesta Brazil Sri Lanka, New GuineaWandering jew Tradescantia fluminensis Brazil New Zealand, CanariesShrubsBabul Acacia nilotica South Asia Antigua, Barbuda, AnguillaRed cinchona Cinchona pubescens Tropical America Tahiti, Galápagos, HawaiiKoster’s curse Clidemia hirta Tropical America Hawaii, Fiji, Vanuatu, Mascarenes, Comores, SeychellesBroom Cytisus scoparius South Europe New ZealandEl marabu Dichrostachys cinerea Africa Cuba, Hispaniola, Guadeloupe, Marie-Galante,

MartiniqueFuchsia Fuchsia boliviana, South America Réunion

F. magellanicaCuban hemp Furcraea cubensis Greater Antiles GalápagosMauritius hemp Furcraea foetida South America Cape VerdeWhite sage Lantana camara Tropical America Hawaii, Cape Verde, Galápagos, New Caledonia,

New ZealandWild tamarind Leucaena leucocephala Central America Hawaii, MascarenesPrivet Ligustrum robustus South Asia Réunion, MauritiusMiconia Miconia calvescens Tropical America Tahiti, Moorea, Raiatea, MarquesasWild tobacco Nicotiana glauca South America St. Helena, CanariesStrawberry guava Psidium cattleianum, Tropical America Hawaii, Norfolk, Mauritius, Galápagos, New Zealand

P. guajavaRhododendron Rhododendron ponticum Mediterranean Great Britain, IrelandBlackberry Rubus argutus North America HawaiiBlackberry Rubus moluccanus Australasia Réunion, MauritiusCrack willow Salix fragilis Eurasia New Zealand

Page 311: Whittaker. Island Biogeography. 2007

pollination and dispersal networks, or hybridiza-tion. We illustrate both these forms of alteration inthis section, reserving mention of other equallyimportant forms of change, such as the develop-ment of extensive thickets of non-native plantspecies (Meyer and Florence 1996; Henderson et al.2006), to later sections.

Predators and browsersAs often noted, it is the introduction of vertebratepredators, such as cats, rats, dogs, and mongooses,which have caused some of the worst problems(Cronk 1989; Groombridge 1992; Benton andSpencer 1995; Keast and Miller 1996). Feral cats area feature of innumerable islands. They are oppor-tunistic predators, eating what is most easilycaught. Cats were introduced to the sub-AntarcticMarion Island in 1949 and at one stage it was

estimated that there were about 2000 of them, eachtaking about 213 birds a year, and thus in totalkilling well over 400 000 birds per annum, mostlyground-nesting petrels. Given these loses, only byeradicating the cat from the island could the long-term survival of the native bird species be ensured(Leader-Williams and Walton 1989). Cats have beenpresent on another sub-Antarctic island, Macquarie,for a lot longer, and in that case have been responsi-ble for the loss of two endemic bird species. Thecombination of cats and mongooses on the twolargest Fijian islands has resulted in the local extinc-tion of two species of ground-foraging skinks of thegenus Emoia: they survive only on mongoose-freeislands. In the Lesser Antilles, three reptiles havebeen extirpated from St Lucia within historical time,coincident with the introduction of the mongoose.Domestic dogs can also be devastating, and feral

298 A N T H R O P O G E N I C L O S S E S A N D T H R E AT S TO I S L A N D E C O S YS T E M S

(continued)

Common name Scientific name Origin Islands invaded

Brazilian pepper tree Schinus terebinthifolia South America Hawaii, Norfolk, Mauritius, St. HelenaChilean guava Ugni molinae Chile Juan Fernández, ChathamGorse Ulex europaeus Western Europe HawaiiSucculentsCook feet Carpobrotus sp. South Africa BalearicsCoirama Kalanchoe pinnata Madagascar Hawaii, Galápagos, KermadecCoastal pricklypear Opuntia dillenii Central America CanariesCommon pricklypear Opuntia ficus-indica Mexico Hawaii, CanariesTreesBlack wattle Acacia mearnsii Australasia New Zealand, HawaiiSycamore Acer pseudoplatanus Europe New Zealand, Great Britain, Ireland, MadeiraShoebutton ardisia Ardisia eliptica South Asia Hawaii, Okinawa, JamaicaAustralian pine Casuarina equisetifolia Australasia Hawaii, RéunionEmbauba Cecropia peltata Tropical America Hawaii, Tahiti, RaiateaRed quinine tree Cinchona succirubra Ecuador Galápagos, St. HelenaCinnamon Cinnamomun zeylanicum East Indies SeychellesPaper bark tree Melaleuca quinquenervia Australasia HawaiiFire tree Myrica faya Macaronesia HawaiiCluster pine Pinus pinaster Mediterranean Hawaii, New ZealandSweet pittosporum Pittosporum undulatum Australia Norfolk, Lord Howe, Jamaica, AzoresAfrican tulip tree Spathodea campanulata Africa Hawaii, Fiji, French Polynesia, SamoaHonduras mahogany Swietenia macrophylla Central America Sri LankaDiseasesAvian malaria Plasmodium relictum unknown? Hawaii

Page 312: Whittaker. Island Biogeography. 2007

T H E AG E N C I E S O F D E S T R U C T I O N 299

populations have been responsible for the localextinction of land iguanas in the Galápagos.

Even invertebrates have led to the extinction ofisland endemics, notably the losses of all endemicspecies of Partula land snails following the mis-guided introduction of the carnivorous snailEuglandina rosea to the island of Moorea(Williamson 1996). Similarly, exotic ants introducedto the Galápagos and Hawaii (from which they arenaturally absent) predate endemic invertebratesand are an increasing cause for concern in botharchipelagos. Browsing animals such as goats andrabbits have also led to losses (Cronk 1989; Johnsonand Stattersfield 1990). Examples blamed on therabbit include the Laysan rail (Porzanula palmeri)and Laysan millerbird (Acrocephalus familiaris), bothof which disappeared during the early twentiethcentury because rabbits denuded their island.

Changes to pollination and dispersal networksSpecies dispersing spontaneously or being intro-duced by humans to remote islands leave behindtheir usual web of interacting species, which mayinclude pollinators, dispersers, competitors, foodplants, prey, predators, parasites, and pathogens(Janzen 1985, Olesen and Valido 2004). If they can-not withstand the loss of evolutionary ‘partners’they may fail to establish on the island, but at leasta fraction of introduced species find a new web ofinteracting species they can interact with success-fully. Indeed, as discussed in Chapter 7, there issome evidence that pollinator networks tend to bemore generalist on remote islands, involving super-generalist endemic species that readily includenon-native species into their networks, thus aidingtheir establishment (Olesen et al. 2002).

The introduction of exotic pollinators such as thehoney bee (Apis mellifera), continental bumblebees(Bombus spp.), or wasps (Vespula spp.) on to anisland, will have consequences for both native plantand pollinator communities, and these conse-quences may be positive or negative for the islandspecies. If the exotic pollinator is more effective intransporting pollen than the native pollinators,then native (but also exotic) plants will increasetheir fruit and seed production in relation to nativeplants not pollinated by the exotic species.

Conversely, if it is less effective in pollinating a par-ticular plant than a displaced native pollinator, thenthis may reduce seed set. For instance, introducedhoney bees have been found to compete with theendemic Canarian bumblebee (Bombus canariensis),resulting in diminished pollen transfer success forseveral Canarian endemic plants (such as Echiumwildpretii) (Valido et al. 2002).

The introduction of predators or disease organ-isms that knock out native pollinator networks canbe expected to have significant negative conse-quences for native plant species. For instance, theintroduction in Hawaii of predatory insects such asVespula pensylvanica and of diseases, like avianmalaria, have extinguished entire groups of pollina-tors, such as 52 endemic species of Nesoprosopis bees(Vitousek et al. 1987a), and many endemic birdspecies, respectively. The disappearance of exclusivepollinators is considered responsible for the extinc-tion of many plants pollinated by them, includingsome 30 endemic Hawaiian Campanulaceae species(Cox and Elmqvist 2000).

The introduction of exotic dispersers can havepositive consequences on native (but also onexotic) plant species. For example, the Balearicendemic plant Cneorum tricoccon, was originallydispersed by Podarcis lilfordi lizards, which occupyelevations of less than 500 m ASL). Following theintroduction to Majorca of an alternative disperser,the European pine marten (Martes martes), Cneorumbegin to colonize altitudes up to 1000 m ASL,driven by the wider altitude range exploited by themarten (Traveset and Santamaría 2004).Conversely, the introduction of less-effective dis-persers can result in negative consequences fornative species. For example, two introduced mam-mal species, rabbit (Oryctolagus cuniculus) andBarbary ground squirrel (Atlantoxerus getulus),now compete for the fruits of the endemic shrubRubia fruticosa in Fuerteventura (Canaries) with theindigenous dispersers (native birds and lizards).The seeds are dispersed less effectively by themammals and have a lower viability. In this casethere are thus negative consequences both for thenative plant (poorer dispersal services) and for thenative frugivores (competition for a food resource)(Nogales et al. 2005).

Page 313: Whittaker. Island Biogeography. 2007

Similarly, the introduction of an exotic plant canalter the insular networks in varying ways. Thenew plant can offer the native dispersers newresources that will produce positive effects forboth of them, but simultaneously have negativeconsequences for the native plants that have tocompete for resources with the exotics. Forinstance, the introduction of the prickly pears(genus Opuntia) on the Canaries has providednative dispersers, including ravens and endemiclizards, a new fleshy fruit to feed on, with a muchlarger year-around availability than the nativeresources (Fig. 11.2). The spread of Opuntia hasthus been aided by native lizards and ravens, sothat Opuntia species have successfully invadedlarge areas of Canarian coastal scrub rich inendemic Euphorbia and other plant genera(Nogales et al. 1999, 2005; Valido et al. 2003).Finally, it is often the case that interactionsbetween two or more exotic species are crucial tothe spread of problematic invasives. For example,within the Galápagos, introduced mammals areentirely responsible for the spread of the intro-duced guava (Psidium guajava), one of a number ofecosystem-transforming non-native plant species(e.g. Henderson et al. 2006).

Hybridization with native speciesThe introduction to an island of an allopatric con-gener, which is not fully reproductively isolatedfrom the native taxon, can lead to a process ofintrogressive hybridization and the genetic dilu-tion and eventual disappearance of the nativeform. This introgression process is currentlyaffecting the Canarian endemic palm Phoenixcanariensis, the plant symbol of the Canaries, as aresult of the introduction of its closest relative, thedate palm Phoenix dactylifera. P. canariensis growswild in the lowlands of each of the main islands ofthe archipelago, whereas P. dactylifera is a fruitpalm, of unknown origin, that has been cultivatedfor at least 5000 years in the Arabian peninsulaand North Africa. Although it was introducedbefore the Spanish settlement of the Canaries, theproblem of introgression is essentially a recentone, following the importation of massive num-bers of date palms to decorate the new touristresorts and paved roads across the archipelago(Morici 2004). Hybrid swarms are now found allover the archipelago, especially in the tourist hon-eypot areas on the eastern islands. In addition tohybridization, the endemic palm is also threat-ened by a lethal pest, the weevil Rhynchophorousferrugineus, introduced with date palms importedfrom the Middle East. As a result of these pres-sures, this particular plant symbol of the Canariesfaces an uncertain future. Genetically pure standsof the Canarian palm populations remain in theinteriors of Gran Canaria and La Gomera, faraway from the resorts (González-Pérez et al. 2004),providing potential seed supply for future conser-vation programmes. A similar process of intro-gression is under way on the Cape Verde islands,between the endemic palm Phoenix atlantica andthe date palm (Morici 2004).

Disease

The problem of disease is closely associated withexposure to exotic competitors and, indeed, theseparation of exotic microbes as a category fromother introduced organisms is largely arbitrary. Themost striking exemplification of the impact of dis-ease on insular populations has, without doubt,

300 A N T H R O P O G E N I C L O S S E S A N D T H R E AT S TO I S L A N D E C O S YS T E M S

80

60

40

20

0S O N D J F M A M J J A S

Neochamaelea

Withania

RubiaOpuntia

Plocama

Month

% o

f rip

e fr

uits

Figure 11.2 Availability of ripe fruits (% of the crop size) in aCanarian subdesert coastal scrub, Teno Bajo, Tenerife, for September1993 through August 1994. The presence of the exotic invasivecactus Opuntia has altered the pattern of fruit availability for thelocal frugivore community, and this in turn may impact on thedispersal services provided to the native plant species. (From Validoet al. 2003)

Page 314: Whittaker. Island Biogeography. 2007

T H E AG E N C I E S O F D E S T R U C T I O N 301

been among the native peoples of islands, particu-larly on first contact with Europeans. This factor, incombination with more purposeful persecution, ledto the decimation of many island peoples, such thatin some cases little or no trace of them remains(Kunkel 1976; Watts 1987; Diamond 2005).

The most commonly cited example of exotic dis-ease afflicting native island birds is from Hawaii.Hypotheses to explain why so many species ofbirds became extinct in the period since Cook’slandfall in AD 1778 include habitat destruction,hunting, competition with introduced birds, andintroduced predators. Undoubtedly, each has hadits role in the degradation of Hawaii’s ecosystems.Early work by Warner (1968) identified avianmalaria (and avian pox), carried by the introducedmosquito Culex quinquefasciatus, as an importantfactor. Warner proposed an ‘imaginary line’ atabout 600 m ASL, above which there were no mos-quitoes and below which native birds were notfound because they had succumbed to the disease.However, van Riper et al. (1986) suggest that avianmalaria probably took some time to build up a largeenough reservoir and that it did not have a majorimpact upon the numbers of Hawaiian birds untilthe 1920s, some time after the extinction pulse ofthe late nineteenth century. Van Riper et al. (1986)also report that the malarial parasite Plasmodium isfound up to treeline elevations, i.e. well aboveWarner’s 600 m threshold. The parasite is nonethe-less concentrated in the mid-elevational ranges,where introduced bird species and native birdshave the greatest distributional overlap. The intro-duced species have been found to be less suscepti-ble to malaria and thus to act as vectors for thedisease. Although a number of native bird specieshave developed immunogenic and behaviouralresponses reducing the impact of the parasite, vanRiper and colleagues concluded that avian malariais currently a major limiting factor, restricting bothabundance and distribution of the endemic avi-fauna.

The omao (Myadestes obscurus) is one of four sur-viving species of thrushes on the Hawaiian islands.It occupies no more than 30% of its former range,and although locally abundant in rain forests inparticular parts of the Big Island of Hawaii, it is

absent from other areas in which it was once com-mon. Its disappearance from the Kona and Kohaladistricts during the early part of the twentieth cen-tury remains an enigma. A plausible hypothesis isthat it failed to develop resistance swiftly in theseparts of its range to avian malaria, but in other partsof its range, such as the Puna district, it did developresistance, thereby allowing its present pattern ofcoexistence with both the mosquitoes and themalarial parasites in that area (Ralph and Fancy1994).

Given the possibilities of local variations in thedevelopment of resistance to disease, it is clearlygoing to be difficult to be certain about the role ofdisease in the demise of Hawaii’s birds. Added towhich, it needs to be remembered that most specieswere lost before European contact, and that a hostof other changes have occurred over the past200 years or so. The pattern of post-European con-tact losses is bimodal, being concentrated between1885 and 1900, and 1915 and 1935. The formerphase of losses included many species historicallyconfined to higher altitudes (� 600 m ASL) andtherefore contradicting Warner’s (1968) originalmalarial scenario. The explanation for this phase oflosses may relate to another introduced disease,avian pox, or to the impact of habitat modificationsand introduced predators. However, the secondperiod does appear to relate to the arrival andspread of malaria. Birds which succumbed duringthis period were found in the mid-elevationalforests in which the highest prevalence of avianmalaria was found during studies around 1977–80(van Riper et al. 1986). As the introduced tropicalmosquito Culex quinquefasciatus slowly adapted tocooler elevations, the die-off of native bird speciesslowly advanced up the mountains, reaching 600 mby the 1950s and 1500 m by the 1970s, sweepingaway several mid-elevation species in the process,and restricting the survivors to ever higher refugia.This process is ominously described by Pratt (2005)as ‘the third and ongoing wave.’

Two things emerge. First, disease has played andcontinues to play a part in the problem, causingrange reductions and precipitating extinctions.Secondly, we are dealing with a complex systeminvolving opportunities for the spread of exotics

Page 315: Whittaker. Island Biogeography. 2007

due to habitat alteration and the interactions of sev-eral introduced organisms. The exotic birds andmosquitoes act as vectors, and the exotic diseaseprovides the proximal cause of mortality. The factthat introduced birds act as vectors illustrates thatimportant interactions between organisms at thesame trophic level are not all obviously ‘competi-tive’ in nature.

Habitat degradation and loss

Probably the most renowned example of insularhabitat degradation is the case of Easter Island, adepressing experiment in unsustainable exploita-tion of natural resources, which some have seen asa warning for the rest of us (Bahn and Flenley 1992).Environmental degradation has actually beenwidespread across the Pacific, and has been linkedto both Polynesian and European expansion (Nunn1990, 1994; Diamond 1991b). Habitat changes arehard to separate from other negative influences, butthere is no doubt that the reduction in area of par-ticular habitats (notably forest) and the disturbanceof that which remains, have been potent forces inthe reduction and loss of numerous islandendemics (Steadman 1997a,b). A specific example isthe loss of the four-coloured flowerpecker (Dicaeumquadricolor) following the almost complete defor-estation of the island of Cebú in the Philippines(Johnson and Stattersfield 1990).

A popular image of pre-industrial hunter-gather-ers and neolithic agriculturalists is that such peo-ples lived in harmony with their environment,practising a conservation ethic and avoiding theexploitative destruction (‘development’) typical oflater, industrial societies. This idea is associatedwith notions of the ‘noble savage’ and a mythical‘golden age’ (Milberg and Tyrberg 1993). Sadly, theevidence for widespread degradation of islandecosystems across the world before they had anycontact with modern European societies is nowoverwhelming (e.g. Nunn 1990; Diamond 1991b;Milberg and Tyrberg 1993; Schüle 1993; Diamond2005).

Prehistoric humans on islands, although depend-ent on limited animal resources, regularly failed toexploit these in a sustainable way. Several cases

where human populations disappeared altogetherfrom Pacific islands were due to overexploitation ofnatural resources, including animals, vegetationand soils (Diamond 2005). It thus seems likely thatthe Polynesian taboo system, which prevents over-exploitation, was started as a result of the extinc-tions and misuse of other resources, such that the‘ecological balance’ only applied to biota that couldnot be profitably overexploited with the availabletechnology and that could survive habitat destruc-tion and the introduction of rats, dogs, and pigs(Rudge 1989; Bahn and Flenley 1992; Milberg andTyrberg 1993; Benton and Spencer 1995). Indeed,Pimm et al. (1995) estimate that the first coloniststypically wiped out about half of the native avi-fauna of an average Pacific island. Yet, there is alsoa danger that some extinctions may be falselyattributed to the first colonists, because intensivecollection often began half a century or so after thedamage initiated by European discovery.Furthermore, while in many systems it appears thatthe first colonists had eventually reached some sortof balance with the native faunas and floras, ongo-ing processes of globalization continue to imperilmany oceanic island species.

11.5 Trends in the causes of decline

It is easiest to identify trends in the nature of thethreat with reference to island birds, for which themost systematic assessments have been carried out.In prehistoric times most losses can be attributed tohunting, followed by the introduction of commen-sal species. Over the last 400 years, it appears thatintroduced species, followed by hunting, have beenthe key forces (Table 11.4). In the contemporary era,Johnson and Stattersfield (1990) cite habitat loss inrespect to over half the threatened island birdspecies, with exotic species the principal threat in afifth of the cases. Many of the potential introduc-tions of predators to once predator-free islandshave already occurred and in at least some cases theimpacts of exotics are being mediated by protectionmeasures. However, it remains the case that manyof the most severely threatened bird species areendangered by exotics (e.g. Box 11.3). Alien speciesalso continue to pose a severe threat to other island

302 A N T H R O P O G E N I C L O S S E S A N D T H R E AT S TO I S L A N D E C O S YS T E M S

Page 316: Whittaker. Island Biogeography. 2007

T R E N D S I N T H E C AU S E S O F D E C L I N E 303

taxa, and for example are regarded as the principalthreat to over 90% of 282 endangered plant specieson Hawaii (Cox and Enquist 2000).

In terms of habitat change, the loss of forest andwoodland is the biggest concern (Fig. 11.3). Johnsonand Stattersfield (1990) estimate that about 402threatened species of birds are restricted to islands (afurther 67 threatened species have both island andcontinental distributions). Of these 402 species, 77%are forest dwellers, divided between 200 rain forest(lowland and montane) species and 113 seasonal/temperate-forest species. A recent assessment ofthreatened birds in insular South-East Asia found aclose relationship between the numbers of birdspecies considered threatened, and the numbers ofextinctions predicted from deforestation using aspecies–area approach (Brooks et al. 1997). Althoughcertain caveats must be attached to such crudelyderived estimates (cf. Chapter 10; Bodmer et al. 1997),there is no doubt that habitat loss is a key drivingforce no matter which means of assessment is used.

Table 11.4 Principal causes of rarity, decline, and extinction ofisland endemic bird species (from data in Johnson and Stattersfield1990)

Threat Number of species

Extinct since Rare or decliningc. AD 1600

Habitat destruction 19 206Limited range – 165Introduced species 34 76Hunting 25 35Trade – 16Human disturbance 1 10Natural causes 1 or 2 11Fisheries – 2Unknown 41 64

For many species in both ‘extinct’ and ‘rare or declining’ categories,more than one cause has been assigned and the values given abovereflect this; the overall totals in the data set were 97 ‘extinct’ and402 ‘rare or declining’.

Box 11.3 The brown tree snake in Guam

The most notorious recent example of the impactthat the accidental introduction of a predator canhave on an oceanic island is provided by thebrown tree snake (Boiga irregularis). A native ofthe Australasian region, it was introducedaccidentally to Guam, possibly as a stowaway inmilitary cargo, soon after the end of the SecondWorld War. Guam is the largest (541 km2) andmost populated (100 000 inhabitants) island ofthe Marianas archipelago, in Micronesia. Thesnake was first sighted in Guam in the 1950s, andby 1968 it had spread all over the island. Itsnocturnal activity, its ability to live in closeproximity to humans and, especially, the widerange of prey it consumes (lizards, rats, fruit bats,birds- including native, introduced and domesticspecies), enabled a demographic explosion, withdensities of 12 000 to 15 000 snakes per squaremile being reported (Patrick 2001).

Before the arrival of the brown tree snake, theonly snake on the island was a small, blind snake,which lives in the soil and feeds on ants and

termites. The vertebrate fauna of Guam, havingevolved for millennia without such nativepredators, lacked the usual defensive behaviourmechanisms of continental forms, and this isthought to have been an important contributoryfactor to their demise (Rodda et al. 1999). Browntree snakes are considered responsible for theextirpation from Guam of all its breedingpopulation of seabirds, as well as 10 of the 13species of native forest birds: Guam rail (Rallusowstoni), white-browed rail (Poliolimnas cinereus),white-throated ground-dove (Gallicolumbaxanthonura), Mariana fruit-dove (Ptilinopusroseicapilla), Micronesian kingfisher (Halcyoncinnamomina), nightingale reed-warbler(Acrocephalus luscinia), Guam flycatcher (Myiagrafreycineti), rufous fantail (Riphidura rufifrons),cardinal honeyeater (Myzomela cardinalis), andbridled white-eye (Zosterops conspicillatus) (Roddaet al. 1999). Two of the extinct endemic species,the Guam rail and the Micronesian kingfisher, arebeing bred in zoos in the hope that they can

Page 317: Whittaker. Island Biogeography. 2007

The species–area relationship overestimates speciesextinction threat in the Lesser Sundas, and this wassuggested by Brooks et al. (1997) to be because thenatural vegetation of these islands is deciduous mon-soon rather than moist tropical forest. Most endemic

birds are found in this deciduous monsoon habitatand they speculate that these birds may adapt betterto secondary growth and scrub than do the moist for-est species. In contrast, in the Philippines, 78% of theendemic birds are confined to the accessible lowland

304 A N T H R O P O G E N I C L O S S E S A N D T H R E AT S TO I S L A N D E C O S YS T E M S

eventually be released back into the wild, oncethe snake is controlled or eradicated.

Furthermore, two out of the three species ofnative mammals: the Pacific sheath-tailed bat(Embollonura semicaudata) and the little Marianasfruit bat (Pteropus tokudae), as well as 5 of the10 species of native lizards: the snake-eyed skink(Cryptoblepharus poecilopleurus), the azure-tailedskink (Emoia cyanura), Slevin’s skink (Emoiaslevini), the island gecko (Gehyra oceanica), andthe Micronesian gecko (Perochirus ateles) havealso been extirpated. It is not possible to becertain that all these losses are solely attributableto the snake, but careful assessments of theevidence suggest that it has had a major role(Rodda et al. 1999), especially for birds andlizards, as shown in the figure.

The ecological (and socio-economic)consequences go beyond this immediatebiodiversity loss. A trophic cascade can beconsidered to be under way, due to the loss of

birds and bats that acted as pollinators anddispersers of trees, and there are concerns thatthere may be consequent outbreaks of insectpopulations that are no longer subject topredation by insectivorous birds and lizards(Rodda et al. 1999). In recent years, largepopulations of mosquitoes have propagateddengue fever among the people of Guam (Frittsand Leasman-Tanner 2001). Economic damagedue to the brown tree snake includes electricaloutages every third day, including island-wideblackouts, that have been costed at $1–4 milliona year. Finally, the central role of Guam in thePacific transportation network has facilitated thefurther extension of the brown tree snake rangeto other oceanic islands, such as several Marianaislands (Saipan, Rota, Tinian), Okinawa, Pohnpei,Oahu, and even Diego García in the Indian Ocean.Considerable efforts are under way to alert peopleto the danger of introducing this species to otherislands.

Due to snake

Extirpated Due to snake

Due to snake

Extirpated

Enda

nger

ed

Endangered

Endangered

Safe

Safe

Extirpated

Status of Guam’s native forest vertebrates (as present in 1950), with estimates of the degree to which their decline may be attributed to theintroduction of the brown tree snake (black segments). The central pie graph represents the vertebrates, and the small pie graphs representthe degree of responsibility for the declines suffered by mammals (upper right), lizards (lower right), and birds (left); heavy lines delineatethose major taxa in the central graph. (Redrawn from Fig 2.13 of Rodda et al. 1999.)

Page 318: Whittaker. Island Biogeography. 2007

PAT T E R N S O F L O S S AC R O S S I S L A N D TA X A 305

forests in which deforestation has been concentrated,hence over half of the Philippines’ 184 endemic birdsare considered threatened.

How these various assessments of risk translateinto future patterns of loss will depend on a varietyof factors; not least, the protection afforded to par-ticular island habitats and species, and conflictingeconomic pressures on the resource base of particu-lar islands. It is one thing to identify causes ofspecies extinction, quite another to provide reme-dies. Solutions to the problems demand politicalwill, legislation, finance, education, and broad pub-lic support (Chapter 12). Nonetheless, the first stepis to assess and articulate the problem.

11.6 A record of passage—patterns ofloss across island taxa

To set the following sections in context, it is worthnoting variations in the timing of human settlement

(Table 11.5). When Europeans began their majorphase of expanding around the world from the six-teenth century onwards, they encountered islandswith varying antiquity of human occupation. Wemight arbitrarily distinguish three groups: (1)palaeoinhabited islands, which were populatedseveral millennia or tens of millennia beforeEuropean expansion (e.g. New Guinea, theSolomons, Tasmania, the Antilles); (2) neoinhabitedislands, which were populated just one or two mil-lennia before European contact (e.g. the Canaries,Madagascar, Marquesas, Hawaii, New Zealand),and finally, (3) previously uninhabited islands, stillpristine when the European expansion began (e.g.the Azores, Madeira, Cape Verde, St Helena,Tristan, Mascarenes, Galápagos, Juan Fernández).

The record of species extinctions from the islandsthat were colonized before modern European con-tact is largely dependent on proxy sources.Important techniques include the analysis of pollen

(a) Most threatened (b) All

Arid1%

Unknown2%

Seasonal/temperate

forest25%

Smallislands

6%

Inland orcoastalwater10%

Grassland,heathlandor scrub

12%

Rain forest44%

Grassland,heathlandor scrub

13%

Inland orcoastalwater10%

Smallislands

9%Seasonal/temperate

forest36%

Rain forest32%

Figure 11.3 Habitats of threatened island endemic bird species: (a) species listed as either endangered or vulnerable, (b) species listed asendangered, vulnerable, rare, indeterminate, or insufficiently known. Some species have been assigned to more than one habitat in this analysis,hence, for example, the relative importance of the combined forest habitat in part (b) (69%) differs slightly from the estimate for forest species(77%) given in the text. Data from Johnson and Stattersfield (1990), according to the then IUCN Red Data Book categories: endangered —those indanger of extinction and whose survival is unlikely if the causal factors continue operating; vulnerable—those likely to move into the endangeredcategory in the near future; rare—taxa with small world populations; indeterminate—taxa known to belong to one of the foregoing categories,but with insufficient data to say which; insufficiently known—taxa suspected but not definitely known to belong to one of the foregoingcategories. (The data given in this figure exclude the category extinct—taxa no longer known to exist in the wild after repeated searches of theirtype localities.)

Page 319: Whittaker. Island Biogeography. 2007

from lake sediments and the analysis of subfossilbones recovered from middens. Such studies haveled to important insights into the past vegetationand animal communities of many oceanic islands.Examples of extinct island birds known to scienceonly from bones include New Zealand’s extinctmoa species (Dinornis novaezealandiae in the NorthIsland and D. robustus in the South Island), giantflightless ratite birds, hunted to extinction by theMaori by about 500 BP (Bunce et al. 2003), and theelephant bird (Aepyornis maximus), extinguished bythe Malagasy people (Groombridge and Jenkins2002). Similarly, two giant rats (Canariomys tamaraniand C. bravoi) endemic to the Canaries, were verylikely wiped out by the first inhabitants (theguanches) or by the animals introduced by them(Cabrera 2001). Considering the diversity of theirendemic biota, some island groups colonized in the

distant past, such as the Canaries or Madagascar,appear to have suffered remarkably few extinctionsafter AD 1600 (Table 11.6), probably because thebiota that were prone to extinction (conspicuous,highly specialized species with low rates of recov-ery or small population sizes) had already beenextinguished by the first colonists.

By contrast, the pristine biota of the previouslyuninhabited group of islands was quite often stud-ied by ships’ naturalists in the early voyages of dis-covery, providing a mix of descriptions, drawings,and specimens, famously including the flightlessdodo. Unfortunately, these earlier voyagers alsointroduced exotic species, such as goats, and alsopredated native species, including giant tortoiseand again the dodo (although it was unpleasant toeat), beginning the slide of many island speciestowards extinction (below).

306 A N T H R O P O G E N I C L O S S E S A N D T H R E AT S TO I S L A N D E C O S YS T E M S

Table 11.5 Approximate date of first human colonization in relation to European contact dates for severalarchipelagoes around the world (from several sources)

Island or archipelagos Pre–European establishment European contact (AD)

Palaeo–inhabitedNew Guinea 40–50 000 BC 1545Tasmania 20–30 000 BC 1642Cyprus c. 8000 BC –Corsica–Sardinia 8600–6700 BC –Great Antilles c. 7000 BC 1511Crete 6250 BC –

Neo–inhabitedMajorca c. 2000 BC –Fiji, Samoa 1500 BC

Canaries c. 500 BC c. 1400Marquesas 200 BC 1595Easter AD 400–900 1722Hawaii AD 400 1778Madagascar AD 800–900 1500New Zealand AD 900 1643

Previously uninhabitedAzores – 1432Cape Verde – 1456St Helena – 1502Galápagos – 1535Juan Fernández – 1574Mascarenes – 1598Tristan da Cunha – 1812

Page 320: Whittaker. Island Biogeography. 2007

PAT T E R N S O F L O S S AC R O S S I S L A N D TA X A 307

Extinctions of island vertebrates since AD 1600have affected all the major oceans, but have beenparticularly concentrated in three nuclei: theMascarenes in the Indian Ocean, the Caribbeanarch (the Antilles), in the Atlantic Ocean, andPolynesia (Table 11.6). The lack of extinctions ofmammals from some remote islands (e.g. Hawaii,St Helena) is simply a reflection of the failure of ter-restrial mammals to colonize them in the first place.

Pacific Ocean birds and the Easter Islandenigma

Birds are typically the most species-rich vertebrateson remote islands. They leave good fossil or sub-fossil evidence, and have been relatively well stud-ied, often from sites associated with humansettlement (i.e. middens). The picture that hasemerged of their demise in prehistoric times canthus be more closely related to human activitiesthan is the case for any other taxon—and theoceanic region where the most dramatic and con-vincing evidence of human involvement has beenfound is the Pacific. Not only has the Pacific lostlarge numbers of bird species in the past, but todayit holds more threatened bird species than anyother oceanic region: approximately 110 species, of

which 31 are classified as endangered and 29vulnerable (the two highest Red Data Book cate-gories; Johnson and Stattersfield 1990).

It is believed that the Australian and New Guinealand masses on the Sahul shelf were colonized bypeople at least 50 000 and 40 000 BP respectively.Voyagers reached New Ireland and the Solomonsby 29 000 BP, but then paused for over25 000 years—islands beyond this zone quite possi-bly were undetectable to these people (Keast andMiller 1996). Then the ancestors of the Polynesiansbecame the first wave of human colonists to spreadacross the Pacific, and they did so very rapidly,between about 1200 BC and AD 1200 (Diamond2005). So extensive were their explorations thatnearly all islands in Oceania (Melanesia,Micronesia, and Polynesia) were inhabited byhumans within the prehistoric period (Fig. 11.4;Pimm et al. 1995; Steadman 1997a). The humancolonists cleared forests, cultivated crops, andraised domesticated animals. Birds providedsources of fat, protein, bones, and feathers for peo-ple and so were hunted as well as being indirectlyaffected by the changes that followed humanarrival. European exploration began in the six-teenth century, but their phase of colonizationstarted much later, such that, for instance, the firstmissionaries arrived in Hawaii in AD 1779 and inTahiti in AD 1795 (Pimm et al. 1995).

In the Hawaiian islands, at least 62 endemicspecies of birds have gone extinct since Polynesiancolonization began some 1600 years BP, and most ofthese were lost before European contact began(Steadman 1997a). The processes involved are man-ifold, including habitat alteration, hunting, dis-eases, and exotic species. The commensalsintroduced by the Polynesians, especially the Maorirat (Rattus exulans) and the feral pig (Sus scofra),have been supplemented since European arrival bymany other exotic species, which have had a pow-erful impact on the native biota, including blackrats, feral dogs, cats, sheep, horses, cattle, goats,mongooses, and an estimated 3200 arthropodspecies (Primack 1993). The extinctions haveincluded flightless geese, ducks, ibises, and rails,and the major part of the radiation of Hawaiianhoneycreepers (Olson and James 1982, 1991; James

Table 11.6 Islands or archipelagoes that have lost at least fiveterrestrial vertebrate species after AD 1600 as a result ofanthropogenic activities (Source: Groombridge and Jenkins 2002)

Island or archipelago Mammals Birds Reptiles Total

Hawaii 0 21 0 21Mauritius 1 10 5 16New Zealand 1 13 1 15Hispaniola 11 1 0 12Rodrigues 0 8 3 11Réunion 1 7 2 10Cuba 8 1 0 9St Helena 0 7 0 7Society Islands 0 6 0 6Cayman 5 1 0 6Jamaica 2 2 1 5Madagascar 4 1 0 5Martinique 1 2 2 5Total 34 80 14 128

Page 321: Whittaker. Island Biogeography. 2007

and Olson 1991). Pimm et al. (1995) believe thatroughly 125–145 bird species once inhabited themain islands, of which 90–110 are extinct, withanother 10 considered imminently threatened withextinction. The original avifauna contained onlythree non-endemic land birds, and all the passer-ines were endemic, yet in consequence of the enor-mous losses and of large numbers of speciesintroductions by humans, only about a third of thepasserines found on Hawaii today are endemic.Nearly all the birds seen now below 1000 m ASL arealiens. Today, some 48% of the land birds of Hawaiiare exotics introduced and released in order toaugment the fauna with game birds and birds of

beautiful plumage or song. Of 94 species known tohave been introduced before 1940, 53 speciesbecame established at least locally and only 41failed completely.

The Hawaiian islands illustrate another featureof prehistoric extinctions, namely, how they alteredthe structure of vertebrate—specifically bird—feed-ing guilds. All but one species of native predatorbecame extinct (James 1995). Such was the fate ofthe terrestrial herbivores, with the exception only ofthe Hawaiian goose (Branta sandvicensis), whichwas rescued from the brink of extinction only by anex situ breeding programme. Losses included thelarge flightless anseriforms (moa-nalos), which

308 A N T H R O P O G E N I C L O S S E S A N D T H R E AT S TO I S L A N D E C O S YS T E M S

NEWZEALAND

CHATHAM IS.

KERMADEC IS.

MariaRanotongaTongatapuS. COOKS IS.Vava’u

TubuaiRaivavae

RapaPitcairn Easter

HendersonMangareva

TUAMOTU IS.

MARQUESAS IS.El’ao

FlintCaroline

MaldenStarbuckSydney

TongarevaVostok

SuwarrowN.COOKS IS.

TOKELAU IS.SAMOA IS.Palmerston

SOCIETY IS.

TON

GA

IS.

FIJI

IS.

TUVALU

HowlandWashingtonLINE IS.

Jarvis

FanningChristmas

PalmyraPOLYNESIA

Johnston

PHOENIX ISBaker

MA

RSH

ALL IS.

MICRONESIACAROLINE IS.

KIRIBATINukuoro

NukumanuOntong Java

SikaianaSANTA RotumaCRUZ

IS.Uvea

FutunaVANUATU

Rennell

MELANESIA

NEW CALEDONIA

Lord Howe

Norfolk

New Guinea

SOLOMONIS.

HAWAILAN IS.

Midway

NeckerNihoa

REVILLA GIGEDO IS.

CocosMalpelo

GALAPAGOS IS.

San FelixSan Ambrosio

JUAN FERNANDEZ IS.

140˚ 160˚ 180˚ 160˚ 140˚ 120˚ 100˚ 80˚

40˚

20˚

20˚

40˚

140˚ 160˚ 180˚ 160˚ 140˚ 120˚ 100˚ 80˚

40˚

20˚

20˚

40˚

Oeno

Figure 11.4 Map of the Pacific showing Polynesia. At the time of European contact, many islands in the eastern and central Pacific wereuninhabited (large dots). Crosses mark islands colonized by the Polynesians prior to AD 1500 but then abandoned, usually after the extinction ofmany native species. Question marks indicate those islands where pre-European colonization was noted by Brown and Lomolino (1998) assuspected but not clearly documented. (Source: Brown and Lomolino 1998, after Terrell 1986.)

Page 322: Whittaker. Island Biogeography. 2007

PAT T E R N S O F L O S S AC R O S S I S L A N D TA X A 309

were browsers on understorey vegetation. The ter-restrial omnivores all but disappeared too, alongwith the Hawaiian land crabs. Prehistoric extinc-tions among insectivores (37%) and arboreal frugi-vores/omnivores (40%) have been proportional tothe losses among passerines generally, but grani-vores have suffered much worse and nectarivoresmuch better in comparison. With predators disap-pearing, habitats shifting, and species richnessfalling during the shake-up following Polynesiansettlement, it is likely that some species, evenguilds, became both relatively and absolutely morewidespread and abundant than under conditionsbefore human contact. James (1995) argues that thedifferential effects of extinction on vertebrate feed-ing guilds in prehistoric times may still be affectingplant communities, i.e. that the losses must collec-tively have implications for the functional charac-teristics of the remaining ecosystems, but such alegacy is difficult to separate from the other drivingforces for change.

The pre-human avifauna of New Zealand wasprobably about twice the size of the native avifaunatoday. The most famous losses were of the giant,flightless moas, the largest of which weighedaround 250 kg (Rudge 1989; Holdaway 1990; Bunceet al. 2003). The moa exhibited exceptional variationin body size in relation to sex and habitat, such thatuntil very recently there were thought to have beenas many as 11 species. Genetic analyses of bonesrecovered from sites around New Zealand revealthere to have been only two distinct genetic line-ages, D. novaezealandiae in North Island and D. robustus in South Island (Bunce et al. 2003).

An interesting story is attached to the loss of oneof the smaller New Zealand birds, the StephensIsland wren, named Traversia lyalli after the light-house keeper (Lyall) who discovered it. StephensIsland is a tiny island in Cook Strait, off SouthIsland. The wren was unknown to science until in1894 the lighthouse keeper’s cat brought in a fewspecimens it had killed. The exotic predator met theisland endemic: end of story. At least, this was theconclusion reached at the time, prompting a corre-spondent to the Canterbury Press to suggest that infuture the Marine Department should see to it thatlighthouse keepers sent to such postings should be

prohibited from taking any cats with them, ‘even ifmouse-traps have to be furnished at the cost of theState’. Although cats have been claimed to have byfar the worst record in exterminating New Zealandbirds (King 1984), we should not place all the blameon the lighthouse keeper and his cat. Before thearrival of the Polynesians, the wren was actually awidespread mainland resident (Holdaway 1989;Rosenzweig 1995). Its loss from mainland NewZealand was probably due to a combination ofhabitat changes and the introduction of rats by thePolynesians; the loss from Stephens Island was butthe final step to extinction. Interestingly, it is justsuch small offshore islands that provide theenclaves that may yet save some of New Zealand’sremaining endemics from extinction as, cleared ofexotic pests, translocated native populations areable to establish and persist. Here, in microcosm,can be found some of the essential ingredients ofthe diversity disaster of oceanic islands, and a life-line of hope for the future.

The Polynesians discovered New Zealandaround AD 950, having voyaged from the area con-taining the Marquesas, the Society Islands, and theSouth Cooks, more than 3000 km to the east. Theybrought with them six species of plants and twospecies of mammals. Polynesian settlement led tothe destruction of half of the lowland montaneforests, widespread soil erosion, and the loss orreduction of much of the vertebrate fauna. Duringthe phase of their expansion, New Zealand lost itssea lions and sea elephants, and at least 25 birdspecies, including the 2 species of moa which, in theabsence of terrestrial mammals, had evolved tooccupy the browser/grazer role normally occupiedby animals such as kudu, bushbuck, or deer. Themost severe modification occurred between 750and 500 years ago, when a rapidly increasinghuman population overexploited animal popula-tions and used fire to clear the land. By aboutAD 1400, the major New Zealand grazing systemshad ceased to exist, their place being taken byunbrowsed systems: the removal of the moas thushad significant effects on the structure and speciescomposition of the vegetation.

Holdaway’s (1990) analyses of subfossil remainsindicate that the pre-human avifauna was

Page 323: Whittaker. Island Biogeography. 2007

dominated by forest species. The high rate ofextinction in the avifauna prior to European settle-ment must have been related to the characteristicsof the forest areas removed by the Polynesians:these being shown by palaeoecological data to havebeen mainly the drier, eastern forests, or thoseinland areas where drought or severe climaterestricted regeneration after clearance. In pre-humantimes, it was the drier, more structurally diverseforests on more fertile soils which supported thegreatest diversities of New Zealand’s birds.

Pacific seabird populations have also suffered, asSteadman (1997a) points out; petrels and shearwa-ters have seen the greatest loss of species, and manyother seabird species have had their distributionand population sizes diminished, including alba-trosses, storm-petrels, frigatebirds, and boobies.Abbott’s booby (Papasula abbotti) is a classic exam-ple: once widespread in the South Pacific, it nowbreeds only on Christmas Island in the easternIndian Ocean.

Easter Island (27�9�S, 109�26�W) was, until thecolonization of Tristan da Cunha, just two centuriesago, the most isolated piece of inhabited land in theworld. When first contacted by Europeans onEaster Day AD 1722, it presented a fascinatingenigma: how could this impoverished treelessisland, with its impoverished people, havesupported the construction of the remarkable giantstatues (moai), and how and why had it all gonewrong? The flora of Easter Island consists today ofover 200 vascular species, of which only 46 arenative. However, formerly the flora was richer, con-taining several native tree and shrub species.Evidence for a forest cover is provided not just bypollen records, and by the discovery of palm nuts,but also by the discovery of extinct endemicachatinellid land snails, and by the identification tospecies or genus of over 2000 fragments of charcoal(Diamond 2005). The forests are now known tohave contained a giant palm tree (Paschalococos dis-perta) and a number of other trees (e.g. Sophoratoromiro), some reaching over 30 m in height.Palynological studies by Flenley and colleaguesdemonstrate that the forests of Easter Island per-sisted for at least 33 000 years (as far back as therecord goes), during the major climatic shifts of the

late Pleistocene and early Holocene. We can thus beclear that deforestation by humans was responsiblefor the treeless state of the island (Flenley et al. 1991;Bahn and Flenley 1992; Diamond 2005).

Dated archaeological evidence indicates thatEaster Island was settled by the Polynesians fromthe Marquesas or Gambier archipelagos (Oliver2003) some time between AD 300 and AD 900, withthe more recent date preferred by Diamond (2005)as the most reliable earliest date of occupation. Theireffect on the forest was soon detectable in the pollenrecord (Bahn and Flenley 1992) and it was entirelyeliminated some time between the early 1400s andthe 1600s AD (Diamond 2005). The human popula-tion reached its peak of about 10 000 (or possibly ashigh as 15 000) around AD 1600, but subsequentlycollapsed along with the megalithic culture that hadsustained the quarrying, sculpting, transport, anderection of the moai. Figure 11.5 sets out a model ofthe sequence of events which may have led to thiscollapse, within which a pattern of overpopulationand continuing overexploitation and thus resourcedepletion was central. This depletion involved notjust the loss of a forest cover, but through it the lossof the raw materials (large trees in particular)needed to make ocean-going canoes, further dimin-ishing resource availability. The pattern of declinemay have also been influenced by the introductionof rats by the Polynesians, which could have had akey role in limiting tree recruitment (Fig 11.5). Theloss of trees and other plant species is matched by amore complete loss of native birds than on any com-parable island in Oceania (Steadman 1997a).Examination of bird bones associated with artefacts600–900 years old has revealed that Easter Islandonce had at least 25 species of nesting seabirds, only7 of which now occur on one or two offshore islets,and just one of which still nests on the main island.Native land birds are known to have included aheron, two rails, two parrots, and an owl. Nonesurvive.

The Easter Island story holds some interestingtwists and turns that remain the subject of somedebate amongst archaeologists. When the Dutchcaptain Jacob Roggeveen and his three vesselsarrived in AD 1772, the overthrow of the statue-building culture and its chiefs and priests had

310 A N T H R O P O G E N I C L O S S E S A N D T H R E AT S TO I S L A N D E C O S YS T E M S

Page 324: Whittaker. Island Biogeography. 2007

PAT T E R N S O F L O S S AC R O S S I S L A N D TA X A 311

already occurred, although the repercussions of thisevent were ongoing, and in fact, some statues werestill standing, with the last European mention of anerect statue being in AD 1838 (Diamond 2005). Thecoup happened around AD 1680, led by militaryleaders called matatoa. After this event, a new cul-ture and some degree of order had developedaround the Birdman cult, which appeared to pro-vide a mechanism for the distribution of resourcesamongst the people of Easter Island. However, itwas a much altered society, in which human strifeand cannibalism were now prevalent features of asociety that had once been so stable and well pro-vided for that it was able to devote surplus labourto the quarrying, carving, transportation and erec-tion of statues weighing between 10 and 90 tons.Possibly hundreds of people would have beeninvolved in some of these operations, and they

would have required large timbers for the sleds,ladders, levers, and ropes needed for the move-ment of the statues. At the time of European con-tact, the human population was estimated at about3000, a figure at which it may or may not have sta-bilized (Erickson and Gowdy 2000). Unfortunately,the interaction of this society with the outsideworld generated a further collapse, this timefuelled by introduced diseases (e.g. smallpoxepidemics, Diamond 2005). In 1862, slave tradersforcibly removed about 1500 people, approximatelyone third of the population at the time. They weretransported to Peru and sold at auction to workin guano mines and other menial jobs. Mostsoon died, but under international pressure, Perurepatriated a dozen survivors in the following year.The survivors brought with them another smallpoxepidemic. By 1872, the population of Easter Island

Forest clearancefor agricultureand firewood

Warfare

Humanpopulation

growth

Insufficient good land

Crop destruction

More food needed

Soilexhaustionand erosion

ForestHuman

immigration

Introduction of rodents

Eating of bird eggs

by rodents

Eating of palm fruits by rodents

Sea Bird Resource

Direct culling of sea birds

Reductionof sea bird resource

Felling palms for statue-

moving and canoes

More land needed for agriculture

FoodshortagePrevention

of palm regeneration

Decline and loss of palm

tree

Decline of fishing and

rodents

Rodentpopulation

growth

Cessationof statue building

Humanpopulation

decline

Figure 11.5 A hypothetical model to indicate the possible course of events on Easter Island as deduced largely from palaeo-ecological data.(Modified from Bahn and Flenley 1992, Fig. 191.)

Page 325: Whittaker. Island Biogeography. 2007

stood at only 111 people. Just 16 years later theChilean government annexed the island, and itbecame in effect a sheep ranch, with the remainingislanders confined to a single village as forcedlabour.

The islands of the Pitcairn group lie between 23.9and 24.7�S and 124.7 and 130.7�W in the SouthPacific Ocean, and consist of volcanic PitcairnIsland, the raised coralline Henderson Island, andthe small coral islands of Ducie and Oeno atolls.Pitcairn is of special historical interest, owing to itsoccupation by the HMS Bounty mutineers, inAD 1790, but before that it was occupied and thenabandoned by the Polynesians. Although the pre-cise timing of the settlement of Pitcairn by thePolynesians remains problematic, it now appearsthat Henderson Island was settled between AD 800and 1050 and deserted some time after AD 1450(Benton and Spencer 1995). During the occupation,marine molluscs, turtles, and birds were heavilypredated. Settlement was accompanied by theintroduction of cultigens and tree species, and byburning for crop cultivation. A number of endemicbird species and at least 6 out of 22 land-snailspecies became extinct during this period. It seemslikely that sustained occupation was only possiblethrough interaction with Pitcairn and, 400 km tothe west, with the island of Mangareva.

Weisler (1995) has outlined a prehistory ofMangareva, which suggests a similar pattern ofresource exploitation and environmental depletionas Easter Island. Mangareva experienced highhuman population levels, depleting the resourcebase to the point where they could no longer sup-port long-distance voyaging, dependent as it wason large ocean-going vessels, skilled crews, foodsupplies, and exchange commodities. As on EasterIsland, the requirement for large trees for canoemanufacture does not appear to have been matchedwith sustainable tree husbandry! Thus, accordingto Weisler’s reconstructions, the overexploitation ofresources by these linked communities resulted inthe severing of voyaging linkages and the eventualabandonment of the most marginal locations,including both Henderson and Pitcairn. With itspoor soils, sporadic rainfall, lack of permanentsources of potable water, karstic topography, and

low diversity of marine and terrestrial life, it ispretty remarkable that Henderson was settled bythe Polynesians in the first place (see illuminatingdiscussion in Diamond 2005).

The Henderson petrel (Pterodroma atrata), onlyrecently described as a distinct species, breedsexclusively on Henderson Island. A study in1991/92 showed that its breeding success wasgreatly reduced by predation by the Pacific rat(Rattus exulans) and the species was judged to bethreatened (Benton and Spencer 1995). It has beencalculated that this petrel has been undergoing aslow decline since the Polynesians introduced therat some 700 years ago, and it is possible that thismay continue until the petrel becomes extinct,although this would seem a remarkably long lagtime for a new ‘equilibrium’ to be reached.Henderson Island has four surviving endemic landbirds, including one fruit dove, Ptilinopus insularis.In the past, this dove shared the island with at leasttwo other Columbiformes: Gallicolumba, a grounddove, and a Ducula pigeon. The dove becameextinct during Polynesian occupation about the latethirteenth century. Indeed, five of Henderson’s nineendemic land birds became extinct as a result of thisoccupation, as did most of the small ground-nest-ing seabirds (Wragg 1995). Human predation, aswell as predation by (and competition with) thecommensal Pacific rat, and habitat alteration bypeople, each had a role in these changes in the avi-fauna. The process of change is ongoing, and todaya number of the endemic land snails of Pitcairnappear to be restricted to remnants of native vege-tation of only about 1 ha or less. These stands arecurrently under great threat from the spread ofinvasive exotic plants, which cast a deep shade andcreate an understorey inimical to the snails (Bentonand Spencer 1995).

Indian Ocean birds

As shown in Table 11.3, the islands of the IndianOcean have also suffered significant losses of birdsin prehistoric and historic times. Most notably, onMadagascar, 6–12 species of ratites became extinct,almost certainly as a result of the arrival of theMalagasy people. These losses included the largest

312 A N T H R O P O G E N I C L O S S E S A N D T H R E AT S TO I S L A N D E C O S YS T E M S

Page 326: Whittaker. Island Biogeography. 2007

PAT T E R N S O F L O S S AC R O S S I S L A N D TA X A 313

bird ever recorded, the giant elephantbird(Aepyornis maximus). Other losses include 2 gianttortoises, pygmy hippo, and at least 14 species oflemur, most of them larger than any survivingspecies (Diamond 1991b; Groombridge 1992).

The islands of Amsterdam and St Paul, some2500 km east of Madagascar in the southern IndianOcean, and 80 km from each other, are known tohave had their own avifauna, but nearly all of thenative species were exterminated through fires andthe introduction of domesticated and commensalmammals, notably feral cattle (Olson and Jouventin1996). The extinct avifauna included a species ofrail, and a different species of duck on each island.The ducks can be assumed to have been differentspecies because remains of the Amsterdam form,Anas marecula, reveal it to have been small andflightless: whatever was on the other island (knownonly from an historical account) must perforce havebeen a distinct form.

Without doubt the most celebrated of all extinctisland birds, the dodo (Raphus cucullatus) was alarge, flightless pigeon (Columbiformes) endemicto Mauritius (Livezey 1993). A related species, thesolitaire (Pezophaps solitaria), occurred onRodrigues, also in the Mascarene group. The exis-tence of a third species on Réunion is indicated byhistorical evidence, but has not been backed up bya specimen. The dodo and its relatives were wipedout by the early eighteenth century by a combina-tion of hunting, habitat destruction, and the nega-tive impacts of introduced vertebrates. Predatorsimplicated include pigs, rats, cats (especially onRodrigues), and monkeys (Mauritius only).Introduced herbivores responsible for significantdestruction of native vegetation were cattle, goats,and deer (Mauritius only). Although they proved tobe vulnerable to the evolutionarily unpredictableintervention of humankind, these giant, ground-dwelling pigeons were formerly dominant con-sumers within their respective island ecosystems.

The native forests and woodlands of Mauritiushave been reduced in extent by 95% to just 5% ofthe land area (Safford 1997). This deforestation,combined with the introductions of many exoticanimals and plants, has led to the extinction of atleast eight other endemic bird species in addition to

the dodo and the reduction to critical levels of eightothers. Five of the six extant forest-living nativepasserines are essentially restricted to native vege-tation. Moreover, only a small fraction of this habi-tat is of good quality because of nest predation,reduced food supply, disease, and, in the past, theuse of organochlorine pesticides. In 1993, each ofthese five species was estimated to have popula-tions of the order of only 100–300 pairs. Safford(1997) recommends management of exoticbrowsers and predators, and weed removal, as partof a broader strategy for conservation on Mauritius.He also suggests that, given the impracticality ofeliminating predators on the mainland, some of theoffshore islets be considered for ecological rehabili-tation and translocation programmes. However, henotes that bird populations on small offshore isletswould be vulnerable to catastrophes, especiallycyclones.

Reptiles

Case et al. (1992) review the data for reptilianextinctions worldwide over the past 10 000 years,noting that the great majority have occurred onislands (as Table 11.2) and as a consequence ofhuman impacts. On small islands, humans haveraised extinction rates by about an order of magni-tude, but rates of loss have been much lower onvery large islands. Introduced predators, princi-pally mongooses, rats, cats, and dogs, have beenthe main agents of human-related extinctions,whereas competition with introduced reptilesappears to have had relatively little impact onnative island reptiles. The reptiles most prone toextinction have been those with relatively largebody size and endemics with a long history ofisland isolation.

Case et al. (1992) subdivide islands into three cat-egories:

● First, they discuss islands colonized in prehistoryby aboriginal people and then colonized later byEuropeans. On such islands, many reptiles areknown only as subfossils, having become extinctduring the aboriginal period. In New Zealand, forexample, three species of lizard known from the

Page 327: Whittaker. Island Biogeography. 2007

Holocene are extinct, and a further nine reptilespecies are today found only on the off-lyingislands, among them the ‘living fossil’ tuatara(Sphenodon punctatus and S. guntheri), the single rep-resentatives of the order Sphenodontia (the rest ofthe lineage died out by 60 Ma). These offshoreisland populations are relictual distributions follow-ing extinctions from the mainland consequent uponPolynesian settlement. A similar pattern is found inthe Caribbean, where several species became extinctduring aboriginal occupation on islands such asHispaniola and Puerto Rico; others have becomeextinct within the past century. The large herbivo-rous iguanine Cyclura has become extinct on a num-ber of islands in the recent past, and the giantCyclura pinguis, lost from Puerto Rico during theHolocene, survives on the small offshore island ofAnegada. Among the lost reptiles of the Caribbeanwere giant tortoises (Geochelone spp.) from theBahamas, Mona island, and Curaçao. Giant tor-toises also once occurred on Sicily, Malta, and theBalearic islands in the Mediterranean, as well as inthe Canaries, and on the Mascarene islands. Theirdisappearance from the latter followed shortly afterhuman contact, but in the Mediterranean the role ofhumans in the losses is not yet so clearly established(Schüle 1993). The Canaries, in the eastern Atlantic,lost a number of lizard species following their colo-nization by the aboriginal Guanches between 2500and 2000 BP. In the Canaries, fossil lizards are foundin association with human artefacts, but lizardswere probably not a major item of the diet, and theintroduction of commensals, such as rats, dogs,goats, and pigs, was probably of greater significancein the demise of the largest species. Interestingly,populations of two Canarian giant lizards (Gallotiasimonyi and G. gomerana) thought to have beenextinct, have been rediscovered on El Hierro andLa Gomera respectively in recent years: examplesof so-called Lazarus species.

● The second category identified by Case et al.(1992) was of islands with a colonial period but noaboriginal history, on which island endemics sur-vived to be described as living species, very often tomeet their demise shortly after. Two spectacularlarge geckos, Phelsuma spp., co-occurred brieflywith Europeans on Rodrigues island in the Indian

Ocean in the late seventeenth century, until huntedto extinction by rats and cats. A similar fate befellmany endemic reptiles on Mauritius. None of theendemic reptile species of the Galápagos hasbecome extinct yet, but population densities havedeclined and local extirpations, for instance, ofgiant tortoises, have occurred. Land iguanas havealso been lost from Baltra and James islands, prob-ably as a result of predation by feral dogs.

● Case et al.’s (1992) third category was of islandswith no permanent human settlement to date. Suchislands are generally too bleak or small to supporthuman settlement. For this reason they also tend tosupport few endemic species. Losses from this cat-egory of islands have thus mostly been local ratherthan global extinctions, and of interest principallyfor the insights they offer into natural turnover.

Caribbean land mammals

Extinctions have dramatically altered the biogeog-raphy of West Indian land mammals over the past20 000 years (Morgan and Woods 1986). Of 76 non-volant mammals, as many as 67 species (88%) havebecome extinct since the late Pleistocene, includingall known primates and Edentata (Table 11.7). LatePleistocene extinctions are attributable to climaticchange and the postglacial rise in sea level, but mostlate Holocene losses are due to humans. Thirty-seven of the losses post-date the arrival of people inthe West Indies, which Morgan and Woods assumedoccurred some 4500 years ago (although it mayhave been rather earlier than this: Table 11.5). Basedon the 4500 years time line, species have been lost atthe average rate of 1 per 122 years since human col-onization, the figure for the preceding 15 500 yearsbeing about 1 per 517 years. In contrast to the datafor non-volant mammals, bats appear to have faredrelatively well, only 8 of the 59 species having dis-appeared over the past 20 000 years.

Many species of extinct West Indian mammals,especially rodents, are common in archaeologicalsites, indicating them to have been an importantpart of the Amerindian diet. In common with otherstudies cited here, humans brought about extinc-tions in the West Indies through predation, habitatdestruction, and (especially in post-Columbian

314 A N T H R O P O G E N I C L O S S E S A N D T H R E AT S TO I S L A N D E C O S YS T E M S

Page 328: Whittaker. Island Biogeography. 2007

PAT T E R N S O F L O S S AC R O S S I S L A N D TA X A 315

times) the introduction of exotic species.Knowledge of the precise timings of species disap-pearances is fragmentary, but it does appear that anumber of species disappeared during theAmerindian period, including at least one speciesof ground sloth, whereas other species undoubt-edly survived to post-Columbian times, includingseveral species of rodent.

Island snails

Snail extinctions from islands appear to be largelyattributable to two factors—the loss of habitat, andthe introduction of exotic species, mostly preda-tory snails or ants. The endodontoid snails are tinytropical land snails, only a few millimetres indiameter. Over 600 species have been describedfrom the Pacific, but one in six of them is thoughtto have gone extinct during the twentieth century(Groombridge 1992). They are mainly grounddwellers in primary forest, and they are threat-ened by habitat loss and by introduced ants thatprey on their eggs. Other important island fami-lies are entirely or largely arboreal, e.g. thePartulidae, a restrictively Pacific family of some120 species, most of which are considered to bethreatened. On Hawaii, populations ofachatinelline snails have been lost because of theirlow fecundity combined with over-collecting andhabitat modification. In the Caribbean and NewCaledonia it is reported that land snails most atrisk are those in dry lowland forests, which maybe lost more rapidly than upland forest to

causes such as cattle grazing and other forms ofdevelopment. In New Zealand, there are manyvery localized endemic snails that are entirelydependent on native plant associations and whichare rapidly being lost. On Madeira, in the easternAtlantic, it is reported that of 34 species of landsnails present in samples of varying ages, 9 havebecome extinct since human settlement in AD 1419,principally, it seems, as a result of habitat loss(Goodfriend et al. 1994). Finally, there is evidenceof at least 19 extinct snail species from theCanaries, although there is some uncertainty as towhether these losses are attributable to humans orto past climate change (Izquierdo et al. 2004).

Griffiths et al. (1993) have studied the diet of theintroduced snail Euglandina rosea in Mauritius,where it was released in the hope that it would con-trol introduced giant African snails, Achatina spp.As in some other cases of failed biological controlschemes, the introduced predator ignored the tar-get species and instead gorged on native snailspecies. Some 30% of Mauritian snails are nowextinct. As most losses pre-dated the introductionof Euglandina, they have been attributed principallyto habitat destruction, but the exotic predator nowrepresents one of the major threats to the remainingendemic snail species. On Hawaii, the introductionof Euglandina has been implicated in the decline of44 species of endemic Achatinella; and it has been amajor factor in the extinction of Partula species inFrench Polynesia, 9 species being lost from Mooreaalone. Thus far, on Mauritius, primary forests donot appear to have been penetrated by the exotic

Table 11.7 Native West Indian land mammals known from living or fossil records fromthe past 20 000 years (source: Morgan and Woods 1986)

Order Total number Living Extinct % extinctionof species species species

Non-volant groupsRodentia 45 7 38 84Insectivora 12 2 10 83Edentata 16 0 3 100Primates 3 0 3 100

Volant groupChiroptera 59 51 8 14

Total 135 60 75 56

Page 329: Whittaker. Island Biogeography. 2007

species, and so habitat conservation remains key tothe conservation of the native species.

Plants in peril

With all the attention given to animals in thesepages, it is important to note, for the sake of bal-ance, that the problems also extend to plants.According to the IUCN 1997 Red List assessment(Walter and Gillet 1998), the total number ofconfirmed plant extinctions globally since AD 1600is 396, with a similar number of unconfirmedlosses, and perhaps as many as 33 500 threatenedspecies. Table 11.2 provides a figure of 384 extinc-tions. The equivalent figures from the 2004 IUCNRed List (�www.redlist.org/�;visited March2006) are 100 extinct, and 8321 threatened, whichamounts to a rather more conservative appraisal.The variation between these assessments high-lights the uncertainty associated with such efforts,and the need for continued efforts to improve theresolution of conservation assessments (cf.Whittaker et al. 2005). With plants, given the lack ofadequate species-level fossil data, it is particularlydifficult to assess extinctions from the prehistoricperiod on islands: there undoubtedly have been

numerous anthropogenic extinctions that willremain long undetected (for an example, seeFlenley 1993). Once again, the figures given inTable 11.8 should thus be taken as of largely indica-tive value. The table summarizes the 1997 Red Listdata on plant extinction and threat in islands orarchipelagoes that have lost more than 5 species.According to this assessment, anthropogenicextinctions of plants from the 12 most affectedislands/archipelagos total 120 species (30% of allthe extinct species), whereas 137 more were con-sidered possibly extinct. In global terms, these fig-ures indicate a disproportionate rate of loss andthreat concentrated on islands. Geographically, thepattern of vascular plant species loss is rather sim-ilar to that of vertebrate animals (Table 11.6), withfairly high numbers of possible or confirmedextinctions in Hawaii, the Mascarenes, theCaribbean, and St Helena.

St Helena provides a classic example of the dam-age to a remote oceanic island brought about byincorporation into European trading routes. It is anisolated island of 122 km2, located in the SouthAtlantic. Since discovery by the Portuguese inAD 1502, the vegetation has been completely trans-formed by browsing, grazing, erosion, cutting for

316 A N T H R O P O G E N I C L O S S E S A N D T H R E AT S TO I S L A N D E C O S YS T E M S

Table 11.8 Islands and archipelagoes that account for more than five extinct or possibly extinct vascular plantspecies since AD 1600, attributable to anthropogenic activities. Sources: Walter and Gillet (1998); Primack and Ros(2002); Groombridge and Jenkins (2002). As noted in the text, the 2004 IUCN assessment is less pessimistic on bothextinctions and threatened species. These figures should therefore be treated with caution

Islands No. of extinct No. of possibly No. of extinct + No. of threatened species extinct species possibly extinct sp. species

Hawaii 3 99 102 623Mauritius 38 4 42 256Cuba 23 0 23 811Madagascar 0 19 19 306French Polynesia 12 0 12 157Rodrigues 9 2 11 51New Caledonia 5 5 10 481St Helena 10 0 10 41Réunion 6 1 7 95Tasmania 7 0 7 142New Zealand 7 0 7 236Fiji 0 7 7 72

Total 120 137 257 3271

Page 330: Whittaker. Island Biogeography. 2007

PAT T E R N S O F L O S S AC R O S S I S L A N D TA X A 317

timber and fuel, the introduction of alien plants,and clearance for cultivation, plantations, and pas-ture (Cronk 1989). No native mammals occurred onthe island, as is typical of remote oceanic locations.The Portuguese introduced goats, pigs, and cattle,principally to supply homeward-bound carracksfrom India, and also brought with them rabbits,horses, donkeys, rats, and mice. Within 75 years oftheir introduction, goats had formed vast herds,devastating the vegetation.

The introduction of exotic plants began in earnestin the nineteenth century by way of British colonialbotanic gardens, in part to provide a new plantresource to replace the exhausted native cover.Large numbers of herb and shrub species wereintroduced, along with timber trees such as Pinuspinaster and Acacia melanoxylon. By the time the firstreliable botanical records were made in the nine-teenth century most of the destruction had alreadyoccurred, but some documentary evidence, and therelict occurrence of endemic plants in isolatedplaces such as cliffs, provide some indications ofthe former cover. Of the 46 endemic plants knownfrom the island, 7 are considered extinct, the restthreatened (Groombridge 1992; compare withTable 11.8). Some species are restricted to only a fewindividuals, and in the early 1980s Nesiota ellipticaand Commidendrum rotundifolium were reduced tojust single plants, the latter becoming extinct in thewild in 1986 (Cronk 1989).

Goats have been particularly important in pre-venting woodland regeneration by grazing andbarking saplings. So, as old trees were cut for fire-wood, areas became converted to a mosaic ofanthropogenic formations, notably grassland andOpuntia scrub. Erosion has been very rapid, and ondecadal timescales, isolated thunderstorms causesheet and gully erosion where the vegetation coverhas been removed by herbivores. After such ero-sion little vegetation cover can re-establish. Goatswere nearly shot out in the early 1970s and alldomestic goats are kept penned by law; however,sheep were still allowed to roam freely. Cronk(1989) considers that a process of rehabilitation ispossible, based on trial plantings of native species,but that initially, at least, plantings require protec-tion by fencing.

Turning to one of the showcases of plant evolu-tion, Wiles et al. (1996) provide estimates for Hawaiiof about 100 extinct taxa, 200 endangered or threat-ened, and a further 150 recommended for listing;however, 282 species appear on the list of federallyendangered plants (Cox and Enquist 2000), and yetanother estimate is given in Table 11.8. Althoughdifferent criteria generate different numbers, it isclear that a disproportionate number of Hawaii’splant species are in trouble: for instance, the figureof 282 contrasts with 486 on an equivalent list forthe mainland USA (Cox and Enquist 2000).Numerous introduced plant species have becomenaturalized on Hawaii, to the point at which theexotic flora outnumbers the endemic flora. Severalhave become serious pest species (Primack 1993).

In the Marianas (Micronesia) the forces involvedin the demise of native vegetation include loss offorest to agriculture; war and development; theintroduction of alien animals, plants, and plantpathogens; alterations to the fire regime; anddeclines in animal pollinators and seed dispersers.The impacts on several native plant species havebeen severe. Serianthes nelsonii (Leguminosae) isone of the largest native trees, and is endemic toRota and Guam. It formerly occurred on volcanicsoils but is now restricted to limestone substrates.Wiles et al. (1996) reported that just 121 adult treessurvived on Rota, and 1 on Guam. The populationswere senescent, with little or no regeneration occur-ring, and the main threats were from browsing byintroduced deer and pigs, plus infestations of her-bivorous insects (Wiles et al. 1996). On Guam, sev-eral plant species were found to be showing poorregeneration for these same reasons, and in onecase at least, germination problems limit re-estab-lishment. It appears that the seeds of Elaeocarpusjoga need to pass through the gut of a bird, yetnative frugivores have been wiped out by the intro-duced brown tree snake (Boiga irregularis)(Box 11.3). Another species, Pisonia grandis, is rarelyseen with fruit, perhaps because of the loss ofnative pollinators such as fruit bats or birds. Inaddition to impacting on seed set and germination,the loss of frugivorous birds and the reduction ofthe one extant pteropid fruit bat species to a rem-nant population of a few hundred animals have

Page 331: Whittaker. Island Biogeography. 2007

had a predictably marked effect on seed dispersalon Guam (Rainey et al. 1995). Thus, while habitatremoval is again a key driving force of plant reduc-tions and extinctions, a broad array of factors andtheir synergistic interactions is at work.

In illustration of how differing factors have pri-macy in different situations, we can turn to Tahiti,where the overriding threat to endemic plants cur-rently takes the form of another plant species. Tahitiis in the Society Islands (Polynesia), and is a large(1045 km2), high (2241 m) island, of diverse habi-tats, with a native flora of 467 species, 45% of whichare endemic (Meyer and Florence 1996). Some 70%of the endemic plants occur in the montane cloudand subalpine forests, which are some of thelargest, finest, and most intact to be found through-out the Pacific. The Polynesians colonized Tahitiperhaps 2000 years ago, bringing with them about30 domestic and 50 other accidental plant introduc-tions. European arrival in the AD 1760s led to mas-sive resource overuse, and habitat loss, as well as agreat flood of exotic plant species—in excess of1500 now being known for the Society Islands.Many of these have become naturalized, and somehave become problem plants. The lowlands andcoastal zones were totally transformed by people,and such native forests as remain on Tahiti areendangered by pressures such as housing, agricul-ture, forestry (e.g. based on exotic Pinus caribbea),feral animals, and hydroelectricity schemes.

Yet, the highlands of Tahiti remained essentiallyintact until the arrival of what has become theworst of all exotic ecosystem transformers, Miconiacalvescens, a small tree species introduced to theornamental gardens in 1937. In less than 50 years ithad spread to all the mesic habitats between 10 and1300 m ASL, including the montane cloud forest,often forming pure stands. It now covers over two-thirds of the island of Tahiti, and has spread to thenearby islands of Moorea and Raiatea. It is a poten-tial problem for other islands in the region, and isalready recognized as one of the 86 plant pests ofHawaii, having been introduced there by horticul-turalists as recently as the early 1970s. It is an effec-tive invader thanks to traits such as: tolerance to awide range of germination conditions, tolerance oflow light levels, fast growth, early reproductive

maturity, prolific year-round seed production, andefficient dispersal by small introduced passerinessuch as the silvereye (Zosterops lateralis). In Tahitialone, 40–50 endemic plant species are consideredto be threatened directly, including 8 species ofCyrtandra, 6 Psychotria, and 2 Fitchia.

What can be done to halt this decorative invader?One approach is to dig it out. In 3 years (1992, 1993,and 1995), nearly half a million plants wereuprooted by hand, and Miconia was cleared out ofits bridgehead sites on the island of Raiatea. In addi-tion to such plain labour, Meyer and Florence (1996)promote the need for a systematic assessment andinventory of alien and endangered native species,for research, monitoring, and control programmes,and for ex situ conservation in the botanic gardens—which still remains only an ornamental garden. Partand parcel of such an integrated approach to con-servation, as commonly advocated in other similarstudies, is a sustained educational programme.

The problem posed by invasive exotic plants maybe first and foremost for native plant species but, ifon sufficient scale, can also present a problem fornative fauna (Cronk and Fuller 1995). For instance,the now pantropical shrub Lantana camara formsdense impenetrable stands in parts of theGalápagos, threatening both some endemic plantspecies and the breeding habitat of an endangeredbird, the dark-rumped petrel (Pterodroma phaeopy-gia) (Trillmich 1992). Invasive plants may alter theenvironment in other, more subtle ways (Cronk andFuller 1995). The shrub Myrica faya, native toMacaronesia, was introduced to Hawaii in the latenineteenth century, and has since been spreadwithin the National Park, especially by an exoticpasserine, the white-eye (Zosterops japonica), and byferal pigs. It has a highly effective nitrogen-fixingsymbiont, which in part accounts for its ability toout-compete the native flora. By increasing nitro-gen inputs relative to native species by 400%, italters the nutritional balance of the Hawaiian soils.It poses considerable management problems and ispredicted to have additional long-term conse-quences because the altered soil nutrient status islikely to facilitate the spread of other alien species(Vitousek et al. 1987b; Cuddihy and Stone 1990;Vitousek 1990). For further details see Box 11.4.

318 A N T H R O P O G E N I C L O S S E S A N D T H R E AT S TO I S L A N D E C O S YS T E M S

Page 332: Whittaker. Island Biogeography. 2007

PAT T E R N S O F L O S S AC R O S S I S L A N D TA X A 319

Box 11.4 The invasion of Myrica faya in the Hawaiian islands

Of the more than 13 000 plant species that havebeen introduced to Hawaii, c.900 have becameestablished in the wild, and over 100 of themhave became serious pests (Cox 1999). Amongthe more problematic tree species are guava(Psidium cattleianum), also a serious pest in theGalápagos, miconia (Miconia calvescens), tropicalash (Fraxinus uhdei) and, especially, the fire tree(Myrica faya, also known as Morella faya). Myricafaya is an interesting example of an oceanic islandendemic species being introduced to otheroceanic islands on the other side of the globewith a similar environment. A valuable constituentand early pioneer tree within protected forests inthe moist zones within Macaronesia, it hasbecome a pernicious forest weed in Hawaii(Cronk and Fuller 1995).

In the Canaries, Myrica faya is the onlynitrogen-fixing tree species of the community, insymbiotic association with the actinomyceteFranckia, and it exhibits a persistent pioneerstrategy (Fernández-Palacios et al. 2004b). It is adioecious species some 20 m in height, possessingmedium-size fleshy fruits dispersed by indigenousbirds, such as the blackbird (Turdus merula).Myrica maintains a local soil seedbank, andindeed its seeds germinate only after theappearance of a large canopy gap. Thus, Myricaplays an important successional role in the returnof forest to previously cleared areas, but unlikesome other Canarian woody pioneers (Erica spp.)also persists after the gap is filled through itsability to re-sprout via suckers from the base(Fernández-Palacios and Arévalo 1998).

Myrica faya was introduced to Hawaii as early asAD 1880 and it was used extensively inreforestation, especially during the 1920s and1930s. In Hawaii its fruits are consumed and seedsdispersed by a large number of both native(Phaeornis obscurus—Hawaiian thrush) andintroduced birds (e.g. Zosterops japonica—theJapanese white-eye, and Carpodacus mexicanus—house finch), as well as feral pigs. Between the1960s and 1980s, aided by its ability to fix

nitrogen, Myrica rapidly invaded native forests ofMetrosideros polimorpha, which it out-competedin reaching the forest canopy and forming purestands (Vitousek and Walker 1989). Myrica becameinvasive within the Hawaiian Volcanoes NationalPark in 1961 and has now spread over more than35 000 ha. Today the new Myrica–Metrosiderosforest extends across a belt of 400–1200 m ASL onthe southern slopes of Hawaii (the Big Island),particularly on recent (� 1000 years) lava flows(Mueller-Dombois and Fosberg 1998), and it is awell-established and problematic invasive in all ofthe main Hawaiian islands.

Cronk and Fuller (1995) listed the followingecological characteristics of the fire tree as possiblereasons for its success as an invader: prolific seedproduction, long-distance dispersal by birds,nitrogen fixation, possible production ofallelopathic substances that may inhibit potentialnative competitors (specifically Metrosiderospolymorpha) and the formation of a dense canopyunder which native species are unable toregenerate. The impacts of Myrica on soilproperties are particularly important, with knock-on consequences for ecosystem form andfunction. Vitousek et al. (1987b) calculate that thenitrogen availability in an unaltered Hawaiianforest is about 5.5 kg/ha per year (mainly throughthe fixation ability of the native Acacia koa). Instands invaded by Myrica this may be increased upto 23.5 kg/ha per year. These substantially alterededaphic conditions, in combination with theactivities of feral pigs, have combined to enablethe invasion of many continental plant species.

Options for the control of Myrica infestationhave been the subject of considerable researchefforts, particularly in the search for a safebiological control agent. However, the mosteffective means of control at present appears tobe cutting and the use of foliar applications ofherbicide, both of which are only really practicalfor dealing with newly establishing populations.The species continues to spread throughout thearchipelago.

Page 333: Whittaker. Island Biogeography. 2007

Exotic plants may alter other ecosystem proper-ties, such as fire regimes or hydrological conditions.Examples of each can be drawn from Hawaii.Bunchgrass (Schizachyrium spp.) increases the flam-mability of the vegetation, and has preventedregeneration of native species that are not adaptedto burning. Andropogon virginicus is another exoticgrass, considered to be one of the most threateningof aliens. It not only carries fire, but also altershydrological properties. It occurs in disturbedgrassland and scrub on the island of Oahu in areason red clay soils where native forest vegetation hasbeen replaced by introduced woody and herba-ceous plants. Its seasonal pattern of growth andtendency to form dense mats of dead matter areresulting in increasing runoff and accelerated ero-sion (Cuddihy and Stone 1990).

11.7 How fragile and invasible areisland ecosystems?

It has become axiomatic to represent oceanic islandsas fragile ecosystems, which, because of their evolu-tion in isolation from continental biotas, are particu-larly susceptible to the introduction of alien species.Are island systems really particularly fragile or havehumans just hit them especially hard? D’Antonioand Dudley (1995) contend that the data supporttwo propositions. First, island ecosystems do typi-cally have a higher representation of alien species intheir biota than do mainland systems. Secondly, theseverity of the impacts of invasions appears in gen-eral to increase with isolation.

Cronk and Fuller (1995) suggest six reasons whyoceanic islands may be more susceptible toinvasion:

● Species poverty. This may mean that there ismore vacant niche space and less competition fromnative species. This is plausible, particularly wherehumans have disturbed habitats, but it is only onepart of the picture.

● Evolution in isolation. It is often assumed thatisland species are competitively inferior to conti-nental species as a resut of their long evolutionaryisolation. The case of Myrica faya, transported from

one set of oceanic islands, in Macaronesia, toanother, Hawaii, illustrates that the ‘continentality’of the land mass from which a species hails may notbe of great moment: simply to be from a differentbiogeographical pool may be sufficient. However,the evolution of flora and fauna in isolation, oftenwithout adaptation to grazing, trampling, or preda-tion by land mammals has led to the loss of defen-sive traits in many oceanic island endemics,expressed in plants as the absence of typical grazingadaptations such as spines, thorns, and pungent leafchemicals. In the presence of browsers these plantsare at a competitive disadvantage and tend to dis-appear, as was the fate of many endemic plantspecies on St Helena following the introduction ofgoats shortly after the island’s discovery in AD 1502.Similarly, many island endemic animals lack a fearof predators, and some birds have reduced flight orare flightless, and build their nests on the ground.These are traits that may have no competitive dis-advantage when faced with a species of the sametrophic level, but which have proven disastrouswhen faced with terrestrial vertebrate predators.

● Exaggeration of ecological release. Alien speciesgenerally arrive on islands without their naturalarray of pests and diseases, and this provides themwith an advantage over native species. A similarecological release mechanism is proposed as part ofthe taxon cycle in Chapter 9, and indeed alsoapplies to many introductions to continents.

● Early colonization. Particularly in the Atlantic,Caribbean, and Indian Oceans, islands were the firstlandfalls and first colonies of the Europeans, hencethey have had a long history of disturbance and ofintroductions. Although this is true, this line of rea-soning fails to account for the severity of impact ofparticular introductions such as the brown treesnake on Guam or Miconia on Tahiti: the outcomehas been a particular property of the speciesinvolved rather than the length of time over whichdisturbance and introductions have occurred.

● Small scale. The small geographical size ofislands means that their history is concentrated in asmall area, within which there are few physicalfeatures large enough to prevent exploitation,

320 A N T H R O P O G E N I C L O S S E S A N D T H R E AT S TO I S L A N D E C O S YS T E M S

Page 334: Whittaker. Island Biogeography. 2007

H O W F R AG I L E A N D I N VA S I B L E A R E I S L A N D E C O S YS T E M S ? 321

disturbance, and introduction on an island-widebasis. They also often sustain surprisingly highhuman populations (Watts 1987); for example,Cronk and Fuller (1995) note that Mauritius, with apopulation of 530 people/km2, may be comparedculturally and historically with India, which hasonly about 240 people/km2.

● Crossroads of intercontinental trade. Particularlyin the days of sailing ships, islands have often beenused as watering points, re-supply posts, and stag-ing posts for trade and for plant transport. As theyhave often changed hands between different colo-nial powers, they have, moreover, had opportuni-ties to receive introductions from differing networksof nations and biogeographical regions.

The deluge of disturbance and of alien specieswhich has swept across so many oceanic islandsgoes some way towards explaining the apparentsusceptibility of islands to invasion. At least onetropical biologist, who has worked for many yearson Hawaii, argues that this is the case:

Contrary to common opinion, many endemic islandspecies are strong competitors. They would not be elimi-nated from their niche in the island ecosystem if it werenot for the new stress factors introduced by man. Theisland species evolved with stress factors associated withvolcanism, fire in seasonally dry habitats, and occasionalhurricanes. The effects of volcanism resulted in superioradaptation of many native species to extreme edaphicconditions existing on volcanic rockland, where soil-waterregimes fluctuate almost instantly in direct relation torainfall. (Mueller-Dombois 1975, p. 364)

This general position is supported by D’Antonioand Dudley (1995) who conclude that, with excep-tions, island habitats are not inherently more easilyinvaded than are continental habitats, but that,nonetheless, exotics have had a greater impact onisland systems. Some more formal analysis pro-vides a measure of support for Mueller-Dombois’sviews. It has been suggested that as a rough rule(the tens rule), about 1 in 10 introductions intoexotic territories becomes established (Williamson1996). The data for Hawaiian birds indicate a farhigher success rate, of over 50% of introducedspecies establishing. However, Williamson (1996)

notes that the majority of introductions havecolonized disturbed lowland habitats, and that thefigure for species establishing into more or lessintact native habitats (some lowland forest,together with upland habitat) falls at between 11and 17%, which is not significantly out of line withthe tens rule. Studies elsewhere within verydisturbed islands, containing large numbers ofnaturalized exotic species, have also reported thatthose habitats that have remained largelyundisturbed quite often contain few exotics (e.g.Watts 1970, on Barbados; Corlett 1992, onSingapore). These examples suggest that thesusceptibility of island ecosystems to invasion isabove all a reflection of the profound alteration ofhabitats. However, other case studies cited in thischapter demonstrate that some invaders have notrequired such assistance.

One of the curious features of particular inva-sions is that invading species sometimes undergo apattern of increase to a peak of density, followed bya crash, sometimes to extinction. This boom-and-bust pattern appears to have occurred fairly oftenon islands (Williamson 1996). It has, for instance,been recorded for a number of the passerines intro-duced to the Hawaiian islands, with their time toextinction varying between 1 and 40 years. Theremay be a number of reasons for these boom-and-bust patterns, but Williamson favours those relat-ing to resource deficiencies as providing the mostgeneral explanation.

From the point of view of island biogeographytheory it is noteworthy that the introduction of asingle exotic species, such as the brown tree snake(Boiga irregularis) on Guam and the shrub Miconiacalvescens on Tahiti, may cause the local extinctionof numerous native island species. Moreover, in thecase of Miconia, the effect operates within the sametaxon, i.e. it is a single species of plant which isimplicated in the loss of numerous native plants.This provides an instance in which an equilibriummodel, involving turnover on a replacement basissubject only to relaxation effects from area reduc-tion (as discussed in Chapter 10), would be awholly inappropriate device for predicting specieslosses.

Page 335: Whittaker. Island Biogeography. 2007

11.8 Summary

As a result of human action, island birds have been40 times more likely to become extinct within thepast 400 years than continental birds. In the sameperiod, the proportion of documented animalextinctions from islands varies from about 60% formammals to 95% for reptiles. Moreover, the lossescaused by humans in prehistoric times exceed thoseof the past 400 years. This is most clearly estab-lished for birds, but applies to other animal taxaand to islands across the world. Island plants havealso been subject to a disproportionate rate of loss.The scale of environmental degradation on certainPacific islands, exemplified by Easter Island andHenderson Island, has been shown to have been sogreat that it led to complete cultural collapse and, inthe latter case, human abandonment of the island.The available evidence indicates that the key turn-ing point in the transformation of island ecologiesis typically the first colonization by humans, butthat successive waves of contact with the outsideworld, and of innovation in resource use, can con-tinue to ramp up the pressure on island biodiver-sity, sometimes generating multiple phases ofextinction.

The major causes of local and global extinctionfrom islands have been: (1) predation by humans,(2) the introduction of exotics, (3) the spread of dis-ease, and (4) habitat loss, of which habitat loss andthe introduction of alien species currently givegreatest overall cause for concern. Predationremains a critical driver of extinction threat for asubset of threatened species and islands. The intro-duction of exotic species, in some cases a single

exotic species, can cause the local extinction ofnumerous native island species. The shrub Miconiacalvescens on Tahiti and the brown tree snake (Boigairregularis) on Guam provide two classic examples.However, case details provided in this chapter fromthe Pacific, Indian Ocean, Caribbean, and Atlanticillustrate that while some losses can indeed beattributed to a sole major cause, many specieslosses have resulted from synergisms betweenseveral causes, such as habitat alteration allowingthe invasion of exotic plants, which in turn arespread by exotic birds, which in turn carry exoticdiseases. The term trophic cascade is used todescribe the situation where the loss of one or asmall number of species triggers the disruption ofecosystem processes more generally. Such knock-oneffects can work through predatory relationshipsand, as just indicated, through other forms of ecological networks, such as pollination ordispersal networks.

While the big problem for continental organismsis the increasing insularization of their habitats, theproblem for oceanic island biotas in recent historyand today is the increasing breakdown of the insu-larity of theirs. As a result, many oceanic islandsnow have a high representation of exotic species. Itdoes not necessarily follow that island habitats,if undisturbed, are always more invasible thancontinental habitats. However, exotic invadershave had a significant impact on many islandecosystems, and this susceptibility to exotic impactcan be related both to peculiarities of human use ofislands and to evolutionary features of the islandbiotas, especially to their evolution in the absenceof terrestrial mammalian browsers and predators.

322 A N T H R O P O G E N I C L O S S E S A N D T H R E AT S TO I S L A N D E C O S YS T E M S

Page 336: Whittaker. Island Biogeography. 2007

Although generally distinct from continental environ-ments, and therefore of interest in their own right,island environments have been generally misunderstood,misinterpreted, and mismanaged. The main reason is thatthey have been interpreted for management purposeslargely by continent-trained observers . . . islands are notsimply miniature continents and . . . continental solutionsdo not simply need to be scaled down in order to besuccessful.

(Nunn 2004, pp. 311 and 319)

In this final short chapter, we first illustrate some ofthe special problems facing contemporary islandsocieties, the pressures on the environments andbiotas of oceanic islands, and constraints withinwhich those interested in promoting conservationmust operate. Conservation in island archipelagoslike the Canaries, the Maldives, the Lesser Antilles,and the Galápagos present distinctive challenges,which arise from the characteristics of the islandenvironments and biota and from the socio-economic and political situation of the islandsocieties (e.g. Nunn 2004). Therefore, secondly, welook at some of the conservation solutions thathave been promoted in the context of the specialcircumstances of remote islands. Solutions ‘para-chuted in’ from the outside without regard to thiscontext are unlikely to produce the desiredoutcomes.

12.1 Contemporary problems on islands

The previous chapter illustrated some of thecommon elements of the problems afflicting islandbiodiversity: habitat loss/degradation and the role

of introduced species foremost amongst them. Westart this section, by contrast, with a short selectionof case studies drawn from different regions of theworld, which illustrate some of the profoundly dif-ferent circumstances and issues that may be of con-cern in different island groups. Although our focusthroughout this final part of the book has been onbiodiversity, we felt it instructive to include at thestart of this chapter a couple of examples of islandswhere the loss of endemic species is not the centralconcern. As will become clear, however, whetherspecies extinctions are at issue or not, the loss ofecosystem goods and services is, in the end, alwaysof concern to island societies.

Maldives: in peril because of climatic change

The Maldives form an extended equatorial archi-pelago of about 1200 small, low coral atolls, extend-ing for 800 km in a line across the Indian Ocean.Being small, low islands they have a low rate ofendemism, with just five endemic plant species (allin the genus Pandanus) within a native flora of 277species (Davis etet alal. 1995). They have a total landarea of about 300 km2 and a population of about350 000 inhabitants. The foreign exchange economyis based almost exclusively on the naturalness andbeauty of these islands, which have become a desti-nation for so-called ‘quality’ tourism.

Together with the Lakshadweep Islands to thenorth and the Chagos Islands to the south, theMaldives form part of a vast submarine mountainrange, on the crest of which coral reefs have grown.Actually, the term ‘atoll’ is derived from the native

323

CHAPTER 12

Island remedies: the conservationof island ecosystems

Page 337: Whittaker. Island Biogeography. 2007

name atholhu, applied to these islands, which areexamples of this structure. It is this mode of originthat explains their low elevation.

Projected twenty-first-century sea-level rise result-ing from global climatic warming could see the com-plete inundation of Malé Atoll, the capital and mostheavily populated island, if not of the entire archi-pelago, which has a maximum elevation of just 2 mabove sea level. Even a lesser rise could be sufficientto jeopardize the continued existence of this societyand its unique culture, along with numerous othersimilarly distinctive societies living on low-lyingislands around the world (Nunn 2004). With fewmitigation options open to them, the Maldivian gov-ernment, with Japanese aid, has undertaken thebiggest project ever seen in the archipelago, the con-struction of sea walls around the island of Malé.

Okino-Tori-Shima: the strategic economicimportance of a rocky outcrop

As islands disappear under the sea by erosion, sub-sidence, or sea-level rise, so do their ExclusiveEconomic Zones (EEZ), the 200 nautical mile exten-sion around a land mass, whereby a nation retainsuse rights to undersea resources, primarily fishingand seabed mining. This aspect of international lawaccounts for the interest countries have in claimingand holding on to desolate chunks of rock in theocean.

A prime example of this is shown in the actions ofJapan with regard to Okino-Tori-Shima (also knownas Parece Vela), Japan’s southernmost island,located some 1750 km south of Tokyo. The island isa largely submerged atoll, which at high tide con-sists of just two rocky outcrops jutting slightly morethan 60 cm above the waters of the Pacific (Chang2000). Typhoons rip through this region annually,and their destructive impact, alongside regularwave action, threatened to submerge the atollaltogether. Because this atoll is the basis for an EEZof 400 000 km2 (larger than Japan itself), of very richfishing waters (alongside mineral rights), Japancould not allow this island to disappear, and in1988 millions of dollars were spent in constructingiron and concrete breakwaters to prevent the loss ofthe atoll—so far successfully.

By contrast, two uninhabited islands, TebuaTarawa and Abanuea (ironically meaning ‘thebeach which is long lasting’), part of the islandnation of Kiribati, were lost in 1999. If sea-level con-tinues to rise over the twenty-first century, as fore-cast, then the further loss of territory in this waywill require the recalculation of the EEZ of Kiribati,which currently constitutes over 3.5 million squarekilometres, almost 5000 times the land area, andequivalent to over a third of the land area of theUSA (Chang 2000).

These examples serve to illustrate the importanceof the loss of land area to island nations, even whenthe area lost is not particularly important in termsof human habitation.

Nauru: the destruction of an island

The island of Nauru (once known as Pleasant Island)is located in the Pacific Ocean, about 500 km north-east of New Guinea. It has a total land area of 21 km2

and a population of about 13 000 as of 2006. Thecoast is fringed by a ring of sand, surrounded by aprotective reef, but the interior is formed of guanodeposits that have mixed and solidified with a corallimestone base. The mining of these deposits hasdevastated the island environmentally, and despitethe initial economic returns, has created severe finan-cial, legal, and cultural problems for the Nauruans.

Guano is a rich source of phosphate, which is nat-urally scarce and of high value as a key constituentof fertilizer. Phosphate mining began in 1908 whenthe Germans, then in control of Nauru, began tomine the island’s substantial deposits. BetweenWorld War II and Nauru’s independence in 1968,Australia became the Administrator of the island.The mining operation in this period produced2 million tons yearly, which was mainly exported toAustralia and New Zealand. Phosphate mining hasbeen concentrated in the central plateau, where themining has left behind deep pits and tall pillars,creating a moon-like scene of barren wasteland,which now comprises over 80% of the island area,with the residents living on a narrow strip along thecoast. As a result of the mining, the vast majority ofsoil and vegetation has been stripped away, makingagriculture impossible.

324 I S L A N D R E M E D I E S : T H E C O N S E R VAT I O N O F I S L A N D E C O S YS T E M S

Page 338: Whittaker. Island Biogeography. 2007

With independence from Australia, the Nauruansbegan to experience the financial benefits of phos-phate mining for the first time. They became richvirtually overnight, creating one of the world’shighest per capita incomes. The new Republic filedsuit with the International Court of Justice againstAustralia in 1989, asking for compensation for theenvironmental damage sustained prior to inde-pendence. Five years later, a settlement wasreached whereby Australia agreed to pay morethan 100 million Australian dollars (Pukrop 1997).The money earned may have been considerable,but the phosphate extraction begun by the colonialpowers and continued by the Nauruans hasdestroyed the traditional and cultural way of life.They no longer practice agriculture or fishing, rely-ing instead on pre-packaged imported fatty foods,which have decreased the life expectancy, in thecase of men to just 55 years (Economist 2001).Phosphate production peaked in the 1980s, andthis, combined with a collapse in the market price,has undermined the main income source for theisland. Moreover, the financial bonanza was badlymishandled, and the money was squandered.Offshore banking was the next big idea, but thisshady trade with next to no internal regulation didnothing for the international reputation of Nauru,which soon found itself in trouble with interna-tional regulators.

With the end of the mining operations in sight,the attention of the island government turned to thefuture rehabilitation of the island. With an expand-ing population crowded into the coastal strip, theNauruans need more space for new buildings,including a hospital and schools. This developmentcan only occur in the central plateau, which, how-ever, remains a barren wasteland of limestone andcoral. One solution is to crush the pillars and toimport topsoil, humus, and other nutrients, thusbeginning a long process of rebuilding the ecosys-tem. The cost is estimated to be double the compen-sation received from the Australian government,and it could take more than 30 years to completethe scheme. Given the worsening financial situationthat the island now finds itself in, such ideas seemunlikely to come to fruition. As The Economist (2001)puts it ‘It is a melancholy sign of the islanders’

desperation that the idea of simply buying anotherisland and starting afresh is once again under dis-cussion. But who in his right mind would let theNauruans get their hands on another island?’ Thismay seem a rather unsympathetic remark, but isfair comment on the quality of leadership sinceindependence, as this once isolated island societyhas attempted to come to terms with the disastrousresults stemming from its interactions with theglobal economy.

The Canaries: unsustainable development in anatural paradise

The Canarian archipelago is one of the most bio-diverse areas within Europe (or at least within theEuropean Union, given that the islands lie closer toAfrica than to Europe). The islands possess morethan 12 500 terrestrial and 5 500 marine species in, oraround, a land area of only 7500 km2. Within thistotal of 18 000 or more species, about 3800 speciesand 113 genera are endemic (Izquierdo et al. 2004).Among them are many examples of spectacular radi-ations of animals (e.g. more than 100 species ofLaparocerus weevils) and plants (e.g. 70 species of suc-culent rosette-forming members of the Crassulaceae).Furthemore, an average rate of description of onespecies new to science every 6 days over the last20 years, suggests that the catalogue of Canarianspecies is still far from complete (Fig 11.1).

The Canaries were first settled by the Guanchepeople, believed to be of Berber stock, around thefirst century BC. The archipelago has been underEuropean influence since the fifteenth century, withextensive deforestation of the mesic zones, goatgrazing of the less productive land, and intensiveagricultural use of whatever land could be put intoproduction. Between 1960 and 1970, as a result ofcheap air travel, the pattern of exploitation of natu-ral resources began perhaps its third major phase,shifting from an agricultural society to a masstourism model.

This tourist boom has given rise to very abruptsocio-economical and cultural changes (Table 12.1),with profound impacts for the natural and semi-natural ecosystems of the archipelago. Some 12 -million tourists now visit the Canaries each year,

C O N T E M P O R A RY P R O B L E M S O N I S L A N D S 325

Page 339: Whittaker. Island Biogeography. 2007

and current population growth, fuelled largelyby immigration, is occurring at the rate of about50 000 a year. The upshot is human populationdensities of about 550/km2 in Gran Canaria and450/km2 in Tenerife. The increased resident andtransient population increasingly demands morespace for houses, hotels, roads, and infrastructure,with increasing demands for food, goods, water,and other resources, while simultaneouslyproducing more waste (e.g. García-Falcón andMedina-Muñoz 1999). It can be estimated that on adaily basis the average Canarian resident con-tributes more than 20 kg CO2 to the atmosphereand produces 5 kg waste (Fernández-Palacios et al.2004a).

In terms of the indigenous resource base, thispattern of development is clearly unsustainable(Fernández-Palacios et al. 2004a). The arid regionsare increasingly being built over, as land is gobbledup by the development instigated by the touristsector. Coastal ecosystems have been fragmentedby buildings, harbours, roads, and golf courses. Thewater table has been depleted as the waterresources have increasingly been mined, and pres-sures have also increased on fisheries in the inshore

waters. As a by-product of tourist-driven urbaniza-tion, about half the agricultural area (50 000 ha) hasbeen abandoned. Agriculture in the Canaries is ofnecessity a labour-intensive activity, with cultiva-tion focused in small units not amenable to large-scale mechanization: easier and seemingly moreattractive livelihoods are to be found in the urbanareas.

The upshot of these changes has been a whole-sale shift in the nature and geography of the humanimpact on Canarian landscapes, alongside a shiftfrom the archipelago acting as a net exporter to anet importer of food. On Tenerife, the upperregions, which were once devastated by overgraz-ing and cutting, have now largely been handedover to replanting (especially in the endemicCanary Island pine belt) and to conservation. Thelaurel forest belt, although reduced to perhaps 10%of its original area, is stable in area and has pro-tected status. Many areas once in cultivation in theless arid parts of the lowlands are gradually beingrecolonized by a mix of native and exotic plants(Box 12.1), or else built on. As current mass tourismfavours the sunniest environments, it is in the drylowlands that the pressure is now greatest, with

326 I S L A N D R E M E D I E S : T H E C O N S E R VAT I O N O F I S L A N D E C O S YS T E M S

Table 12.1 The shift of the economic development model (1960–2000) on the Canaries (Source: Fernández-Palacios et al. 2004a).

Property 1960 1970 1980 1990 2000

Population (M) 0.94 1.17 1.44 1.64 1.78Number of tourists (M) 0.07 0.79 2.23 5.46 12.0Population density (inhabitants/km2) 130 155 189 206 231Cultivated area (K ha) 95 68 60 49 46Oil consumption (K oil equiv. ton.) – 827 1442 2473 3155Electric energy consumption (GW) – 890 1680 3423 6292Concrete consumption (M ton.) – 0.76 1.22 1.57 2.65Number of cars (M) 0.02 0.08 0.28 0.5 1.08Active population in agriculture (%) 54 28 17 7 6Active population in services (%) 27 46 55 62 70Unemployment (%) 2 1 18 26 13Female life expectancy (y) 65 75 77 80 82Literacy (%) 36.2 – 91.7 95.7 96.4Per capita income (K dollars) 4.3 8.8 11.4 15.4 17.2

Page 340: Whittaker. Island Biogeography. 2007

C O N T E M P O R A RY P R O B L E M S O N I S L A N D S 327

Box 12.1 Invasive plants in the Canaries: the incorporation of prickly pear (Opuntia) into thelandscape

During the nineteenth century, several prickly pearspecies (Opuntia, Cactaceae) of Mexican originwere introduced into the Canaries, principally forthe natural red-purple colorant (carminic acid)produced by the females of the parasitic insectDactylopius coccus (Coccoidea, Homoptera), alsointroduced from Mexico, which feed on theirsucculent leaf-shaped stems. True cacti are notnaturally found outside the New World. Opuntiarapidly naturalized and became invasive within thenatural and semi-natural ecosystems of theCanarian lowlands (Otto et al. 2001, 2006), wherethey found similar arid, warm conditions to theiroriginal distribution. One of the keys to the successof prickly pears as invaders on the Canaries is theyear-round availability of their large, fleshy, sweetfruits, which have proved attractive to both nativeand exotic dispersers (Chapter 11). They alsoregenerate and spread vegetatively, from fallingleaves taking root: behaviour not found in thenative Canarian Euphorbia, which are theindigenous dominants of this scrub vegetation.Finally, another clue to their success is that whereasall but one species of Euphorbia lose their leaves insummer drought conditions, Opuntia are still ableto photosynthesize, using the crassulacean acidmetabolism (CAM) photosynthetic pathway, whichagain is not possessed by the local Euphorbiaspecies.

Although several prickly pear species grow wildtoday on the Canaries (O. ficus-barbarica,O. dillenii, O. tomentosa, O. tuna, O. robusta, andO. vulgaris) (Izquierdo et al. 2004) only two ofthem (O. ficus-barbarica and O. dillenii) are widelydistributed. O. dillenii form part of the subdesertcoastal scrub of the islands, whereas O. ficus-barbarica thrives in the disturbed zones at midaltitudes, where it forms dense, almostmonospecific patches. Their succulent leaves wereeaten by goats, providing them with a valuablewater intake, whereas their fleshy fruits are notonly eaten by humans, but also by endemic

Gallotia lizards (Valido and Nogales 1994) and several native birds, especially the raven(Corvus corax). Nogales et al. (1999) found O.ficus-barbarica to be the most common seed in the pellets regurgitated by ravens, and both O. ficus-barbarica and O. dilleniishowed improved germination after passingthrough raven guts. Furthermore, Sturnusvulgaris has been observed feeding on theparasitic insect Dactylopius coccus growing on the prickly pear (Martín and Lorenzo 2001).Opuntia thus provides an illustration of howexotic plant species can compete successfully with native and endemic species, whilst in this case simultaneously contributing to the food resources of both native and exotic animal populations.

Besides Opuntia, other invasive plants competing successfully in otherwise well-preservedCanarian ecosystems include Pennisetumsetaceum, Nicotiana glauca, and Agave americana,in the subdesert coast scrub; Cytisus scoparius,Ulex europaeus, Tradescantia fluminensis,Ageratina adenophora, Ailanthus altissima,Zantedeschia aethiopica, Eschscholtzia californica(Californian poppy) in the native laurel and pineforest ecosystem and several Acacia species atdifferent altitudes. Altogether, alongsideapproximately 1300 native plant species in theCanaries (see Table 3.2), Izquierdo et al. (2004) listsome 673 plant species as being introduced (orprobably introduced), of which 79 are currentlyconsidered invasive. This amounts to 11.7% of thetotal introduced set, in line with Williamson’s (1996) so-called ‘tens rule’ (Chapter 11).

Today, after several centuries of presence in theCanaries, prickly pears are considered part of theCanarian landscape and no attempt has beenmade to eradicate or control them. In contrast, a lot of money has been invested in ineffectiveefforts to control some other exotics, such as thegrass Pennisetum setaceum.

Page 341: Whittaker. Island Biogeography. 2007

some of the biggest tourist developments severelyimpacting on the most arid areas, previously onlysparsely populated.

Contemporary problems in the Galápagos: athreatened evolutionary showcase

On the Galápagos islands, the main terrestrial con-servation threats include introduced mammals(goats, pigs, cats, rats, dogs, cattle and donkeys),aggressive alien plants (e.g. Lantana camara, Cedrelaodorata, Cinchona succirubra, Rubus sp., and Psidiumguajava), habitat fragmentation, agriculturalencroachment, over-exploitation and replacementof native woody species, and fires (Jackson 1995;Desender et al. 1999; Cruz et al. 2005). The possibil-ity of the introduction of exotic disease, as hasafflicted Hawaii’s honeycreepers (Chapter 11), alsoconstitutes a latent threat. The problems of theGalápagos are not solely land-based. Overfishing(e.g. for sea cucumber and lobster) and illegal fish-ing (especially for shark) by the artisanal fishingfleet in response to growing overseas markets (e.g.Merlen 1995) have led to the virtual collapse of themost lucrative legal fisheries and have resulted inviolent and prolonged disputes between fishermenand conservationists.

As has happened in other island archipelagos, thegrowth of the tourist sector has brought mixed ben-efits to the Galápagos. Although many touristswould count themselves environmentalists, theirinterest in the islands has stimulated the local econ-omy and boosted immigration. This populationgrowth has, in turn, brought additional pressures tobear on the environment and biota, further fuellingconflicts between pro-sustainability and pro-develop-ment and extraction interests (Trillmich 1992; Daviset al. 1995).

It has been estimated that tourism in theGalápagos generates in the order of US $350 millionper year but that over 90% of it is retained in thenational economy rather than locally. The numberof visitors to the islands has increased from about41 000 in 1990 to about 110 000 in 2005, most ofwhom are non-Ecuadorians. Each foreign visitorpays an entrance fee of $100 (as of 2006), and the

additional revenue is generated through tours,boats, hotels, and other local services (ScottHenderson, personal communication). Given thetotemic significance of the Galápagos in the develop-ment of evolutionary theory, it might be anticipatedthat there would be a matching conservation focusand effort. Indeed, the Galápagos National Parkcovers 97% of the archipelago and the island havealso been designated a World Heritage Site, aBiosphere Reserve, and a priority in schemes suchas the 2004 CI hotspots programme.

Research in the park is coordinated by theCharles Darwin Research Station and the ParkService. The management of the park aims to pre-serve the biological uniqueness and intactness ofthe islands and surrounding seas, through regulat-ing resource use and research activities. A compre-hensive zoning system is a key tool in bothterrestrial and marine realms. Introduced animalsremain the greatest problem, and although feralgoats have been eliminated completely from anumber of smaller islands—with notable recoveryof vegetation—the destruction brought about bypigs, goats, and donkeys on the large island ofSantiago has been so great that it is doubtedwhether a full recovery is now possible (Davis et al.1995). Pigs eat plants, invertebrates, the eggs andhatchlings of endemic tortoises, lava lizards, andGalápagos petrels, amongst other species. As aresult of one of the most ambitious conservationprogrammes, the last pig was eliminated from thelarge island of Santiago in 2000, at the end of a30 year campaign. Santiago is now believed to befree of goats too (Campbell and Donlan 2005) andnone has been spotted recently on Isabela, follow-ing a campaign in which over 100 000 were eradi-cated. Over the years, other ambitious conservationinitiatives have been launched, the goals of whichhave included fencing against feral mammals,inventory of the worst alien plants, development ofcontrol and eradication methods, a programme todevelop sustainable timber use, and an ex situbreeding programme for threatened plants to com-plement in situ efforts.

In May 1989, seven feral goats were observed onthe Alcedo volcano on Isabela: the first confirmed

328 I S L A N D R E M E D I E S : T H E C O N S E R VAT I O N O F I S L A N D E C O S YS T E M S

Page 342: Whittaker. Island Biogeography. 2007

report of goats in the habitat of the most intact raceof Galápagos tortoise. By 1997, the goat populationwas close to 100 000, and the habitat of the tortoisesand other fauna on Alcedo was reported to be col-lapsing. It is indicative of the problem facing manyconservationists that appeals for donations tofinance the goat control programme had to beissued, for example, in the journal ConservationBiology, as the Charles Darwin Foundation at thattime lacked the core funding needed (Herrero1997). One former director of the Charles DarwinResearch Station, Chantal M. Blanton, noted withdismay that the national financial commitment toconservation research in the Galápagos slipped tozero in 1995, while tourist concessions doubledbetween 1992 and 1996 (Galápagos Newsletter 1996).When asked to name the most important problemsfacing the islands, she listed: first, introduced ani-mals and the lack of a functioning quarantine sys-tem; secondly, inadequate political and financialsupport from government; and thirdly, the politi-cizing of the National Park Service and of postsrequiring technical expertise. Although recent suc-cesses in reducing and in some cases eradicatinggoats and pigs from islands in the Galápagos archi-pelago are greatly encouraging—it seems thatmethods now exist for efficient eradication pro-grammes (Box 12.2)—funding remains a key limit-ing factor in island goat eradications generally(Campbell and Donlan 2005).

The basic conservation problem of the Galápagosis the ever-accelerating breakdown of the islands’former isolation as a result of of dramatic changes inhuman population size, activity patterns, andmobility (Trillmich 1992). The real impact of thetourist business occurs not through the direct dis-turbance to animals by tourists armed with camerasbut indirectly via the socio-economic changes on thefour inhabited islands. The resident human popula-tion has grown rapidly due to the tourist and fish-ing industries, from about 10 000 in 1990 to 28 000by 2005. This growth has come largely from immi-gration from mainland Ecuador, with migrants gen-erating increased pressure on infrastructure and anongoing influx of alien species. Indeed, in a 5 yearperiod, about 100 new plant species were detected,

and although this sudden rise reflects new effortsby botanists to record non-native and invasivespecies (Tye 2006), there is no doubt that theincreased flow of people from the mainland hascontributed greatly to the non-native flora, as peo-ple have brought in ornamental plants in the urbanareas, and have tested new agricultural plants inthe upland areas, from where several have alreadyspread to become ecologically and economicallyexpensive invasives (Scott Henderson, personalcommunication).

The growth of the human population has meantmore pets. A survey in the main settlement on SantaCruz established that the dog population was about300 at the start of the 1990s, with unwanted pupsbeing released to become feral. Feral dogs prey ontortoise hatchlings and eggs, and in the late 1970swild dogs decimated a large colony of land iguanason Santa Cruz, killing over 500 animals (Jackson1995). Remnants of endemic birds, reptiles, andeven insects in cat faeces bear witness to theimpacts these animals have in both feral anddomesticated conditions. Although the NationalParks Service is working to keep the feral popula-tion low, it will be difficult to reduce the number ofdomestic dogs as long as robberies in the villagescontinue to increase, as this creates an incentive forsettlers to keep dogs (Trillmich 1992). An intensivecommunity-based programme is underway toincrease owner responsibility through educationprogrammes, designed to demonstrate the long-term benefits of maintaining the archipelago’sunique biological diversity and heritage (ScottHenderson, personal communication). The casearticulated by numerous conservationists overmore than 25 years for a quarantine system is over-whelming. But, in common with feral animal con-trol and other conservation measures, this will onlyultimately be successful with sustained politicaland societal support.

12.2 Some conservation responses

Adsersen (1995) notes the prevalence of the sameinvasive plants in archipelagos as diverse andremote from one another as the Mascarenes, the

S O M E C O N S E R VAT I O N R E S P O N S E S 329

Page 343: Whittaker. Island Biogeography. 2007

330 I S L A N D R E M E D I E S : T H E C O N S E R VAT I O N O F I S L A N D E C O S YS T E M S

Box 12.2 The arrival and eventual removal of ecosystem transformers: feral goats and Judas goats

Goats in the CanariesIntroduced from North Africa in the firstmillennium BC by the Guanches, goats (Caprahircus) enjoyed a feral status on all of the CanaryIslands until just a few decades ago. Today herdsof semi-wild goats roam freely only onFuerteventura. With the exception of the nowextinct giant tortoises (Geochelone burchardii andG. tamaranae), the Canarian flora evolved withoutlarge grazers for several million years, and thusgenerally lacked the defensive spines and toxiccompounds typical of continental flora. The longhistory of goat grazing has thus undoubtedlybeen hugely significant in transforming thevegetation cover of the islands, and most likely inreducing native endemic species to extinction. It isdifficult to be certain how many speciespopulations were lost in the absence of fossildata, but it is clear that goats and otherintroduced herbivores have been directlyresponsible for reducing some persisting endemicsto the very brink of final extinction (e.g. Marrero-Gómez et al. 2003), just as also occurredfor example on St Helena (Cronk 1989).

Following the declaration of the summit regionof Tenerife as a National Park in 1954, goats wereremoved from the uplands. After the shift of theeconomic development model from agriculture tomass tourism during the 1960s, the goats slowlybegan to disappear from the rest of the Canarianlandscape. Finally, in 1980, the last wild goatswere successfully extirpated from the uninhabited10 km2 island of Alegranza, marking the end(except on Fuerteventura) of an influence thatlasted for more than two millennia. In response tothe removal of the goats, in those areas escapingdevelopment, a trend of vegetation recovery canbe seen. In the extreme (arid) climates of the highuplands this has not been a particularly rapidprocess, but nonetheless, several rare plantspecies have since greatly increased in abundance(e.g. the endemic Pterocephalus lasiospermus).Conservation efforts have focused on assessingthe key life-history phases and autoecological (e.g. germination) requirements of highlythreatened plant species. Breeding programmes

have been established and efforts are being madeto reintroduce populations into the wild. Thisapproach requires a good and sustained level offunding, alongside the availability of suitablytrained scientists, appropriate infrastructure, andfinely tuned management plans for protectedareas (e.g. see Marrero-Gómez et al. 2003).

Removal of feral mammalsGoats have been introduced on numerous islandsaround the world, sometimes to provide a foodresource on uninhabited islands for visiting sailors,other times by settlers. They are highly effectiveecosystem transformers, and present hugeconservation problems in many islands. However,they have now been eradicated successfully fromover 120 islands across the world (Campbell andDonlan 2005). Whereas in the past, such campaignstypically took many years to complete and oftenstruggled to eradicate the population altogether,modern technology has now been harnessed toenable swift and effective eradication of populationsof goats numbering many thousands.

An illustration is the use of so-called Judas goats.Animals are trapped and radio-collared beforebeing re-released, upon which (being socialanimals) they typically find and remain with othergoats. They are then tracked and the other animalsaccompanying the Judas goat are shot with high-powered rifles. The Judas goat is then releasedagain, and the process repeated. This approachallows the last members of the feral population tobe tracked down and eliminated. Similarly, feralpigs, which were causing huge ecological impactson Santiago island in the Galápagos, were finallyeliminated from the island at the end of a 30 yearcampaign, during which some 18 000 animalswere killed (Cruz et al. 2005), with many lessonslearnt towards increasing eradication efficiency infuture efforts.

Although some environmentalists raiseprincipled objections to such culling programmes,there is no doubting their central importance inprotecting the native biota of many oceanicislands (Desender et al. 1999, Campbell andDonlan 2005, Cruz et al. 2005).

Page 344: Whittaker. Island Biogeography. 2007

Galápagos, the Canaries, and the Bahamas.Similarly, many other problems, such as feral ani-mals, habitat loss, and burning, are common tonumerous islands (e.g. Cronk 1989; Vitousek et al.1995; Rodríguez-Estrella et al. 1996). Thus, althoughsolutions have to be tailored to the particular cir-cumstances of each island system, field managersand conservationists often face the same problems.There are thus still many benefits to be gained sim-ply from improved information exchange betweenthose involved in conservation management onislands (Adsersen 1995, Campbell and Donlan2005).

The large catalogue of extinctions of native islandspecies is bound to increase. What can be done tostem the tide? If we cannot save the remaining evo-lutionary wonders of the showcase islands such asGalápagos and Hawaii—and the signs are nothugely encouraging—then the prospect for manyother less totemic islands and island endemics mustbe dismal indeed. Yet, it is not a hopeless task.Species can be saved by means of rigorous habitatprotection, pest or predator control, and transloca-tion of endangered island species (Franklin andSteadman 1991, Marrero-Gómez et al. 2003), givenadequate political and financial support.

Biological control—a dangerous weapon?

Biological control is the term given to pest controlnot by means of chemical agents but by the intro-duction of a species that targets the pest organismwithout seriously affecting non-target species.Cronk and Fuller (1995) state that it is the long-termgoal of conservation managers in Hawaii to achievethe biological control of most of the introducedweeds, but as yet only 21 are controlled biologically.Although in many cases biological control is theonly practical approach, there are associated risks(above; Williamson 1996). A classic example in ear-lier times was the introduction of the mongoose asa pest-control agent. The mongoose has shown lim-ited success in controlling rats but has beenextremely effective in devastating many nativeisland bird and reptile species, particularly ground-foraging skinks and snakes (Case et al. 1992). So, itis vital that biological control programmes begin

with rigorous screening and testing to ensure thatthe control agent’s effects are highly specific to thetarget species (e.g. Causton 2005).

Translocation and release programmes

Translocation and repatriation from ex situ breedingand rearing programmes have been used as meas-ures to rescue highly threatened species on islands,with some success. Translocation is used to removea species from an overwhelming local threat or tore-establish a population on an offshore islandwhere it may be safe from exotic predators. In gen-eral, translocated wild animals have been found toestablish more successfully than captive-bred ani-mals. Griffith et al. (1989) surveyed translocationprogrammes undertaken in Australia, Canada,Hawaii, mainland USA, and New Zealand over theperiod 1973–86. They found that only about 46% ofrelease programmes of threatened/endangeredspecies were successful, a far lower percentage thanfor translocations of native game species, for whichthe success rate was 86%. They attempted to iden-tify the factors influencing success for 198 bird andmammal translocations from within their survey.The single most important factor was the numberof animals released. They also found that herbi-vores were significantly more likely to be success-fully translocated than carnivores or omnivores,that it was better to put an animal back into the cen-tre of its historical range than the periphery, andthat wild-caught animals fared better than captive-reared animals.

In New Zealand, a number of threatened birdspecies have been translocated into small offshoreislands that either lacked or have been cleared ofintroduced predators, such as the ship rat (Rattusrattus), which are prevalent on the mainland(Lovegrove 1996). By this means the extinction ofseveral species, including the South Island saddle-back (Philesturnus carunculatus), have been averted.The kakapo (Strigops habroptilus), a flightless, noc-turnal parrot, was once considered to be effectivelyextinct, until the discovery of a small population onStewart Island (Clout and Craig 1995). They contin-ued to decline following rediscovery, to fewer than50 individuals, as a result principally of predation

S O M E C O N S E R VAT I O N R E S P O N S E S 331

Page 345: Whittaker. Island Biogeography. 2007

by feral cats. All known kakapo (i.e. the entirespecies) were therefore transferred to three preda-tor-free island refuges, where, supported by sup-plementary feeding, the kakapo appeared to havebetter prospects. However, recovery was reportedto have been initially disappointing due to the biasof survivors to older birds, a delay in resumingbreeding after translocation, and periodic invasionsof stoats on one of the islands.

Assessing the success of 45 release programmes,Lovegrove (1996) concluded that failures occurredwhere either (1) predators were still present, orarrived after release, or (2) too few birds, with anunbalanced sex ratio, were released. All releasesinvolving more than 15 birds on islands lackingpredators were successful, at least in the short term.Other New Zealand studies reportedly show thatreleases of as few as 5 birds on predator-free islandsare likely to succeed, and, more generally, thereview of Griffith et al. (1989) indicates that releasesof 40 or more birds into good habitat are generallysuccessful.

Clearly, the elimination of exotic predatory mam-mals is essential. Moreover, as on occasion evenindividual animals can deal a heavy blow to smallpopulations of threatened species, it may be neces-sary to monitor islands closely to ensure that pred-ators do not return. Much effort has beenchannelled into developing effective methods ofpredator control. Conservationists have been devel-oping a wide range of techniques, often at consid-erable expense, including the introduction of newrodenticides, and the spreading of bait from heli-copters, in order to target particular pest species togreatest effect (e.g. Clout and Craig 1995; Micol andJouventin 1995). Such schemes are often unafford-able (Herrero 1997).

It is obvious that it is important to translocatepopulations into suitable habitat, but what this maybe is not so obvious, as for many species popula-tions the reasons for their original demise are notfully understood (e.g. Hambler 1994). Given therole of humans in the demise of island species, it isimportant that translocations take place into areasin which harmful human influences are appropri-ately managed, or mitigated. Because the prehis-toric human impact was so intense in Polynesia,

Franklin and Steadman (1991) argue that the lack ofmany taxa from uninhabited, forested islands maybe due to events that occurred centuries ago, andthat with human abandonment and forest regener-ation, such islands may once again be able to sup-port the extirpated species. They therefore advocatethe translocation of species into such islands, fol-lowing careful assessment of the resources andproblems involved. From their preliminary assess-ments they argue that low makatea islands, such asthe Cooks, hold limited potential, but that theremay be greater potential among the 20 or so Tonganislands that are both uninhabited and greater than2 km2 in area. Moreover, the fossil record demon-strates the former presence on one such island,‘Eua, of species that still survive in Fiji, Samoa, andother Tongan islands, including the parrots Viniaustralis and Phigys solitarius. Establishing multi-ple populations will greatly improve the prospectsof a species surviving events such as the loss of alocal population in a hurricane. Such initiativesrequire a sound biogeographical and ecologicalbaseline, together with the political and legalframeworks to support the efforts.

The only wild giant tortoises in the Indian Ocean,Geochelone gigantea, survive on Aldabra atoll. Here,the culling of goats has enabled their population topersist in reasonable numbers. Between 1978 and1982, 250 tortoises were translocated to Curieuse inthe granitic Seychelles (Hambler 1994), but in adetailed survey in 1990 it was found that only 117animals remained. Poaching probably accountedfor most of the losses, although some of the surviv-ing adults were diseased and low growth andreproductive rates suggested possible saturation ofresources and nutrient limitation (particularly cal-cium deficiencies). Both feral cats and feral rats areabundant on the island, and probably account forthe very low rate of recruitment. To ensure the per-sistence of the new colony, it became necessary toestablish a facility to rear hatchlings until they areover 5 years old and thus relatively safe from pre-dation (if not from poaching). In contrast, a translo-cation to another of the granitic Seychelles, Frégate,has been comparatively successful. The reasons forthe difference in outcome on the two islands are asyet uncertain, but it is likely that both habitat

332 I S L A N D R E M E D I E S : T H E C O N S E R VAT I O N O F I S L A N D E C O S YS T E M S

Page 346: Whittaker. Island Biogeography. 2007

differences and differences in predation andpoaching are involved. This demonstrates that thesuccess of translocation is not simply a function ofinitial size of the translocated population.Eradication of the feral predators, and reduction inpoaching might be the answer on Curieuse, but it isalso possible that the island is basically unsuitablefor the long-term persistence of a viable population.

Programmes based on ex situ breeding andtranslocation back into the wild thus have theirplace in the conservation of island endemics.However, they have to be undertaken as part of ahighly organized, managed programme. The hap-hazard removal of animals from the wild suppos-edly for captive breeding in zoos may otherwise becounter-productive. Christian (1993) notes thatalthough several parrots have been taken out of StVincent and the Grenadines for captive breeding,there have been only three known successes, and todate there is no evidence of any captive-bred par-rots being returned to St Vincent for re-release.Thus, not only is it necessary to ‘fix’ the local envi-ronmental problems, but it is also necessary to con-trol the ex situ part of the process.

Protected area and species protection systems:the Canarian example

The combination of high biodiversity value andsevere pressure on natural environments and biotawithin the Canaries has been recognized by govern-mental and non-governmental organizations at alllevels. For example, Birdlife International’sEndemic Bird Areas scheme rates the islands a ‘highpriority’, and Conservation International hasrecently included the islands within an expandedMediterranean hotspot in its 2005 revision of the‘hotspots’ scheme (CI 2005; see Chapter 3). UnderUNESCO designation systems, the Canaries containfour Biosphere Reserves (the islands of Lanzarote,La Palma, El Hierro, and the south-western part ofGran Canaria), a World Heritage site (GarajonayNational Park, in La Gomera), and one RamsarConvention wetland (El Matorral, Fuerteventura).The archipelago also has four National Parks(IUCN management category II): Timanfaya inLanzarote, Teide in Tenerife, Garajonay in La

Gomera, and Taburiente in La Palma, out of aSpanish network of just 14 National Parks,although the archipelago constitutes just 1.5% ofSpain’s land area. There are also two importantconservation networks: the Canarian ENP network(Red Canaria de Espacios Naturales Protegidos)and the European Union Natura 2000 network ofsites. Today, the archipelago has about 150 pro-tected areas (Table 12.2), covering 45% of theCanarian land area. These are impressive figures,especially for such a densely populated and heavilyvisited archipelago.

The Canarian ENP system of protected areadesignation (Table 12.2) is a slightly modifiedimplementation of the IUCN protected areasmodel. It includes both strictly protected sites(e.g. ‘integral natural reserves’ and ‘natural monu-ments’), and ‘rural parks’ in which conservationand sustainable development go hand in hand(Martín-Esquivel et al. 1995). In the latter, theemphasis is on sustaining the rural populations andtheir culture while improving livelihoods andstandards of living. Thus, the goals of species andhabitat conservation within these parks are to beachieved within ‘cultural landscapes’ of traditionalresources use.

Although the Canarian National Parks have hith-erto been co-managed by the state and regionaladministrations, the rest of the protected area sys-tem has been managed by each island government.In total, the Canarian ENP network provides somemeasure of protected area designation for about40% of the land area of the archipelago, with per-centages for each island ranging from about 30%for Fuerteventura up to 60% for El Hierro. Together,the large islands of Gran Canaria and Tenerife (Fig. 12.1) contribute more than the half of thewhole protected area in the archipelago. ThisCanarian network does not include protectedmarine areas.

The recent establishment of the EuropeanUnion’s Natura 2000 network on the Canaries hasbeen used by the regional administration as anopportunity to complement the Canarian ENP net-work in those terrestrial habitats not previouslywell-represented (mainly thermophyllous wood-lands). Otherwise, almost all the terrestrial Natura

S O M E C O N S E R VAT I O N R E S P O N S E S 333

Page 347: Whittaker. Island Biogeography. 2007

334 I S L A N D R E M E D I E S : T H E C O N S E R VAT I O N O F I S L A N D E C O S YS T E M S

Table 12.2 Designations within the Canarian Protected Areas system (sources: Martín-Esquivel et al. 1995;Fernández-Palacios et al. 2004a).

Name Characteristics No.

National Park Large areas relatively untransformed by human activity, with high 4(Spanish and Canarian networks) importance due to the singularity of their biota, geology or

geomorphology and representing the main natural Spanish ecosystems

Natural Park Large areas with similar characteristics as the National Parks, being 12(Canarian network) representatives of the Canarian natural heritage

Rural Park Large areas where agricultural and livestock activities coexist with 7(Canarian network) zones of a great natural and ecological interest

Integral Natural Reserve Small natural areas protecting populations, communities, 11(Canarian network) ecosystems or geological elements deserving a special value for

their rareness or fragility. Only scientific activities allowed

Special Natural Reserve Similar to the former, but allowing educational and recreational 15(Canarian network) activities together with scientific ones

Site of Scientific Interest Small isolated sites comprising populations of threatened species 21(Canarian network)

Natural Monument Small areas characterized by unusual geological or palaeontological 52(Canarian network) elements

Protected Landscape Areas with outstanding aesthetic or cultural values 27(Canarian network)

Special Area of Conservation Areas which contribute in a valuable manner in maintaining or 174(EU Natura 2000 network) restoring natural habitat types or species in a favourable

conservation status. The 151 terrestrial SACs largely overlap withthe Canarian network of protected areas, adding 300 km2

more, but the 23 marine SACs contributed an additional1765 km2 of protected marine ecosystems

Special Protected Area Areas which contribute to preserve, maintain or restore the diversity 27(EU Natura 2000 network) and extension of the proper habitats for the 44 Canarian

bird species included in the European directives

Biosphere Reserves Areas protecting spaces where human activity constitutes an 4(UNESCO network) integral component of the territory, and where the management

should focus on the sustainable development of the resources

World Heritage Site Outstanding natural areas with unique features on a worldwide scale 1(UNESCO network)

Ramsar Convention Protected Wetlands offering important ecological services, such as the 1Wetlands (UNESCO network) regulation of water regimes as well as important sources of biodiversity

Page 348: Whittaker. Island Biogeography. 2007

2000 protected areas were already covered by theCanarian ENP network. The additional land desig-nated through this network is significant only inGran Canaria, La Gomera, and La Palma, togethercontributing a little over 300 km2 land area. Thisfigure has increased the protected land area withinthe archipelago from 40.4 to 44.5% (Table 12.3).However, its main additional benefit is in incorpo-

rating 23 marine areas of high natural value,totalling 1765 km2, into the protected area estate.

In addition to these ‘area-based’ schemes, there aretwo different protected species catalogues, one desig-nated at national (Spanish state) level, seeking to pro-tect 173 Canarian species and the other at theCanarian (autonomous region) level, seeking toprotect 450 endemic species or Canarian populations

S O M E C O N S E R VAT I O N R E S P O N S E S 335

Figure 12.1 The Canarian ENP network (Red Canaria de Espacios Naturales Protegidos) and the Natura 2000 networks of Protected Areas on and around Tenerife. Some zones correspond to two different protection categories, for example, the highest peak of the island (Teide) isprotected as a Natural Monument within a National Park, similarly, some zones of the Anaga massif (in the northeast corner of the island) areprotected as Natural Reserves within a Rural Park. Also, the Natura 2000 terrestrial sites largely overlap with the ENP network. The Natura 2000marine protected areas can be easily distinguished from the coastline of the island by the simplicity of their form. See Table 12.2 for explanationof the categories. (Map drawn by Ángel Vera, based on a figure in Martín-Esquivel et al. 1995.)

Page 349: Whittaker. Island Biogeography. 2007

of non-endemic species with a high emblematic valueor global conservation, like whales, dolphins, or birds.

Recently, a governmental proposal for the with-drawal of about 56% of the species included in theCanarian catalogue has been published (Martín-Esquivel et al. 2005). This action was based on a4 year monitoring and assessment project of theconservation status of species listed in theCatalogue. The authors concluded that in about 200cases, while the species concerned were restrictedto only a few populations, these populations werehealthy and stable. In addition, some 40 speciespreviously listed as endangered were removedfrom that list on the basis of changed assessmentsof whether they were endemic, taxonomic clarifica-tion of status, etc.

Although some species populations may behealthier than was once thought, and intensive con-servation efforts may be pulling some endangeredspecies back from the brink of extinction, many oth-ers remain acutely at risk (e.g. Marrero-Gómez et al.2003). The political imperatives within the archipel-ago remain focused on short-term economic interestsand a model of increased tourism, development, andurbanization. This economic model casts a denseshadow of uncertainty over the future of the natural

resource base and biodiversity of the archipelago,particularly of the warm, dry climate belt so popularwith European tourists. Despite the attempts toprovide legal protection for biodiversity and toinvest in environmental conservation at varyingpolitical levels, the pressures on the naturalresources of the Canaries continue to increase(García-Falcón and Medina-Muñoz 1999). Thesepressures include efforts to reduce the protectionafforded to particular protected areas that havepotential commercial value for development.Nonetheless, the various networks of protected areastatus and the other conservation measures dis-cussed in this section show how it is possible to tai-lor protected area models to an insular context:without these legal instruments, the future of manyCanarian endemic species and ecosystems wouldindeed be bleak.

12.3 Sustainable development onislands: constraints and remedies

The elements involved in the many case studiesdiscussed in this and the previous chapter arerepeated in varying combinations and with varyingemphasis for numerous other islands for which

336 I S L A N D R E M E D I E S : T H E C O N S E R VAT I O N O F I S L A N D E C O S YS T E M S

Table 12.3 Protected terrestrial areas designated by the Canarian and European Union Natura 2000 networks on the Canary Islands (sources:Martín Esquivel et al. 1995, Fernández-Palacios et al. 2004a).

Item Lanzarote Fuerteventura Gran Tenerife La La El CanariesCanaria Palma Gomera Hierro

Island area (km2) 846 1660 1560 2034 708 370 269 7447Number of Canarian 13 13 32 43 20 17 7 145

network protected areasCanarian network 350 477 666 989 250 123 156 3011

protected area (km2)Additional protected area 2.9 1.5 103 10.7 116 63.4 1.3 298.8

contributed by the Natura2000 network (km2)

Island area protected (km2) 352.9 478.5 769 999.7 366 186.4 157.3 3309.8Island area protected (%) 41.7 28.8 49.3 49.1 51.7 50.4 58.5 44.5Island contribution 10.7 14.5 23.2 30.2 11.1 5.6 4.8 100

to Canarian protected area system (%)

Page 350: Whittaker. Island Biogeography. 2007

conservationists have made management assess-ments. The remedies identified are often the same:control exotics, halt or reverse habitat loss, preventhunting, assist breeding of endangered species, enlistpolitical and legislative support, educate and enthusethe people to improve compliance and cooperation inmeeting management goals. It is important, finally, toconsider whether there are any special considerationson the human side of the equation which must beincluded in conservation thinking.

In many respects, oceanic islands are peculiarenvironments, both biotically (high proportion ofendemics, disharmonic, etc.) and abiotically (expe-riencing wave action from all sides, tending to havesmall hydrological catchments, peculiar geologies,etc.) (Nunn 2004). Insular peoples have commonlyevolved distinctive cultures and retain a strongallegiance to both home and culture (Beller et al.1990). It may be worth noting also that insularitymay provide essentially ecological problems forhumans. Examples include the extermination ofentire cultures, such as the Caribs of the Caribbean,killed off by disease introduced by Europeans; theintroduction of malaria into Mauritius in the 1860s,which led to the death of 20% of the capital’s popu-lation; and a decline of perhaps 50% in the popula-tion of Hawaii in the 25 years following the arrivalin 1778 of Europeans, who brought diseases such assyphilis, tuberculosis, and influenza (Cuddihy andStone 1990). Less directly, an epidemic of swinefever hit São Tomé and Principe in 1978, wiping outtheir entire pig population, and causing an 11-foldincrease in meat imports in 3 years. Likewise, oneof the most precious resources with respect toattracting outside finance is the possession of eco-logical interest, to attract international support, ortourist dollars. So, in many respects the problems ofthe plants and animals of islands are faced also bypeople on islands.

Yet, as islands include a vast array of climatic,geographic, economic, social, political, and culturalconditions, it is extremely difficult to offer meaning-ful generalizations as to the problems facing islandpeoples and how they influence and constrain theoptions from a conservation management perspec-tive. One approach might be to focus on thoseislands holding most threatened species. For

instance, over 90% of threatened island species areendemic to single geopolitical units and just 11 suchunits are home to over half the threatened islandbirds (Groombridge 1992, p. 245). However, eventhis does not allow much in the way of generaliza-tion, as the 11 units concerned are Cuba, Hawaii,Indonesia, the Marquesas, Mauritius, New Zealand,Papua New Guinea, the Philippines, São Tomé andPríncipe, the Seychelles, and the Solomons. This is apretty mixed bag of islands and societies.

For the human societies and economies ofsmaller islands (up to about 10 000 km2) the prob-lems include the following (Beller et al. 1990):

● Lack of water. This provides a particular problemfor very small, low, narrow atolls, where salinityconditions the entire land area, but can also affectsubstantial, high islands such as Tenerife, where thetourist and horticultural industries have been sus-tained only on the back of the unsustainable miningof geological reserves of water (see e.g. Ecker 1976).

● Susceptibility to rising sea level. For example,the Bimini islanders have formally requestedadvice from the UN on how to plan for their entireland area being below the projected sea levelaccording to greenhouse-warming predictions. It isnot simply the loss of area that may be critical, butfor small states like the Bahamas, the potential lossof much of their low-lying reservoirs of fresh water(Stoddart and Walsh 1992) and/or of their EEZs.

● Vulnerability to storm or tectonic damage.Hurricanes and cyclones can badly damage theeconomy of a small island (where population andkey elements of infrastructure are often concen-trated in just a few coastal localities), forcing thehuman populations to exploit biological resourcesthemselves damaged by such events. Certain typesof islands are particularly prone to volcanic and tec-tonic hazards and to related phenomena such astsunamis (Chapter 2). Such phenomena can deal asavage blow to the sustainable human use of anisland, as evident from the devastation inflicted onthe island of Montserrat by eruptions in 1997.

● Climatic variation. The agricultural base of asmall island may be hard hit by atypical weatherconditions. The El Niño event of 1983 caused

S U S TA I N A B L E D E V E L O P M E N T O N I S L A N D S : C O N S T R A I N T S A N D R E M E D I E S 337

Page 351: Whittaker. Island Biogeography. 2007

immense disruption to marine and terrestrial habi-tats, both proximal and distal to the tropics wherethe unusually high water temperatures wererecorded. Central Pacific seabird communities hada dramatic reproduction failure and briefly aban-doned atolls such as Christmas Island. Island life isclearly vulnerable to short-term weather events aswell as long-term climatic change (Stoddart andWalsh 1992). Because of the limited options provided by the discrete area of a small island,agricultural/fishing economies are similarly constrained.

● Narrow agricultural base. The constraints of theoften narrow agricultural base of island societies isclearly demonstrated by the history of Barbados,which has been dominated until modern times bythe boom-and-bust fortunes of the sugar industrysince colonization of the islands by Europeans in1492 (Watts 1987). When sugar was first establishedas the export crop, the island’s ecology was devas-tated in just 20 years, at the end of which time 80%of the land area was cultivated for sugar cane.Many indigenous species were lost in this first, dra-matic land-use conversion. Subsequently, changesin the global market for sugar, and occasional pestproblems, have resulted in very significant fluctua-tions in the area of cane, and in the security andwealth of the island. The narrow environmentalrange provided by such small size, and the smallland area, means that it is difficult for the localeconomy to diversify, and hard to produceexportable quantities of more than a limited rangeof agricultural crops. The boom-and-bust cycle ofperiodic export booms, subsequent deflation andresource exhaustion, and chronic emigrationappears to be a feature common to many smallisland histories (Beller et al. 1990).

● Lack of suitable land for agriculture. Very muchtied in to the previous point, is that there may befew crops suitable for the island soils and climate.The Bahamian island of Andros, for instance, ischaracterized by a viciously dissected coral ragsubstrate. Mechanized agriculture is not feasible,and the land essentially does not provide a living.

● Heavy dependence on the seas. Dependence onthe seas is all very well when the fish are plentiful.

Problems arise not only from anomalies in climateor ocean currents, but also from the activities oflarger, more efficient foreign fishing fleets, overwhich local peoples may have limited sway (cf.Merlen 1995).

● Small or fragmented states. Tied in to all of theforegoing is the political geography of small orfragmented states. In the case of countries such asthe Bahamas, or Jamaica, with a shallow naturalresource base, a fluctuation in the environment (e.g.a hurricane) can deal a significant blow to theresources of the state.

In his review of resource development on Pacificislands, Hamnett (1990) identifies five major trendswithin the region: (1) increasing pressure on theland and freshwater resources; (2) intensification ofagriculture; (3) loss of native forest resources; (4)river and stream siltation and the loss of aquaticresources; and (5) increasing use of coastal areasand the degradation of harbours and lagoon envi-ronments. We might wish to add others to this list,notably the increasing interest from tourists, seek-ing an idyllic desert island on which to holiday, butexpecting the facilities of a modern society—thehotels, roads, fresh water, and other amenities—aswell, perhaps, as the exotic wildlife. This sixth trendis powerfully evident in the Caribbean, which, inthe period after the Second World War, has wit-nessed rapid urbanization, a significant increase intourism, and marine degradation from the con-struction of transport (air and sea) and petroleumfacilities and hotels (Beller et al. 1990). Many of theislands of the Mediterranean, and certain Atlanticislands, notably the Canaries, have experiencedsimilar patterns of expansion in the tourist indus-tries. Tourist pressures are also extremely signifi-cant on Hawaii and the Galápagos (Cuddihy andStone 1990; above) and indeed on other tropical andsubtropical islands. This is just one way in which,in common with the problems facing native plantsand animals, exogenously driven forces (social,economic, and technological) are affecting and dis-rupting many island cultures, and eroding tradi-tional skills.

The task of confronting these pressures is not asimple one. Some of the themes involved in

338 I S L A N D R E M E D I E S : T H E C O N S E R VAT I O N O F I S L A N D E C O S YS T E M S

Page 352: Whittaker. Island Biogeography. 2007

sustainable development planning for small islandsare listed in Table 12.4. In many ways the ecologicalmanagement goals are the easiest part of the equa-tion to specify, while managing the societal, eco-nomic, and political issues is very difficult (Belleret al. 1990). As an illustration, concepts of owner-ship among island nations can be both extremelyvaried and complex. In Papua New Guinea, theSolomon Islands, and Vanuatu, ownership isdefined by oral traditions, land is typically ownedby family groups rather than individuals, and theremay be secondary and tertiary levels of ‘ownership’providing rights of use but not implying actualownership of the land (Keast and Miller 1996).Success in such circumstances will depend on howwell insular resource planning and managementcan provide solutions tailored to local circum-stances and cultures (Nunn 2004), and whetherthey carry with them the support of the island com-munities. The involvement of major users, farmers,fishermen, charter operators, etc., is one element inthis. The mobilization of non-governmental organi-zations (NGOs) in island problems is another ele-ment, whereby energies and expertise exist thatcould be harnessed provided interest can be raised.Perhaps above all, it is necessary to foster a generalpublic awareness of the environmental and ecolog-ical constraints and problems, and the significanceto the island societies themselves of the ecological‘goods’ at stake (Christian 1993).

The importance of public awareness and supportcan be illustrated by reference to parrot conserva-tion in the West Indies. Since the arrival ofColumbus at the end of the fifteenth century, 14species and two genera of parrots have been lost.The remaining nine endemic members of the genusAmazona are each confined to individual islands.The case of the Puerto Rican parrot, Amazona vittata,and its conservation was outlined in Chapter 10.Within the Lesser Antilles, seven endemic parrotshave become extinct in historical times, with justfour species remaining. They are now threatenedby deforestation, predation, illegal hunting and col-lecting, competition with exotics, and natural disas-ters (Christian 1993; Christian et al. 1996). Impacts on the parrots resulting from habitat alterationinclude the loss of nesting cavities, food, and shelter.Given these pressures, management intervention isessential for their survival.

Measures for parrot conservation in the LesserAntilles consist of environmental education, habitatprotection, enforcement of appropriate legislation,and enhancement of wild breeding and captivebreeding programmes. Expensive programmes,such as those adopted for the conservation of thePuerto Rican parrot, are not feasible in the LesserAntilles because of shortage of funds. Despite this,effective targeting of conservation resources hasdelivered some degree of success. Nonetheless,more needs to be done; first, to protect habitats;

S U S TA I N A B L E D E V E L O P M E N T O N I S L A N D S : C O N S T R A I N T S A N D R E M E D I E S 339

Table 12.4 Sustainable development options for small islands should aim at increasing self-reliance, and can be categorizedunder six basic headings (after Hess 1990).

Categories Examples

Resource preservation Conservation zones, multiple-use options, control of huntingResource restoration Replanting, re-introduction, alien herbivore removalResource enhancement Freshwater re-use, seawater desalinationSustainable resource development Small-scale diversified, closely managed forms of resource-

based enterprises (agriculture, fisheries, tourism)Provision of human services Alternative energy generation sources and distribution

systems, waste disposalNon-resource-dependent development options Financial services (tax havens), light industries processing

imported materials, rental of fishing rights

Page 353: Whittaker. Island Biogeography. 2007

secondly, to discontinue the export (legal or illegal)of live parrots; and thirdly, to involve local peopleas much as possible in both the rationale of and theemploy of the management programmes.

Christian (1993) has described the efforts made bythe government of St Vincent and the Grenadines toensure the long-term survival of the Vincentian par-rot, Amazona guildingii, which has been reduced toabout 450 birds. Aided by financial support frombodies such as the World Wildlife Fund, the RARECentre for Tropical Bird Conservation, and notablythe local St Vincent Brewery, the ForestryDepartment has been able to bring the conservationmessage to the local population and enlist their sup-port. A significant amount of the island’s limitedresources have been channelled into activities andprogrammes that support and complement parrotconservation. Thus this small island nation hasalready made a big contribution and shown its com-mitment to biodiversity conservation. Christianstresses the importance of continuing to demonstratethe links between habitat protection for the parrot,and the opportunities for ecotourism, extraction ofminor forest products, and soil and watershed pro-tection. He cautions that failure to demonstrate suchdirect linkages would risk the loss of public cooper-ation and support so vital for conservation.

This final section has served merely to introducea few of the topics of sustainable development ofislands and to present, briefly, the case that islandshave many problems different in kind from those oflarger land masses (for a fuller account see Belleret al. 1990). There is in this a parallel with the under-lying ecological or nature conservation problems ofoceanic islands. In short, small or remote oceanicislands are ecologically special: they constitute evo-lutionary treasure-houses, threatened by contactwith other regions and especially the continents.The nature of the ecological problems involved isnow well understood and the pattern of continuingdegradation and species loss is all too obvious and,in the general sense, predictable. Dealing with theseproblems requires a sensitivity to, and respect for,the island condition of the human societies thatoccupy them (Nunn 2004). There are remedies, butthey require solutions tailored to the islands andnot simply exported from the continents.

12.4 Summary

Island societies face many problems, some commonto those of societies everywhere, and others that arespecific to the island context. We illustrate a few ofthese special problems by reference to the implica-tions of the loss of very small islands to rights ofaccess to marine resources, and by reference to thepotential loss of entire island societies through(projected) future sea-level rise.

In an increasingly globalized world, the effectiveisolation of many oceanic islands has been brokenin the last few hundred years and especially in themost recent decades. The devastation of Nauruthrough mining of phosphate-rich guano deposits,and the huge socio-economic changes (and conse-quent environmental pressures) brought by masstourism within the Canaries, provide evidence ofthese processes, which can be particularly difficultfor small island societies and governments (likethose of Nauru) to cope with. The societal dimen-sion of these processes and phenomena is a crucialpart of any serious effort to plan for biodiversityconservation in island settings.

In considering some of the conservation responsesthat have been attempted on remote islands, webriefly review the use of biological control, translo-cation, captive breeding and re-release programmes,and of ‘high-tech’ approaches to the removal offeral animals from islands: just a small selectionof the wildlife conservation tools that may be crucialto conservation efforts. Integrated programmesinvolving the re-insularization of populations, forinstance by transferring them to small offshoreislands from which predators have been eradicated,offer some hope for the persistence of particularthreatened island endemics. We also illustratethrough the Canarian example, how for largeroceanic islands, it is possible—and arguablycrucial—to adapt the legal instruments of protectedarea planning and species protection, for applicationin an insular context.

Many practical problems continue to face islandconservationists, and they are illustrated here byreference to the Galápagos, wherein the political,financial, and socio-economic problems are arguedto be the key drivers of conservation problems. The

340 I S L A N D R E M E D I E S : T H E C O N S E R VAT I O N O F I S L A N D E C O S YS T E M S

Page 354: Whittaker. Island Biogeography. 2007

strategy required for effective conservation of suchisland systems must be tailored to meet the particularfeatures of island environments and their humansocieties, and the particular factors that threatentheir environments and their biodiversity (sea-level rise, exotic predators, exotic grazers, mining,over-fishing, over-hunting, habitat degradation,

etc.). To be successful, conservation actions mustultimately be part of broader management for thesustainable development of island ecosystems. It isstressed that it is crucial to enlist political andlegislative support, and to build educational pro-grammes to engender the interest and support ofisland peoples.

S U M M A RY 341

Page 355: Whittaker. Island Biogeography. 2007

actinomorphic Having flowers with radialsymmetry.adaptive radiation The evolutionary process bywhich an ancestral species gives rise to an array ofdescendant species exhibiting great ecological,morphological or behavioural diversity.alien species A foreign (exotic, or non-native)species moved by humans to a region outside itsnatural geographical range or environment.alleles Two or more forms of a gene occupyingthe same locus on a chromosome.allelopathy A form of interference competition bymeans of compounds produced by one species ofplant that reduce the germination, establishment,growth, survival, or fecundity of another plant species.allochthonous Having originated outside thearea where it now occurs.allogamy Fertilization involving pollen andovules from: (1) different flowers (whether on thesame plant or not); (2) genetically distinct individu-als of the same species.allopatric speciation The differentiation of geo-graphically isolated populations into distinct species.allozyme One of several forms of an enzymecoded for by different alleles at a locus.anacladogenesis Evolutionary change within alineage where the progenitor species survives withlittle change alongside the derived species.anagenesis Evolutionary change within a lineagewhere the progenitor species becomes extinct.anemochory Dispersal by wind.anemophily Pollination by wind.area cladogram A cladogram in which the termi-nal taxa have been replaced by their respective geo-graphical distributions.assemblage The set of plant or animal speciesfound in a specific geographical unit or area.

assembly rules Identified structure in thecomposition of a series of islands (or otheroperational units), typically determined within eco-logical guilds. Assembly rules include: (1) compati-bility rules, (2) incidence rules, and (3) combinationrules.assortative equilibrium In the context of theequilibrium theory of island biogeography, thespecies number having initially equilibrated, mayincrease over time to a new equilibrium valuereflecting the ecological interactions within thebiota.atoll Coral reefs built around subsiding volca-noes.auto-compatibility Plant species that can self-fertilize (also termed simply ‘compatible’).autogamy Self-fertilization.auto-succession The assemblage of a communitywith no obvious environmental change or transition.bauplan A particular body architectural plan.benthic Living at, in, or associated with, struc-tures on the bottom of a body of water.biodiversity The variability of life from allsources, including within species, between species,and of ecosystems.biodiversity hotspot May mean a geographicarea with high levels of species richness and/orendemism. Also used to refer to areas that combinehigh biodiversity with high threat to that biodiver-sity (e.g. due to habitat loss).biogeographical regions, kingdoms, andprovinces Major biogeographical subdivisions ofthe world based largely on shared composition atthe family levelbiomass The total weight of living tissue of organ-isms accumulated over time, usually expressed asdry weight of organic matter per unit area.

342

Glossary

Page 356: Whittaker. Island Biogeography. 2007

biosphere reserve Areas of terrestrial or coastalecosystems that are internationally recognizedwithin the framework of the UNESCO’s Man andthe Biosphere Programme (MAB).biota All species of plants, animals, and microbesinhabiting a specified region.bottleneck A severe reduction in populationnumber that is often associated with reducedgenetic diversity and heterozygosity and reducedadaptability of that population.caldera A large crater (typically 5–20 km diame-ter) formed by the collapse of a volcano followingwithdrawal of magma from an underlying storagechamber.Cenozoic An era of geological time comprisingthe Tertiary and the Quaternary periods, i.e. the last65 million years.character displacement The divergence of afeature of two similar species where their rangesoverlap so that each uses different resources.chequerboard distribution Pattern shown by twoor more species that have mutually exclusive butinterdigitating distributions across a series of iso-lates, such that each island (or habitat) supportsonly one species.clade Any evolutionary branch in a phylogeny,especially one that is based in genealogical rela-tionships.cladogram (or phylogenetic tree) A line diagramderived from a cladistic analysis showing thehypothesized branching sequence of a monophyletictaxon and using shared derived character states todetermine when each branch diverged.cladogenesis Evolutionary change within a line-age where the progenitor species is partitioned intotwo or more lineages and becomes extinct.coenocline Vegetation continuum along a gradient.colonization The relatively lengthy persistence ofan immigrant species on an island, especiallywhere breeding and population increase are accom-plished.competition Negative, detrimental interactionbetween organisms caused by their need for a com-mon resource. Competition may occur betweenindividuals of the same species (intraspecific com-petition) or of different species (interspecific com-petition).

competitive exclusion The principle that whentwo species with similar resource requirementsco-occur, one eventually outcompetes and causesthe extinction of the other.connectance Fraction of all possible pairs ofspecies within a community that interact directly asfeeder and food. In other words, number of actualconnections in a food (or pollination) web dividedby the total number of possible connections.conservation Human intervention with the goalof maintaining valued biodiversity (genetic varia-tion, species, ecosystems, landscapes, naturalresources).continental fragments or micro-continentsMainland fragments separated by tectonic driftfrom their continents millions of years ago, with thespecies they carried.continental or land bridge islands Emergentfragments of the continental shelf, separated fromthe continents by narrow, shallow waters. This sep-aration is often recent, as a result of the postglacialrise of sea level, which isolated the species on theisland from conspecific mainland populations.continental shelf The area of shallow sea floor(�200 m depth) adjacent to continents, underlainby continental crust and effectively a submergedextension of the continents.convergent evolution Evolution of similarfeatures independently in unrelated taxa, usuallyfrom different antecedent features or by differentdevelopmental pathways.corridor A dispersal route of favourable habitattype, which permits the direct spread of speciesfrom one region or habitat fragment to another.counter-adaptation The evolutionary reaction ofan island’s native species to a newly arrivedspecies, such that over time they begin to exploit orcompete with the immigrant more effectively, thuslowering its competitive ability.crassulacean acid metabolism (CAM) plantsUsually succulent xerophytes in which carbondioxide taken up during the night is stored as malicacid until fixed using the C3 pathway the followingday.cryptoturnover Turnover of species (extinctionfollowed by immigration) not detected becauseoccurring between two surveys.

G L O S S A RY 343

Page 357: Whittaker. Island Biogeography. 2007

density compensation Higher than normaldensities of a species on an island: typicallyexplained as an outcome of the lower overall rich-ness of the island assemblage.density overcompensation A situation wheredensity compensation occurs to an apparentlyexcessive degree in one or more species.density stasis A situation where the overallpopulation of the community on the island is lessthan that of the reference mainland system, suchthat population sizes per species are the same as themainland.diffuse competition Community-wide interspe-cific competition among a group of species, subject-ing each species to a range of competitive pressuresexerted by other species.dioecious species Having the sexes in separateindividuals.diplochory Plant species with two different dis-persal mechanisms.disassembly The process of a community/habitatpatch losing species as, for instance, a result ofhabitat loss and increased isolation due to habitatfragmentation.disharmony A characteristic of island biotas,such that they are biased in their representation ofhigher taxa compared to nearby mainland sourceareas.disjunction A distribution pattern exhibited bytaxa that have populations that are geographicallyseparated.dispersal The movement of organisms away fromtheir point of origin.disturbance Any relatively discrete event in timethat removes organisms and opens up space whichcan be colonized by individuals of the same or dif-ferent species.dwarfism The tendency for some isolated popu-lations to undergo significant decreases in bodysize in comparison to their conspecifics on themainland (also termed nanism).ecological release A colonizing speciesencounters an environment in which particularcompetitors or other interacting organisms, suchas predators, are absent, and being releasedfrom these constraining forces responds, e.g. by

expanding its feeding niche, or losing defensivetraits.ecological trap The preference of some species forbreeding in edge habitats even though mortality ishigher there than in the fragment.ecoregion Major ecosystem type prevalent acrossa region, resulting from large-scale predictable pat-terns of solar radiation and moisture, which in turnaffect the kinds of local ecosystems and animalsand plants found there.ecotone The transition zone between two adja-cent communities/habitats, which typically showsa mix of the properties of each.ecosystem transformer An introduced speciesthat becomes naturalized, and invasive, and whichsignificantly alters ecosystem properties, thusimpacting on native species.ecotourism Tourism based on an interest in observ-ing nature, preferably with minimal ecologicalimpact.edge effect In fragmented landscapes, the edge ofa habitat patch typically has altered physical andbiotic properties, which extend some way towardsthe core of the habitat patch.effective population size The proportion ofthe adult population that actually participates inbreeding.El Niño–Southern Oscillation (ENSO) Climaticanomalies yielding significant changes in rainfall,temperature, humidity, and storm patterns overmuch of the tropics and subtropics, due to the inva-sion of warm surface waters from the western partof the equatorial Pacific basin to the eastern part,disrupting the upwelling along the western coast ofSouth America.empty niche In the island context, refers to thelack of representation of a particular mainlandecological guild or niche, providing evolutionaryopportunities for colonizing taxa to exploit thevacancy.en echelon lines Short lines of crustal weakness,which are sub-parallel to each other.endemic A taxon restricted to a given area,whether a mountain top, island, archipelago, conti-nent, or zoogeographic region.entomochory Dispersal by insects.

344 G L O S S A RY

Page 358: Whittaker. Island Biogeography. 2007

entomophily Pollination by insects.equilibrium A condition of balance betweenopposing forces, such as between rates of immigra-tion and extinction.establishment The successful start or founding ofa population.eustatic sea level changes Sea-level changes orig-inated by the changing volume of water in the sea.explosive radiation Evolutionary process result-ing in an exceptional number of new monophyleticspecies, e.g. the Hawaiian drosophilid flies or LakeMalawi’s cichlid fishes.extinction The loss of all populations of a speciesfrom an island, or the global loss of a species.extinction debt The anticipated eventual speciesloss from an area of fragmented habitat followingfragmentation.extirpation The loss of a local (insular) popula-tion, not implying global loss of the species.faunal drift The random sampling of speciesreaching an ‘empty’ habitat, and thus providing anovel (disharmonic) biotic environment in whichselection drives divergence into a variety of newniches.fitness The number of offspring an individualproduces over its lifetime.founder effect Genetic loss that occurs when anewly isolated population is founded, by a smallnumber of colonists.founder event The colonization of an island bythe first representatives of a species.frugivory Action of those animals feeding on fruits.geitonogamy Pollen transfer occurring amongdifferent flowers of the same individual.gene flow The movement of alleles within apopulation or between populations caused by thedispersal of gametes or offspring.generalist species A species that uses a relativelylarge proportion, or in extreme cases all, of theavailable resource types.genetic drift Changes in gene frequency within apopulation caused solely by chance without anyinfluence of natural selection.Ghyben–Herzberg lens Rainwater accumulationthat floats on the denser salt or brackish water thatpermeates the base of an island.

gigantism The tendency for some isolatedpopulations to undergo significant increases inbody size in comparison to their conspecifics on themainland.Gondwanaland The southern super-continent,which split apart from Pangaea (the last uniqueland mass) some 160 Ma. It was composed of theareas known today as South America, Africa,Antarctica, Australia, India, Madagascar, NewZealand, and New Caledonia.guild A group of species exploiting the same classof resources in a similar way.guyot A type of seamount with a flat top,formed by accretions of carbonate sediments on thesummit of a subsiding volcano.habitat fragmentation Anthropogenic process bywhich a large, continuous habitat is diminished inarea and divided into numerous isolated habitatfragments.habitat island Discrete patch of a particularhabitat type surrounded by a matrix of stronglycontrasting habitat(s).halophytes Salt-tolerant plant species, commonlyfound in salt marsh, drylands, and saline lakeenvironments.hermaphrodites Individuals having both maleand female reproductive structures.herpetofauna Amphibian and reptile fauna.heterozygote An individual organism that pos-sesses different alleles at a locus.hybridization The production of offspring byparents of two different species, populations, orgenotypes.hyperdynamism An increase in the frequencyand/or amplitude of population, community, andlandscape dynamics in fragmented habitats.hydrochory Dispersal by water.Holocene The latest epoch, commencing about11 500 years ago, at the end of the Pleistocene.homozygote An individual organism that has thesame allele at each of its copies of a gene locus.immigration In island ecology, refers to theprocess of arrival of a propagule on an island notoccupied by that species (may be distinguishedfrom ‘supplementary immigration’—subsequentreinforcements).

G L O S S A RY 345

Page 359: Whittaker. Island Biogeography. 2007

impoverishment The characteristic by whichislands have fewer species per unit area than themainland in comparable ecosystems, a distinctionthat is more marked the smaller the island.inbreeding depression A reduction of the fitnessin a normally outbreeding population as aconsequence of increased homozygosity due toinbreeding.incidence functions Biogeographical tool fordescribing how the probability of occurrence of aspecies varies with selected characteristics ofislands (species richness, area, isolation, etc.).introduction The voluntary or accidental human-assisted movement of individual species into a sitethat lies outside their natural, historical range.introgressive hybridization (or introgression)The incorporation of genes of one species into thegene pool of another species.invader complex Whereby introduced speciespreferentially establish interactions with otherintroduced species.invasion The spread of a naturalized exoticspecies into natural or semi-natural habitats, hence‘invasive species’.island rule The tendency for a graded series ofchanges in the size of island vertebrate species inrelation to mainland congeners, such that small-bodied species tend to get larger, and vice versa.island sterilization Process occurring whendestructive volcanic ash flows are deposited overan entire island, completely eliminating the existingbiota. It implies that colonization has to begin againfrom nothing.isolating mechanism Any structural, physiologi-cal, ecological, or behavioural mechanism thatblocks or strongly interferes with gene exchangebetween two populations.isostatic sea level changes Changes in sea leveldue to the relative adjustment of the elevation ofthe land surface (e.g. (1) due to the removal of massfrom the land causing uplift, as when an icecapmelts, or (2) by tectonic uplift).keystone species Species supplying a vitalresource or holding a crucial function in an ecosys-tem, such that alteration in its numbers has a radi-cal impact on key functional properties of theecosystem.

kipuka Hawaiian term for a remnant of an oldforest ecosystem that has been isolated by a newlava flow.lag time The time taken for the relaxation ordecrease of species richness following fragmenta-tion (i.e. time taken to pay the extinction debt).landscape Spatial concept, in ecology referring tothe interacting complex of systems on and close tothe Earth’s surface, including parts of the atmos-phere, biosphere, hydrosphere, lithosphere, andpedosphere.landslides Gravitational instabilities that in themost spectacular cases can lead to the collapse ofthe slopes of an island, producing debris ava-lanches of hundreds of km3 into the sea. Such col-lapses may occur unexpectedly and suddenly andmay lead to the disappearance of a significant partof an island in minutes.late successional species Species that occurprimarily in, or are dominant in, the late stages ofsuccession.laurisilva Relict subtropical evergreen montanecloud forest typical of the Macaronesian islands,dominated by Laureaceae tree species: formerlymuch more widely distributed.lignification The acquirement of a woodiness ininsular plant species derived from herbaceouscontinental ancestors.Macaronesia Biogeographical region of thePalearctic comprising the Atlantic Ocean volcanicarchipelagoes of the Azores, Madeira, theSalvage Islands, the Canaries, and the Cape VerdeIslands.Makatea islands Islands formed by young volca-noes, with raised coral reef uplifted to a few metresabove sea level.marine transgression The flooding of the land bythe sea due to a rise in the sea level, either by anabsolute sea-level change (interglacial periods) orby tectonic subsidence.mesopredator release The increasing number ofsmaller omnivores and predators due to theabsence of larger predators: a phenomenon that canresult from habitat fragmentation.metapopulation A population of geographicallyseparated subpopulations interconnected bypatterns of gene flow, extinction and recolonization.

346 G L O S S A RY

Page 360: Whittaker. Island Biogeography. 2007

minimum viable area (MVA) The area neededfor maintaining the minimum viable population ofa species.minimum viable metapopulation (MVM) Theminimum number of interacting local populationsneeded for the long-term persistence of a species inan area.minimum viable population (MVP) The mini-mum size of a population that will ensure its long-term survival, for instance, providing a 95% ofpersistence probability for the next 100 or1000 years.molecular clock Based on the idea that proteinsand DNA evolve at a more or less constant rate, theextent of molecular divergence (attributable to ran-dom mutations) is used as a metric for the timing ofevents within lineage development: greater confi-dence can be placed in such molecular clocks if theycan be independently calibrated.monoecy Refers to plant species that haveseparate male and female flowers on the sameindividual.monophyletic group A group of taxa that share acommon ancestor and which includes all descen-dants of that ancestor, also referred to as a clade.natural selection The process of eliminating froma population through differential survival andreproduction those individuals with inferior fitness.naturalized species A non-native species that hasestablished a breeding population.neoendemic A species exclusive to an area(island or island group) that has evolved itsdistinctive characteristics in situ, having originallycolonized from a continental ancestor (contrastwith palaeoendemic).nestedness Where a set of islands is rankedaccording to species richness, the tendency for eachsuccessively smaller assemblage to be a subset ofricher sets.niche The total requirements of a population or aspecies for resources and physical conditions.niche expansion An increase in the range of habi-tats or resources used by a population, which mayoccur when a potential competitor is absent.non-adaptive radiation A radiation processdriven not by natural selection, but rather bygenetic drift or sexual selection.

non-equilibrium A characteristic of communitiesthat change constantly in structure and compositiondue to the vagaries of the environment and wherestable end-points are not achieved.nuées ardentes See pyroclastic flowoceanic islands Islands originated from subma-rine volcanic activity, mostly with basaltic founda-tions, that have never been connected to thecontinents. Initially populated by species thathave dispersed to the islands from elsewhere aresubsequently enriched by speciation.palaeoendemic A species exclusive to an area(island or island group) that has apparentlychanged little since colonization, and which hasdisappeared from its continental source area (con-trast with neoendemic).Pangaea The supercontinent, comprising allpresent continents, which coalesced in the lateCarboniferous or early Permian and which brokeup some 70 million years later in the late Triassic.panmictic population Population where matingoccurs randomly with respect to the distribution ofthe genotypes in the population.parallel evolution The evolution of similar oridentical features independently in unrelatedlineages.parapatric speciation A mode of speciation inwhich differentiation occurs when two populationshave contiguous but non-overlapping ranges, oftenrepresenting two different habitat types.paraphyletic group An incomplete evolutionaryunit in which one or more descendants of a partic-ular ancestor have been excluded from the group:also referred to as a grade.parthenogenesis The development of eggs with-out fertilization by a male gamete.peninsular effect The hypothesized tendency forspecies richness to decrease along a gradient fromthe axis to the most distal point of a peninsula.phylogeny The evolutionary relationshipsbetween an ancestor and all its known descen-dants.phylogeographical analysis The analysis ofgeographically structured genetic signal within asingle taxon (e.g. species); the reconstructedgenealogical history is used to infer past episodesof population expansion, bottlenecks, vicariance,

G L O S S A RY 347

Page 361: Whittaker. Island Biogeography. 2007

and colonization, e.g. in an island context, thesequence of interisland dispersal events.pioneer species Early colonizing species within asuccession.plate boundary The edge of one of the Earth’slithospheric plates: boundaries can be eitherconstructive (divergent), destructive (convergent),or conservative.Pleistocene The first epoch of the Quaternaryperiod, formally taken as beginning about 1.8 Ma(although the climate cooling that marked theperiod began perhaps as early as 2.5 Ma), whichended about 11 500 years BP, with the transition tothe present Holocene epoch.Pliocene The final epoch of the Tertiary period,spanning approximately 5.3 to 1.8 Ma.pollination The transfer of pollen grains to areceptive stigma, e.g. by wind or by pollinating ani-mals.polyphyletic group A taxonomic group compris-ing several monophyletic groups.population viability analyses (PVA) Populationanalyses based on data for key demographic andgenetic parameters, which set out to assess theprobability that a population of a given size willpersist for a particular time period.productivity rate The rate at which a communityof organisms manufactures new biomass.propagule The minimal number of individuals ofa species capable of successfully colonizing a habit-able island.pseudoturnover In island ecological analyses,species appearing to turn over (undergo extinctionand re-immigration) due to incomplete censusdata, when they have actually been residentsthroughout or, alternatively, never properlycolonized.pyroclastic flow High-speed avalanche of hot ashmobilized by expanding gases and travelling atspeeds in excess of 100 km/h.Quaternary The most recent period of the geolog-ical timescale, following the end of the Pliocene andcontinuing to the present day; and comprising thePleistocene and the Holocene epochs.radiation In evolutionary biology, a term for theexpansion of a group, implying that many newspecies have been produced.

radiation zone The zone within an ocean, nearthe effective dispersal limits of a higher taxon(e.g. terrestrial mammals), where few lineagescolonize and where radiation in those lineages iscorrespondingly greater as a consequence of the‘empty niche space’ encountered.red (data) book A catalogue listing species thatare rare or in danger of becoming extinct locally,nationally or globally (e.g. the IUCN red databooks).refugia Areas of survival of species duringadverse conditions, most notably during glaciationepisodes.relictualism The possession by some islands ofpalaeoendemic, i.e. relict taxa, which formerly hada continental distribution, and which have goneextinct from the source region since colonizing theisland, e.g. due to climatic or geological changes.reproductive isolation Inability of individualsfrom different populations to produce viable off-spring.rescue effect The prevention of extinction of asmall insular population by the occasional influx ofindividuals from another (e.g. mainland) area.Sahul Pleistocene continent constituted byAustralia, New Guinea, and Tasmania.seamount A mountain beneath the sea (see alsoguyot).sexual selection Depends on the success ofcertain individuals over others of the same sex, inrelation to the propagation of the species, and canresult in the evolution of character traits that carrya fitness cost, and would otherwise not seemadvantageous.shield volcanoes Wide volcanoes, of graduallysloping sides, build up in layers by basaltic magmaflows erupting from lengthy fissures or vents.sink population A term used in the metapopula-tion literature for a population occupying anunfavourable habitat, which maintains its presenceby supplementary immigration of offspring fromsource populations.SLOSS debate The debate as to which strategy isthe best for protected area systems: Single Large OrSeveral Small reserves.source population A population occupying afavourable habitat and thus producing an excess of

348 G L O S S A RY

Page 362: Whittaker. Island Biogeography. 2007

offspring that will emigrate to less favourablehabitats where sink populations, with negativedemography, can be maintained.specialist species A species that uses a relativelysmall proportion of the available resource types.speciation The process in which two or morecontemporaneous species evolve from a singleancestral population.species swarm A large number of closely relatedspecies occurring together in an area (e.g. an archi-pelago) and derived by multiple splitting of anancestral stock.species–area relationship (SAR) The relationshipbetween the number of species and the area of thesample or island; within this simple idea, a varietyof important distinctions exist, which are often notproperly recognized.species turnover The product of opposing ratesof species immigration to and species extinctionfrom an island. Where these rates precisely balancethis forms a dynamic equilibrium, with speciesrichness remaining constant through time butspecies composition continually altering.stepping stones Island or island groups thatfacilitate the dispersal of species across oceans bytheir intermediate geographic position between acontinental source pool and an island of interest.stratovolcano The composite cone built by avolcano that emits both molten and solid materials,and builds up a steep-sided cone.subduction The process by which lithosphere-bearing oceanic crust is destroyed in plate tectonicsby the movement of one plate beneath another toform a marine trench. The process leads to theheating and subsequent re-melting of the lowerplate, in turn generating volcanic activity, typicallyforming an arc of volcanic islands above the area ofsubduction.subfossil A dead organism that is not truly fos-silized.subsidence The downward movement ofan object relative to its surroundings. Subsidenceof the lithosphere can be due to increased mass(e.g. increased ice, water, or rock loading) or due tothe movement of the island away from mid-oceanridges and other areas that can support anomalousmass.

succession Refers to directional change in(mostly) plant communities following significantecosystem disturbance, or following the creation ofan entirely new land surface area.successional turnover In an island ecologicalcontext, refers to species turnover attributable tosuccessional processes.supertramp A species that has excellent coloniz-ing abilities but is a poor competitor in diversecommunities.sustainable development Development thatmeets the needs of the present generation withoutcompromising the ability of future generations tomeet their own needs.sweepstake dispersal route The least favourableof potential migration pathways in that it can onlybe traversed very infrequently and with greatdifficulty, or only under circumstances that occurvery rarely.sympatric speciation The differentiation of tworeproductively isolated species from one initial pop-ulation within the same local area, in which muchgene flow potentially could or actually does occur.taxon cycle A proposed series of ecological andevolutionary changes in a newly arrived colonist onan isolated island, from the state of being indistin-guishable from its mainland relatives to that of ahighly differentiated endemic, to extinction andreplacement by new colonists.tephra A collective term for all the unconsoli-dated, primary pyroclastic products of a volcaniceruption, independent of grain size.territoriality Behaviour related to the defence of aspecified area (the territory) against intruders.Tertiary A period of geological time lasting fromc.65 Ma to the beginning of the Quaternary.translocation programmes Attempts at speciesconservation involving the movement of a numberof individuals into an area from which the speciesis currently lacking (typically having once beenpresent).trophic cascade The chain of knock-on extinc-tions following the loss of one or a few species thatplay a critical role (e.g. as a pollinator) in ecosystemfunctioning.trophic level Position in a food chain determinedby the number of energy transfer steps to that level.

G L O S S A RY 349

Page 363: Whittaker. Island Biogeography. 2007

350 G L O S S A RY

turnover See species turnover.vagility The ability to move actively from oneplace to another.vicariance The separation of a population or aspecies into disjunct groups resulting from the for-mation of a physical barrier.Wallace’s line A boundary, described by AlfredRussel Wallace, across the Indonesian archipelago,separating faunas characteristic of the islands of theSunda continental shelf from those of the Sahulshelf. Many variants of Wallace’s line have been

suggested in this region, which has been termedWallacea.widow species A species that has lost its uniquepollinator (or disperser, prey, host, etc.) and is thusheading for extinction.zoochory Dispersal of seeds by animals, either through their gut (endozoochous) orattached to their skin, hair or feathers (exozoo-chorous).zygomorphic Having flowers with bilateralsymmetry.

Page 364: Whittaker. Island Biogeography. 2007

Abbott, I. (1983). The meaning of z in species/arearegressions and the study of species turnover in islandbiogeography. Oikos, 41, 385–390.

Abbott, I. and Black, R. (1980). Changes in species compo-sition of floras on islets near Perth, Western Australia.Journal of Biogeography, 7, 399–410.

Abbott, I. and Grant, P. R. (1976). Nonequilibrial birdfaunas on islands. American Naturalist, 110, 507–528.

Abdelkrim, J., Pascal, M., and Samadi, S. (2005). Islandcolonization and founder effects: the invasion of theGuadeloupe islands by ship rats (Rattus rattus).Molecular Ecology, 14, 2923–2931.

Adler, G. H. (1992). Endemism in birds of tropical Pacificislands. Evolutionary Ecology, 6, 296–306.

Adler, G. H. (1994). Avifaunal diversity and endemism ontropical Indian Ocean islands. Journal of Biogeography,21, 85–95.

Adler, G. H., Austin, C. A., and Dudley, R. (1995).Dispersal and speciation of skinks among archipelagosin the tropical Pacific Ocean. Evolutionary Ecology, 9,529–541.

Adler, G. H. and Dudley, R. (1994). Butterfly biogeogra-phy and endemism on tropical Pacific Islands. BiologicalJournal of the Linnaean Society, 52, 151–162.

Adler, G. H. and Levins, R. (1994). The island syndrome inrodent populations. Quarterly Review of Biology, 69,473–490.

Adler, G. H. and Seamon, J. O. (1991). Distribution andabundance of a tropical rodent, the spiny rat, on Islandsin Panama. Journal of Tropical Ecology, 7, 349–360.

Adler, G. H. and Wilson, M. L. (1985). Small mammals onMassachusetts islands: the use of probability functionsin clarifying biogeographic relationships. Oecologia, 66,178–186.

Adler, G. H. and Wilson, M. L. (1989). Insular distribu-tions of voles and shrews: the rescue effect and implica-tions for species co-occurence patterns. Coenoses, 4,69–72.

Adler, G. H. Wilson, M. L. and DeRosa, M. J. (1986).Influence of island area and isolation on populationcharacteristics of Peromyscus leucopus. Journal ofMammalogy, 67, 406–409.

Adsersen, H. (1995). Research on islands: classic, recent,and prospective approaches. In Islands: biological diver-sity and ecosystem function (ed. P. M. Vitousek,L. L. Loope, and H. Adsersen), Ecological Studies 115,pp. 7–21. Springer-Verlag, Berlin.

Aguilera, F., Brito, A., Castilla, C., Díaz, A., Fernández-Palacios, J.M., Rodríguez, A. Sabaté, F., and Sánchez, J.(1994). Canarias: economía ecología y medio ambiente.Francisco Lemus Editor, La Laguna.

Alonso, J. C., Alonso, J. A., and Muñoz-Pulido, R. (1994).Mitigation of bird collisions with transmission linesthrough groundwire marking. Biological Conservation,67, 129–134.

Anciães, M. and Marini, M. Â. (2000). The effects of frag-mentation on fluctuating asymmetry in passerine birdsof Brazilian tropical forests. Journal of Applied Ecology,37, 1013–1028.

Ancochea, E., Hernán, F., Cendrero, A., Cantagrel, J. M.,Fúster, J. M., Ibarrola, E., and Coello, J. (1994).Constructive and destructive episodes in the buildingof a young oceanic island, La Palma, Canary Islands,and the genesis of the Caldera de Taburiente. Journal ofVolcanology and Geothermal Research, 60, 243–262.

Anguita, F. and Hernán, F. (1975). A propagating fracturemodel versus a hot-spot origin for the Canary Islands.Earth and Planetary Science Letters, 24, 363–368.

Anguita, F. and Hernán, F. (2000).The Canary Islandsorigin: a unifying model. Journal of Volcanology andGeothermal Research, 103, 1–26.

Anguita, F., Márquez, A., Castiñeiras, P., and Hernán, F.(2002). Los volcanes de Canarias. Guía geológica e itinerar-ios. Ed. Rueda, Madrid.

Araújo, M. B., Thuiller, W., Williams, P. H., and Reginster,I. (2005a). Downscaling European species atlas distribu-tions to a finer resolution: implications for conservationplanning. Global Ecology and Biogeography, 14, 17–30.

Araújo, M. B., Whittaker, R. J., Ladle, R. J., and Erhard, M.(2005b). Reducing uncertainty in projections of extinc-tion risk from climate change. Global Ecology andBiogeography, 14, 529–538.

Arechavaleta, M., Zurita, N, Marrero, M. C., and Martín, J. L.(2005). Lista preliminar de especies silvestres de Cabo Verde

351

References

Page 365: Whittaker. Island Biogeography. 2007

(hongos, plantas y animales terrestres). Consejería deMedio ambiente y Ordenación Territorial, Gobierno deCanarias, Santa Cruz de Tenerife.

Arendt, W. J., Gibbons, D. W., and Gray, G. (1999). Statusof the volcanically threatened Montserrat Oriole Icterusoberi and other forest birds in Montserrat, West Indies.Bird Conservation International, 9, 351–372.

Arévalo, J. R., Delgado, J. D., Otto, R., Naranjo, A., Salas, M.,and Fernández-Palacios, J. M. (2005). Distribution ofalien vs. native plant species in roadside communitiesalong an altitudinal gradient in Tenerife and GranCanaria (Canary Islands). Perspectives in Plant Ecology,Evolution and Systematics, 7, 185–202.

Armstrong, P. (1982). Rabbits (Oryctolagus cuniculus) onislands: a case study of successful colonization. Journalof Biogeography, 9, 353–362.

Arnold, E. N. (1979). Indian Ocean giant tortoises: theirsystematics and island adaptations. PhilosophicalTransactions of the Royal Society of London, B 286, 127–145.

Arrhenius, O. (1921). Species and area. Journal of Ecology,9, 95–99.

Ashmole, N. P. (1963). The regulation of numbers oftropical oceanic birds. Ibis, 103, 458–478.

Ashmole, N. P. and Ashmole, M. J. (1988). Insect dispersalon Tenerife, Canary Islands: high altitude fallout andseaward drift. Arctic and Alpine Research, 20, 1–12.

Asquith, A. (1995). Evolution of Sarona (Heroptera,Miridae): Speciation on geographic and ecologicalislands. In Hawaiian Biogeography: evolution on a hot spotarchipelago (ed. W. L. Wagner and V. A. Funk), pp.90–120. Smithsonian Institution Press, Washington, DC.

Axelrod, D. I. (1975). Evolution and biogeography ofMadrean–Tethyan sclerophyll vegetation. Annals of theMissouri Botanical Garden, 62, 280–334.

Ayers, J. M., Bodmer, R. E., and Mittermeier, R. A. (1991).Financial considerations of reserve design in countrieswith high primate diversity. Conservation Biology, 5,109–114.

Backer, C. A. (1929). The problem of Krakatoa as seen by abotanist. Published by the author, Surabaya.

Badano, E. I., Regidor, H. A., Núñez, H. A., Acosta, R., andGianoli, E. (2005). Species richness and structure of antcommunities in a dynamic archipelago: effects of islandage and area. Journal of Biogeography, 32, 221–227.

Báez, M., Martín, J. L. and Oromí, P (2001). Diversidadtaxonómica terrestre. In Naturaleza de las Islas Canarias:Ecología y conservación. (ed. J. M. Fernández-Palaciosand J. L. Martín Esquivel), pp. 119–125. TurquesaEdiciones, Santa Cruz de Tenerife.

Bahn, P. and Flenley, J. R. (1992). Easter Island, Earth island.Thames & Hudson, London.

Baker, H. G. (1955). Self-compatibility and establish-ment after long-distance dispersal. Evolution, 9,347–349.

Barrett, S. C. H. (1989). Mating system evolution andspeciation in heterostylous plants. In Speciation and itsconsequences (ed. D. Otte and J. A. Endler), pp. 257–283.Sinauer, Sunderland, MA.

Barrett, S. C. H. (1996). The reproductive biology andgenetics of island plants. Philosophical Transactions of theRoyal Society of London, B 351, 725–733.

Barton, N. H. (1989). Founder effect speciation. InSpeciation and its consequences (ed. D. Otte andJ. A. Endler), pp. 229–256. Sinauer, Sunderland, MA.

Barton, N. H. and Charlesworth, B. (1984). Geneticrevolutions, founder effects, and speciation. AnnualReview of Ecology and Systematics, 15, 133–164.

Bawa, K. S. (1980). Evolution of dioecy in floweringplants. Annual Review of Ecology and Systematics, 11,15–19.

Bawa, K. S. (1982). Outcrossing and incidence of dioecismin island floras. American Naturalist, 119, 866–871.

Begon, M., Harper, J. L., and Townsend, C. R. (1986).Ecology. Blackwell, Oxford.

Begon, M., Harper, J. L., and Townsend, C. R. (1990).Ecology (2nd edn). Blackwell, Oxford.

Bell, M. and Walker, M. J. C. (1992). Late Quaternaryenvironmental change: physical and human perspectives.Longman, Harlow.

Beller, W., d’Ayala, P., and Hein, P. (ed.) (1990). Sustainabledevelopment and environmental management of smallislands, Vol. 5, Man and the Biosphere Series.UNESCO/Parthenon Publishing, Paris.

Bennett, A. F. and Gorman, G. C. (1979). Population den-sity and energetics of lizards on a tropical island.Oecologia, 42, 339–358.

Benton, T. and Spencer, T. (ed.) (1995). The Pitcairn Islands:biogeography, ecology and prehistory. Academic Press,London.

Berry, R. J. (1986). Genetics of insular populations ofmammals with particular reference to differentiationand founder effects in British small mammals. BiologicalJournal of the Linnean Society, 28, 205–230.

Berry, R. J. (1992). The significance of island biotas.Biological Journal of the Linnean Society, 46, 3–12.

Berry, R. J., Berry, A. J., Anderson, T. J. C., and Scriven, P.(1992). The house mice of Faray, Orkney. Journal ofZoology (London), 228, 233–246.

Bevanger, K. (1996). Estimates and population conse-quences of teraonid mortality caused by collisions withhigh tension power lines in Norway. Journal of AppliedEcology, 32, 745–753.

352 R E F E R E N C E S

Page 366: Whittaker. Island Biogeography. 2007

Bibby, C. J. (1995). Recent past and future extinctions inbirds. In Extinction rates (ed. J. H. Lawton and R. M. May),pp. 98–110. Oxford University Press, Oxford.

Bierdermann, R. (2003). Body size and area–incidencerelationships: is there a general pattern? Global Ecologyand Biogeography, 12, 381–387.

Blackburn, T. M., Cassey, P., Duncan, R. P., Evans, K. L.,and Gaston, K. J. (2004). Avian extinctions and mam-malian introductions on oceanic islands. Science, 305,1955–1958.

Blackburn, T. M. and Duncan, R. P. (2001). Establishmentpatterns of exotic birds are constrained by non-randompatterns in introduction. Journal of Biogeography, 28,927–939.

Blake, J. G. (1991). Nested subsets and the distribution ofbirds on isolated woodlots. Conservation Biology, 5,58–66.

Blondel, J. (1985). Breeding strategies of the Blue Tit andCoal Tit (Parus) in mainland and island Mediterraneanhabitats: A comparison. Journal of Animal Ecology, 54,531–556.

Blondel, J. and Aronson, J. (1999). Biology and wildlife ofthe Mediterranean Region. Oxford University Press,Oxford.

Bodmer, R. E., Eisenberg, J. F., and Redford, K. H. (1997).Hunting and the likelihood of extinction of Amazonianmammals. Conservation Biology, 11, 460–466.

Boecklen, W. J. (1986). Effects of habitat heterogeneity onthe species–area relationships of forest birds. Journal ofBiogeography, 13, 59–68.

Boecklen, W. J. and Gotelli, N. J. (1984). Island biogeo-graphic theory and conservation practice: species–areaor specious–area relationships? Biological Conservation,29, 63–80.

Boggs, C. L. and Murphy, D. D. (1997). Community com-position in mountain ecosystems: climatic determinantsof montane butterfly distributions. Global Ecology andBiogeography Letters, 6, 39–48.

Böhle, U.-R., Hilger, H. H., and Martin, W. F. (1996). Islandcolonization and evolution of the insular woody habitin Echium L. (Boraginaceae). Proceedings of the NationalAcademy of Sciences of the USA, 93, 11740–11745.

Borgen, L. (1979). Karyology of the Canarian flora. InPlants and islands (ed. D. Bramwell), pp. 329–346.Academic Press, London.

Borges, P. A. V., Cunha, R., Gabriel, R., Frias Martins, A,Silva L., and Vieira, V. (eds.) (2005). A list of the terrestrialfauna (Mollusca and Arthropoda) and flora (Byrophyta,Pteridophyta and Spermatophyta) from the Azores. DirecçaoRegional de Ambiente and Universidade dos Açores,Horta.

Bramwell, D. (1972). Endemism in the flora of the CanaryIslands. In Taxonomy, phytogeography and evolution (ed.D. H. Valentine) pp. 141–159. Academic Press, London.

Bramwell, D. (ed.) (1979). Plants and islands. AcademicPress, London.

Bramwell, D. and Bramwell, Z. I. (1974). Wild flowers of theCanary Islands. Stanley Thornes, Cheltenham.

Bravo, T. (1962). El Circo de las Cañadas y sus dependen-cias. Boletín de la Real Sociedad Española de HistoriaNatural, 60, 93–108.

Brochmann, C., Rustan, O. H., Lobin, W., and Kilian, N.(1997). The endemic plants of the Cape Verde Islands,W. Africa. Sommerfeltia, 24, 1–356.

Bromham, L. and Woolfit, M. (2004). Explosive radiationsand the reliability of molecular clocks: island endemicradiations as a test case. Systematic Biology, 53, 758–766.

Brooks, T. and Balmford, A. (1996). Atlantic forest extinc-tions. Nature, 380, 115.

Brooks, T. M., Mittermeier, R. A., Mittermeier, C. G., daFonseca, G. A. B., Rylands, A.B., Konstant, W. R. Flick,P., Pilgrim, J., Oldfield, S., Magin, G., and Hilton-Taylor,C. (2002) Habitat loss and extinction in the hotspots ofbiodiversity. Conservation Biology, 16, 909–923.

Brooks, T. M., Pimm, S. L., and Collar, N. J. (1997).Deforestation predicts the number of threatened birds ininsular Southeast Asia. Conservation Biology, 11, 382–394.

Brooks, T. M., Pimm, S. L., and Oyugi, J. O. (1999). Timelag between deforestation and bird extinction in trop-ical forest fragments. Conservation Biology, 13, 1140–1150.

Brothers, T. S. and Spingarn, A. (1992). Forest fragmentationand alien plant invasion of Central Indiana old-growthforests. Conservation Biology, 6, 91–100.

Brown, J. H. (1971). Mammals on mountaintops: nonequi-librium insular biogeography. American Naturalist, 105,467–478.

Brown, J. H. (1981). Two decades of homage to SantaRosalia: toward a general theory of diversity. AmericanZoologist, 21, 877–888.

Brown, J. H. (1986). Two decades of interaction betweenthe MacArthur–Wilson model and the complexities ofmammalian distribution. Biological Journal of the LinneanSociety, 28, 231–251.

Brown, J. H. (1995). Macroecology. Chicago UniversityPress, Chicago.

Brown, J. H. (1999). Macroecology: progress and prospect.Oikos, 87, 3–13.

Brown, J. H. and Gibson, A. C. (1983). Biogeography.Mosby, St Louis.

Brown, J. H. and Kodric-Brown, A. (1977). Turnover ratesin insular biogeography: effect of immigration onextinction. Ecology, 58, 445–449.

R E F E R E N C E S 353

Page 367: Whittaker. Island Biogeography. 2007

Brown, J. H. and Lomolino, M. V. (1989). Independentdiscovery of the equilibrium theory of island biogeog-raphy. Ecology, 70, 1954–1957.

Brown, J. H. and Lomolino, M. V. (1998). Biogeography, 2ndedn. Sinauer, Sunderland, MA.

Brown, J. W. (1987). The peninsular effect in BajaCalifornia: an entomological assessment. Journal ofBiogeography, 14, 359–366.

Brown, J. W. and Opler, P. A. (1990). Patterns of butterflyspecies density in peninsular Florida. Journal ofBiogeography, 17, 615–622.

Brown, M. and Dinsmore, J. J. (1988). Habitat islands andthe equilibrium theory of island biogeography: testingsome predictions. Oecologia, 75, 426–429.

Brown, P., Sutikna, T., Morwood, M. J., Soejono, R. P.,Jatmiko, Saptomo, E. W., and Due, R. A. (2004). A newsmall-bodied hominin from the late Pleistocene ofFlores, Indonesia. Nature, 431, 1055–1061.

Brown, W. L., Jr and Wilson, E. O. (1956). Characterdisplacement. Systematic Zoology, 7, 49–64.

Browne, J. and Neve, M. (1989). Introduction to theabridged version of Charles Darwin’s 1839 Voyage of theBeagle. Penguin Classics, London.

Brualdi, R. A. and Sanderson, J. G. (1999). Nested speciessubsets, gaps, and discrepancy. Oecologia, 119, 256–264.

Bruijnzeel, L. A., Waterloo, M. J., Proctor, J., Kuiters, A. T.,and Kotterink, B. (1993). Hydrological observations inmontane rain forests on Gunung Silam, Sabah, Malaysia,with special reference to the ‘Massenerhebung’ effect.Journal of Ecology, 81, 145–167.

Buckley, R. C. (1981). Scale-dependent equilibrium onhighly heterogeneous islands: plant geography ofnorthern Great Barrier Reef sand cays and shingle islets.Australian Journal of Ecology, 6, 143–148.

Buckley, R. C. (1982). The Habitat-unit model of islandbiogeography. Journal of Biogeography, 9, 339–344.

Buckley, R. C. (1985). Distinguishing the effects of areaand habitat type on island plant species richness byseparating floristic elements and substrate types andcontrolling for island isolation. Journal of Biogeography,12, 527–535.

Buckley, R. C. and Knedlhans, S. B. (1986). Beachcomberbiogeography: interception of dispersing propagules byislands. Journal of Biogeography, 13, 68–70.

Bunce, M., Worthy, T. H., Ford, T., Hoppitt, W., Willerslev, E.,Drummond, A., and Cooper, A. (2003). Extremereversed sexual size dimorphism in the extinct NewZealand moa Dinornis. Nature, 425, 172–175.

Burkey, T. V. (1993). Edge effects in seed and egg preda-tion at two neotropical rainforest sites. BiologicalConservation, 66, 139–143.

Bush, M. B. (1994). Amazonian speciation—a necessarilycomplex model. Journal of Biogeography, 21, 5–17.

Bush, M. B. (1996). Amazonian conservation in a changingworld. Biological Conservation, 76, 219–228.

Bush, M. B. (2002). Distributional change and conserva-tion on the Andean flank: a palaeoecological perspec-tive. Global Ecology and Biogeography, 11, 463–473.

Bush, M. B. and Whittaker, R. J. (1991). Krakatau: colo-nization patterns and hierarchies. Journal of Biogeography,18, 341–356.

Bush, M. B. and Whittaker, R. J. (1993). Non-equilibrationin island theory of Krakatau. Journal of Biogeography, 20,453–458.

Bush, M. B., Whittaker, R. J., and Partomihardjo, T. (1992).Forest development on Rakata, Panjang and Sertung:contemporary dynamics (1979–1989). Geojournal, 28,185–199.

Buskirk, R. E. (1985). Zoogeographic patterns and tectonichistory of Jamaica and the northern Caribbean. Journalof Biogeography, 12, 445–461.

Cabrera, J. C. (2001). Poblamiento e impacto aborigen. InNaturaleza de las Islas Canarias: Ecología y conservación. (ed.J. M. Fernández-Palacios and J. L. Martín Esquivel), pp.241–245. Turquesa Ediciones, Santa Cruz de Tenerife.

Cadena, C. D., Ricklefs, R. E., Jiménez, I., andBermingham, E. (2005). Is speciation driven by speciesdiversity? Arising from B. C. Emerson & N. KolmNature 434, 1015–1017 (2005). Nature, 438, doi:10.1038 /nature04308.

Cameron, R. A. D., Cook, L. M., and Hallows, J. D. (1996).Land snails on Porto Santo: adaptive and non-adaptiveradiation. Philosophical Transactions of the Royal Society ofLondon, B 351, 309–337.

Canals, M., Urgelés, R., Masson, D. G., and Casamor, J. L.(2000). Los deslizamientos submarinos de las IslasCanarias. Makaronesia, 2, 57–69.

Carine, M. A., Russell, S. J., Santos-Guerra, A., andFrancisco-Ortega, J. (2004). Relationships of theMacaronesian and Mediterranean floras: molecularevidence for multiple colonizations into Macaronesiaand back-colonization of the continent in Convolvulus(Convolvulaceae). American Journal of Botany, 91,1070–1085.

Carlquist, S. (1965). Island life: a natural history of the islandsof the world. Natural History Press, New York.

Carlquist, S. (1974). Island biology. Columbia UniversityPress, New York.

Carlquist, S. (1995). Introduction. In Hawaiian Bio-geography: evolution on a hot spot archipelago (ed.W. L. Wagner and V. A. Funk), pp. 1–13. SmithsonianInstitution Press, Washington, DC.

354 R E F E R E N C E S

Page 368: Whittaker. Island Biogeography. 2007

Campbell, K. and Donlan, C. J. (2005). Feral goat eradicationson islands. Conservation Biology, 19, 1362–1374.

Carr, G. D., Powell, E. A., and Kyhos, D. W. (1986). Self-incompatibility in the Hawaiian Madiinae (Compositae):an exception to Baker’s rule. Evolution, 40, 430–434.

Carr, G. D., Robichaux, R. H, Witter, M. S., and Kyhos D.W. (1989). Adaptive radiation of the Hawaiian silver-sword alliance (Compositae—Madiinae): a comparsionwith Hawaiian picture-winged Drosophila. In Genetics,speciation and the founder principle (ed. L. Y. Giddings, K.Y. Kaneshiro, and W. W. Anderson), pp. 79–95. OxfordUniversity Press, New York.

Carracedo, J. C. and Tilling, R. I. (2003). Geología yvolcanología de islas volcánicas oceánicas. Canarias–Hawai.CajaCanarias–Gobierno de Canarias.

Carracedo, J. C. (2003). El volcanismo de las IslasCanarias. In Geología y volcanología de islas volcánicasoceánicas. Canarias-Hawai (ed. J. C. Carracedo and R. I.Tilling), pp. 17–38. CajaCanarias–Gobierno de Canarias.

Carracedo, J. C., Day, S., Guillou, H., Badiola, E. R.,Canas, J. A., and Torrado, F. J. P. (1998). Hotspot volcan-ism close to a passive continental margin: the Canaryislands. Geological Magazine, 135, 591–604.

Carracedo, J. C., Day, S. J., Guillou, H., and Torrado, F. J. P.(1999). Giant Quaternary landslides in the evolution ofLa Palma and El Hierro, Canary Islands. Journal ofVolcanology and Geothermal Research, 94, 169–190.

Carranza, S., Harris, D. J., Arnold., E. N., Batista, V., andGonzalez de la Vega, J. P. (2006). Phylogeography of thelacertid lizard Psammodromos algirus, in Iberia andacross the Strait of Gibraltar. Journal of Biogeography,33, 1279–1288.

Carson, H. L. (1983). Chromosomal sequences and interis-land colonizations in Hawaiian Drosophilidae. Genetics,103, 465–482.

Carson, H. L. (1992). Genetic change after colonization.Geojournal, 28, 297–302.

Carson, H. L., Hardy, D. E., Spieth, H. T., and Stone, W. S.(1970). The evolutionary biology of the HawaiianDrosophilidae. In Essays in evolution and genetics in honorof Theodosius Dobzhansky (ed. M. K. Hecht and W. C.Steere), pp. 437–543. Appleton-Century-Crofts, NewYork.

Carson, H. L., Lockwood, J. P., and Craddock, E. M.(1990). Extinction and recolonization of local popula-tions on a growing shield volcano. Proceedings of theNational Academy of Sciences of the USA, 87, 7055–7057.

Carson, H. L. and Templeton, A. R. (1984). Genetic revo-lutions in relation to speciation phenomena: the found-ing of new populations. Annual Review of Ecology andSystematics, 15, 97–131.

Carvajal, A. and Adler, G. H. (2005). Biogeography ofmammals on tropical Pacific islands. Journal ofBiogeography, 32, 1561–1569.

Case, T. J., Bolger, D. T., and Richman, A. D. (1992).Reptilian extinctions: the last ten thousand years. InConservation Biology: the theory and practice of nature con-servation and management (ed. P. L. Fielder and S. K.Jain), pp. 91–125. Chapman & Hall, New York.

Case, T. J. and Cody, M. L. (1987). Testing theories ofisland biogeography. American Scientist, 75, 402–411.

Cassey, P., Blackburn, T. M., Duncan, R. P., and Lockwood,J. L. (2005a). Lessons from the establishment of exoticspecies: a meta-analytical case study using birds.Journal of Animal Ecology, 74, 250–258.

Cassey, P., Blackburn, T. M., Duncan, R. P., and Gaston, K.J. (2005b). Causes of exotic bird establishment acrossoceanic islands. Proceedings of the Royal Society of London,B, 272, 2059–2063.

Caswell, H. (1978). Predator-mediated coexistence: a non-equilibrium model. American Naturalist, 112,127–154.

Caswell, H. (1989). Life-history strategies. In Ecologicalconcepts: the contribution of ecology to an understanding ofthe natural world (ed. J. M. Cherrett), pp. 285–307.Blackwell Scientific Publications, Oxford.

Causton, C. (2005). Galapagos News, Winter 2005, p.1.Chang, M. (2000). Exclusive economic zones. In:

�http://geography.about.com/library/misc/uceez.htm� (visited March 2006)

Christensen, U. (1999). Fixed hotspots gone with thewind. Nature, 391, 739–740.

Christian, C. S. (1993). The challenge of parrot conserva-tion in St Vincent and the Grenadines. Journal ofBiogeography, 20, 463–469.

Christian, C. S., Lacher, T. E., Jr, Zamore, M. P., Potts, T. D.,and Burnett, G. W. (1996). Parrot conservation in theLesser Antilles with some comparison to the PuertoRican efforts. Biological Conservation, 77, 159–167.

Christiansen, M. B. and Pitter, E. (1997). Species loss in a for-est bird community near Lagoa Santa in SoutherasternBrazil. Biological Conservation, 80, 23–32.

CI (2005). Conservation International hotspots.�www.biodiversityhotspots.org/xp/Hotspots/hotspotsScience/hotspots_revisited.xml�. Visited 8 June2005.

Clague, D. A. (1996). The growth and subsidence of theHawaiian-Emperor volcanic chain. In The origin andevolution of Pacific Island Biotas, New Guinea toEastern Polynesia: patterns and processes (ed. A. Keast,and S. E. Miller), pp. 35–50. SPB Academic Publishing,Amsterdam.

R E F E R E N C E S 355

Page 369: Whittaker. Island Biogeography. 2007

Clarke, B. and Grant, P. R. (ed.) (1996). Evolution onislands. Philosophical Transactions of the Royal Society ofLondon, B, 351, 723–854.

Clarke, B., Johnson, M. S., and Murray, J. (1996). Clines inthe genetic distance between two species of island landsnails: how ‘molecular leakage’ can mislead us aboutspeciation. Philosophical Transactions of the Royal Societyof London, B, 351, 773–784.

Clegg, S. M., Degnan, S. M., Kikkawa, J., Moritz, C.,Estoup, A., and Owens, I. P. F. (2002). Geneticconsequences of sequential founder events by anisland-colonizing bird. Proceedings of the NationalAcademy of Sciences of the USA, 99, 8127–8132.

Cody, M. L. (1971). Ecological aspects of reproduction. InAvian Biology (ed. D. S. Farner and J. R. King).pp. 461–512. Academic Press, London.

Cody, M. L. and Overton, J. McC. (1996). Short-term evo-lution of reduced dispersal in island plant populations.Journal of Ecology, 84, 53–61.

Colinvaux, P. A. (1972). Climate and the GalápagosIslands. Nature, 219, 590–594.

Clout, M. N. and Craig, J. L. (1995). The conservation ofcritically endangered flightless birds in New Zealand.Ibis, 137, S181–190.

Colwell, R. K. and Winkler, D. W. (1984). A null model fornull models in biogeography. In Ecological communities:conceptual issues and the evidence (ed. D. R. Strong,Jr. D. Simberloff, L. G. Abele, and A. B. Thistle), pp.344–359. Princeton University Press, Princeton, NJ.

Compton, S. G., Ross, S. J., and Thornton, I. W. B. (1994).Pollinator limitation of fig tree reproduction on theisland of Anak Krakatau (Indonesia). Biotropica, 26,180–186.

Conant, S. (1988). Geographic variation in the Laysanfinch (Telepyza cantans). Evolutionary Ecology, 2, 270–282.

Connell, J. H. (1980). Diversity and the coevolution ofcompetitors, or the ghost of competition past. Oikos, 35,131–138.

Connor, E. F. and McCoy, E. D. (1979). The statistics andbiology of the species–area relationship. AmericanNaturalist, 113, 791–833.

Connor, E. F., McCoy, E. D., and Cosby, B. J. (1983). Modeldiscrimination and expected slope values inspecies–area studies. American Naturalist, 122, 789–796.

Connor, E. F. and Simberloff, D. (1979). The assembly ofspecies communities: chance or competition? Ecology,60, 1132–1140.

Cook, L. G. and Crisp, M. D. (2005). Directional asymme-try of long-distance dispersal and colonization couldmislead reconstructions of biogeography. Journal ofBiogeography, 32, 741–754.

Cook, R. R. (1995). The relationship between nestedsubsets, habitat subdivision, and species diversity.Oecologia, 101, 204–210.

Cook, R. R. and Quinn, J. F. (1995). The influence of colo-nization in nested species subsets. Oecologia, 102,413–425.

Cook, W. M., Lane, K. T., Foster, B. L., and Holt, R. D.(2002). Island theory, matrix effects and species richnesspatterns in habitat fragments. Ecology Letters, 5,619–623.

Cordeiro, N. J. and Howe, H. F. (2001). Low recruitment oftrees dispersed by animals in African forest fragments.Conservation Biology, 15, 1733–1741.

Corlett, R. T. (1992). The ecological transformation ofSingapore, 1819–1990. Journal of Biogeography, 19,411–420.

Cox, C. B. (2001). The biogeographic regions reconsidered.Journal of Biogeography, 28, 511–523.

Cox, C. B. and Moore, P. D. (1993). Biogeography: an ecolog-ical and evolutionary approach (5th edn). BlackwellScientific Publications, Oxford.

Cox, G. W. (1999). Alien species in North America andHawaii: Impacts on natural ecosystems. Island Press,Washington, DC.

Cox, G. W. and Ricklefs, R. E. (1977). Species diversity,ecological release, and community structuring inCaribbean land bird faunas. Oikos, 28, 113–122.

Cox, P. A. and Elmqvist, T. (2000). Pollinator extinction inthe Pacific Islands. Conservation Biology, 14, 1237–1239.

Cox, P. A., Elmqvist, T., Pierson, E. D., and Rainey, W. E.(1992). Flying foxes as pollinators and seed dispersersin Pacific Island ecosystems. In Pacific island flying foxes:Proceedings of an international conservation conference,Biological Report 90 (23), July 1992 (ed. D. E. Wilsonand G. L. Graham), pp. 18–23. US Department of theInterior, Fish and Wildlife Service, Washington, DC.

Cramp, S. and Perrins, C. M. (eds.) (1994). Handbook of thebirds of Europe, the Middle East and North Africa. The birdsof the Western Palearctic. Vol. VIII. Crows to finches.Oxford University Press, Oxford.

Crawford, D. J., Sang, T., Stuessy, T. F., Kim, S. C., andSilva, M. (1998). Dendroseris (Asteraceae: Lactucaceae)and Robinsonia (Asteraceae: Senecioneae) on the JuanFernández Islands: similarities and differences in biol-ogy and phylogeny. In Evolution and speciation of islandplants. (ed. T. F. Stuessy and M. Ono) pp. 97–119.Cambridge University Press, Cambridge.

Crawford, D. J., Stuessy, T. F., Haines, D. W., Cosner, M. B.O., Silva, M., and López, P. (1992). Allozyme diversitywithin and divergence among four species of Robinsonia(Asteraceae: Senecioneae), a genus endemic to the Juan

356 R E F E R E N C E S

Page 370: Whittaker. Island Biogeography. 2007

Fernandez islands, Chile. American Journal of Botany, 79,962–966.

Cronk, Q. C. B. (1987). History of the endemic flora of St.Helena: a relictual series. New Phytologist, 105, 509–520.

Cronk, Q. C. B. (1989). The past and present vegetation ofSt. Helena. Journal of Biogeography, 16, 47–64.

Cronk, Q. C. B. (1990). History of the endemic flora ofSt. Helena: late Miocene ‘Trochetiopsis-like’ pollen fromSt. Helena and the origin of Trochetiopsis. NewPhytologist, 114, 159–165.

Cronk, Q. C. B. (1992). Relict floras of Atlantic islands:patterns assessed. Biological Journal of the LinneanSociety, 46, 91–103.

Cronk, Q. C. B. and Fuller, J. L. (1995). Plant invaders.Chapman & Hall, London.

Crooks, K. R. and Soulé, M. E. (1999). Mesopredatorrelease and avifaunal extinctions in a fragmentedsystem. Nature, 400, 563–566.

Crowell, K. L. (1962). Reduced interspecific competitionamong the birds of Bermuda. Ecology, 43, 75–88.

Crowell, K. L. (1986). A comparison of relict versus equi-librium models for insular mammals of the Gulf ofMaine. Biological Journal of the Linnean Society, 28, 37–64.

Cruz, F., Donlan, C. J., Campbell, K., and Carrión, V.(2005). Conservation action in the Galápagos: feral pig(Sus scrofa) eradication from Santiago Island. BiologicalConservation, 121, 473–478.

Cuddihy, L. W. and Stone, C. P. (1990). Alteration of nativeHawaiian vegetation: effects of humans, their activities andintroductions. University of Hawaii CooperativeNational Park Resources Studies Unit, 3190 Maile Way,Honolulu, Hawaii.

Cutler, A. (1991). Nested fauna and extinction in frag-mented habitats. Conservation Biology, 5, 496–505.

Daily, G. C., Ceballos, G., Pacheco, J., Suzán, G., andSánchez-Azofeifa, A. (2003). Countryside biogeographyof Neotropical mammals: conservation opportunities inagricultural landscapes of Costa Rica. ConservationBiology, 17, 1814–1826.

Daily, G. C., Ehrlich, P. R., and Sánchez-Azofeifa, A.(2001). Countryside biogeography: use of human-dom-inated habitats by the avifauna of southern Costa Rica.Ecological Applications, 11, 1–13.

D’Antonio, C. M. and Dudley, T. L. (1993). Alien species:the insidious invasion of ecosystems by plants and ani-mals from around the world has become a major envi-ronmental problem. Pacific Discovery 1993, Summer,9–11.

D’Antonio, C. M. and Dudley, T. L. (1995). Biological inva-sions as agents of change on islands versus mainlands.In Islands: biological diversity and ecosystem function (ed.

P. M. Vitousek, L. L. Loope, and H. Adsersen),Ecological Studies, 115, pp. 103–121. Springer-Verlag,Berlin.

Dammerman, K. W. (1948). The fauna of Krakatau(1883–1933). Verhandelingen der Konninklijke NederlandseAkademie Van Wetenschappen Afdeling Natuurkunde(II), 44, 1–594. N. V. Noord-Hollandsche UitgeversMaatschappij, Amsterdam.

Damuth, J. (1993). Cope’s rule, the island rule and thescaling of mammalian population density. Nature, 365,748–750.

Danley, P. D. and Kocher, T. D. (2001). Speciation inrapidly diverging systems: Lessons from Lake Malawi.Molecular Ecology, 10, 1075–1086.

Darlington, P. J. (1957). Zoogeography: the geographicaldistribution of animals. Wiley, New York.

Darwin, C. (1842). The structure and distribution of coralreefs. Being the first part of the geology of the voyage of theBeagle, under the command of Capt. Fitzroy, R. N., duringthe years 1832–36. Smith, Elder, & Company, London.

Darwin, C. (1845). Voyage of the Beagle. (First published1839. Abridged version edited by J. Browne andM. Neve (1989). Penguin Books, London.)

Darwin, C. (1859). On the origin of species by means ofnatural selection. J. Murray, London. (Page numberscited here are from an edition published by Avenel,New York, in 1979 under the title The origin of species.)

Davis, S. D., Heywood, V. H., and Hamilton, A. C. (ed.)(1994). Centres of plant diversity: a guide and strategy fortheir conservation. Volume 1: Europe, Africa, South WestAsia and the Middle East. WWF and IUCN, Cambridge.

Davis, S. D., Heywood, V. H., and Hamilton, A. C. (ed.)(1995). Centres of plant diversity: a guide and strategy fortheir conservation. Volume 2: Asia, Australasia and thePacific. WWF and IUCN, Cambridge.

Dawson, D. (1994). Are habitat corridors conduits foranimals and plants in a fragmented landscape?A review of the scientific evidence. English NatureResearch Reports, No. 94.

Decker, R. W. and Decker, B. B. (1991). Mountains of fire:The nature of volcanoes. Cambridge University Press,Cambridge.

DeJoode, D. R. and Wendel, J. F. (1992). Genetic diversityand the origin of the Hawaiian Islands cotton Gossypiumtomentosum. American Journal of Botany, 79, 1311–1319.

Delcourt, H. R. and Delcourt, P. A. (1988). Quaternarylandscape ecology: relevant scales in space and time.Landscape Ecology, 2, 23–44.

Delcourt, H. R. and Delcourt, P. A. (1991). Quaternaryecology: A paleoecological perspective. Chapman & Hall,London.

R E F E R E N C E S 357

Page 371: Whittaker. Island Biogeography. 2007

Delgado, J. D. (2000). Patterns of insect flower visitation inLavandula buchii Webb (Lamiaceae), an endemic shrubof Tenerife (Canary Islands). Journal of Natural History,34, 2145–2155.

Delgado, J. D., Arévalo, J. R., and Fernández-Palacios, J. M.(2001). Road and topography effects on invasion: edgeeffects in rat foraging patterns in two oceanic islandforests (Tenerife, Canary Islands). Ecography, 24,539–546.

Delgado, J. D., Arévalo, J. R., and Fernández-Palacios, J. M.(2005). Patterns of artificial avian nest predation byintroduced rats in a fragmented laurel forest (Tenerife,Canary Islands). Journal of Natural History, 38, 2661–2669.

Delgado, J. D., Arévalo, J. R. and Fernández-Palacios, J. M.(in press) Road edge effect on the abundance of thelizard Gallotia galloti (Sauria: Lacertidae) in two CanaryIslands forests. Biodiversity and Conservation. In press.

Desender, K., Baert, L., Maelfait, J.-P. and Verdyck, P.(1999). Conservation on Volcán Alcedo (Galápagos): ter-restrial invertebrates and the impact of introduced feralgoats. Biological Conservation, 87, 303–310.

Deshaye, J. and Morisset, P. (1988). Floristic richness, area,and habitat diversity in a hemiarctic archipelago.Journal of Biogeography, 15, 721–728.

Diamond, J. M. (1969). Avifaunal equilibria and speciesturnover rates on the Channel Islands of California.Proceedings of the National Academy of Sciences of the USA,64, 57–63.

Diamond, J. M. (1972). Biogeographical kinetics: estima-tion of relaxation times for avifaunas of southwestPacific islands. Proceedings of the National Academy ofSciences of the USA, 69, 3199–3203.

Diamond, J. M. (1974). Colonization of exploded volcanicislands by birds: the supertramp strategy. Science, 184,803–806.

Diamond, J. M. (1975a). Assembly of species communities.In Ecology and evolution of communities (ed. M. L. Codyand J. M. Diamond), pp. 342–444. Harvard UniversityPress, Cambridge, MA.

Diamond, J. M. (1975b). The island dilemma: lessons ofmodern biogeographic studies for the design of naturalpreserves. Biological Conservation, 7, 129–146.

Diamond, J. M. (1977). Continental and insular speciationin Pacific land birds. Systematic Zoology, 26, 263–268.

Diamond, J. M. (1984). ‘Normal’ extinctions of isolatedpopulations. In Extinctions (ed. M. H. Nitecki), pp.191–246. University of Chicago Press, Chicago.

Diamond, J. M. (1986). The design of a nature reserve sys-tem for Indonesian New Guinea. In Conservation biology:the science of scarcity and diversity (ed. M. Soulé), pp.485–503. Sinauer Associates, Sunderland, MA.

Diamond, J. M. (1991a). A new species of rail from theSolomon islands and convergent evolution of insularflightlessness. Auk, 108, 461–470.

Diamond, J. M. (1991b). The rise and fall of the third chim-panzee. Vintage, London.

Diamond, J. M. (1996). Daisy gives an evolutionaryanswer. Nature, 380, 103–104.

Diamond, J. M. (2005). Collapse: how societies choose to fail orsurvive. Allen Lane/Penguin, London.

Diamond, J. M. and Gilpin, M. E. (1982). Examination ofthe ‘null’ model of Connor and Simberloff for speciesco-occurrences on islands. Oecologia, 52, 64–74.

Diamond, J. M. and Gilpin, M. E. (1983). Biogeographicalumbilici and the origin of the Philippine avifauna.Oikos, 41, 307–321.

Diamond, J. M. and May, R. M. (1977). Species turnoverrates on islands: dependence on census interval. Science,197, 266–270.

Diamond, J. M. and May, R. M. (1981). Island biogeographyand the design of nature reserves. In Theoretical Ecology(2nd edn) (ed. R. M. May), pp. 228–252. Blackwell, Oxford.

Diamond, J. and Pimm, S. (1993). Survival times of birdpopulations: a reply. American Naturalist, 142, 1030–1035.

Diamond, J. M., Pimm, S. L., Gilpin, M. E., and LeCroy, M.(1989). Rapid evolution of character displacement inMyzomelid Honeyeaters. American Naturalist, 134,675–708.

Dias, E. (1996). Vegetaçao natural dos Açores. Ecologia e sin-taxonomia das florestas naturais. Doctoral thesis, AzoresUniversity.

Diehl, S. and Bush, G. L. (1989). The role of habitat prefer-ence in adaptation and speciation. In Speciation and itsconsequences (ed. D. Otte and J. A. Endler), pp. 345–365.Sinauer, Sunderland, MA.

Doak, D. F. and Mills, L. S. (1994). A useful role for theoryin conservation. Ecology, 75, 615–626.

Docters van Leeuwen, W. M. (1936). Krakatau 1883–1933.Annales du jardin botanique de Buitenzorg, 46–47, 1–506.

Drotz, M. K. (2003). Speciation and mitochondrial DNAdiversification of the diving beetles Agabus bipustulatusand A. wollastoni (Coleoptera, Dytiscidae) withinMacaronesia. Biological Journal of the Linnean Society, 79,653–666.

Duncan, R. P. and Blackburn, T. M. (2002). Morphologicalover-dispersion in game birds (Aves: Galliformes) suc-cessfully introduced to New Zealand was not caused byinterspecific competition. Evolutionary Ecology Research,4, 551–561.

Duncan, R. P., Blackburn, T. M., and Sol, D. (2003). Theecology of bird introductions. Annual Review of Ecology,Evolution and Systematics, 34, 71–98.

358 R E F E R E N C E S

Page 372: Whittaker. Island Biogeography. 2007

Dunn, C. P. and Loehle, C. (1988). Species–area parameterestimation: testing the null model of lack of relation-ship. Journal of Biogeography, 15, 721–728.

Dunstan, C. E. and Fox, B. J. (1996). The effects of frag-mentation and disturbance of rainforest on ground-dwelling small mammals on the Robertson Plateau,New South Wales, Australia. Journal of Biogeography, 23,187–201.

Ecker, A. (1976). Groundwater behaviour in Tenerife,volcanic island (Canary Islands, Spain). Journal ofHydrology, 28, 73–86.

Economist (2001). Nauru: Paradise well and truly lost. TheEconomist, 20 December2001. �www.economist.com/ dis-playstory.cfm?story_id � 884045� (visited March 2006)

Ehrendorfer, E. (1979). Reproductive biology in islandplants. In Plants and islands (ed. D. Bramwell), pp.293–306. Academic Press, London.

Ehrlich, P. R. and Hanski, I. (eds) (2004). On the wings ofcheckerspots: a model system for population biology. OxfordUniversity Press, Oxford.

Elisens, W. J. (1992). Genetic divergence in Galvezia(Scrophulariaceae): evolutionary and biogeographicrelationships among South American and Galápagosspecies. American Journal of Botany, 79, 198–206.

Elmqvist, T., Rainey, W. E., Pierson, E. D., and Cox, P. A.(1994). Effects of tropical cyclones Ofa and Val on thestructure of a Samoan lowland rain forest. Biotropica, 26,384–391.

Elvers, I. (1977). Flower-visiting lizards on Madeira.Botaniska Notiser, 130, 231–234.

Emerson, B. C. (2002). Evolution on oceanic islands:molecular phylogenetic approaches to understandingpattern and process. Molecular Ecology, 11, 951–966.

Emerson, B. C. and Kolm, N. (2005a). Species diversity candrive speciation. Nature, 434, 1015–1017.

Emerson, B. C. and Kolm, N. (2005b). Emerson and Kolmreply. doi: 10.1038/nature04309.

Erickson, J. D. and Gowdy, J. M. (2000). Resource use,institutions, and sustainability: a tale of two Pacificisland cultures. Land Economics, 76, 345–354.

Ernst, A. (1908). The new flora of the volcanic island ofKrakatau (translated into English by A. C. Seward).Cambridge University Press, Cambridge.

Erwin, R. M., Hatfield, J. S., and Winters, T. J. (1995). Thevalue and vulnerability of small estuarine islands forconserving metapopulations of breeding waterbirds.Biological Conservation, 71, 187–191.

Estrada, A., Coates-Estrada, R., and Meritt, D. (1993). Batspecies richness and abundance in tropical rain forestfragments and in agricultural habitats at Los Tuxtlas,Mexico. Ecography, 16, 309–318.

Ewers, R. M. and Didham, R. K. (2005). Confoundingfactors in the detection of species responses to habitatfragmentation. Biological Reviews, doi: 10.1017/S1464793105006949.

Fahrig, L. and Merriam, G. (1994). Conservation of frag-mented populations. Conservation Biology, 8, 50–59.

Feeley, K. (2003). Analysis of avian communities in LakeGuri, Venezuala, using multiple assembly rule models.Oecologia, 137, 104–113.

Fernández-Palacios, J. M. (1992). Climatic responses ofplant species on Tenerife, the Canary Islands. Journal ofVegetation Science, 3, 595–602.

Fernández-Palacios, J. M. (2004). Introducción a las islas.In Ecología Insular/Island Ecology (ed. J. M. Fernández-Palacios and C. Morici), pp. 21–55. Asociación Españolade Ecología Terrestre-Cabildo Insular de La Palma,Santa Cruz de La Palma.

Fernández-Palacios, J. M. and Andersson, C. (2000).Geographical determinants of the biological richness inthe Macaronesian region. Acta Phytogeographica Suecica,85, 41–49.

Fernández-Palacios, J. M. and Arévalo, J. R. (1998). Treestrategies regeneration of the trees in the laurel forest ofTenerife (The Canary Islands). Plant Ecology, 127, 21–29.

Fernández-Palacios, J. M., Arévalo J. R., Delgado, J. D.,and Otto, R. (2004a). Canarias: Ecología, medio ambiente ydesarrollo. Centro de la Cultura Popular de Canarias, LaLaguna.

Fernández-Palacios, J. M., Arévalo, J. R., González-Delgado, G., Delgado, J. D., and Otto, R. (2004b).Estrategias de regeneración en la laurisilva. Makaronesia,6, 90–101.

Fernández-Palacios, J. M. and Dias, E. (2001). El marcobiogeográfico macaronésico. In Naturaleza de las IslasCanarias: Ecología y conservación. (ed. J. M. Fernández-Palacios and J. L. Martín Esquivel), pp. 45–52. TurquesaEdiciones, Santa Cruz de Tenerife.

Fernández-Palacios, J. M. and de Nicolás, J. P. (1995).Altitudinal pattern of vegetation variation on Tenerife.Journal of Vegetation Science, 6, 183–190.

Fernandopullé, D. (1976). Climate characteristics of theCanary Islands. In Biogeography and ecology in the CanaryIslands (ed. G. Kunkel), pp. 185–206. W. Junk, TheHague.

Fiedler, P. L. and Jain, S. K. (ed.) (1992). ConservationBiology: The theory and practice of nature conservation,preservation and management. Chapman & Hall, London.

Field, R., O’Brien, E. M., and Whittaker, R. J. (2005). Globalmodels for predicting woody plant richness from cli-mate: development and evaluation. Ecology, 86,2263–2277.

R E F E R E N C E S 359

Page 373: Whittaker. Island Biogeography. 2007

Fischer, J. and Lindenmayer, D. B. (2005). Nestedness infragmented landscapes: a case study on birds, arborealmarsupials and lizards. Journal of Biogeography, 32,1737–1750.

Fisher, R. A., Corbet, A. S., and Williams, C. B. (1943). Therelation between the number of species and the numberof individuals in a random sample of an animal popu-lation. Journal of Animal Ecology, 12, 42–58.

Flenley, J. R. (1993). The palaeoecology of Easter Island,and its ecological disaster. In Easter Island studies.Contributions to the history of Rapanui in memory ofWilliam T. Mulloy (ed. S. R. Fisher), pp. 27–45. Oxbow,Oxford.

Flenley, J. R., King, S. M., Jackson, J., Chew, C., Teller, J. T.,and Prentice, M. E. (1991). The Late Quaternary vegeta-tional and climatic history of Easter Island. Journal ofQuaternary Science, 6, 85–115.

Forsman, A. (1991). Adaptive variation in head size inVipera berus L. populations. Biological Journal of theLinnean Society, 43, 281–296.

Fosberg, F. R. (1948). Derivation of the flora ofthe Hawaiian Islands. In Insects of Hawaii, Vol. 1 (ed.E. C. Zimmerman), pp. 107–119. University of HawaiiPress, Honolulu.

Fox, B. J. and Fox, M. D. (2000). Factors determining mam-mal species richness on habitat islands and isolates:habitat diversity, disturbance, species interactions andguild assembly rules. Global Ecology and Biogeography,9, 19–37.

Francisco-Ortega, J., Jansen, R. K., and Santos-Guerra, A.(1996). Chloroplast DNA evidence of colonization,adaptive radiation and hybridization in the evolution ofthe Macaronesian flora. Proceedings of the NationalAcademy of Sciences of the USA, 24, 249–266.

Francisco-Ortega, J. and Santos-Guerra, A. (2001). Genes yconservación de plantas vasculares. In Naturalezade las Islas Canarias: Ecología y conservación. (ed.J. M. Fernández-Palacios and J. L. Martín Esquivel),pp. 357–365. Turquesa Ediciones, Santa Cruz deTenerife.

Francisco-Ortega, J., Santos-Guerra, A., Kim, S. C., andCrawford, D. J. (2000). Plant genetic diversity in theCanary Islands: a conservation perspective. AmericanJournal of Botany, 87, 909–919.

Frankham, R. (1997). Do island populations have lessgenetic variation than mainland populations? Heredity,78, 311–327.

Franklin, J. and Steadman, D. W. (1991). The potential forconservation of Polynesian birds through habitat map-ping and species translocation. Conservation Biology,5, 506–521.

Fritts, T. H. and Leasman-Tanner, D. (2001). The browntreesnake on Guam: How the arrival of one invasivespecies damaged the ecology, commerce, electrical sys-tems, and human health on Guam: A comprehensiveinformation source. Available online: �www.fort.usgs.gov/resources/education/bts/bts_home.asp� (visitedJune 2005).

Fu, Y. (2000). Toward the phylogeny of the familyLacertidae—why 4708 base pairs of mtDNA sequencescannot draw the picture. Biological Journal of the LinneanSociety, 71, 203–217.

Funk, V. A. and Wagner, W. L. (1995). Biogeographic pat-terns in the Hawaiian Islands. In Hawaiian biogeography.Evolution on a hot spot archipelago (ed. W. L. Wagner andV. L. Funk). pp. 379–419. Smithsonian, Washington, DC.

Fujita, M. S. and Tuttle, M. D. (1991). Flying foxes(Chiroptera: Pteropodidae): threatened animals of keyecological and economic importance. ConservationBiology, 5, 455–463.

Futuyma, D. (1986). Evolutionary biology. SinauerAssociates, Sunderland, MA.

Galapagos Newsletter (1996). No. 3, Autumn. GalapagosConservation Trust, 18 Curzon Street, London.

Galley, C. and Linder, H. P. (2006). Geographical affinitiesof the Cape flora, South Africa. Journal of Biogeography,33, 236–250.

García-Falcón, J. M. and Medina-Muñoz, D. (1999).Sustainable tourism development in islands: a casestudy of Gran Canaria. Business Strategy and theEnvironment, 8, 336–357.

García-Talavera, F. (1999). La Macaronesia. Consider-aciones geológicas, biogeográficas y paleoecológicas.In Ecología y cultura en Canarias.(ed. J. M. Fernández-Palacios, J. J. Bacallado, and J. A. Belmonte), pp. 39–63.Ed. Organismo Autónomo de Museos y Centros,Cabildo Insular de Tenerife, Santa Cruz de Tenerife,Tenerife.

Gardner, A. S. (1986). The biogeography of the lizards of theSeychelles Islands. Journal of Biogeography, 13, 237–253.

Gascon, C., Lovejoy, T. E., Bierregaard, Jr. R. O., Malcolm,J. R., Stouffer, P. C., Vasconcelos, H., Laurance, W. F.,Zimmerman, B., Tocher, M., and Borges, S. (1999).Matrix habitat and species persistence in tropical forestremnants. Biological Conservation, 91, 223–230.

Gaston, K. J. (1991). Estimates of the near imponderable: areply to Erwin. Conservation Biology, 5, 564–566.

Gates, S. and Donald, P. F. (2000). Local extinction ofBritish farmland birds and the prediction of furtherloss. Journal of Applied Ecology, 37, 806–820.

Gentile, G. and Argano, R. (2005). Island biogeographyof the Mediterranean sea: the species-area relationship

360 R E F E R E N C E S

Page 374: Whittaker. Island Biogeography. 2007

for terrestrial isopods. Journal of Biogeography, 32,1715–1726.

Gibbs, H. L. and Grant, P. R. (1987). Ecological con-sequences of an exceptionally strong El Niño event onDarwin’s finches. Ecology, 68, 1735–1746.

Gibson, C. W. D. (1986). Management history in relation tochanges in the flora of different habitats on anOxfordshire estate, England. Biological Conservation, 38,217–232.

Gilbert, F. S. (1980). The equilibrium theory of islandbiogeography, fact or fiction? Journal of Biogeography, 7,209–235.

Gilpin, M. E. and Diamond, J. M. (1982). Factors con-tributing to non-randomness in species co-occurenceson islands. Oecologia, 52, 75–84.

Gittenberger, E. (1991). What about non-adaptive radia-tion? Biological Journal of the Linnean Society, 43, 263–272.

Givnish, T. J. (1998). Adaptive plant evolution on islands.In: Evolution on islands (ed. P. Grant) pp. 281–304. OxfordUniversity Press, Oxford.

Godwin, H. (1975). The history of the British flora: a factualbasis for phytogeography (2nd edn). CambridgeUniversity Press, Cambridge.

Golinski, M. and Boecklen, W. J. (2006). A model-inde-pendent test for the presence of regulatory equilibriumand non-random structure in island species trajectories.Journal of Biogeography, 33, 1566–1570.

González, P., Pinto, F., Nogales, M., Jiménez-Asensio, J.,Hernández, M., and Cabrera, V.M. (1996). Phylogeneticrelationships of the Canary Islands endemic lizardgenus Gallotia (Sauria: Lacertidae), inferred from mito-chondrial DNA sequences. Molecular Phylogenetics andEvolution, 6, 63–71.

González-Pérez, M. A., Caujapé-Castells, J., and Sosa, P. A.(2004). Molecular evidence of hybridisation betweenthe endemic Phoenix canariensis and the widespreadPhoenix dactylifera with Random AmplifiedPolymorphic DNA (RAPD). Plant Systematics andEvolution, 247, 165–175.

Goodfriend, G. A., Cameron, R. A. D., and Cook, L. M.(1994). Fossil evidence of recent human impact on theland snail fauna of Madeira. Journal of Biogeography, 21,309–320.

Goodwillie, C. (1999). Wind pollination and reproductiveassurance in Linanthus parviflorus (Polemoniaceae), aself-compatible annual. American Journal of Botany, 86,948–954.

Gorchov, D. L., Cornejo, F., Ascorra, C., and Jaramillo, M.(1993). The role of seed dispersal in the natural regener-ation of rain forest after strip-cutting in the PeruvianAmazon. Vegetatio, 107/108, 339–349.

Gorman, M. L. (1979). Island ecology. Chapman & Hall,London.

Gotelli, N. J. (1991). Metapopulation models: the rescueeffect, the propagule rain, and the core-satellite hypoth-esis. American Naturalist, 138, 768–776.

Gotelli, N. J. and Entsminger, G. L. (2001). Swap and fillalgorithms in null model analysis: rethinking theknight’s tour. Oecologia, 129, 281–291.

Gotelli, N. J. and Graves, G. R. (1996). Null models in ecol-ogy. Smithsonian Institution Press, Washington, DC.

Gotelli, N. J. and McCabe, D. J. (2002). Species co-occurrence:a meta-analysis of J. M. Diamond’s assembly rulesmodel. Ecology, 83, 2091–2096.

Grainger, A. (1993). Rates of deforestation in the humidtropics: estimates and measurements. GeographicalJournal, 159, 33–44.

Grant, B. R. and Grant, P. R. (1996a). High survival ofDarwin’s finch hybrids: effects of beak morphology anddiets. Ecology, 77, 500–509.

Grant, P. R. (1981). Speciation and the adaptive radiationof Darwin’s finches. American Scientist, 69, 653–663.

Grant, P. R. (1984). The endemic land birds. In Galapagos(ed. R. Perry), Key Environments Series, pp. 175–189.IUCN/Pergamon Press, Oxford.

Grant, P. R. (1986). Interspecific competition in fluctuatingenvironments. In Community Ecology (ed. J. M. Diamondand T. J. Case), pp. 173–191. Harper & Row, New York.

Grant, P. R. (1994). Population variation and hybridiza-tion: comparison of finches from two archipelagos.Evolutionary Ecology, 8, 598–617.

Grant, P. R. (1998). Competition exposed by knight?Nature, 396, 216–217.

Grant, P. R. (2002). Founder effects and silvereyes.Proceedings of the National Academy of Sciences of the USA,99, 7818–7820.

Grant, P. R. and Grant, B. R. (1989). Sympatric speciationand Darwin’s finches. In Speciation and its consequences(ed. D. Otte and J. A. Endler), pp. 433–457. SinauerAssociates, Sunderland, MA.

Grant, P. R. and Grant, B. R. (1994). Phenotypic andgenetic effects of hybridization in Darwin’s finches.Evolution, 48, 297–316.

Grant, P. R. and Grant, B. R. (1996b). Speciation andhybridization in island birds. Philosophical Transactionsof the Royal Society of London, B 351, 765–772.

Graves, G. R. and Gotelli, N. J. (1983). Neotropical land-bridge avifaunas: new approaches to null hypotheses inbiogeography. Oikos, 41, 322–333.

Gray, J. S., Ugland, K. I., and Lambshead, J. (2004). Onspecies accumulation and species-area curves. GlobalEcology and Biogeography, 13, 567–568.

R E F E R E N C E S 361

Page 375: Whittaker. Island Biogeography. 2007

Grayson, D. L. and Livingston, S. D. (1993). Missingmammals on Great Basin mountains: Holocene extinc-tions and inadequate knowledge. Conservation Biology,7, 527–532.

Greenslade, P. J. M. (1968). Island patterns in the SolomonIslands bird fauna. Evolution, 22, 751–761.

Grelle, C. E. V., Alves, M. A. S., Bergallo, H. G., Geise, L.,Rocha, C. F. D., Van Sluys, M., and Caramaschi, U.(2005). Prediction of threatened tetrapods based on thespecies–area relationship in Atlantic forest, Brazil.Journal of the Zoological Society of London, 265, 359–364.

Griffith, B., Scott, J. M., Carpenter, J. W., and Reed, C.(1989). Translocation as a species conservation tool: sta-tus and strategy, Science, 245, 477–480.

Griffiths, O., Cook, A., and Wells, S. M. (1993). The diet ofthe introduced carnivorous snail Euglandina rosea inMauritius and its implications for threatened island gas-tropod faunas. Journal of Zoology (London), 229, 78–89.

Groombridge, B. (ed.) (1992). Global biodiversity: status ofthe Earth’s living resources. (A report compiled by theWorld Conservation Monitoring Centre.) Chapman &Hall, London.

Groombridge, B. and Jenkins, M. D. (2002). World atlas ofbiodiversity. Earth’s living resources in the 21st century.United Nations Environmental Programme–WorldConservation Monitoring Center, University ofCalifornia Press, Berkeley.

Guevara, S. and Laborde, J. (1993). Monitoring seed dis-persal at isolated standing trees in tropical pastures:consequences for local species availability. PlantEcology, 107/108, 319–338.

Haas, A. C., Dunski, J. F., and Maxson, L. R. (1995).Divergent lineages within the Bufo margaritiferaComplex (Amphibia: Anura; Bufonidae) revealed byalbumin immunology. Biotropica, 27, 238–249.

Haila, Y. (1990). Towards an ecological definition of anisland: a northwest European perspective. Journal ofBiogeography, 17, 561–568.

Haila, Y. and Hanski, I. K. (1993). Birds breeding on smallBritish island and extinction risks. American Naturalist,142, 1025–1029.

Haila, Y. and Järvinen, O. (1983). Land bird communitieson a Finnish island: species impoverishment and abun-dance patterns. Oikos, 41, 255–273.

Hair, J. B. (1966). Biosystematics of the New Zealand flora1945–64. New Zealand Journal of Botany, 4, 559–595.

Hambler, C. (1994). Giant tortoise Geochelone giganteatranslocation to Curieuse Island (Seychelles): Success orfailure? Biological Conservation, 69, 293–299.

Hamnett, M. P. (1990). Pacific island resource develop-ment and environmental management. In Sustainable

development and environmental management of smallislands. Vol. 5, Man and the Biosphere Series (ed.W. Beller, P. d’Ayala, and P. Hein), pp. 227–257.UNESCO/Parthenon Publishing, Paris.

Hanley, K. A., Bolger, D. T., and Case, T. J. (1994).Comparative ecology of sexual and asexual geckospecies (Lepidodactylus) in French Polynesia.Evolutionary Ecology, 8, 438–454.

Hanski, I. (1986). Population dynamics of shrews on smallislands accord with the equilibrium model. BiologicalJournal of the Linnean Society, 28, 23–36.

Hanski, I. (1992). Inferences from ecological incidencefunctions. American Naturalist, 138, 657–662.

Hanski, I. (1996). Metapopulation ecology. In Populationdynamics in ecological space and time (ed. O. E. Rhodes, Jr,R. K. Chesser, and M. H. Smith), pp. 13–43. ChicagoUniversity Press, Chicago.

Hanski, I., Moilanen, A., and Gyllenberg, M. (1996).Minimum viable metapopulation size. AmericanNaturalist, 147, 527–541.

Hansson, L. (1998). Nestedness as a conservation tool:plants and birds of oak-hazel woodland in Sweden.Ecology Letters, 1, 142–145.

Harris, L. D. (1984). The fragmented forest: island biogeographytheory and the preservation of biotic diversity. University ofChicago Press, Chicago.

Harrison, S. (1994). Metapopulations and conservation.In Large-scale ecology and conservation biology (ed.P. J. Edwards, R. M. May, and N. R. Webb), pp. 111–128.Blackwell Scientific Publications, Oxford.

Harrison, S., Murphy, D. D., and Ehrlich, P. R. (1988).Distribution of the Bay Checkerspot Butterfly, Euphydryaseditha bayensis: evidence for a metapopulation model.American Naturalist, 132, 360–382.

Hart, K., Carey, S., Sigurdsson, H., Sparks, R. S. J., andRobertson, R. E. A. (2004). Discharge of pyroclasticflows into the sea during the 1996–1998 eruptions of theSoufriere Hills volcano, Montserrat. Bulletin ofVolcanology, 66, 599–614.

Hawkins, B. A., Field, R., Cornell, H. V., Currie, D. J.,Guegan, J-F.,. Kaufman, D. M. Kerr, J. T., Mittelbach, G. G.,Oberdorff, T., O’Brien, E. M., Porter, E. E., and Turner,J. R. G. (2003). Energy, water, and broad-scale geographicpatterns of species richness. Ecology, 84, 3105–3117.

Heads, M. J. (1990). Mesozoic tectonics and the decon-struction of biogeography: a new model of Australasianbiology. Journal of Biogeography, 17, 223–225.

Heads, M. J. (2004). What is a node? Journal ofBiogeography, 32, 1883–1891.

Heaney, L. R. (1978). Island area and body size of insularmammals: evidence from the tri-colored squirrel

362 R E F E R E N C E S

Page 376: Whittaker. Island Biogeography. 2007

(Callosciurus prevosti) of Southeast Asia. Evolution, 32,29–44.

Heaney, L. R. (1986). Biogeography of mammals in SEAsia: estimates of rates of colonization, extinction andspeciation. Biological Journal of the Linnean Society, 28,127–165.

Heaney, L. R. (2000). Dynamic disequilibrium: a long-term, large-scale perspective on the equilibrium modelof island biogeography. Global Ecology and Biogeography,9, 59–74.

Heaney, L. R., Walsh, J. S., Jr., and Peterson, A. T. (2005).The roles of geological history and colonization abilitiesin genetic differentiation between mammalian popula-tions in the Philippine archipelago. Journal ofBiogeography, 32, 229–247.

Hearty, J. P. (1997). Boulder deposits from large wavesduring the last interglaciation on North EleutheraIsland, Bahamas. Quaternary Research, 48, 326–338.

Henderson, S., Dawson, T. P., and Whittaker, R. J. (2006).Progress in invasive plants research. Progress in PhysicalGeography, 30, 25–46.

Herben, T., Suda, J., and Munclinger, P. (2005). The ghostof hybridization past: niche pre-emption is not the onlyexplanation of apparent monophyly in island endemics.Journal of Ecology, 93, 572–575.

Hernández, E., Nogales, M., and Martín, A. (2000).Discovery of a new lizard in the Canary Islands, with amultivariate analysis of Gallotia. Herpetelogica, 56, 63–76.

Herrera, C. M. (1989). Pollinator abundance, morphologyand floral visitation rate: analysis of the quantity compo-nent in a plant-pollinator system. Oecologia, 80, 241–248.

Herrero, S. (1997). Galapagos tortoises threatened.Conservation Biology, 11, 305.

Hess, A. L. (1990). Overview: sustainable developmentand environmental management of small islands. InSustainable development and environmental management ofsmall islands. Vol. 5, Man and the Biosphere Series (ed.W. Beller, P. d’Ayala, and P. Hein), pp. 3–14.UNESCO/Parthenon Publishing, Paris.

Hess, G. R. (1996). Linking extinction to connectivity andhabitat destruction in metapopulation models.American Naturalist, 148, 226–236.

Hess, J., Kadereit, J. W., and Vargas, P. (2000). The colo-nization history of Olea europaea L. in Macaronesiabased on internal transcribed spacer 1 (ITS-1)sequences, randomly amplified polymorphic DNAs(RAPD), and intersimple sequence repeats (ISSR).Molecular Ecology, 9, 857–868.

Higuchi, H. (1976). Comparative study on the breeding ofmainland and island subspecies of the varied tit (Parusvarius). Tori, 25, 11–20.

Hille, S. M., Nesje, M., and Segelbacher, G. (2003). Geneticstructure of kestrel populations and colonization of theCape Verde archipelago. Molecular Ecology, 12,2145–2151.

Hilton, G. M., Atkinson, P. W., Gray, G. A. L., Arendt, W.J.,and Gibbons, D. W. (2003). Rapid decline of the vol-canically threatened Montserrat oriole. BiologicalConservation, 111, 79–89.

Hinsley, S. A., Bellamy, P. E., Newton, I., and Sparks, T. H.(1994). Factors influencing the presence of individualbreeding bird species in woodland fragments. EnglishNature Research Reports, No. 99.

Hnatiuk, S. H. (1978). Plant dispersal by the AldabranGiant tortoise, Geochelone gigantea (Schweigger)Oecologia, 36, 345–350.

Hobbs, R. J. and Huenneke, L. F. (1992). Disturbance,diversity, and invasion: implications for conservation.Conservation Biology, 6, 324–337.

Hohmann, H., La Roche, F., Ortega, G., and Barquín, J.(1993). Bienen, Wespen und Ameisen der KanarischenInseln, I-II. Übersee-Museum, Bremen.

Holdaway, R. N. (1989). New Zealand’s pre-human avi-fauna and its vulnerability. In Rudge, M. R. (ed.), Moas,man and climate in the ecological history of NewZealand. New Zealand Journal of Ecology, 12 (Suppl.), 11–25.

Holdaway, R. N. (1990). Changes in the diversity of NewZealand forest birds. New Zealand Journal of Zoology, 17,309–321.

Holloway, J. D. (1996). The Lepidoptera of Norfolk Island,actual and potential, their origins and dynamics. In Theorigin and evolution of Pacific Island Biotas, New Guinea toEastern Polynesia: Patterns and processes (ed. A. Keast,and S. E. Miller), pp. 123–151. SPB AcademicPublishing, Amsterdam.

Hommel, P. W. F. (1990). Ujung Kulon: landscape surveyand land evaluation as a habitat for the Javan rhinoc-eros. ITC Journal, 1990–1, 1–15.

Honnay, O., Hermy, M., and Coppin, P. (1999). Nestedplant communities in deciduous forest fragments:species relaxation or nested habitats? Oikos, 84, 119–129.

Hoogerwerf, A. (1953). Notes on the vertebrate fauna ofthe Krakatau Islands, with special reference to the birds.Treubia, 22, 319–348.

Hubbell, S. P. (2001). The unified neutral theory of biodiversityand biogeography. Princeton University Press, Princeton,NJ.

Huggett, R. J. (1995). Geoecology: an evolutionary approach.Routledge, London.

Hughes, J. B., Daily, G. C., and Ehrlich, P. R. (2002).Conservation of tropical forest birds in countrysidehabitats. Ecology Letters, 5, 121–129.

R E F E R E N C E S 363

Page 377: Whittaker. Island Biogeography. 2007

Hughes, L., Dunlop, M., French, K., Leishman, M. R., Rice,B., Rodgerson, L., and Westoby, M. (1994). Predictingdispersal spectra: a minimal set of hypotheses based onplant attributes. Journal of Ecology, 82, 933–950.

Humphreys, W. F. and Kitchener, D. J. (1982). The effect ofhabitat utilization on species–area curves: implicationsfor optimal reserve area. Journal of Biogeography, 9,381–396.

Humphries, C. J. (1979). Endemism and evolution inMacaronesia. In Plants and islands (ed. D. Bramwell), pp.171–199. Academic Press, London.

Hunt, G. L., Jr and Hunt, M. W. (1974). Trophic levels andturnover rates: the avifauna of Santa Barbara island,California. Condor, 76, 363–369.

Huntley, B., Berry, P. M., Cramer, W., and McDonald, A. P.(1995). Modelling present and potential future ranges ofsome European higher plants using climate responsesurfaces. Journal of Biogeography, 22, 967–1001.

Hürlimann A., Martí J., and Ledesma, A. (2004).Morphological and geological aspects related to largeslope failures on oceanic islands—The huge La Orotavalandslides on Tenerife, Canary Islands. Geomorphology,62, 143–158.

Hutchinson, G. E. (1957). Concluding remarks. ColdSpring Harbor Symposia on Quantitative Biology, 22,415–427.

Inoue, K. and Kawahara, T. (1996). Evolution of Campanulaflowers in relation to insect pollinators on islands. InFloral biology: Studies on floral evolution in animal-pollinated plants (ed. D. G. Lloyd and S. C. H. Barrett) pp. 377–400. Chapman & Hall, New York.

Itow, S. (1988). Species diversity of mainland- and islandforests in the Pacific area. Vegetatio, 77, 193–200.

IUCN/SSC Invasive Species Specialist Group (2004).�www.issg.org� (visited June 2005).

Izquierdo, I, Martín, J. L., Zurita, N., and Arechavaleta, M.(eds) (2001). Lista de especies silvestres de Canarias (hongos,plantas y animales terrestres). Consejería de PolíticaTerritorial y Medio Ambiente, Gobierno de Canarias, LaLaguna, Tenerife.

Izquierdo, I, Martín, J. L., Zurita, N., and Arechavaleta, M.(eds.) (2004). Lista de especies silvestres de Canarias (hon-gos, plantas y animales terrestres) 2004. Consejería deMedio Ambiente y Ordenación Territorial, Gobierno deCanarias, Santa Cruz de Tenerife.

Jackson, E. D., Silver, E. A., and Dalrymple, G. B. (1972).Hawaiian-Emperor chain and its relation to Cenozoiccircumpacific tectonics. Geological Society of America,Bulletin, 83, 601–618.

Jackson, M. H. (1995). Galápagos: a natural history. Universityof Calgary Press, Calgary.

James, H. F. (1995). Prehistoric extinctions and ecologicalchanges on Oceanic islands. In Islands: biological diver-sity and ecosystem function (ed. P. M. Vitousek, L. L.Loope, and H. Adsersen), Ecological Studies 115, pp.87–102. Springer-Verlag, Berlin.

James, H. F. and Olson, S. L. (1991). Descriptions of thirty-two new species of birds from the Hawaiian Islands:Part II. Passeriformes. Ornithological Monographs, 46,1–88.

Janzen, D. H. (1979). How to be a fig. Annual Review ofEcology and Systematics, 10, 13–51.

Janzen, D. H. (1983). No park is an island: increase ininterference from outside as park size increases. Oikos,41, 402–410.

Janzen, D. H. (1985). On ecological fitting. Oikos, 45,308–310.

Järvinen, O. and Haila, Y. (1984). Assembly of land birdcommunities on Northern Islands: a quantitative analy-sis of insular impoverishment. In Ecological communities:conceptual issues and the evidence (ed. D. R. Strong,Jr, D. Simberloff, L. G. Abele, and A. B. Thistle), pp.138–147. Princeton University Press, Princeton, NJ.

Jenkyns, H. C. and Wilson, P. A. (1999). Stratigraphy, pale-oceanography, and evolution of Cretaceous Pacific guy-ots: relics from a greenhouse earth. American Journal ofScience, 299, 341–392.

Johnson, D. L. (1980). Problems in the land vertebrate zoo-geography of certain islands and the swimming powerof elephants. Journal of Biogeography, 7, 383–398.

Johnson, T. C., Scholz, C. A., and Talbot, M. R. (1996). LatePleistocene desiccation of Lake Victoria and rapid evo-lution of cichlid fishes. Science, 273, 1091–1093.

Johnson, T. H. and Stattersfield, A. J. (1990). A globalreview of island endemic birds. Ibis, 132, 167–180.

Juan, C., Emerson, B. C., Oromí, P., and Hewitt, G. M.(2000). Colonization and diversification: towards a phy-logenetic synthesis for the Canary Islands. Trends inEcology and Evolution, 15, 104–109.

Juvig, J. O. and Austring, A. O. (1979). The Hawaiianavifauna: biogeographical theory in evolutionary time.Journal of Biogeography, 6, 205–224.

Kadmon, R. (1995). Nested species subsets and geo-graphic isolation: a case study. Ecology, 76, 458–465.

Kalmar, A. and Currie, D. J. (2006). A global model ofisland biogeography. Global Ecology and Biogeography,15, 72–81.

Kaneshiro, K. Y. (1989). The dynamics of sexual selectionand founder effects in species formation. In Genetics,speciation and the founder principle (ed. L. V. Giddings, K.Y. Kaneshiro, and W. W. Anderson), pp. 279–296.Oxford University Press, New York.

364 R E F E R E N C E S

Page 378: Whittaker. Island Biogeography. 2007

Kaneshiro, K. Y. (1995). Evolution, speciation, and thegenetic structure of island populations. In Islands: bio-logical diversity and ecosystem function (ed. P. M.Vitousek, L. L. Loope, and H. Adsersen), EcologicalStudies 115, pp. 22–23. Springer-Verlag, Berlin.

Kaneshiro, K. Y., Gillespie, R. G., and Carson, H. L. (1995).Chromosomes and male genitalia of HawaiianDrosophila: tools for interpreting phylogeny and geogra-phy. In Hawaiian biogeography: evolution on a hot spotarchipelago (ed. W. L. Wagner and V. A. Funk), pp. 57–71.Smithsonian Institution Press, Washington, DC.

Kapos, V. (1989). Effects of isolation on the water status offorest patches in the Brazilian Amazon. Journal ofTropical Ecology, 5, 173–185.

Kayanne, H., Ishi, T., Matsumoto, E., and Yonekura, N.(1993). Late Holocene sea-level change on Rota andGuam, Mariana Islands, and its contraint on geophysi-cal predictions. Quaternary Research, 40, 189–200.

Keast, A. and Miller, S. E. (ed.) (1996). The origin and evolutionof Pacific Island biotas, New Guinea to Eastern Polynesia:patterns and processes. SPB Academic Publishing,Amsterdam.

Kellman, M. (1996). Redefining roles: plant communityreorganization and species preservation in fragmentedsystems. Global Ecology and Biogeography Letters, 5,111–116.

Kelly, B. J., Wilson, J. B., and Mark, A. F. (1989). Causes ofthe species–area relation: a study of islands inLake Manapouri, New Zealand. Journal of Ecology, 77,1021–1028.

Kim, S.-C., Crawford, D. J., Francisco-Ortega, J., and Santos-Guerra, A. (1996). A common origin for woody Sonchusand five related genera in the Macaronesian islands:Molecular evidence for extensive radiation. Proceedings ofthe National Academy of Sciences of the USA, 93, 7743–7748.

Kindvall, O. and Ahlén, I. (1992). Geometrical factors andmetapopulation dynamics of the Bush Cricket,Metrioptera bicolor Philippi (Orthoptera: Tettigoniidae).Conservation Biology, 6, 520–529.

King, C. (1984). Immigrant killers. Oxford University Press,Auckland.

King, K. J., Young, K. D., Waters, J. M., and Wallis, G. P.(2003). Preliminary genetic analysis of koaro (Galaxiasbrevipinnis) in New Zealand lakes: Evidence forallopatric differentiation among lakes but little popula-tion subdivision within lakes. Journal of the Royal Societyof New Zealand, 33, 591–600.

Klein, N. K. and Brown, W. M. (1994). Intraspecific molec-ular phylogeny in the Yellow Warbler (Dendroicapetechia), and implications for avian biogeography inthe West Indies. Evolution, 48, 1914–1932.

Kohn, D. D. and Walsh, D. M. (1994). Plant speciesrichness—the effect of island size and habitat diversity.Journal of Ecology, 82, 367–377.

Koopowitz, H. and Kaye, H. (1990). Plant extinction: aglobal crisis (2nd edn). Christopher Helm, London.

Korn, H. (1994). Genetic, demographic, spatial, environ-mental and catastrophic effects on the survival proba-bility of small populations of mammals. In Minimumanimal populations (ed. H. Remmert), pp. 39–49.Ecological Studies 106. Springer-Verlag, Berlin.

Krebs, J. R. (1970). Regulation of numbers in the greattit. Journal of the Zoological Society of London, 162,317–333.

Kunkel, G. (ed.) (1976). Biogeography and ecology of theCanary Islands. W. Junk, The Hague.

Kunkel, G. (1993). Die Kanarischen Inseln und ihrePflanzenwelt. Gustav Fisher, Stuttgart.

Kvist, L., Broggi, J., Illera, J. C., and Koivula, K. (2005).Colonisation and diversification of the blue tits (Paruscaeruleus teneriffae-group) in the Canary Islands. MolecularPhylogenetics and Evolution, 34, 501–511.

Lack, D. (1947a). Darwin’s finches: an essay on the general bio-logical theory of evolution. Cambridge University Press,Cambridge

Lack, D. (1947b). The significance of clutch size. Ibis, 89,302–352.

Lack, D. (1969). The numbers of bird species on islands.Bird Study, 16, 193–209.

Lack, D. (1970). The endemic ducks of remote islands.Wildfowl, 21, 5–10.

Lack, D. (1976). Island biology illustrated by the land birds ofJamaica. Blackwell Scientific Publications, Oxford.

Lacy, R. C. (1992). The effects of inbreeding onisolated populations: are minimum viable populationsizes predictable? In Conservation Biology: the theoryand practice of nature conservation and management (ed.P. L. Fiedler and S. K. Jain), pp. 276–320. Chapman &Hall, New York.

Lambshead, P. J. D. and Boucher, G. (2003). Marine nema-tode deep-sea biodiversity—hyperdiverse or hype?Journal of Biogeography, 30, 475–485.

Laurance, W. F. (2002). Hyperdynamism in fragmentedhabitats. Journal of Vegetation Science, 13, 595–602.

Law, R. and Watkinson, A. R. (1989). Competition. InEcological concepts: the contribution of ecology to an under-standing of the natural world (ed. J. M. Cherrett), pp. 243–284. Blackwell Scientific Publications, Oxford.

Lawesson, J. E., Adsersen, H., and Bentley, P. (1987). Anupdated and annotated check list of the vascular plantsof the Galápagos Islands. Reports from The BotanicalInstitute, University of Aarhus, 16.

R E F E R E N C E S 365

Page 379: Whittaker. Island Biogeography. 2007

Lawlor, T. E. (1986). Comparative biogeography ofmammals on islands. Biological Journal of the LinneanSociety, 28, 99–125.

Lawton, J. H. and May, R. M. (ed.) (1995). Extinction rates.Oxford University Press, Oxford.

Le Friant, A., Harford, C. L., Deplus, C., Boudon, G.,Sparks, R. S. J., Herd, R. A. and Komorowski, J. C.(2004). Geomorphological evolution of Montserrat(West Indies): importance of flank collapse anderosional processes. Journal of the Geological Society, 161,147–160.

Leader-Williams, N. and Walton, D. (1989). The isle andthe pussycat. New Scientist, 121 (1651), 11 February,48–51.

Leberg, P. L. (1991). Influence of fragmentation andbottlenecks on genetic divergence of wild turkey popu-lations. Conservation Biology, 5, 522–530.

Leuschner, C. (1996). Timberline and alpine vegetation onthe tropical and warm-temperate oceanic islands of theworld: elevation, structure, and floristics. Vegetatio, 123,193–206.

Lipman, P. W., Normark, W. R., Moore, J. G., Wilson, J. B.and Gutmacher. C. E. (1988). The giant Alika debrisslide, Mauna Loa, Hawaii. Journal of GeophysicalResearch, 93, 4279–4299.

Lister, A. M. (1993). Mammoths in miniature. Nature, 362,288–289.

Livezey, B. C. (1993). An ecomorphological review of thedodo (Raphus cucullatus) and solitaire (Pezophaps soli-taria), flightless Columbiformes of the MascareneIslands. Journal of Zoology (London), 230, 247–292.

Lobin, W. (1982). Untersuchung Über flora, vegetationund biogeographische beziehungen der KapverdischenInseln. Courier Forschung Institut Senckenberg, 53, 1–112.

Lockwood, J. L. and Moulton, M. P. (1994). Eco-morphological pattern in Bermuda birds: the influenceof competition and implications for nature preserves.Evolutionary Ecology, 8, 53–60.

Loehle, C. and Li, B. L. (1996). Habitat destruction and theextinction debt revisited. Ecological Applications, 6,784–789.

Lomolino, M. V. (1984a). Immigrant selection, predation,and the distribution of Microtus pennsylvanicus andBlarina brevicauda on islands. American Naturalist, 123,468–483.

Lomolino, M. V. (1984b). Mammalian island biogeogra-phy: effects of area, isolation, and vagility. Oecologia, 61,376–382.

Lomolino, M. V. (1985). Body size of mammals on islands:the island rule reexamined. American Naturalist, 125,310–316.

Lomolino, M. V. (1986). Mammalian community structureon islands: the importance of immigration, extinctionand interactive effects. Biological Journal of the LinneanSociety, 28, 1–21.

Lomolino, M. V. (1990). The target area hypothesis: theinfluence of island area on immigration rates of non-volant mammals. Oikos, 57, 297–300.

Lomolino, M. V. (1996). Investigating causality of nested-ness of insular communities: selective immigrations orextinctions? Journal of Biogeography, 23, 699–703.

Lomolino, M. V. (2000a). A call for a new paradigm ofisland biogeography. Global Ecology and Biogeography, 9,1–6.

Lomolino, M. V. (2000b). A species-based theory of insularzoogeography. Global Ecology and Biogeography, 9, 39–58.

Lomolino, M. V. (2000c). Ecology’s most general, yet pro-tean pattern: the species-area relationship. Journal ofBiogeography, 27, 17–26.

Lomolino, M. V. (2002). ‘ . . . there are areas too small, andareas too large, to show clear diversity patterns . . . ‘ R.H. MacArthur (1972: 191). Journal of Biogeography, 29,555–557.

Lomolino, M. V. (2005). Body size evolution in insular ver-tebrates: generality of the island rule. Journal ofBiogeography, 32, 1683–1699.

Lomolino, M. V., Brown, J. H., and Davis, R. (1989). Islandbiogeography of montane forest mammals in theAmerican southwest. Ecology, 70, 180–194.

Lomolino, M. V. and Davis, R. (1997). Biogeographic scaleand biodiversity of mountain forest mammals of west-ern North America. Global Ecology and BiogeographyLetters, 6, 57–76.

Lomolino, M. V., Riddle, B. R. and Brown, J. H. (2005).Biogeography (3rd edn), Sinauer Associates, Sunderland,MA.

Lomolino, M. V., and Weiser, M. D. (2001). Towards amore general species-area relationship: diversity on allislands, great and small. Journal of Biogeography, 28,431–445.

Long, A. J., Crosby, M. J., Stattersfield, A. J., and Wege, D.C. (1996). Towards a global map of biodiversity: pat-terns in the distribution of range-restricted birds. GlobalEcology and Biogeography Letters, 5, 281–304.

Losos, J. B. (1990). A phylogenetic analysis of characterdisplacement in Caribbean Anolis lizards. Evolution, 44,558–569.

Losos, J. B. (1994). Integrative approaches to evolutionaryecology: Anolis lizards as model systems. Annual Reviewof Ecology and Systematics, 25, 467–493.

Losos, J. B. (1996). Phylogenetic perspectives on communityecology. Ecology, 77, 1344–1354.

366 R E F E R E N C E S

Page 380: Whittaker. Island Biogeography. 2007

Losos, J. B., Jackman, T. R., Larson, A., de Queiroz, K., andRodríguez-Schettino, L. (1998). Contingency and deter-minism in replicated adaptive radiations of islandlizards. Science, 279, 2115–2118.

Losos, J. B., Warheit, K. I., and Schoener, T. W (1997).Adaptive differentiation following experimental islandcolonization in Anolis lizards. Nature, 387, 70–73.

Lovegrove, T. G. (1996). Island releases of SaddlebacksPhiesturnus carunculatus in New Zealand. BiologicalConservation, 77, 151–157.

Lovejoy, T. E., Bierregaard, R. O., Rylands, A. B., Malcolm,J. R., Quintela, C. E., Harper, L. H., Brown, K. S., Jr.,Powell, A. H., Powell, G. V. N., Schubart, H. O. R., andHays, M. B. (1986). Edge and other effects of isolationon Amazon forest fragments. In Conservation biology: thescience of scarcity and diversity (ed. M. Soulé),pp. 257–285. Sinauer Associates, Sunderland, MA.

Ludwig, D. (1996). The distribution of population survivaltimes. American Naturalist, 147, 506–26.

Lugo, A. E. (1988). Ecological aspects of catastrophes inCaribbean Islands. Acta Cientifica, 2, 24–31.

Lüpnitz, D. (1995). Kanarischen Inseln. Florenvielfalt aufengen Raum. Palmengarten Sonderheft, 23, 1–117.

Lynch, J. D. and Johnson, N. V. (1974). Turnover and equi-libria in insular avifaunas, with special reference to theCalifornia Channel Islands. Condor, 76, 370–84.

Mabberley, D. J. (1979). Pachycaul plants and islands. InPlants and islands (ed. D. Bramwell), pp. 259–277.Academic Press, London.

MacArthur, R. H., Diamond, J. M., and Karr, J. (1972).Density compensation in island faunas. Ecology, 53,330–342.

MacArthur, R. H. and Wilson, E. O. (1963). An equilibriumtheory of insular zoogeography. Evolution, 17, 373–387.

MacArthur, R. H. and Wilson, E. O. (1967). The theory of islandbiogeography. Princeton University Press, Princeton, NJ.

Mackay, R. (2002). The atlas of endangered species. Threatenedplants and animals of the world. Earthscan, London.

Madsen, T., Stille, B., and Shine, R. (1996). Inbreedingdepression in an isolated population of adders Viperaberus. Biological Conservation, 75, 113–118.

Magurran, A. E. (2004). Measuring biological diversity.Blackwell, Oxford.

Mallet, J. (1995). A species definition for the modernsynthesis. Trends in Ecology and Evolution, 10, 294–299.

Mangel, M. and Tier, C. (1994). Four facts every conserva-tion biologist should know about persistence. Ecology,75, 607–614.

Manne, L. L., Pimm, S. L., Diamond, J. M., and Reed, T. M.(1998). The form of the curves: a direct evaluation ofMacArthur & Wilson’s classic theory. Journal of AnimalEcology, 67, 784–794.

Marra, A. C. (2005). Pleistocene mammals of Mediterraneanislands. Quaternary International, 129, 5–14.

Marrero, A. and Francisco-Ortega, J. (2001a). Evolución enislas: la metáfora espacio-tiempo-forma. In Naturaleza delas Islas Canarias. Ecología y Conservación (ed. J. M.Fernández-Palacios and J. L. Martín Esquivel), pp.133–140. Turquesa Ediciones, Santa Cruz de Tenerife.

Marrero, A. and Francisco-Ortega, J. (2001b). Evolución enislas. La forma en el tiempo. In Naturaleza de las IslasCanarias. Ecología y conservación. (ed. J. M. Fernández-Palacios and J. L. Martín Esquivel) pp. 141–150.Turquesa Ediciones, Santa Cruz de Tenerife.

Marrero-Gómez, M. V., Bañares-Baudet, A., and Carqué -Alamo, E. (2003). Plant resource conservation planningin protected natural areas: an example from the CanaryIslands, Spain. Biological Conservation, 113, 399–410.

Marshall, H. D. and Baker, A. J. (1999). Colonization his-tory of Atlantic Island common chaffinches (Fringillacoelebs) revealed by mitochondrial DNA. MolecularPhylogenetics and Evolution, 11, 201–212.

Martin, J.-L., Gaston, A. J., and Hitier, S. (1995). The effectof island size and isolation on old growth forest habitatand bird diversity in Gwaii Haanas (Queen CharlotteIslands, Canada). Oikos, 72, 115–131.

Martín, A. and Lorenzo, J. A. (2001). Aves del ArchipiélagoCanario. Francisco Lemus Ed., La Laguna.

Martin, T. E. (1981). Species-area slopes and coefficients: acaution on their interpretation. American Naturalist, 188,823–837.

Martín-Esquivel, J.L., Fajardo, S., Cabrera, M.A.,Arechavaleta, M., Aguiar, A., Martín, S., and Naranjo,M. (2005). Evaluación 2004 de especies amenazadas deCanarias. Especies en peligro de extinción, sensibles a laalteración de su hábitat y vulnerables. Consejería de Medioambiente y ordenación territorial, Gobierno deCanarias. Santa Cruz de Tenerife.

Martín-Esquivel, J. L., García, H., Redondo, C., García, I.,and Carralero, I. (1995). La red Canaria de espacios natu-rales protegidos. Viceconsejería de Medio Ambiente,Gobierno de Canarias. Santa Cruz de Tenerife.

Martínez-Garza, C. and Howe, H. F. (2003). Restoringtropical diversity: beating the time tax on species loss.Journal of Applied Ecology, 40, 423–429.

Masson, D. G., Watts, A. B., Gee, M. J. R., Urgelés, R.,Mitchell, N. C., Le Bas, T. P., and Canals, M. (2002).Slope failures in the flanks of the western CanaryIslands. Earth-Science Reviews, 57, 1–35.

Matthews, J. A., Bridges, E. M., Caseldine, C. J., Luckman,A. J., Owen, G., Perry, A. H., Shakesby, R. A., Walsh,R. P. D., Whittaker, R. J., and Willis, K. J. (eds) (2001).The encyclopaedic dictionary of environmental change.Arnold, London.

R E F E R E N C E S 367

Page 381: Whittaker. Island Biogeography. 2007

Mawdsley, N. A., Compton, S. G., and Whittaker, R. J.(1998). Population persistence, pollination mutualisms,and figs in fragmented tropical landscapes. ConservationBiology, 12, 1416–1420.

Mayr, E. (1942). Systematics and the origin of species.Columbia University Press, New York.

Mayr, E. (1954). Change of genetic environment andevolution. In Evolution as a process (ed. J. S. Huxley,A. C. Hardy, and E. B. Ford), pp. 156–180. Allen &Unwin, London.

Mayr, E. (1963). Animal species and evolution. HarvardUniversity Press, Cambridge, MA.

McCall, R. A. (1997). Implications of recent geologicalinvestigations of the Mozambique Channel for themammalian colonization of Madagascar. Proceedings ofthe Royal Society of London, B 264, 663–665.

McCall, R. A. (1998). The role of wing length in the evolu-tion of avian flightlessness. Evolutionary Ecology, 12,569–580.

McDowall, R. M. (2005). Falklands: fact, fiction or fiddle-sticks? Journal of Biogeography, 32, 2187.

McGlone, M. S., Duncan, R. P. and Heenan, P. B. (2001).Endemism, species selection and the origin and distri-bution of the vascular plant flora of New Zealand.Journal of Biogeography, 28, 199–216.

McGlone, M. S. (2005). Goodbye Gondwana. Journal ofBiogeography, 32, 739–740.

McGuinness, K. A. (1984). Equations and explanations inthe study of species–area curves. Biological Reviews, 59,423–440.

McMullen, C. K. (1987). Breeding systems of selectedGalápagos islands angiosperms. American Journal ofBotany, 74, 1694–1705.

Means, D. B. and Simberloff, D. (1987). The peninsulaeffect: habitat-correlated species decline in Florida’sherpetofauna. Journal of Biogeography, 14, 551–568.

Meiri, S., Dayan, T. and Simberloff, D. (2006). Thegenerality of the island rule re-examined. Journal ofBiogeography, 33, 1571–1577.

Menard, H. W. (1986). Islands. Scientific American Library,New York.

Menges, E. S. (1992). Stochastic modeling of extinction inplant populations. In Conservation Biology: the theory andpractice of nature conservation, preservation and manage-ment (ed. P. L. Fiedler and S. K. Jain), pp. 253–275.Chapman & Hall, London.

Merlen, G. (1995). Use and misuse of the seas around theGalápagos Archipelago. Oryx, 29, 99–106.

Meyer, J.-Y. and Florence, J. (1996). Tahiti’s native floraendangered by the invasion of Miconia calvescens DC.(Melastomataceae). Journal of Biogeography, 23, 775–781.

Micol, T. and Jouventin, P. (1995). Restoration ofAmsterdam Island, South Indian Ocean, followingcontrol of feral cattle. Biological Conservation, 73,199–206.

Mielke, H. W. (1989). Patterns of life: biogeography of a chang-ing world. Unwin Hyman, Boston, MA.

Milberg, P. and Lamont, B. B. (1995). Fire enhances weedinvasion of roadside vegetation in SouthwesternAustralia. Biological Conservation, 73, 45–49.

Milberg, P. and Tyrberg, T. (1993). Naive birds and noblesavages—a review of man-caused prehistoric extinc-tions of island birds. Ecography, 16, 229–250.

Millien-Parra, V. and Jaeger, J.-J. (1999). Island biogeogra-phy of the Japanese terrestrial mammal assemblages: anexample of a relict fauna. Journal of Biogeography, 26,959–972.

Miskelly, C. M. (1990). Effects of the 1982–83 El Niño eventon two endemic landbirds on the Snares Islands, NewZealand. Emu, 90, 24–27.

Moore, D. M. (1979). The origins of temperate island flo-ras. In Plants and islands (ed. D. Bramwell), pp. 69–86.Academic Press, London.

Moore, G. J. and Clague, D. A. (1992). Volcano growth andevolution of the island of Hawaii. Bulletin of theGeological Society of America, 104, 1471–1484.

Moore, G. J, Clague, D. A., Holcomb, R. T., Lipman, P. W.,Normark, W. R., and Torressan, M. E. (1989). Prodigioussubmarine landslides on the Hawaiian ridge. Journal ofGeophysical Research, 94, 14465–14484.

Moore, G. J., Normark, W. R., Holcomb, R. T. (1994).Giant Hawaiian underwater landslides. Science, 264,46–47.

Morand, S. (2000). Geographic distance and the role ofisland area and habitat diversity in the species–arearelationships of four Lesser Antillean faunal groups: acomplementary note to Ricklefs and Lovette. Journal ofAnimal Ecology, 69, 1117–1119.

Morgan, G. S. and Woods, C. A. (1986). Extinction andzoogeography of West Indian land mammals. BiologicalJournal of the Linnean Society, 28, 167–203.

Morici, C. (2004). Palmeras e islas: la insularidad en unade las familias más diversas del reino vegetal. InEcología Insular/Island Ecology (ed. J. M. Fernández-Palacios and C. Morici), pp. 81–122. AsociaciónEspañola de Ecología Terrestre-Cabildo Insular de LaPalma, Santa Cruz de La Palma.

Morin, M. P. (1992). The breeding biology of an endan-gered Hawaiian Honeycreeper, the Laysan Finch.Condor, 94, 646–667.

Morrison, L. W. (1997). The insular biogeography of smallBahamian cays. Journal of Ecology, 85, 441–454.

368 R E F E R E N C E S

Page 382: Whittaker. Island Biogeography. 2007

Morrison, L. W. (2002a). Island biogeography andmetapopulation dynamics of Bahamian ants. Journal ofBiogeography, 29, 387–394.

Morrison, L. W. (2002b). Determinants of plant speciesrichness on small Bahamian islands. Journal ofBiogeography, 29, 931–941.

Morrone, J. J. (2005). Falklands: facts and fiction. Journal ofBiogeography, 32, 2183–2187.

Morwood, M. J., Soejono, R. P., Roberts, R. G., Sutikna, T.,Turney, C. S. M., Westaway, K. E., Rink, W. J., Zhao, J.-X., van Den Bergh, G. D., Rokus Awe Due, Hobbs, D. R.,Moore, M. W., Bird, M. I., and Fifield, L. K. (2004).Archeology and age of a new hominin from Flores inEastern Indonesia. Nature, 431, 1087–1091.

Moulton, M. P., Sanderson, J. G., and Labisky, R. F. (2001).Patterns of success in game bird (Aves: Galliformes)introductions to the Hawaiian islands and NewZealand. Evolutionary Ecology Research, 3, 507–519.

Moya, O., Contreras-Díaz, H. G., Oromí, P., and Juan, C.(2004). Genetic structure, phylogeography and demog-raphy of two ground-beetle species endemic to theTenerife laurel forest (Canary Islands). MolecularEcology, 13, 3153–3167.

Mueller-Dombois, D. (1975). Some aspects of islandecosystem analysis. In Tropical ecological systems: trendsin terrestrial and aquatic research (ed. F. B. Golley and E.Medina), pp. 353–366. Springer-Verlag, New York.

Mueller-Dombois, D. and Fosberg, F. R. (1998). Vegetationof the tropical Pacific islands. Springer-Verlag, New York.

Munroe, E. G. (1948). The geographical distribution ofbutterflies in the West Indies. PhD dissertation, CornellUniversity, Ithaca, NY.

Munroe, E. G. (1953). The size of island faunas. InProceedings of the Seventh Pacific Science Congress of thePacific Sciences Association, vol IV, Zoology, pp. 52–53.Whitcome and Tombs, Auckland, New Zealand.

Myers, A. A. (1991). How did Hawaii accumulate itsbiota? A test from the Amphipoda. Global Ecology andBiogeography Letters, 1, 24–29.

Myers, A. A. and Giller, P. S. (ed.) (1988). Analyticalbiogeography. Chapman & Hall, London.

Myers, N., Mittermeier, R. A., Mittermeier, C. G., daFonseca, G. A. B., and Kent, J. (2000). Biodiversityhotspots for conservation priorities. Nature, 403, 853–859.

Newmark, W. D. (1991). Tropical forest fragmentation andthe local extinction of understorey birds in the EasternUsambara Mountains, Tanzania. Conservation Biology, 5,67–78.

Nicholson, K. E., Glor, R. E., Kolbe, J. J., Larson, A., Hedges,S. B., and Losos, J. B. (2005). Mainland colonization byisland lizards. Journal of Biogeography, 32, 929–938.

Nicolás, J. P. de, Fernández-Palacios, J. M., Ferrer, F. J., andNieto, E. (1989). Inter-island floristic similarities in theMacaronesian Region. Vegetatio, 84, 117–125.

Nilsson, I. N. and Nilsson, S. G. (1985). Experimentalestimates of census efficiency and pseudo-turnover onislands: error trend and between-observer variationwhen recording vascular plants. Journal of Ecology, 73,65–70.

Nogales, M., Hernández, E. C., and Valdés, F. (1999). Seeddispersal by common ravens Corvus corax among islandhabitats (Canarian Archipelago). Ecoscience, 5, 56–61.

Nogales, M., Nieves, C., Illera, J. C., Padilla, D. P., andTraveset, A. (2005). Effect of native and alien vertebratefrugivores on seed viability and germination patterns ofRubia fruticosa (Rubiaceae) in the eastern CanaryIslands. Functional Ecology, 19, 429–436.

Nores, M. (1995). Insular biogeography of birds onmountain-tops in north western Argentina. Journal ofBiogeography, 22, 61–70.

Nores, M. (2004). The implications of Tertiary andQuaternary sea level rise events for avian distributionpatterns in the lowlands of northern South America.Global Ecology and Biogeography, 13, 149–161.

Nunn, P. D. (1990). Recent environmental changes onPacific Islands. Geographical Journal, 156, 125–140.

Nunn, P. D. (1994). Oceanic islands. Blackwell, Oxford.Nunn, P. D. (1997). Late Quaternary environmental

changes on Pacific islands: controversy, certainty andconjecture. Journal of Quaternary Science, 12, 443–450.

Nunn, P. D. (2000). Illuminating sea-level fall around AD1220–1510 (730–440 cal yr BP) in the Pacific Islands:implications for environmental change and culturaltransformation. New Zealand Geographer, 56, 46–54.

Nunn, P. D. (2004). Through a mist on the ocean: humanunderstanding of island environments. Tijdschrift voorEconomische en Sociale Geografie, 95, 311–325.

Nyhagen, D. R., Kragelund, C., Olesen, J. M., and Jones, C.G. (2001). Insular interactions between lizards and flow-ers: flower visitation by an endemic Mauritian gecko.Journal of Tropical Ecology, 17, 755–761.

O’Brien, E. M. (1993). Climatic gradients in woody plantspecies richness: towards an explanation based on ananalysis of southern Africa’s woody flora. Journal ofBiogeography, 20, 181–198.

O’Brien, E. M. (1998). Water–energy dynamics, climate, andprediction of woody plant species richness: an interimgeneral model. Journal of Biogeography, 25, 379–398.

O’Dea, N., Araújo, M. B., and Whittaker, R. J. (2006). Howwell do Important Bird Areas represent species andminimise conservation conflict in the tropical Andes?Diversity and Distributions, 12, 205–214.

R E F E R E N C E S 369

Page 383: Whittaker. Island Biogeography. 2007

O’Dowd, D. J., Green, P. T. and Lake, P. S. (2003).Invasional ‘meltdown’ on an oceanic island. EcologyLetters, 6, 812–817.

Olesen, J. M., Eskildsen, L. I., and Venkatasamy, S. (2002).Invasion of pollination networks on oceanic islands:importance of invader complexes and endemic super-generalists. Diversity and Distributions, 8, 181–192.

Olesen, J. M. and Jordano, P. (2002). Geographical patternsin plant-pollinator mutualistic networks. Ecology, 83,2416–2424.

Olesen, J. M. and Valido, A. (2003). Lizards as pollinatorsand seed dispersers: an island phenomenon. Trends inEcology and Evolution, 18, 177–181.

Olesen, J. M. and Valido, A. (2004). Lizards and birds asgeneralized pollinators and seed dispersers of islandplants. In Ecología Insular/Island Ecology (ed. J. M.Fernández-Palacios and C. Morici), pp. 229–249.Asociación Española de Ecología Terrestre–CabildoInsular de La Palma, Santa Cruz de La Palma.

Oliver, D. L. (2003). Las islas del Pacífico (3rd Ed). Melusina,Barcelona.

Ollier, C. D. (1988). Volcanoes. Basil Blackwell, Oxford.Olson, S. L. and James, H. F. (1982). Fossil birds from the

Hawaiian Islands: evidence for wholesale extinction byman before western contact. Science, 217, 633–635.

Olson, S. L. and James, H. F. (1991). Descriptions of thirty-two new species of birds from the Hawaiian Islands:Part I. Non-passeriformes. Ornithological Monographs,45, 1–88.

Olson, S. L. and Jouventin, P. (1996). A new species ofsmall flightless duck from Amsterdam Island, SouthernIndian Ocean (Anatidae: Anas). Condor, 98, 1–9.

Oromí, P. and Báez, M. (2001). Fauna invertebrada nativaterrestre. In Naturaleza de las Islas Canarias. Ecología yConservación (ed. J. M. Fernández-Palacios and J. L.Martín Esquivel). pp. 205–211. Editorial Turquesa, SantaCruz de Tenerife.

Otte, D. (1989). Speciation in Hawaiian crickets. In Speciationand its consequences (ed. D. Otte and J. A. Endler), pp.482–526. Sinauer Associates, Sunderland, MA.

Otte, D. and Endler, J. A. (ed.) (1989). Speciation and its con-sequences. Sinauer Associates, Sunderland, MA.

Otto, R., Fernández-Palacios, J. M., and Krüsi, B. O. (2001).Variation in species composition and vegetation structureof succulent scrub on Tenerife in relation to environ-mental variation. Journal of Vegetation Science, 12,237–248.

Otto, R., Krüsi, B. O., Burga, C. A., and Fernández-Palacios, J. M. (2006). Old-field succession along a pre-cipitation gradient in the semi-arid coastal region ofTenerife. Journal of Arid Environments, 65, 156–178.

Paine, R. T. (1985). Re-establishment of an insular winterwren population following a severe freeze. Condor, 87,558–559.

Panero, J. L., Francisco-Ortega, J., Jansen, R. K., andSantos-Guerra, A. (1999). Molecular evidence for multi-ple origins of woodiness and a New World biogeo-graphic connection of the Macaronesian Islandsendemic Pericallis (Asteraceae: Senecioneae).Proceedings of the National Academy of Sciences of the USA,96, 13886–13891.

Pannell, C. (1989). The role of animals in natural regener-ation and the management of equatorial rain forests forconservation and timber production. CommonwealthForestry Review, 68, 309–313.

Parrish, T. (2002). Krakatau: genetic consequences of islandcolonization. University of Utrecht and NetherlandsInstitute of Ecology, Heteren.

Parrish, T. L., Koelewijn, H. P., and van Dijk, P. J. (2003).Genetic evidence for natural hybridization betweenspecies of dioecious Ficus on island populations.Biotropica, 35, 333–343.

Partomihardjo, T., Mirmanto, E., and Whittaker, R. J.(1992). Anak Krakatau’s vegetation and flora circa 1991,with observations on a decade of development andchange. Geojournal, 28, 233–248.

Paton, P. W. C. (1994). The effect of edge on avian nest suc-cess: how strong is the evidence? Conservation Biology, 8,17–26.

Patrick, L. (2001). Introduced Species Summary Project:Brown tree snake (Boiga irregularis). �www.columbia.edu/itc/cerc/danoff-burg/invasion_bio/inv_spp_summ/boiga_irregularis.html�. (Last visitedMarch 2006).

Patterson, B. D. (1990). On the temporal development ofnested subset patterns of species composition. Oikos, 59,330–342.

Patterson, B. D. and Atmar, W. (1986). Nested subsets andthe structure of insular mammalian faunas and archipel-agos. Biological Journal of the Linnean Society, 28, 65–82.

Paulay, G. (1994). Biodiversity on oceanic islands: its ori-gin and extinction. American Zoologist, 34, 134–144.

Pearson, R. G. and Dawson, T. P. (2003). Predicting theimpacts of climate change on the distribution of species:are bioclimatic envelope models useful? Global Ecologyand Biogeography, 12, 361–371.

Peres, C. A. (2001). Synergistic effects of subsistence hunt-ing and habitat fragmentation on Amazonian forest ver-tebrates. Conservation Biology, 15, 1490–1505.

Peltonen, A. and Hanski, A. (1991). Patterns of islandoccupancy explained by colonization and extinctionrates in shrews. Ecology, 72, 1698–1708.

370 R E F E R E N C E S

Page 384: Whittaker. Island Biogeography. 2007

Pérez-Mellado, V. and Traveset, A. (1999). Relationshipsbetween plants and Mediterranean lizards. NaturaCroatica, 8, 275–285.

Perry, R. (ed.) (1984). Galapagos, Key Environments Series.IUCN/Pergamon Press, Oxford.

Petren, K., Grant, B. R., and Grant, P. R. (1999). A phy-logeny of Darwin’s finches based on microsatelliteDNA variation. Proceedings of the Royal Society of London,B 266, 321–329.

Pickett, S. T. A., Parker, V. T., and Fiedler, P. L. (1992). Thenew paradigm in ecology: implications for conservationbiology above the species level. In Conservation biology:the theory and practice of nature conservation and manage-ment (ed. P. L. Fiedler and S. K. Jain), pp. 91–125.Chapman & Hall, New York.

Pickett, S. T. A. and Thompson, J. N. (1978). Patch dynam-ics and the design of nature reserves. BiologicalConservation, 13, 27–37.

Pickett, S. T. A. and White, P. S. (ed.) (1985). The ecology ofnatural disturbance and patch dynamics. Academic Press,Orlando, FL.

Pierson, E. D., Elmqvist, T., Rainey, W. E., and Cox, P. A.(1996). Effects of tropical cyclonic storms on flying foxpopulations on the South Pacific islands of Samoa.Conservation Biology, 10, 438–451.

Pimm, S. L. (1991). The balance of nature? Ecological issues inthe conservation of species and communities. University ofChicago Press, Chicago.

Pimm. S. L. and Askins, R. A. (1995). Forest losses predictbird extinctions in eastern North America. Proceedings ofthe National Academy of Sciences of the USA, 92,9343–9347.

Pimm, S. L., Jones, H. L., and Diamond, J. (1988). On therisk of extinction. American Naturalist, 132, 757–785.

Pimm, S. L., Moulton, M. P., and Justice, L. J. (1995). Birdextinctions in the central Pacific. In Extinction rates (ed.J. H. Lawton and R. M. May), pp. 75–87. OxfordUniversity Press, Oxford.

Pole, M. (1994). The New Zealand flora—entirely long-distance dispersal? Journal of Biogeography, 21, 625–635.

Porter, D. M. (1979). Endemism and evolution inGalapagos islands vascular plants. In Plants and islands(ed. D. Bramwell), pp. 225–256. Academic Press, London.

Porter, D. M. (1984). Endemism and evolution in terres-trial plants. In Galapagos, Key Environments Series(ed. R. Perry), pp. 85–99. IUCN/Pergamon Press,Oxford.

Power, D. M. (1972). Numbers of bird species on theCalifornia islands. Evolution, 26, 451–463.

Pratt, H. D. (2005). The Hawaiian honeycreepers. OxfordUniversity Press, Oxford.

Pratt, H. D., Bruner, P. L., and Berrett, D. G. (1987). A fieldguide to the birds of Hawaii and the tropical Pacific.Princeton University Press, Princeton, NJ.

Pregill, G. K. (1986). Body size of insular lizards: a patternof Holocene dwarfism. Evolution 40, 997–1008.

Pregill, G. K. and Olson, S. L. (1981). Zoogeography ofWest Indian vertebrates in relation to Pleistocene cli-mate cycles. Annual Review of Ecology and Systematics,12, 75–98.

Preston, F. W. (1948). The commonness, and rarity, ofspecies. Ecology, 48, 254–283.

Preston, F. W. (1962). The canonical distribution of com-monness and rarity. Ecology, 43, part I, pp. 185–215; partII, pp. 410–432.

Price, J. P. and Clague, D. A. (2002). How old is theHawaiian biota? Geology and phylogeny suggest recentdivergence. Proceedings of the Royal Society of London, B269, 2429–2435.

Price, J. P. and Elliott-Fisk, D. (2004). Topographic historyof the Maui Nui complex, Hawai’i, and its implicationsfor biogeography. Pacific Science, 58, 27–45.

Price, J. P. and Wagner, W. L. (2004). Speciation inHawaiian angiosperm lineages: cause, consequence,and mode. Evolution, 58, 2185–2200.

Price, O., Woinarski, J. C. Z., Liddle, D. L., and Russell-Smith, J. (1995). Patterns of species composition andreserve design for a fragmented estate: monsoon rain-forests in the Northern Territory, Australia. BiologicalConservation, 74, 9–19.

Primack, R. B. (1993). Essentials of conservation biology.Sinauer Associates, Sunderland, MA.

Primack, R. B. and Miao, S. L. (1992). Dispersal can limitlocal plant distribution. Conservation Biology, 6, 513–519.

Primack, R. B. and Ros, J. (2002). Introducción a la biologíade la conservación. Ariel Ciencia, Barcelona.

Pukrop, M. E. (1997). Phosphate mining in Nauru.�www.american.edu/projects/mandala/TED/nauru.htm� (Visited June 2006.)

Pulliam, H. R. (1996). Sources and sinks: empiricalevidence and population consequences. In Populationdynamics in ecological space and time (ed. O. E. Rhodes,Jr. R. K. Chesser, and M. H. Smith), pp. 45–69. ChicagoUniversity Press, Chicago.

Racine, C. H. and Downhower, J. F. (1974). Vegetative andreproductive strategies of Opuntia (Cactaceae) in theGalapagos Islands. Biotropica, 6, 175–186.

Rafe, R. W., Usher, M. B., and Jefferson, R. G. (1985). Birdson reserves: the influence of area and habitat on speciesrichness. Journal of Applied Ecology, 22, 327–335.

Rainey, W. E., Pierson, E. D., Elmqvist, T., and Cox, P. A.(1995). The role of flying foxes (Pteropodidae) in

R E F E R E N C E S 371

Page 385: Whittaker. Island Biogeography. 2007

oceanic island ecosystems of the Pacific. Symposium ofthe Zoological Society of London, 67, 47–62.

Ralph, C. J. and Fancy, S. G. (1994). Demography andmovements of the Omao (Myadestes obscurus). Condor,96, 503–511.

Rassmann, K., Trillmich, F., and Tautz, D. (1997).Hybridization between the Galápagos land and marineiguana (Conlophus subcristatus and Amblyrhynchus crista-tus) on Plaza Sur. Journal of Zoology (London), 242,729–739.

Ratcliffe, D. A. (ed.) (1977). A nature conservation review.Cambridge University Press, Cambridge.

Rawlinson, P. A., Zann, R. A., van Balen, S., and Thornton,I. W. B. (1992). Colonization of the Krakatau islands byvertebrates. Geojournal, 28, 225–231.

Reed, D. H., O’Grady, J. J., Brook, B. W., Ballou, J. D., andFrankham, R. (2003). Estimates of minimum viable pop-ulation sizes for vertebrates and factors influencingthose estimates. Biological Conservation, 113, 23–34.

Rees, D. J., Emerson, B. C., Oromí, P., and Hewitt, G. M.(2001). The diversification of the genus Nesotes(Coleoptera: Tenebrionidae) in the Canary Islands: evi-dence from mtDNA. Molecular Phylogenetics andEvolution, 21, 321–326.

Reijnen, R., Foppen, R., and Meeuwsen, H. (1996). Theeffects of traffic on the density of breeding birds inDutch agricultural grasslands. Biological Conservation,75, 255–260.

Renvoize, S. A. (1979). The origins of Indian Ocean islandfloras. In Plants and islands (ed. D. Bramwell), pp.107–129. Academic Press, London.

Rey, J. R. (1984). Experimental tests of island biogeo-graphic theory. In Ecological communities: conceptualissues and the evidence (ed. D. R. Strong, Jr, D. Simberloff,L. G. Abele, and A. B. Thistle), pp. 101–112. PrincetonUniversity Press, Princeton, NJ.

Rey, J. R. (1985). Insular ecology of salt marsh arthropods:species level patterns. Journal of Biogeography, 12, 97–107.

Reyment, R. A. (1983). Palaeontological aspects of islandbiogeography: colonization and evolution of mammalson Mediterranean islands. Oikos, 41, 299–306.

Ribon, R., Simon, J. E., and de Mattos, G. T. (2003). Birdextinctions in Atlantic forest fragments of the Viçosaregion, southeastern Brazil. Conservation Biology, 17,1827–1839.

Ricketts, T. H. (2001). The matrix matters: effective isola-tion in fragmented landscapes. American Naturalist, 158,87–99.

Ricketts, T. H. (2004). Tropical forest fragments enhancepollinator activity in nearby coffee crops. ConservationBiology, 18, 1262–1271.

Ricklefs, R. E. (1980). Geographical variation in clutch sizeamong passerine birds: Ashmole’s hypothesis. Auk, 97,38–49.

Ricklefs, R. E. (1989). Speciation and diversity: the inte-gration of local and regional processes. In Speciation andits consequences (ed. D. Otte and J. A. Endler), pp.599–622. Sinauer Associates, Sunderland, MA.

Ricklefs, R. E. and Bermingham, E. (2001). Nonequilibriumdiversity dynamics of the Lesser Antillean avifauna.Science, 294, 1522–1524.

Ricklefs, R. E. and Bermingham, E. (2002). The concept ofthe taxon cycle in biogeography. Global Ecology andBiogeography, 11, 353–362.

Ricklefs, R. E. and Cox, G. W. (1972). Taxon cycles inthe West Indian avifauna. American Naturalist, 106,195–219.

Ricklefs, R. E. and Cox, G. W. (1978). Stage of taxon cycle,habitat distribution, and population density in the avi-fauna of the West Indies. American Naturalist, 112,875–895.

Ricklefs, R. E. and Lovette, I. J. (1999). The roles of islandarea per se and habitat diversity in the species-area rela-tionships of four Lesser Antillean faunal groups. Journalof Animal Ecology, 68, 1142–1160.

Ridley, H. N. (1930). The dispersal of plants throughout theworld. Reeve, Ashford, England.

Ridley, M. (ed.) (1994). A Darwin selection. Fontana Press,London.

Ridley, M. (1996). Evolution (2nd edn). Blackwell Science,Cambridge, MA.

Roberts, A. and Stone, L. (1990). Island-sharing by archi-pelago species. Oecologia, 83, 560–567.

Rocha, S., Carretero, M. A., Vences, M., Glaw, F., and Harris,D. J. (2006). Deciphering patterns of trans-oceanicdispersal: the evolutionary origin and biogeography ofcoastal lizards (Cryptoblepharus) in the western IndianOcean region. Journal of Biogeography, 33, 13–22.

Rodda, G. H., Fritts, T. H., McCoid, M. J., and Campbell,E. W. III (1999). An overview of the biology of thebrown treesnake (Boiga irregularis), a costly introducedpest on Pacific Islands. In Problem snake management: thehabu and the brown treesnake, (eds Rodda, G. H., Sawai,Y., Chiszar, D., and Tanaka, H.), pp 44–80. CornellUniversity Press, Ithaca, NY.

Rodríguez de la Fuente, F. (1980). Enciclopedia Salvat de laFauna. Salvat Ediciones, Pamplona.

Rodríguez-Estrella, R., Leon de la Luz, J. L., Breceda, A.,Castellanos, A., Cancino, J., and Llinas, J. (1996). Status,density and habitat relationships of the endemic terres-trial birds of Socorro island, Revillagigeo Islands,Mexico. Biological Conservation, 76, 195–202.

372 R E F E R E N C E S

Page 386: Whittaker. Island Biogeography. 2007

Rodríguez-Gironés, M. A. and Santamaría, L. (2006). Anew algorithm to calculate the nestedness temperatureof presence-absence matrices. Journal of Biogeography,33, 924–935.

Roff, D. A. (1991). The evolution of flightlessness ininsects. Ecological Monographs, 60, 389–421.

Roff, D. A. (1994). The evolution of flightlessness: is his-tory important? Evolutionary Ecology, 8, 639–657.

Rogers, R. R., Hartman, J. H., and Krause, D. W. (2000).Stratigraphic analysis of Upper Cretaceous rocks in theMahajanga Basin, northwestern Madagascar: implica-tions for ancient and modern faunas. Journal of Geology,108, 275–301.

Rosenzweig, M. L. (1978). Competitive speciation.Biological Journal of the Linnean Society, 10, 275–289.

Rosenzweig, M. L. (1995). Species diversity in space and time.Cambridge University Press, Cambridge.

Rosenzweig, M. L. (2003). Win-win ecology: How the Earth’sspecies can survive in the midst of human enterprise. OxfordUniversity Press, New York.

Rosenzweig, M. L. (2004). Applying species-area relation-ships to the conservation of diversity. In Frontiers of bio-geography: new directions in the geography of Nature. (edsM. V. Lomolino and L. R. Heaney), pp. 325–343. SinauerAssociates, Sunderland, MA.

Rosenzweig, M. L. and Clark, C. W. (1994). Island extinc-tion rates from regular censuses. Conservation Biology, 8,491–494.

Ross, K. A., Fox, B. J., and Fox, M. D. (2002). Changes toplant species richness in forest fragments: fragment age,disturbance and fire history may be as important asarea. Journal of Biogeography, 29, 749–765.

Roughgarden, J. (1989). The structure and assembly ofcommunities. In Perspectives in Ecological Theory(ed. J. Roughgarden, R. M. May, and S. A. Levin), pp.203–226. Princeton University Press, Princeton, NJ.

Roughgarden, J. and Pacala, S. (1989). Taxon cycle amongAnolis lizard populations: review of evidence. InSpeciation and its consequences (ed. D. Otte and J. A.Endler), pp. 403–432. Sinauer Associates, Sunderland,MA.

Rudge, M. R. (ed.) (1989). Moas, man and climate in theecological history of New Zealand. New Zealand Journalof Ecology, 12 (Suppl.), 11–25.

Rummel, J. D. and Roughgarden, J. D. (1985). A theory offaunal buildup for competition communities. Evolution,39, 1009–1033.

Runhaar, J., van Gool, C. R., and Groen, C. L. G. (1996).Impact of hydrological changes on nature conservationareas in the Netherlands. Biological Conservation, 76,269–276.

Russell, G. J., Diamond, J. M., Pimm, S. L., and Reed, T. M.(1995). A century of turnover: community dynamicsat three timescales. Journal of Animal Ecology, 64,628–641.

Russell, R. W., Carpenter, F. L., Hixon, M. A., and Paton,D. C. (1994). The impact of variation in stopover habitatquality on migrant rufous hummingbirds. ConservationBiology, 8, 483–490.

Safford, R. J. (1997). Distribution studies on the forest-living native passerines of Mauritius. BiologicalConservation, 80, 189–198.

Sakai, A. K., Wagner, W. L., Ferguson, D. M., and Herbst,D. R. (1995a). Origins of dioecy in the Hawaiian flora.Ecology, 76, 2517–2529.

Sakai, A. K., Wagner, W. L., Ferguson, D. M., and Herbst,D. R. (1995b). Biogeographical and ecological correlatesof dioecy in the Hawaiian flora. Ecology, 76, 2530–2543.

Sanderson, J. G., Moulton, M. P., and Selfridge, R. G.(1998). Null matrices and the analysis of speciesco-occurrences. Oecologia, 116, 275–283.

Sanmartín, I. and Ronquist, F. (2004). Southern hemi-sphere biogeography inferred by event-based models:plant versus animal patterns. Systematic Biology, 53,216–243.

Santos-Guerra, A. (1990). Bosques de laurisilva en la regiónMacaronésica. Consejo de Europa, Strasbourg.

Santos-Guerra, A. (1999). Origen y evolución de laflora canaria. In Ecología y cultura en Canarias (ed. J. M.Fernández-Palacios, J. J. Bacallado, and J. A. Belmonte),pp. 107–129. Museo de las Ciencias y el Cosmos,Cabildo Insular de Tenerife, Santa Cruz de Tenerife.

Santos-Guerra, A. (2001). Flora vascular nativa. InNaturaleza de las Islas Canarias. Ecología y Conservación.(ed. J. M. Fernández-Palacios and J. L. Martín Esquivel),pp 185–192. Turquesa Ediciones, Santa Cruz deTenerife.

Sarre, S. and Dearn, J. M. (1991). Morphological variationand fluctuating asymmetry among insular populations ofthe Sleepy Lizard, Trachydosurus rugosus Gray (Squamata:Scincidea). Australian Journal of Zoology, 39, 91–104.

Sarre, S., Schwaner, T. D., and Georges, A. (1990). Geneticvariation among insular populations of the SleepyLizard, Trachydosaurus rugosus Gray (Squamata:Scincidea). Australian Journal of Zoology, 38, 603–616.

Sato, A., O’hUigin, C., Figueroa, F., Grant, P. R., Grant, B.R., Tichy, H., and Klein, J. (1999). Phylogeny of Darwin’sfinches as revealed by mtDNA sequences. Proceedings ofthe National Academy of Sciences of the USA, 96,51010–5106.

Sauer, J. D. (1969). Oceanic island and biogeographic theory:a review. Geographical Review, 59, 582–593.

R E F E R E N C E S 373

Page 387: Whittaker. Island Biogeography. 2007

Sauer, J. D. (1990). Allopatric speciation: deduced but notdetected. Journal of Biogeography, 17, 1–3.

Saunders, D. A. and Hobbs, R. J. (1989). Corridors for con-servation. New Scientist, 121 (1648), 28 January, 63–68.

Saunders, D. A., Hobbs, R. J., and Margules, C. R. (1991).Biological consequences of ecosystem fragmentation: areview. Conservation Biology, 5, 18–32.

Saunders, N. E. and Gibson, D. J. (2005). Breeding system,branching processes, hybrid swarm theory, and thehumped-back diversity relationship as additional expla-nations for apparent monophyly in the Macaronesianisland flora. Journal of Ecology, 93, 649–652.

Sax, D. F., Brown, J. H., and Gaines, S. D. (2002). Speciesinvasions exceed extinctions on islands world-wide: acomparative study of plants and birds. AmericanNaturalist, 160, 776–783.

Scatena, F. N. and Larsen, M. C. (1991). Physical aspects ofHurricane Hugo in Puerto Rico. Biotropica, 23, 317–323.

Scheiner, S. M. (2003). Six types of species-area curves.Global Ecology and Biogeography, 12, 441–447.

Scheiner, S. M. (2004). A mélange of curves—further dia-logue about species–area relationships. Global Ecologyand Biogeography, 13, 479–484.

Schluter, D. (1988). Character displacement and the adap-tive divergence of finches on islands and continents.American Naturalist, 131, 799–824.

Schmidt, N. M. and Jensen, P. M. (2003). Changes in mam-malian body length over 175 years—adaptations to afragmented landscape? Conservation Ecology, 7, 6 (onlinejournal).

Schmitt, S. and Whittaker, R. J. (1998). Disturbance andsuccession on the Krakatau Islands, Indonesia. InDynamics of tropical communities. British EcologicalSociety Symposium, Vol. 37 (ed. D. M. Newbery, H. N.T. Prins, and N. D. Brown), pp. 515–548. BlackwellScience, Oxford.

Schoener, T. W. (1975). Presence and absence of habitatshift in some widespread lizard species. EcologicalMonographs, 45, 233–258.

Schoener, T. W. (1983). Rate of species turnover decreasesfrom lower to higher organisms: a review of the data.Oikos, 41, 372–377.

Schoener, T. W. (1986). Patterns in terrestrial vertebrateversus arthropod communities: do systematic differ-ences in regularity exist? In Community ecology (ed. J. M.Diamond and T. J. Case), pp. 556–586. Harper & Row,New York.

Schoener, T. W. (1988). Testing for non-randomness insizes and habitats of West Indian lizards: choice ofspecies pool affects conclusions from null models.Evolutionary Ecology, 2, 1–26.

Schoener, T. W. (1989). The ecological niche In Ecologicalconcepts: the contribution of ecology to an understanding ofthe natural world (ed. J. M. Cherrett), pp. 79–113.Blackwell Scientific Publications, Oxford.

Schoener, T. W. and Gorman, G. C. (1968). Some nichedifferences in three Lesser Antillean lizards of the genusAnolis. Ecology, 49, 819–830.

Schoener, T. W. and Schoener, A. (1983). Distribution ofvertebrates on some very small islands. I. Occurrencesequences of individual species. Journal of AnimalEcology, 52, 209–235.

Schoener, T. W. and Spiller, D. A. (1987). High populationpersistence in a system with high turnover. Nature, 330,474–477.

Schumm, S. A. (1991). To interpret the Earth: ten ways to bewrong. Cambridge University Press, Cambridge.

Schüle, W. (1993). Mammals, vegetation and the initialhuman settlement of the Mediterranean islands: apalaeoecological approach. Journal of Biogeography, 20,399–411.

Schwaner, T. D. and Sarre, S. D. (1988). Body size of TigerSnakes in Southern Australia, with particular referenceto Notechis ater serventyi (Elapidae) on Chappell Island.Journal of Herpetology, 22, 24–33.

Scott, T. A. (1994). Irruptive dispersal of black-shoulderedkites to a coastal island. Condor, 96, 197–200.

Sequeira, A. S., Lanteri, A. A., Scataglini, M. A.,Confalonieri, V. A., and Farrell, B. D. (2000). Are flight-less Galapaganus weevils older than the GalápagosIslands they inhabit? Heredity, 85, 20–29.

Sfenthourakis, S. (1996). The species–area relationship ofterrestrial isopods (Isopoda; Oniscidea) from theAegean archipelago (Greece): a comparative study.Global Ecology and Biogeography Letters, 5, 149–157.

Shafer, C. L. (1990). Nature reserves: island theory and con-servation practice. Smithsonian Institution Press,Washington, DC.

Shennan, I. (1983). Flandrian and late Devensian sea-level changes and crustal movements in Englandand Wales. In Shorelines and isostacy (ed. D. E. Smithand A. G. Dawson), pp. 255–284. Academic Press,London.

Shepherd, U. L. and Brantley, S. L. (2005). Expanding onWatson’s framework for classifying patches: when is anisland not an island? Journal of Biogeography, 32,951–960.

Shine, T., Böhme, W. Nickel, H., Thies, D. F., and Wilms, T.R. (2001). Rediscovery of relict populations of the Nilecrocodile Crocodylus niloticus in south-easternMauritania, with observations on their natural history.Oryx, 35, 260–262.

374 R E F E R E N C E S

Page 388: Whittaker. Island Biogeography. 2007

Shilton, L. A. (1999). Seed dispersal by fruit bats on theKrakatau Islands, Indonesia. Ph.D. thesis, School ofBiology, University of Leeds.

Shilton, L. A., Altringham, J. D., Compton, S. G., andWhittaker, R. J. (1999). Old World fruit bats can be long-distance seed dispersers through extended retention ofviable seeds in the gut. Proceedings of the Royal Society ofLondon, B 266, 219–223.

Short, J. and Turner, B. (1994). A test of the vegetationmosaic hypothesis: a hypothesis to explain the declineand extinction of Australian mammals. ConservationBiology, 8, 439–449.

Shrader-Frechette, K. S. and McCoy, E. D. (1993). Methodin ecology: strategies for conservation. CambridgeUniversity Press, Cambridge.

Silvertown, J. (2004). The ghost of competition past in thephylogeny of island endemic plants. Journal of Ecology,92, 168–173.

Silvertown, J., Francisco-Ortega, J., and Carine, M. (2005).The monophyly of island radiations: an evaluation ofniche pre-emption and some alternative explanations.Journal of Ecology, 93, 653–657.

Simberloff, D. (1976). Species turnover and equilibriumisland biogeography. Science, 194, 572–578.

Simberloff, D. (1978). Using island biogeographic distri-butions to determine if colonization is stochastic.American Naturalist, 112, 713–726.

Simberloff, D. (1983). When is an island community inequilibrium? Science, 220, 1275–1277.

Simberloff, D. (1992). Do species–area curves predictextinction in fragmented forests? In Tropical deforestationand species extinction (ed. T. C. Whitmore and J. A.Sayer), pp. 119–142. Chapman & Hall, London.

Simberloff, D., Farr, J. A., Cox, J., and Mehlman, D. W.(1992). Movement corridors: conservation bargains orpoor investments? Conservation Biology, 6, 493–504.

Simberloff, D. and Levin, B. (1985). Predictable sequences ofspecies loss with decreasing island area—land birds intwo archipelagoes. New Zealand Journal of Ecology, 8, 11–20.

Simberloff, D. and Martin, J.-L. (1991). Nestedness ofinsular avifaunas: simple summary statistics maskingcomplex species patterns. Ornis Fennica, 68, 187–192.

Simberloff, D. and Wilson, E. O. (1969). Experimental zoo-geography of islands. The colonisation of emptyislands. Ecology, 50, 278–296.

Simberloff, D. and Wilson, E. O. (1970). Experimental zoo-geography of islands. A two year record of colonization.Ecology, 51, 934–937.

Simkin, T. (1984). Geology of Galápagos Islands. InGalapagos, Key Environments Series (ed. R. Perry), pp.15–41. IUCN/Pergamon Press, Oxford.

Simpson, B. B. (1974). Glacial migration of plants: islandbiogeographical evidence. Science, 185, 698–700.

Sohmer, S. H. and Gustafson, R. (1993). Plants and flowersof Hawaii (3rd printing). University of Hawaii Press,Honolulu.

Sorensen, M. F. L. (1977). Niche shifts of Coal Tits Parusater in Denmark. Journal of Avian Biology, 28, 68–72.

Sosa, P. (2001). Genes, poblaciones y especies. In:Naturaleza de las Islas Canarias. Ecología y conservación(ed. J. M. Fernández-Palacios and J. L. Martín-Esquivel)pp. 151–155. Turquesa Ediciones, Santa Cruz deTenerife.

Soulé, M. (ed.) (1986). Conservation biology: the science ofscarcity and diversity. Sinauer Associates, Sunderland,MA.

Soulé, M. E., Bolger, D. T., Alberts, A. C., Wright, J., Sorice,M., and Hill, S. (1988). Reconstructed dynamics of rapidextinctions of Chaparral-requiring birds in urban habi-tat islands. Conservation Biology, 2, 75–92.

Spellerberg, I. F. and Gaywood, M. J. (1993). Linear fea-tures: linear habitats and wildlife corridors. EnglishNature Research Reports, No. 60.

Spencer-Smith, D., Ramos, S. J., McKenzie, F., Munroe, E.,and Miller, L. D. (1988). Biogeographical affinities of thebutterflies of a ‘forgotten’ island: Mona (Puerto Rico).Bulletin of the Allyn Museum, No. 121, pp. 1–35.

Spiller, D. A. and Schoener, T. W. (1995). Longterm varia-tion in the effect of lizards on spider density is linked torainfall. Oecologia, 103, 133–139.

Stace, C. A. (1989). Dispersal versus vicariance—no con-test! Journal of Biogeography, 16, 201–202.

Stamps, J. A. and Buechner, M. (1985). The territorialdefence hypothesis and the ecology of insular verte-brates. Quarterly Review of Biology, 60, 155–181.

Stanley, S. M. (1999). Earth system history. W. H. Freeman,New York.

Steadman, D. W. (1997a). Human-caused extinctions ofbirds. In Biodiversity II: understanding and protecting ourbiological resources (ed. M. L. Reaka-Kudla, W. E. Wilson,and W. O. Wilson), pp. 139–161. Joseph Henry Press,Washington, DC.

Steadman, D. W. (1997b). The historic biogeography andcommunity ecology of Polynesian pigeons and doves.Journal of Biogeography, 24, 737–753.

Steadman, D. W., Pregill, G. K., and Burley, D. V. (2002).Rapid prehistoric extinction of iguanas and birds inPolynesia. Proceedings of the National Academy of Sciencesof the USA, 99, 3673–3677.

Steadman, D. W., White, J. P. and Allen, J. (1999).Prehistoric birds from New Ireland, Papua NewGuinea: extinctions on a large Melanesian island.

R E F E R E N C E S 375

Page 389: Whittaker. Island Biogeography. 2007

Proceedings of the National Academy of Sciences of the USA,96, 2563–2568.

Stearn, W. T. (1973). Philip Barker Webb and CanarianBotany. Monographiae Biologicae Canarienses, 4, 15–29.

Steers, J. A. and Stoddart, D. R. (1977). The origin of fring-ing reefs, barrier reefs and atolls. In Biology and geologyof coral reefs (ed. O. A. Jones and R. Endean), pp. 21–57.Academic Press, New York.

Stoddart, D. R. and Walsh, R. P. D. (1992). Environmentalvariability and environmental extremes as factors in theisland ecosystem. Atoll Research Bulletin, No. 356.

Stone, L. and Roberts, A. (1990). The checkerboard scoreand species distributions. Oecologia, 85, 74–79.

Stone, L. and Roberts, A. (1992). Competitive exclusion, orspecies aggregation? Oecologia, 91, 419–424.

Storey, B. C. (1995). The role of mantle plumes in conti-nental breakup: case histories from Gondwanaland.Nature, 377, 310–318.

Stouffer, P. C. and Bierregaard, R. O., Jr (1995). Use ofAmazonian forest fragments by understory insectivo-rous birds. Ecology, 76, 2429–2445.

Stouffer, P. C. and Bierregaard, R. O., Jr (1996). Effects offorest fragmentation on understorey hummingbirds inAmazonian Brazil. Conservation Biology, 9, 1085–1094.

Streelman, J. T. and Danley, P. D. (2003). The stages of ver-tebrate evolutionary radiation. Trends in Ecology andEvolution, 18, 126–131.

Strong, D. R., Jr, Simberloff, D., Abele, L. G., and Thistle,A. B. (ed.) (1984). Ecological communities: conceptual issuesand the evidence. Princeton University Press, Princeton, NJ.

Stuessy, T. F., Crawford, D. J., and Marticorena, C. (1990).Patterns of phylogeny in the endemic vascular flora ofthe Juan Fernandez Islands, Chile. Systematic Botany, 15,338–346.

Stuessy, T. F., Crawford, D. J., and Marticorena, C. andRodríguez, R. (1998). Island biogeography ofangiosperms of the Juan Fernandez archipelago. InEvolution and speciation of island plants. (ed. T. F. Stuessyand M. Ono), pp. 121–138. Cambridge University Press,Cambridge.

Stuessy, T. F., Jakubowsky,G. Salguero Gómez, R., Pfosser,M., Schlüter, P. M., Fer, T., Sun, B.-Y., and Kato, H.(2006). An alternative model for plant evolution inislands. Journal of Biogeography, 33, 1259–1265.

Sugihara, G. (1981). S � CAz, z � 1/4: a reply to Connorand McCoy. American Naturalist, 117, 790–793.

Sunding, P. (1979). Origins of the Macaronesian Flora. InPlants and islands (ed. D. Bramwell), pp. 13–40.Academic Press, London.

Sziemer, P. (2000). Madeira’s natural history in a nutshell.Ribeiro & Filhos, Funchal.

Tarbuck, E. J. and Lutgens, F. K. (1999). Earth: introductionto physical geology, 6th edn. Prentice-Hall, EnglewoodCliffs, NJ.

Tarr, C. L. and Fleischer, R. C. (1995). Evolutionary rela-tionships of the Hawaiian Honeycreepers (Aves,Drepanidae). In Hawaiian biogeography: evolution on a hotspot archipelago (ed. W. L. Wagner and V. A. Funk), pp.147–159. Smithsonian Institution Press, Washington, DC.

Tauber, C. A. and Tauber, M. J. (1989). Sympatric specia-tion in insects: perception and perspective. In Speciationand its consequences (ed. D. Otte and J. A. Endler), pp.307–344. Sinauer Associates, Sunderland, MA.

Taylor, B. W. (1957). Plant succession on recent volcanoesin Papua. Journal of Ecology, 45, 233–243.

Tellería, J. L. and Santos, T. (1995). Effects of forest frag-mentation on a guild of wintering passerines: the role ofhabitat selection. Biological Conservation, 71, 61–67.

Templeton, A. R. (1981). Mechanisms of speciation—apopulation genetic approach. Annual Review of Ecologyand Systematics, 12, 23–48.

Terborgh, J. (1992). Maintenance of diversity in tropicalforests. Biotropica, 24, 283–292.

Terrell, J. (1986). Prehistory in the Pacific Islands: A study ofvariation in language, customs, and human biology.Cambridge University Press, Cambridge.

Thomas, C. D. (1994). Extinction, colonization, andmetapopulations: environmental tracking by rarespecies. Conservation Biology, 8, 373–378.

Thomas, C. D., Cameron, A., Green, R E., Bakkenes, M.,Beaumont, L. J., Collingham, Y. C., Erasmus, B. F. N., deSiquiera, M. F., Grainger, A., Hannah, L., Hughes, L.,Huntley, B., van Jaarsveld, A. S., Midgley, G. F., Miles, L.,Ortega-Huerta, M. A., Peterson, A. T., Phillips, O., andWilliams, S. E. (2004). Extinction risk from climatechange. Nature, 427, 145–148.

Thomas, C. D. and Harrison, S. (1992). Spatial dynamics ofa patchily distributed butterfly species. Journal ofAnimal Ecology, 61, 437–446.

Thomas, C. D., Singer, M. C., and Boughton, D. A. (1996).Catastrophic extinction of population sources in a but-terfly metapopulation. American Naturalist, 148, 957–975.

Thornton, I. W. B. (1992). K. W. Dammerman—fore-runner of island equilibrium theory. Global Ecology andBiogeography Letters, 2, 145–148.

Thornton, I. W. B. (1996). Krakatau—the destruction andreassembly of an island ecosystem. Harvard UniversityPress, Cambridge, MA.

Thornton, I. W.B., Cook, S., Edwards, J. S., Harrison, R. D.,Schipper, C., Shanahan, M., Singadan, R., and Yamuna,R. (2001). Colonization of an island volcano, LongIsland, Papua New Guinea, and an emergent island,

376 R E F E R E N C E S

Page 390: Whittaker. Island Biogeography. 2007

Motmot, in its caldera lake. VII. Overview and discus-sion. Journal of Biogeography, 28, 1389–1408.

Thornton, I. W. B., Partomihardjo, T., and Yukawa, J.(1994). Observations on the effects, up to July 1993, ofthe current eruptive episode of Anak Krakatau. GlobalEcology and Biogeography Letters, 4, 88–94.

Thornton, I. W. B. and Walsh, D. (1992). Photographic evi-dence of rate of development of plant cover on theemergent island Anak Krakatau from 1971 to 1991 andimplications for the effect of volcanism. Geojournal, 28,249–259.

Thornton, I. W. B., Ward, S. A., Zann, R. A., and New, T. R.(1992). Anak Krakatau—a colonization model within acolonization model? Geojournal, 28, 271–286.

Thornton, I. W. B., Zann, R. A., and van Balen, S. (1993).Colonization of Rakata (Krakatau Is.) by non-migrantland birds from 1883 to 1992 and implications for thevalue of island equilibrium theory. Journal ofBiogeography, 20, 441–452.

Thornton, I. W. B., Zann, R. A., and Stephenson, D. G.(1990). Colonization of the Krakatau islands by landbirds, and the approach to an equilibrium number ofspecies. Philosophical Transactions of the Royal Society ofLondon, B 328, 55–93.

Thorpe, R. S. and Malhotra, A. (1996). Molecular and mor-phological evolution within small islands. PhilosophicalTransactions of the Royal Society of London, 351, 815–822.

Thorpe, R. S., McGregor, D. P., Cumming, A. M., andJordan, W. C. (1994). DNA evolution and colonizationsequence of island lizards in relation to geological his-tory: mtDNA RFLP, cytochrome B, cytochrome oxidase,12S RRNA sequence and nuclear RAPD analysis.Evolution, 48, 230–240.

Tjørve, E. (2003). Shapes and functions of species–areacurves: a review of possible models. Journal ofBiogeography, 30, 827–835.

Tidemann, C. D., Kitchener, D. J., Zann, R. A., andThornton, I. W. B. (1990). Recolonization of theKrakatau islands and adjacent areas of West Java,Indonesia, by bats (Chiroptera) 1883–1986. PhilosophicalTransactions of the Royal Society of London, B 328, 121–130.

Toft, C. A. and Schoener, T. W. (1983). Abundance and diver-sity of orb spiders on 106 Bahamian islands: biogeogra-phy at an intermediate trophic level. Oikos, 41, 411–426.

Townsend, C.R., Begon, M., and Harper J.L. (2003).Essentials of ecology, 2nd edn. Blackwell Publishing,Oxford.

Tracy, C. R. and George. L. (1992). On the determinants ofextinction. American Naturalist, 139, 102–122.

Traveset, A. (2001). Ecología reproductiva de plantas encondiciones de insularidad: consecuencias ecológicas y

evolutivas del aislamiento geográfico. In Ecosistemasmediterráneos: análisis funcional (ed. R. Zamora and F.Puignaire). pp. 269–289. Consejo Superior deInvestigaciones Científicas–Asociación Española deEcología Terrestre, Granada.

Traveset, A. and Sáez, E. (1997). Pollination of Euphorbiadendroides by lizards and insects: spatio-temporal varia-tion in patterns in flower visitation. Oecologia, 111,241–248.

Traveset, A. and Santamaría, L. (2004). Alteración demutualismos planta-animal debido a la introducción deespecies exóticas en ecosistemas insulares. In EcologíaInsula /Island Ecology (ed. J. M. Fernández-Palacios andC. Morici), pp. 251–276. Asociación Española deEcología Terrestre–Cabildo Insular de La Palma, SantaCruz de La Palma.

Trewick, S. A. (1997). Flightlessness and phylogenyamongst endemic rails (Aves: Railidae) of the NewZealand region. Philosophical Transactions of the RoyalSociety of London, B 352, 429–456.

Triantis, K. A., Mylonas, M., Lika, K., and Vardinoyannis,K. (2003). A model for the species–area–habitat relation-ship. Journal of Biogeography, 30, 19–27.

Triantis, K. A., Mylonas, M., Weiser, M. D., Lika, K., andVardinoyannis, K. (2005). Species richness, environmen-tal heterogeneity and area: A case study based on landsnails in Skyros archipelago (Aegean Sea, Greece).Journal of Biogeography, 32, 1727–1735.

Triantis, K. A., Vardinoyannis, K., Tsolaki, E. P., Botsaris, I.Lika, K., and Mylonas, M. (2006). Re-approaching thesmall island effect. Journal of Biogeography, 33, 914–923.

Trillmich, F. (1992). Conservation problems on Galápagos:the showcase of evolution in danger. Naturwissenschaften,79, 1–6.

Trusty, J. L., Olmstead, R. G., Santos-Guerra, A.,Sá-Fontinha, S., and Francisco-Ortega, J. (2005).Molecular phylogenetics of the Macaronesian-endemicgenus Bystropogon (Lamiaceae): palaeo-islands, ecologi-cal shifts and interisland colonizations. MolecularEcology, 14, 1177–1189.

Tudge, C. (1991). Time to save rhinoceroses. New Scientist,131 (1788), 30–35.

Turner, I. M. (1996). Species loss in fragments of tropicalrain forest: a review of the evidence. Journal of AppliedEcology, 33, 200–209.

Turner, W. R. and Tjørve, E. (2005). Scale-dependence inspecies-area relationships. Ecography, 28, 721–730.

Tye, A. (2006). Can we infer island introduction and natu-ralization rates from inventory data? Evidence fromintroduced plants in Galápagos. Biological Invasions, 8,201–215.

R E F E R E N C E S 377

Page 391: Whittaker. Island Biogeography. 2007

Valido, A., Dupont, Y. L., and Hansen, D. M. (2002).Native birds and insects, and introduced honey beesvisiting Echium wildpretii (Boraginaceae) in the CanaryIslands. Acta Oecologica, 23, 413–419.

Valido, A. and Nogales, M. (1994). Frugivory and seed dis-persal by the lizard Gallotia galloti (Lacertidae) in a xerichabitat of the Canary Islands. Oikos, 70, 403–411.

Valido, A., Nogales, M., and Medina, F. M. (2003). Fleshyfruits in the diet of Canarian lizards Gallotia galloti(Lacertidae) in a xeric habitat of the island of Tenerife.Journal of Herpetology, 37, 741–747.

Valido, A., Rando, J. C., Nogales, M., and Martín, A.(2000). ‘Fossil’ lizard found alive in the Canary Islands.Oryx, 34, 71–72.

van Balgooy, M. M. J., Hovenkamp, P. H., and van Welzen,P. C. (1996). Phytogeography of the Pacific—floristicand historical distribution patterns in plants. In The ori-gin and evolution of Pacific island biotas, New Guinea toEastern Polynesia: patterns and processes (ed. A. Keast andS. E. Miller), pp. 191–213. SPB Academic Publishing,Amsterdam.

van Riper, C. III, van Riper, S. G., Goff, M. L., and Laird,M. (1986). The epizootiology and ecological significanceof Malaria in Hawaiian land birds. EcologicalMonographs, 56, 327–344.

van Steenis, C. G. G. J. (1972). The mountain flora of Java. E.J. Brill, Leiden.

van Valen, L. (1973). A new evolutionary law. EvolutionaryTheory, 1, 1–30.

Vartanyan, S. L., Garutt, V. E., and Sher, A. V. (1993).Holocene dwarf mammoths from Wrangel Island in theSiberian Arctic. Nature, 362, 337–340.

Villa, F., Rossi, O., and Sartore, F. (1992). Understandingthe role of chronic environmental disturbance in thecontext of island biogeographic theory. EnvironmentalManagement, 16, 653–666.

Vincek, V., O’Huigin, C., Satta, Y., Takahata, Y., Boag, P. T.,Grant, P. R., Grant, B. R., and Klein, J. (1997). How largewas the founding population of Darwin’s finches?Proceedings of the Royal Society of London, B 264, 111–118.

Vitousek, P. M. (1990). Biological invasion and ecosystemprocesses: towards an integration of population biologyand ecosystem studies. Oikos, 57, 7–13.

Vitousek, P. M., Loope, L. L., and Adsersen, H. (ed.)(1995). Islands: biological diversity and ecosystem function,Ecological Studies 115. Springer-Verlag, Berlin.

Vitousek, P, Loope, L. L., and Stone, C. P. (1987a).Introduced species in Hawaii: biological effects andopportunities for ecological research. Trends in Ecologyand Evolution, 2, 224–227.

Vitousek, P.M. and Walker, L.R. (1989). Biological invasionby Myrica faya in Hawai’i: plant demongraphy, nitrogen

fixation, ecosystem effects. Ecological Monographs, 59,247–265.

Vitousek, P. M., Walker, L. R., Whiteaker, L. D., Mueller-Dombois, D., and Matson, P. A. (1987b). Biological inva-sion by Myrica faya alters ecosystem development inHawaii. Science, 238, 802–804.

Wagner, W. L. and Funk, V. A. (ed.) (1995). Hawaiian bio-geography: evolution on a hot spot archipelago. SmithsonianInstitution Press, Washington, DC.

Waide, R. B. (1991). Summary of the responses of animalpopulations to hurricanes in the Caribbean. Biotropica,23, 508–512.

Walker, L. R., Brokaw, N. V. L., Lodge, D. J., and Waide, R.B. (ed.) (1991b). Special issue: ecosystem, plant, and ani-mal responses to hurricanes in the Caribbean. Biotropica,23, 313–521.

Walker, L. R., Lodge, D. J., Brokaw, N. V. L., and Waide, R.B. (1991a). An introduction to hurricanes in theCaribbean. Biotropica, 23, 313–316.

Walker, L. R., Silver, W. L., Willig, M. R., and Zimmerman,J. K. (ed.) (1996). Special issue: long term responses ofCaribbean ecosystems to disturbance. Biotropica, 28,414–613.

Wallace, A. R. (1878). Tropical nature and other essays.Macmillan, London.

Wallace, A. R. (1902). Island life (3rd edn). Macmillan,London.

Walter, H. (1998). Driving forces of island biodiversity: anappraisal of two theories. Physical Geography, 19,351–377.

Walter, K. S. and Gillet. H. J. (eds.) (1998). 1997 IUCN RedList of Threatened Plants. Compiled by theWCMC–IUCN–The World Conservation Union, Gland.(See also �www.redlist.org�.)

Ward, S. N., and Day, S. (2001). Cumbre Vieja Volcano—Potential collapse and tsunami at La Palma, CanaryIslands. Geophysical Research Letters, 28, 3397–3400.

Wardle, D. A., Zackrisson, O., Hörnberg, G., and Gallet, C.(1997). The influence of island area on ecosystem prop-erties. Science, 277, 1296–1299.

Warner, R. E. (1968). The role of introduced diseases in theextinction of the endemic Hawaiian avifauna. Condor,70, 101–120.

Waterloo, M. J., Schelleken, J., Bruijnzeel, L. A., Vugts, H.F., Assenberg, P. N., and Rawaqa, T. T. (1997). Chemistryof bulk precipitation in southwestern Viti Levu, Fiji.Journal of Tropical Ecology, 13, 427–447.

Watson, J. E. M. (2004). Bird responses to habitat fragmen-tation: illustrations from Madagascan and Australiancase studies. D.Phil. thesis, University of Oxford.

Watson, J. E. M., Whittaker, R. J., and Dawson, T. P.(2004a). Habitat structure and proximity to forest edge

378 R E F E R E N C E S

Page 392: Whittaker. Island Biogeography. 2007

affect the abundance and distribution of forest-depend-ent birds in tropical coastal forests of southeasternMadagascar. Biological Conservation, 120, 311–327.

Watson, J. E. M., Whittaker, R. J., and Dawson, T. P.(2004b). Avifaunal responses to habitat fragmentation inthe threatened littoral forests of south-easternMadagascar. Journal of Biogeography, 31, 1791–1807.

Watson, J. E. M., Whittaker, R. J., and Freudenberger, D.(2005). Bird community responses to habitat fragmenta-tion: how consistent are they across landscapes? Journalof Biogeography, 32, 1353–1370.

Watts, D. (1970). Persistence and change in the vegetationof oceanic islands: an example from Barbados, WestIndies. Canadian Geographer, 14, 91–109.

Watts, D. (1987). The West Indies: patterns of development,culture and environmental change since 1492. CambridgeUniversity Press, Cambridge.

Weaver, M. and Kellman, M. (1981). The effects of forestfragmentation on woodlot tree biotas in SouthernOntario. Journal of Biogeography, 8, 199–210.

Webb, C. J. and Kelly, D. (1993). The reproductive biologyof the New Zealand flora. Trends in Ecology andEvolution, 8, 442–447.

Weiher, E. and Keddy, P. A. (1995). Assembly rules, nullmodels, and trait dispersion: new questions from oldpatterns. Oikos, 74, 159–164.

Weiher, E. and Keddy, P. A. (eds) (1999). Ecological assem-bly rules: perspectives, advances, retreats. CambridgeUniversity Press, Cambridge.

Weins, J. (1984). On understanding a non-equilibriumworld: myth and reality in community patterns andprocesses. In Ecological communities: conceptual issues andthe evidence (ed. D. R. Strong, Jr, D. Simberloff, L. G.Abele, and A. B. Thistle), pp. 439–457. PrincetonUniversity Press, Princeton, NJ.

Weisler, M. I. (1995). Henderson Island prehistory: colo-nization and extinction on a remote Polynesian island.Biological Journal of the Linnean Society, 56, 377–404.

Werner, T. K. and Sherry, T. W. (1987). Behavioural feedingspecialization in Pinaroloxias inornata, the ‘Darwin’sFinch’ of Cocos Island, Costa Rica. Proceedings of theNational Academy of Sciences of the USA, 84, 5506–5510.

Westerbergh, A. and Saura, A. (1994). Genetic differentiationin endemic Silene (Caryophyllaceae) on the Hawaiianislands. American Journal of Botany, 81, 1487–1493.

Whelan, F. and Kelletat, D. (2003). Submarine slides onvolcanic islands—a source for mega-tsunamis in theQuaternary. Progress in Physical Geography, 27, 198–216.

Whitaker, A. H. (1987). The roles of lizards in NewZealand plant reproductive strategies. New ZealandJournal of Botany, 25, 315–328.

White, G. B. (1981). Semispecies, sibling species andsuperspecies. In The evolving biosphere (chance, chance andchallenge) (ed. P. L. Forey), pp. 21–28. British Museum(Natural History) and Cambridge University Press,Cambridge.

Whitehead, D. R. and Jones, C. E. (1969). Small islandsand the equilibrium theory of insular biogeography.Evolution, 23, 171–179.

Whitmore, T. C. (1984). Tropical rain forests of the Far East(2nd edn). Clarendon Press, Oxford.

Whitmore, T. C. (ed.) (1987). Biogeographical evolution of theMalay Archipelago, Oxford Monographs onBiogeography No. 4. Oxford University Press, Oxford.

Whitmore, T. C. and Sayer, J. A. (ed.) (1992). Tropical defor-estation and species extinction. Chapman & Hall, London.

Whittaker, R. H. (1977). Evolution of species diversity inland communities. In Evolutionary Biology, vol. 10 (ed.M. K. Hecht, W. C. Steere, and B. Wallace), pp. 250–268.Plenum Press, New York.

Whittaker, R. J. (1992). Stochasticism and determinism inisland ecology. Journal of Biogeography, 19, 587–591.

Whittaker, R. J. (1995). Disturbed island ecology. Trends inEcology and Evolution, 10, 421–425.

Whittaker, R. J. (2004). Dynamic hypotheses of richness onislands and continents. In Frontiers of Biogeography: newdirections in the geography of Nature (eds M. V. Lomolinoand L. R. Heaney), pp. 211–231. Sinauer Associates,Sunderland, MA.

Whittaker, R. J., Araújo, M. B., Jepson, P., Ladle, R. J.,Watson, J. E. M., and Willis, K. J. (2005). Conservationbiogeography: assessment and prospect. Diversity andDistributions, 11, 3–23.

Whittaker, R. J. and Bush, M. B. (1993). Anak Krakatauand old Krakatau: a reply. Geojournal, 29, 417–420.

Whittaker, R. J., Bush, M. B., Partomihardjo, T., Asquith,N. M., and Richards, K. (1992a). Ecological aspects ofplant colonisation of the Krakatau Islands. Geojournal,28, 201–211.

Whittaker, R. J., Bush, M. B., and Richards, K. (1989). Plantrecolonization and vegetation succession on the KrakatauIslands, Indonesia. Ecological Monographs, 59, 59–123.

Whittaker, R. J., Field, R., and Partomihardjo, T. (2000).How to go extinct: lessons from the lost plants ofKrakatau. Journal of Biogeography, 27, 1049–1064.

Whittaker, R. J. and Jones, S. H. (1994a). Structure inre-building insular ecosystems: an empirically derivedmodel. Oikos, 69, 524–529.

Whittaker, R. J. and Jones, S. H. (1994b). The role of fru-givorous bats and birds in the rebuilding of a tropicalforest ecosystem, Krakatau, Indonesia. Journal ofBiogeography, 21, 689–702.

R E F E R E N C E S 379

Page 393: Whittaker. Island Biogeography. 2007

Whittaker, R. J., Jones, S. H., and Partomihardjo, T. (1997).The re-building of an isolated rain forest assemblage:how disharmonic is the flora of Krakatau? Biodiversityand Conservation, 6, 1671–1696.

Whittaker, R. J., Partomihardjo, T., and Riswan, S. (1995).Surface and buried seed banks from Krakatau,Indonesia: implications for the sterilization hypothesis.Biotropica, 27, 346–354.

Whittaker, R. J., Schmitt, S. F., Jones, S. H., Partomihardjo,T., and Bush, M. B. (1998). Stand biomass and treemortality from permanent forest plots on Krakatau,1989–1995. Biotropica, 30, 519–529.

Whittaker, R. J., Walden, J., and Hill, J. (1992b). Post-1883ash fall on Panjang and Sertung and its ecologicalimpact. Geojournal, 28, 153–171.

Whittaker, R. J., Willis, K. J., and Field, R. (2001). Scale andspecies richness: towards a general, hierarchical theoryof species diversity. Journal of Biogeography, 28, 453–470.

Wiggins, D. A., Moller, A. P., Sorensen, M. F. L., and Brand,L. (1998). Island biogeography and the reproductiveecology of great tits Parus major. Oecologia, 115, 478–482.

Wilcove, D. S., McLellan, C. H., and Dobson, A. P. (1986).Habitat fragmentation in the temperate zone. InConservation biology: the science of scarcity and diversity(ed. M. Soulé), pp. 237–256. Sinauer Associates.,Sunderland, MA.

Wiles, G. J., Schreiner, I. H., Nafus, D., Jurgensen, L. K.,and Manglona, J. C. (1996). The status, biology, andconservation of Serianthes nelsonii (Fabaceae), an endan-gered Micronesian tree. Biological Conservation, 76,229–239.

Williams, E. E. (1972). The origin of faunas. Evolution oflizard congeners in a complex island fauna: A trialanalysis. Evolutionary Biology, 6, 47–88.

Williamson, M. H. (1981). Island populations. OxfordUniversity Press, Oxford.

Williamson, M. H. (1983). The land-bird community ofSkokholm: ordination and turnover, Oikos, 41, 378–384.

Williamson, M. H. (1984). Sir Joseph Hooker’s lecture oninsular floras. Biological Journal of the Linnean Society, 22,55–77.

Williamson, M. H. (1988). Relationship of species numberto area, distance and other variables. In AnalyticalBiogeography, an integrated approach to the study of animaland plant distributions (ed. A. A. Myers and P. S. Giller),pp. 91–115. Chapman & Hall, London.

Williamson, M. H. (1989a). The MacArthur and Wilsontheory today: true but trivial. Journal of Biogeography, 16,3–4.

Williamson, M. H. (1989b). Natural extinction on islands.Philosophical Transactions of the Royal Society of London, B325, 457–468.

Williamson, M. H. (1996). Biological invasions. Populationand Community Biology Series 15, Chapman & Hall,London.

Williamson, M., Gaston, K. J., and Lonsdale, W. M. (2001).The species–area relationship does not have an asymp-tote! Journal of Biogeography, 28, 827–830.

Williamson, M., Gaston, K. J., and Lonsdale, W. M. (2002).An asymptote is an asymptote and not found inspecies–area relationships. Journal of Biogeography, 29,1713.

Willis, K. J., Gillson, L., and Brncic, T. M. (2004). How ‘vir-gin’ is virgin rainforest? Science, 304, 402–403.

Willis, K. J. and Whittaker, R. J. (2002). Species diversity-scale matters. Science, 295, 1245–1248.

Wilson, D. E. and Graham, G. L. (ed.) (1992). Pacific islandflying foxes: proceedings of an international conservationconference. US Fish and Wildlife Service, BiologicalReport 90 (23), Washington, DC.

Wilson, E. O. (1959). Adaptive shift and dispersal in atropical ant fauna. Evolution, 13, 122–144.

Wilson, E. O. (1961). The nature of the taxon cycle in theMelanesian ant fauna. American Naturalist, 95, 169–193.

Wilson, E. O. (1969). The species equilibrium. BrookhavenSymposia in Biology, 22, 38–47.

Wilson, E. O. (1995). Naturalist. Allen Lane, The PenguinPress, London.

Wilson, E. O. and Bossert, W. H. (1971). A primer of popula-tion biology. Sinauer Associates, Stamford, CT.

Wilson, E. O. and Willis, E. O. (1975). Applied biogeogra-phy. In Ecology and evolution of communities (ed. M. L.Cody and J. M. Diamond), pp. 522–534. HarvardUniversity Press, Cambridge, MA.

Wilson, J. B. and Roxburgh, S. H. (1994). A demonstrationof guild-based assembly rules for a plant community,and the determination of intrinsic guilds. Oikos, 69,267–276.

Wilson, J. B. and Whittaker, R. J. (1995). Assembly rulesdemonstrated in a saltmarsh community. Journal ofEcology, 83, 801–807.

Wilson, J. T. (1963). A possible origin of the Hawaiianislands. Canadian Journal of Physics, 41, 863–870.

Wilson, M. H., Kepler C. B., Snyder, N. F. R., Derrickson,S. R., Dein, F. J., Wiley, J. W., Wunderle, J. M., Lugo, A.E., Graham, D. L., and Toone, W. D. (1994). Puerto Ricanparrots and potential limitations of the metapopulationapproach to species conservation. Conservation Biology,8, 114–123.

Winkworth, R. C., Wagstaff, S. J., Glenny, D., andLockhart, P. J. (2002). Plant dispersal NEWS from NewZealand. Trends in Ecology and Evolution, 17, 514–520.

Woinarski, J. C. Z., Whitehead, P. J., Bowman, D. M. J. S.,and Russell-Smith, J. (1992). Conservation of mobile

380 R E F E R E N C E S

Page 394: Whittaker. Island Biogeography. 2007

species in a variable environment: the problem ofreserve design in the Northern Territory, Australia.Global Ecology and Biogeography Letters, 2, 1–10.

Worthen, W. B. (1996). Community composition andnested-subset analyses: basic descriptors for commu-nity ecology. Oikos, 76, 417–426.

Wragg, G. M. (1995). The fossil birds of Henderson Island,Pitcairn Group: natural turnover and human impact, asynopsis. Biological Journal of the Linnean Society, 56,405–414.

Wright, D. H. (1983). Species–energy theory: an extensionof species–area theory. Oikos, 41, 496–506.

Wright, D. H., Currie, D. J., and Maurer, B. A. (1993).Energy supply and patterns of species richness on localand regional scales. In Species diversity in ecologicalcommunities: historical and geographical perspectives (ed.R. Ricklefs and D. Schluter), pp. 66–74. University ofChicago Press, Chicago.

Wright, D. H., Patterson, B. D., Mikkelson, G. M., Cutler,A., and Atmar, W. (1998). A comparative analysis ofnested subset patterns of species composition.Oecologia, 113, 1–20.

Wright, D. H. and Reeves, J. J. (1992). On the meaning andmeasurement of nestedness of species assemblages.Oecologia, 92, 416–428.

Wright, S. (1932). The roles of mutation, inbreeding,crossbreeding, and selection in evolution. Proceedingsof the XI International Congress of Genetics 1, 356–366.

Wright, S. D., Yong, C. G., Dawson, J. W., Whittaker, D. J.,and Gardner, R. C. (2000). Riding the ice age El Niño?Pacific biogeography and evolution of Metrosideros

subg. Metrosideros (Myrtaceae) inferred from nuclearribosomal DNA. Proceedings of the National Academy ofSciences of the USA, 97, 4118–4123.

Wright, S. J. (1980). Density compensation in islandavifaunas. Oecologia, 45, 385–389.

Wyatt-Smith, J. (1953). The vegetation of Jarak Island,Straits of Malacca. Journal of Ecology, 41, 207–225.

Wylie, J. L. and Currie, D. J. (1993a). Species–energy the-ory and patterns of species richness: I. Patterns of bird,angiosperm, and mammal richness on islands. BiologicalConservation, 63, 137–144.

Young, T. P. (1994). Natural die-offs of large mammals:implications for conservation. Conservation Biology, 8,410–418.

Zann, R. A. and Darjono (1992). The birds of Anak Krakatau:the assembly of an avian community. Geojournal, 28,261–270.

Zavaleta, E. S., Hobbs, R. J., and Money, H. A. (2001).Viewing invasive species removal in a whole-ecosystemcontext. Trends in Ecology and Evolution, 16, 454–459.

Zavodna, M., Arens, P., van Dijk, P. J., Partomihardjo, T.,Vosman, B., and van Damme, J. M. M. (2005).Pollinating fig wasps: genetic consequences of islandrecolonization. Journal of Evolutionary Biology, 18,1234–1243.

Zimmerman, B. L. and Bierregaard, R. O., Jr (1986).Relevance of the equilibrium theory of island biogeog-raphy and species–area relations to conservation with acase from Amazonia. Journal of Biogeography, 13, 133–143.

Zink, R. M. (2002). A new perspective on the evolutionaryhistory of Darwin’s finches. Auk, 119, 864–871.

R E F E R E N C E S 381

Page 395: Whittaker. Island Biogeography. 2007

This page intentionally left blank

Page 396: Whittaker. Island Biogeography. 2007

A-tramp categories 110abiotic forces 239Acacia melanoxylon (tree) 317Accipiter gentilis (goshawk) 286Achatina (snail) 315Achatinella (land snail) 225, 315Acrocephalus familiaris (millerbird) 299actual evapotranspiration (AET) 97adaptive differences 200adaptive landscapes 205adaptive radiation 5, 217–30, 235, 239

Darwin’s finches and Hawaiian honeycreeper-finches219–25

emergent models of evolution 208, 230, 233, 234, 243,245, 247, 248

Hawaiian crickets and drosophilids 225–8island-hopping allopatric radiations 233–7island-hopping on grand scale 237–8non-adaptive radiation 230plants 228–30speciation within an archipelago 230–3

Adeje (Canary islands) 28, 29Aegean islands 230Aegithalos (tit) 258, 275Aeonium 62, 66, 180, 228, 229, 234, 235, 237Aepyornis maximus (bird) 306, 313Africa and islands 186, 285African Great Lakes 231African rift valley 293Agadir (Morocco) 60Aglycyderidae 201agricultural base, narrow 338agricultural land, lack of 338Aichryson 66, 228, 229Aigrettes, Ile aux (Mauritius) 180Akialoa 223Åland islands 122, 147–8Albinaria (land snail) 230Aldabra atoll 68, 181, 186, 332Alegranza (Canary islands) 29, 231Aleutia Islands 27Alisinidendron 236allopatry/allopatric 199, 201, 207, 228, 229

distributions 196

phase 248populations 224–5radiations, island–hopping 233–7

allopolyploid species 205allozymes 169–70altitudinal zones, contraction/telescoping of 34Amaranthaceae 184Amazona (parrot) 255–6, 339, 340Amereiva polyps (lizard) 189American warblers 122amphibians 296Amsterdam Island 238, 313anacladogenesis 206, 207, 208Anaga (Canary islands) 28, 29anagenesis 204, 206, 207, 208–9, 248Anak Krakatau 105, 132, 141, 143, 154Anas (duck) 183, 313Andropogon virginicus 320Anegada 314anemophily 178angiosperms 65, 66, 97, 98, 206, 240Anguilla 215, 216anoline lizards 59, 68, 176, 179, 215–17, 234, 240Anseranas semipalmata (goose) 282Anthophora allouardii (bee) 180anthropogenic changes 152–3anthropogenic experiments: competitive effects 125–6anthropogenic losses and threats to ecosystems

290–322Caribbean land mammals 314–15decline, trends in causes of 302–5disease 300–2extinctions, current 290–1fragility and invasibility 320–1habitat degradation and loss 302Indian Ocean birds 312–13introduced species 295–300island losses, scale of globally 292–5Pacific Ocean birds and Easter Island enigma 307–12plants 316–20predation by humans 295reptiles 313–14snails 315–16stochastic versus deterministic extinctions 291–2

383

Index

Page 397: Whittaker. Island Biogeography. 2007

Antigua 215, 216, 217Antilles 59, 70, 117, 247, 305

see also Greater Antilles; Lesser Antillesants 209–11, 219, 299, 315Apia mellifera (honey bee) 299Apteribis (ibis) 183aquatic species 292Aquifoliaceae 237Aquila chrysaetos (eagle) 267Araliaceae 237area 247, 276

cladograms 198, 236diversity 89–90effect on immigration rate 102effects, minimum 121–isolation 246minimum dynamic 274–5not always first variable in model 90–1per se effect 89requirements, differential 128see also minimum viable areas; species–area

Argyranthemum 62, 67, 228, 229, 234Argyroxiphium 66, 228, 235, 237Arrhenius plots 81arrival and change 167–94

character displacement 176–7empty niche space 172–6founder effects, genetic drift and bottlenecks 167–72pollination and dispersal networks 179–81reproduction 177–9see also niche shifts and syndromes

arrival rate 120Artamus insignis (bird) 108Arthrophyllum javancium (tree) 142arthropods 103Aruba island 122Ascension island 22assembly process 141–3assembly rules 108, 111–13, 114–15, 116, 126assortative equilibrium 103Asteraceae 50, 62, 66–7, 131, 135, 183–4, 197

emergent models of island evolution 228, 229, 234,235, 236, 237, 240, 241

Atlantic Ocean and islands 184, 201, 203, 320, 322, 339Atlantoxerus getulus (squirrel) 299Atlas tectonics 22atolls 22–4, 30, 32, 37, 67, 160

conservation of ecosystems 323, 324emergent models of evolution 239, 243

Auckland Islands 183Australia and islands 17, 19

anthropogenic losses and threats to ecosystems 290arrival and change 169, 178, 189–90community assembly and dynamics 123, 128

conservation of ecosystems 331island theory and conservation 258–9, 268, 277, 279,

280–1, 282–3, 287Perth islands 90scale and ecological theory 158, 163speciation and island condition 206species numbers 97, 102zoogeographic regions 52–3, 54

autecological requirements 283autopolyploid species 205Azores 19, 180, 203, 305

biodiversity ‘hotspots’ 50, 60, 62, 63–4emergent models of evolution 232–3, 245

Azorina vidalii 180

back-colonization 248back-dispersal 235Bahamas 102, 261, 314, 331, 338Baja California 273balance of nature paradigm 253Balearic islands 72, 181, 299, 314Baltic Sea islands 189banks 24Barbados 321, 338barbs and hairs, loss of 183Barbuda 217Barkley Sound (Canada) 177, 183barrier formation 203Barringtonia association 131, 133Barro Colorado island (Panama) 269–70, 274, 286basaltic continental shelf 12Bass Straits islands 121bats 69, 71, 72, 135–6, 143, 214, 304

see also fruit batsbecard 213beech forest 88beetles 219, 227behaviour, changes in 190–2Bermuda 126, 174Bidens 183bimaculatus group 215–16, 217Bimini islands 337biodiversity, global significance of 46–9biological control 331Biosphere Reserves 328, 333biotas, scale and dynamics of 147–9biotic collapse 264biotic forces 239Birdlife International 48

Endemic Bird Area scheme 47, 333birds 52, 65, 72, 73, 74, 191, 219

anthropogenic losses and threats to ecosystems 292–4,296, 301, 302–3, 304–5, 321–2

arrival and change 176–7

384 I N D E X

Page 398: Whittaker. Island Biogeography. 2007

biodiversity ‘hotspots’ 68–70, 71Caribbean 211–15community assembly and dynamics 108, 110–11, 122,

126, 128–9, 135–6, 139–40, 143Indian Ocean 312–13island theory and conservation 258–9, 265, 267,

269–70, 275, 278, 284, 287macroecology of island biotas 79, 84, 86, 98, 99, 100Pacific Ocean 307–12scale and ecological theory 152, 153see also flightless birds

Bismarck islands 209biodiversity ‘hotspots’ 68, 69, 71community assembly and dynamics 108, 109, 110, 111,

112, 116, 118, 143Blarina brevicauda (shrew) 121–2, 173block effects 120blue tit 62Boiga irregularis (brown tree snake) 303–4, 317, 320,

321, 322Bombus (bumbleebee) 180, 299boom-and-bust patterns 321Boraginaceae 67, 228bottlenecks 167–72, 194, 203, 253, 254–5boundary zone 276Branta sandvicensis (goose) 308–9Brazil (Amazonia) 254, 270–1, 272, 273, 292British Isles 12, 14, 26, 27, 246

island theory and conservation 255, 258, 262–3, 266,267, 268, 274

scale and ecological theory 158species numbers 101

browsers 298–9Bubo virginianus (owl) 286Bucconidae 122Bufo margaritifera (toad) 196buglosses 184bumble bee 299butterflies 69–70, 139–40, 141, 197, 259–60, 262–3Bystropogon 62

C-score 117, 125Cabra Corral dam islands 125Cactaceae 327Cactospiza (finch) 219Calathus (beetles) 237calderas 42, 43California islands 49, 91Callosciurus prevosti (squirrel) 186Calyptorhyncus funereus latirostrus (cockatoo) 268, 279Camarhynchus (finch) 197, 219, 220Campanula 178Campanulaceae 50, 236, 237, 299Canada 274, 331

Canadian woodlots 151Canariomys (rat) 306Canary islands 21–2, 43

anthropogenic losses and threats to ecosystems 291,299–300, 305–7, 314–16, 319–20, 322

arrival and change 171, 177, 178, 180, 181, 184, 189, 191

biodiversity ‘hotspots’ 50, 60, 61, 62, 63–4, 67, 68, 71conservation of ecosystems 323, 327, 331, 337, 339,

340–1developmental history 27–9emergent models of evolution 213–14, 217, 228, 231–2,

234–7, 239, 242–3lizards 176protected area and species protection systems 333–7speciation and island condition 198, 199, 203, 206spiders, beetles and snails 227unsustainable development 325–8

Cape Verde islands 21, 22, 184anthropogenic losses and threats to ecosystems

300, 305biodiversity ‘hotspots’ 48, 60, 61, 62emergent models of evolution 232–3

Caribbean and islands 33, 39arrival and change 179biodiversity ‘hotspots’ 47, 68birds 211–15disturbance phenomena 40land mammals 314–15lizards 215–17scale and ecological theory 156–7, 158tectonic plate 59

Carolines 69Carophyllaceae 236Carpodacus mexicanus (finch) 319carrying capacities, temporal

variation in 156–61endemics, implications for 159–61intense disturbance events 156–7long term non-equilibrium systems 158–9species number variation in short and medium term

157–8Casuarina equisetifolia (tree) 131catastrophic losses 243caterpillars 219cats 298, 307, 313, 314, 328, 332cattle 307, 313, 317, 328cays 160Cebú island 302Cecropia 274Central America 196, 214Centropus violaceus (bird) 108Certhia (treecreeper) 275Certhidea (finch) 219

I N D E X 385

Page 399: Whittaker. Island Biogeography. 2007

chaffinches 62, 203Chagos Islands 323Chalcides (skink) 181, 237Chalopsitta cardinalis (parrot) 116chance 115, 142Channel Islands 100Chappell Island (Bass Strait) 190character displacement 176–7, 215, 216, 217, 240, 245Charandrius dubius (plover) 276Charles Darwin Research Station 328, 329Chasiempis sandwichensis (elepaio) 222Cheirodendron 226Cheirolophus 67, 230, 237Chenopoidiaceae 184chequerboard distributions 111–12, 115–18, 125, 127, 143Chiroptera (flying fox) 69Chiroxiphia caudate 254Chitwan National Park (Nepal) 254–5choros model 90, 97Christmas Island 22, 24, 33, 55, 310, 338Cicadae 227cichlid fish in African Great Lakes 231, 240, 293Cikobia island 20CITES (Convention on International Trade in

Endangered Species) 287clades:

progressive 236unresolved 237

cladistics 198cladogenesis 206–8cladograms 198, 237Clarks Hill Reservoir islands (USA) 128Clermontia (bellflower) 234, 237Clethrionomys 119climate/climatic 115, 185

change 26–7, 286, 290, 323–4characteristics 32–6filter 73fluctuations 156, 160, 161trade winds 35, 39variation 338wind 268see also El Niño-Southern Oscillation (ENSO)

phenomena; weatherclustered island groups 21–2clutch sizes, small 190–1Cneorum tricoccon (plant) 299Coccoidea 327cockatoos 279Cocos Island 48, 197, 220, 221

finch 174, 204Coenocorypha aucklandica (snipe) 159–60colonization 85, 86

of another island 244

back-colonization 248curve 85, 86, 92deterministic 262–3early 320island theory and conservation 289Krakatau 136–41pattern, repeated 248prehistoric 313–14rates 100, 103, 119recent 237repeated from outside archipelago 237scale and ecological theory 158see also recolonization

Columba (bird) 191, 213, 276Columbidae 122Columbiformes 312, 313combination rules 112–13, 114, 144Commidendrum rotundifolium 317community assembly and dynamics 107–44

anthropogenic experiments: competitive effects 125–6assembly patterns and habitat factors 122–5assembly rules 108chequerboard distributions 111combination and compatibility 111–13criticisms, null hypothesis and responses 113–18dynamics of island assembly 110–11incidence functions 118–22incidence functions and tramps 108–10nestedness 126–9see also Krakatau: succession, dispersal structure and

hierarchiesComoros 69compatibility 111–13, 144compensatory effects 120, 121competition/competitive 105

arrival and change 173community assembly and dynamics 114, 115, 116, 117,

119, 144diffuse 104effects 125–6emergent models of evolution 238–9hypothesis 185speciation model 207

Compositae: see Asteraceaecomposite cone volcano 42conservation:

biogeography 48, 252biology 251see also conservation of ecosystems; theory and

conservationconservation of ecosystems 323–41

biological control 331Canary islands: protected area and species protection

systems 333–7

386 I N D E X

Page 400: Whittaker. Island Biogeography. 2007

Canary islands: unsustainable development 325–8Galápagos islands: threatened evolutionary showcase

328–30Maldives: climatic change 323–4Nauru: destruction of the island 324–5Okino-Tori-Shima: strategic economic importance 324sustainable development 337–40translocation and release programmes 331–3

Conservation International 47–8, 333hotspots programme 328

continental fragments 10–11, 14, 57–8, 203continental islands 14, 47, 57continental shelf islands 11, 12Convolvulus 62, 235Cook Islands 32, 40, 67, 69, 332Coprosma 177, 178coral growth 30coral islands 32core-satellite 261core-sink model variant 261–2corridors 278–9, 281, 284, 289Corsica 174–5Corvus corax (raven) 327Costa Rica 287counter-adaptation 212, 239Crambe 62, 234Crassulaceae 66, 228, 325Crete 230crickets 176, 225–8, 234Crocodylus niloticus (crocodile) 187Crotophaga ani (ani) 213cryptic island endemics 71–3cryptoturnover 100–1Cuba 234, 337Culex quinquefasciatus (mosquito) 301culling programmes 330Curaçao island 314Curieuse 332–3Cyanea 236Cyanocitta cristata (jay) 278Cyclades archipelago (Aegean Sea) 257Cyclodina alani (skink) 181cyclonic storms 246Cyclura (iguanine) 314Cylindroiulus 227Cyperraceae 50Cyprus 72, 189Cyrptoblepharus (lizard) 237Cyrtandra (shrub) 66, 74, 133, 179, 198, 225, 318

Dactylopius coccus (insect) 327Daphne Major island 160, 179Darwin island (Galápagos islands) 220Darwin’s finches see Galápagos islands finches

deer 188, 313, 317defence hypothesis 192Dendroica petechia (yellow warbler) 214density compensation 174–6, 179, 191, 193density overcompensation 174, 175density stasis 193Desventuradas, Las islands 48determinism 142Dicaeum quadricolor (flowerpecker) 302dicots 241differentiation 148dimorphy 177dioecy 177–8Diospyros egrettarum (ebony) 181diplochores 136diplopods 227disease 298, 299, 300–2, 322disharmony 50–2, 73dispersal:

ability 240back–dispersal 235community assembly and dynamics 115differences 114distances/range 248, 289filters 119limits 50–2networks 179–81, 299–300, 322powers, loss of 182–4scale and ecological theory 162-structured model of recolonization 134–6sweepstake 52, 73traits 241

distance effect 83–4, 127, 262and species numbers 91–2

distributional context and speciation 199–201disturbance:

hypothesis 88see also natural disturbance

diversity begets diversity model 244dodo 173, 182, 306, 313dogs 298–9, 302, 307, 313, 314, 328, 330Dolichoiulus 227Dominica 217donkeys 317, 328double invasions 203, 245doves 118Dracaena (tree) 62, 291Drepanidinae 222, 235Drepanis pacifica (mamo) 295Drosophila (fruit fly) 67, 74, 168–9, 201, 203, 254–5

emergent models of island evolution 225–6, 228, 234,239, 240

drosophilids 217, 225–8Dubautia 66, 228, 236

I N D E X 387

Page 401: Whittaker. Island Biogeography. 2007

Ducie atoll 312Ducula (fruit pigeon) 113, 118, 149, 312dung-beetles 227dunnock 258dwarfism see nanismdynamic equilibrium 77, 150, 155, 161dynamic non-equilibrium 151, 155dynamics see community assembly and dynamicsDysdera (spider) 227, 237Dysoxylum gaudichaudianum (tree) 142

earthquakes 39East Indies 84East Melanesian islands 47Easter Island 27, 50, 302, 307–12, 322Eastern Wood 101echelon lines 21Echium (plant) 62, 67, 171, 181, 185, 228, 299ecological niche 172ecological release 173–4, 320ecological trap 278ecosystem transformers 292, 295ecotone 276, 277edge effects 276–8, 289edge habitats 289edge-related changes 269El Hierro (Canary islands) 22, 28–9, 43–4

anthropogenic losses and threats to ecosystems 314conservation of ecosystems 333emergent models of evolution 231, 232speciation and island condition 199

El Niño-Southern Oscillation (ENSO) phenomena 34, 39,40–1, 179, 338

emergent models of evolution 240, 246scale and ecological theory 159–60, 161, 164

Elaeocarpus joga 317Elanus caeuleus (kite) 156elephants 188, 189, 286Emberiza citrinella (yellowhammers) 258Emberizinae 219Emoia (skink) 68, 298Emperor Seamount Chain 21, 30empty niche space 172–6, 204

emergent models of evolution 218, 222, 239, 243, 245

end-Cretaceous event 290endemism/endemic 46, 63–71

animals 67–71carrying capacities, temporal variation in 159–61emergent models of evolution 240–1insular between taxa, variation in 240–5multiple-island 231–2, 244, 245neo- and palaeoendemism 63–5plants 65–7

super-generalists 180–1see also single-island endemics

energy reallocation (Cody) hypothesis 191ENP (Espacios Naturales Protegidos) network (Canary

islands) 333, 335environmental catastrophe 255environmental changes over long

timescales 22–32Canary islands, developmental history of 27–9climate change 26–7Hawaii, developmental history of 29–31Jamaica, developmental history of 31–2sea level changes 22–4sea level changes, eustatic 24–6

Eopsaltria australis (robin) 123equilibrium 101–2, 103–4, 150–6, 159

static 150, 155see also equilibrium model of island biogeography

(EMIB); equilibrium theoryequilibrium model of island

biogeography (EMIB) 6, 78–9, 84–7, 88–9, 90, 92community assembly and dynamics 114, 120, 126, 131,

137, 140, 141island theory and conservation 252, 260, 263–5,

269–70, 272, 274, 283, 288limitations 145–9scale and ecological theory 146, 149, 150, 153–4, 155,

158, 161, 164species numbers 99, 101, 102, 103–4, 106

equilibrium theory 79–87, 105distance effect 83–4island species-area relationship (ISAR) 81–2species abundance distributions 82–3turnover and equilibrium model of island

biogeography (EMIB) 84–7Erica 319Erysimum 230establishments 152‘Eua 73, 332Euglandina rosea (snail) 299, 315Euphorbia 181, 300, 327Euphydryas editha bayensis (checkerspot butterfly)

259–60European Union Natura 2000 333, 334, 335, 336Eutrichopus (beetle) 237evolution 92

in isolation 320scale 148, 164scale and ecological theory 162see also evolution models

evolution models 208–48anagenesis: speciation with little or no radiation

208–9forcing factors 238–40

388 I N D E X

Page 402: Whittaker. Island Biogeography. 2007

hierarchies 245–7insular endemism between taxa, variation in 240–5see also adaptive radiation; taxon cycle

Exclusive Economic Zones 324, 338exotics, introduction of 322extinct island endemics 71–3extinction 85

anthropogenic losses and threats to ecosystems 293causes 104–6community assembly and dynamics 120, 127, 129, 138,

140, 141current 290–1curves 149debt 8, 269, 272, 291deterministic 262–3, 291–2differential 275emergent models of evolution 237, 244events, mass 290final 292island theory and conservation 256, 259–60, 269–71,

276, 284–5, 288–9local species 292scale and ecological theory 152, 154, 155, 158–9,

161, 162species numbers 84, 86, 99, 102, 106stochastic 262, 291–2see also extinction rates

extinction rates 85community assembly and dynamics 118, 119,

130, 137island theory and conservation 255, 264, 266, 274species numbers 100, 101, 103, 104, 105

Fabaceae 67Faial–Pico fusion 233Falco (bird) 116, 191, 286Falkland islands 50Faray (Orkneys) 168Farne Islands 101Farquhar Group 55faunal drift 169fecundity, changes in 190–2fence effect 193Fernandina (Galápagos islands) 219–20, 233ferns 50, 131, 133, 134, 135, 141–2, 240–1Ficus (fig) 133, 136, 197fig wasps 170, 179Fiji 12, 69, 172, 209, 219, 298, 332filtering effects 50–2, 289finches 173–4, 197

Laysan 160, 177, 240see also Galápagos islands finches; Hawaii

honeycreeper-finchesFringilla (chaffinch) 191

Finland islands 118fire regimes 268, 275Fitchia 183, 318fitness 253flightless birds 182–3, 188, 189, 190, 306, 307,

308–9, 313flightless insects 93, 182, 225–8, 245flightless rails 71, 173, 294floras 152Flores (Azores) 61, 180Flores (Indonesia) 187Florida and islands 273Florida Keys 103, 267flower attractiveness, loss of 178fluctuating asymmetry 254flux of nature paradigm 253, 283flycatchers 122flying fox (fruit) bats 69, 160–1, 295food chain links 189forcing factors of evolution 238–40forest 135, 136, 270–1, 287

closure 133fragmentation 269succession 149

founder effects 167–72, 184, 188, 209, 214, 218founder events 203, 233, 246founder principle 167, 194Fragaria chiloensis 177fragility of ecosystems 320–1fragmentation/fragmented:

forest system 254island theory and conservation 266, 270, 271, 274, 283,

284, 289landscapes 285–6landscapes and succession 274–6states 338systems, physical changes and hyperdynamism

of 268–9Frankia 319Fraxinus uhdei (ash) 319freshwater bodies, islands occurring in 11Fringilla (chaffinch) 203, 235Frisian Islands 12frogs 189, 219, 266fruit bats 72, 74, 135, 148–9, 160–1, 173, 295, 304

see also flying foxfruit flies 201fruit-pigeons 136, 149Fuerteventura (Canary islands) 21, 27–8, 29, 43, 171

anthropogenic losses and threats to ecosystems 299

biodiversity ‘hotspots’ 60emergent models of evolution 231Ramsar Convention wetland (El Matorral) 333

I N D E X 389

Page 403: Whittaker. Island Biogeography. 2007

Galápagos islands 3, 5, 22, 27, 33, 37, 40anthropogenic losses and threats to ecosystems 295,

299, 300, 305, 314, 318arrival and change 174, 177–81, 183, 184,

186, 189biodiversity ‘hotspots’ 47, 50, 64–5, 68conservation of ecosystems 323, 331, 339, 341emergent models of evolution 217, 219, 232–3, 240–1,

247, 248National Park 328scale and ecological theory 152, 160speciation and island condition 197, 203–4threatened evolutionary showcase 328–30see also Galápagos islands finches

Galápagos islands finches (Darwin’sfinches) 68, 160, 179, 203–4, 219–25emergent models of evolution 217, 240speciation and island condition 195, 200, 205

Gallicolumba (dove) 312galliforms 126Gallirallus (rail) 71, 173, 183Gallotia (lizard) 68, 71, 181, 198, 199, 327

emergent models of evolution 234, 236, 237losses and threats to ecosystems 291, 314

Gallotinae 199Gatun Lake and islands 119, 286geckos 70, 178–9, 181, 304genetic clock 198genetic drift 167–72, 203, 288

emergent models of evolution 209, 218, 230, 246,247, 248

Genovesa, Isla 205Geochelone (giant tortoise) 181, 186, 314, 329, 332–3geoid surface 25Geospiza (finch) 179, 197, 205, 219, 220, 224Geospizinae 203–4Geraniaceae 237Geranium 235, 237Gesneriaceae 50, 66, 198Ghyben-Herzberg lens 36giant tortoises 181, 186, 189, 306, 313, 314, 332–3gigantism 186, 189–90gleaning flycatchers 113glossary 342–50goats 299, 306, 307, 313, 314, 317Gomera, La (Canary islands) 21, 29

anthropogenic losses and threats to ecosystems 300,314

biodiversity ‘hotspots’ 64, 71conservation of ecosystems 335emergent models of evolution 231, 232Garajonay National Park 333pollination networks 180speciation and island condition 199

Gondwanaland 17–19, 56Gonopteryx (butterfly) 234Gough Island (Tristan da Cunha) 22, 56, 58grades, progressive 236grain 267

size 271Gran Canaria (Canary islands) 28, 41, 43, 232, 300

conservation of ecosystems 326, 333, 335grasses 131, 133, 134, 135, 141–2Great Barrier Reef 156Great Basin (North America) 92, 121, 159, 276great spotted woodpecker 258great tit 258Greater Antilles 176, 214, 234Greater Sundas 13Greenovia 66Grenadines 333, 340grizzly bears 254ground squirrel 299Guadeloupe 22, 169, 215, 217Guam 303–4, 317–18, 320, 321, 322Guri, Lake islands (Venezuela) 125guyots 12, 22–4, 30gymnosperms 65

habitat 115degradation and loss 302diversity 88, 89–90, 231loss 322-unit model of island biogeography 90variable 119see also habitat islands

habitat islands 5, 10, 11, 12, 252–3community assembly and dynamics 122island theory and conservation 259, 273–4, 280, 282,

283, 285, 286, 287scale and ecological theory 158

Halcyon cinnamomina (kingfisher) 303–4half-life (species losses) 271Halictus (bee) 180Haplochromis 293Happy islands (Macaronesia) 60–3Harpagornis moorei (eagle) 295Hawaii 3

anthropogenic losses and threats to ecosystems 295,299, 301, 303, 305, 307–9, 315–21

arrival and change 174, 178, 179, 182, 183, 184biodiversity ‘hotspots’ 50, 54, 65, 66, 67, 68, 69, 70,

72, 74community assembly and dynamics 126conservation of ecosystems 331, 337, 339crickets 176, 225–8developmental history 29–31drosophilids 168–9, 225–8

390 I N D E X

Page 404: Whittaker. Island Biogeography. 2007

emergent models of evolution 218–19, 227, 234–7,239–40, 241, 242–3, 245, 247–8

eruptions 41honeycreeper finches 68, 93, 209, 217, 219–25, 240, 328island environments 13, 15, 19, 20–1, 32, 37, 42, 43plants 177scale and ecological theory 152, 153speciation and island condition 198, 201, 203species numbers 93

Hedera 237Hegeter (beetle) 234Helianthemum 230Hemignathus 223Henderson Island 24, 72, 312, 322herbaceous plant lineages, woodiness in 184–6herbs 297hermaphroditic plants 177Hermes Reef 177Hesperomannia (plant) 234Heteralocha acutirostris (huia) 295hierarchical interdependency 148–9hierarchies 245–7, 285–6high-S (sedentary) species 108, 110, 143Himatione sanguinea (apapane) 223–4hippopotami 188Hispaniola 214, 234, 314historical context and speciation 201–2hominids, dwarf 187Homo erectus 187Homo floresiensis 187Homoptera 327honey bee 299Hoplodactylus (gecko) 181horses 307, 317‘hotspots’ 4, 5, 46–74, 333

biodiversity, global significance of 46–9cryptic and extinct island endemics 71–3filtering effects, dispersal limits and disharmony 50–2island environments 13–14, 16, 19, 21, 22, 29Macaronesia: Happy Islands 60–3regionalism and vicariance/dispersalism debate 52–60species poverty 49–50see also endemism/endemic

human activity 290, 291–2hurricanes 39–41, 156–7, 255–6Hybiscadelphus 236hybridization 179, 194, 196, 197, 206, 207

emergent models of evolution 230, 240with native species 300see also introgressive hybridization

hyperdynamism of fragment systems 268–9Hypochaeris radicata 183

Iceland 13, 15, 19, 42

icelandic eruptions 41Icterus oberi (oriole) 256Ilex 237immigration:

community assembly and dynamics 120, 127, 129,136, 140

curves 149filters 119immigration ratesisland theory and conservation 269scale and ecological theory 155selection 183species numbers 84, 85–6, 92, 99, 103supplementary 102see also immigration rates

immigration rates:community assembly and dynamics 131, 137and island area 102island theory and conservation 264scale and ecological theory 152, 154species numbers 85, 100, 101, 104

Imperata cylindrica (grass) 133inbreeding 253–4incidence functions 88, 257–9, 288

community assembly and dynamics 108–10, 116,118–22, 123–5

incidence patterns 120–1incidence rules 112Indian Ocean and islands 56

anthropogenic losses and threats to ecosystems320, 322

arrival and change 186biodiversity ‘hotspots’ 69birds 312–13conservation of ecosystems 323emergent models of evolution 237–8

Indian subcontinent 14individual scale 147, 164individuals, numbers necessary 253–7Indonesia and islands 19, 35, 209, 337insects 65, 67–8, 218, 293, 296, 317

see also flightness insectsintense disturbance events 156–7Interagency Spotted Owl Scientific

Committee 260intercontinental trade 321International Union for the Conservation of Nature 271,

284, 295protected areas model 333, 334Red List assessment (1997) 316

intraplate islands/clusters 16, 19, 20–2introduced species 295–300introgressive hybridization 168–9, 198, 224–5,

230, 300

I N D E X 391

Page 405: Whittaker. Island Biogeography. 2007

invader complexes 180invasibility of ecosystems 320–1invasive alien species 296–8, 327Ireland 158Irian Jaya 266Isabela (Galápagos islands) 219–20, 233, 328island:

area effect on immigration rate 102area-isolation 246condition see speciation and island conditionextirpation 292-hopping allopatric radiations 233–7-hopping on grand scale 237–8immaturity–speciation pulse model 239, 242–4, 248rule 186–90, 194sterilization 41see also island species–arearelationship (ISAR)

island assembly 6island rule 186–90island species-area relationship (ISAR) 79–84, 87, 88,

89–99, 105, 106area and habitat diversity 89–90area not always first variable in model 90–1distance and species numbers 91–2island theory and conservation 267, 271, 272remote archipelagos 92–3scale and ecological theory 153, 159scale effects 93–7species-energy theory 97–9

isolation 22, 93–4, 99, 247, 281area versus island-or archipelago 248benefits 279–81community assembly and dynamics 119, 120, 121,

122, 128-incidence functions 127intraisland 248reproductive 196species numbers 102

Isoplexis 62, 181, 206isostacy 22isostatic changes 25, 26

Jamaica 24, 59, 97, 338developmental history 31–2emergent models of evolution 212, 213, 234

Jandía (Canary islands) 28Japan 47Jarak island 49Jasper Ridge Preserve (USA) 259Juan Fernández islands 3, 47, 305

arrival and change 178, 184biodiversity ‘hotspots’ 47–8

emergent models of evolution 208, 209, 240speciation and island condition 197, 206, 207

Judas goats 329–30

K-selected species 87, 110, 261Kakamega forest 271Kapingamarangi atoll 89Kauai (Hawaii) 20, 30, 222, 226, 228, 234Kenya 271Kerguelen Island 17, 27, 50keystone species 286kipukas 228Kiribati 324Kokia 234, 235Krakatau islands (Indonesia) 6, 34, 41, 42

arrival and change 170, 176, 179, 182biodiversity ‘hotspots’ 50community assembly and dynamics 107, 111, 113, 128,

130, 132, 137, 139, 143–4emergent models of evolution 241island theory and conservation 262, 285residency and hierarchical interdependency 148–9scale and ecological theory 151, 154, 157, 161speciation and island condition 201species numbers 81, 86, 89, 104succession, dispersal structure and hierarchies

131–43colonization and turnover: dynamics of species lists

136–41dispersal-structured model of recolonization 134–6organization in assemblyprocess 141–3succession 131–4

Lactoridaceae 197, 208Lactoris fernandeziana 197, 208Lactuca muralis 177, 183, 184laderas 43lag effects 272, 273lakes (negative islands) 292–3Lakshadweep Islands 323Lamiaceae 50, 67Lanai (Hawaii) 30land crabs 309land snails 67, 74, 198, 227, 230, 245, 299land-bridge islands 11, 14, 167, 203

community assembly and dynamics 108, 111, 114, 122,128, 129

scale and ecological theory 156, 158, 159landscape effects 289Lantana camara (shrub) 318, 328Lanzarote (Canary islands) 21, 28, 29, 244, 333Laparocerus (beetle) 67, 227, 325

392 I N D E X

Page 406: Whittaker. Island Biogeography. 2007

Lates niloticus (perch) 293Laupala (cricket) 176, 236laurel (Laurus) 62laurel-forest tree 184Lavandula 62, 180Lavatera 237Laysan island (Hawaii) 30Leeward Islands 67Leguminosae 317lemurs 18, 201, 313Lepidodactylus (gecko) 178–9Lesser Antilles 89, 298

conservation of ecosystems 323, 339–40emergent models of evolution 211, 213, 214,

215–16, 217Lesser Sundas islands 13, 52, 304lime, short-leaved 266limestone islands 32Limonium 62, 67, 230, 237Line Islands 21lineage structure 206–7linear island groups 20–1Linnean shortfall 290lizards 68, 72, 105, 178, 189, 190,

191, 219, 300, 313–14Canary islands 71, 170, 198, 236Caribbean 176, 215–17Greater Antilles 234Indian Ocean islands 237Lesser Antilles 170Pacific 70as pollinators 181Seychelles 93see also anoline; Gallotia

llicura militaris 254lobelioids 184Loboptera (cockroach) 227, 237locational context and speciation 201–2log-log models 267log-normal series of abundance 82–3Loihi Seamount (Hawaii) 21Long Island (New Guinea) 105, 176–7

community assembly and dynamics 108, 111, 130, 131

longevity hypothesis 185Lotus 67, 237Loxigilla portoricensis grandis (bullfinch) 256Luscinia megarhynchos (nightingale) 258

Macaranga tanarius (tree) 133Macaronesia 319

arrival and change 180, 184, 185biodiversity ‘hotspots’ 47, 48, 56, 63–4, 66, 74

emergent models of evolution 228–9, 230, 234, 235,240, 248

speciation and island condition 203, 206Macquarie Island 177, 298Macrolepidoptera 262Macropygia (bird) 109, 111–13Madagascar 13, 17, 18, 186, 278

anthropogenic losses and threats to ecosystems 305,306, 312–13

biodiversity ‘hotspots’ 47, 56, 66, 74speciation and island condition 201, 203

Madeira 184, 203, 227anthropogenic losses and threats to ecosystems 305, 315biodiversity ‘hotspots’ 60, 61, 62, 63–4

Madia 228Madiinae 226magpie goose 282–3Majorca 299Makatea Island 24, 32malaria 301Maldives 323–4Malé Atoll 324Malpelo islands 47Malta 314Malvaceae 234, 235, 236, 237mammals 69, 71, 72, 98, 119–20, 190, 191, 219, 314–15

island theory and conservation 279, 280–1, 286losses and threats to ecosystems 292, 293, 296–7

Manacus manacus 254Manaus habitat fragmentation study 284, 286Mangareva island 23, 312manuka scrub 88Marcetella 62Margarita island 122Mariana (Micronesia) 317Marion Island 27, 298Marquesas islands 21, 69, 305, 337marsh tit 258marsupials 69, 70–1, 129Martes martes (pine marten) 299Martinique (Caribbean) 41Masafuera island 3–4, 208Masatierra island 3, 4, 208, 240Mascarenes 241, 331

anthropogenic losses and threats to ecosystems 305,307, 314, 316

arrival and change 186, 189biodiversity ‘hotspots’ 47, 69

mass-extinction events 290Massachusetts islands 119matrix effects 287matrix habitats 289Maui (Hawaii) 30, 169

I N D E X 393

Page 407: Whittaker. Island Biogeography. 2007

Maui-Nui complex (Hawaii) 30–1, 226Mauritius 68, 337

anthropogenic losses and threats to ecosystems 313,314, 315–16, 321

arrival and change 180, 181, 182meadow vole 173Mecyclothorax (beetle) 240Mediterranean islands 16

arrival and change 188, 189biodiversity ‘hotspots’ 71–2conservation of ecosystems 339species numbers 96

mega-disturbance phenomena 244mega-landslides 42–4mega-tsunamis 43–4Melanesia 73, 84, 118, 307

ants 82, 209–11emergent models of evolution 210, 211

Melitaea cinxia (butterfly) 270mesopredator release 270metapopulation 270, 278, 280, 286, 288–9

dynamics 259–63, 264minimum viable 260scenarios 261, 262

Metrosideros (tree) 218, 224, 238, 247, 319mice 317Michigan, Lake islands 121Miconia (tree) 318, 319, 320, 321, 322Micronesia 47, 73, 84, 307Micropsitta bruijnii (bird) 108Microtus 119, 121–2, 173, 189mid-ocean ridge islands 16, 19Midway island (Hawaii) 25, 30Mimus 220minimum area effects 121Minimum Critical Size of Ecosystem project 269, 273minimum dynamic areas 274–5minimum viable area 253–9, 288minimum viable metapopulation 260minimum viable population 252, 253–9, 260, 264, 288Miridae 236mitochondrial DNA 214moas 72, 295, 306, 309mobile species 282–3moderate insular climate hypothesis 185molecular clocks 198–9, 222, 238molecular leakage 198molluscs 65, 292, 293Molokai (Hawaii) 30, 160Mona (Antilles) 37, 56, 314Monanthes 66, 228, 229Monarcha telescophthalmus (flycatcher) 111mongooses 298, 307, 313, 331monkeys 313

monocots 241monophyletic lineage 203, 214, 217–18Monte Carlo analysis 176Montserrat (Caribbean) 39, 41, 158, 256, 338Moorea (French Polynesia) 23, 198, 299, 318Morocco 191Motmot island 131motu (sand islands) 40multiple-island endemics 231–2, 244, 245multiple-island occupancy 233Mustela erminea (weasel) 278Myadestes (thrush) 222, 301Myrica (tree) 291, 318–19, 320Myzomela (honeyeater) 176–7myzomelid sunbirds 113

nanism (dwarf forms) 186–90natural catastrophes 290natural disturbance on islands 37–44

magnitude and frequency 39–41mega-landslides 42–4volcanic eruptions 41–2

natural laboratory paradigm 3–9Nauru 324–5, 340Nazca plate 4neoendemism 63–5neoinhabited islands 305Neonauclea calycina (tree) 133Nesiota elliptica 317Nesomimus 220Nesoprosopis (bee) 299Nesotes (beetle) 232nested design 95nestedness 6, 126–9, 144, 275–6

area-ordered 127distance-ordered 127habitat 127island theory and conservation 272, 277richness-ordered 127

New Britain 84New Caledonia 18, 47, 182–3, 219, 315

biodiversity ‘hotspots’ 68, 69, 70New Georgia 183New Guinea 6, 19, 34

anthropogenic losses and threats to ecosystems 305biodiversity ‘hotspots’ 52, 53community assembly and dynamics 107, 108, 111, 113,

114, 130, 143conservation of ecosystems 337, 339emergent models of evolution 209, 210, 247species numbers 84, 97

New Hebrides see VanuatuNew Ireland 307New Providence Island 214

394 I N D E X

Page 408: Whittaker. Island Biogeography. 2007

New Zealand 12, 13, 16, 17, 18, 19, 32, 34anthropogenic losses and threats to ecosystems 295,

305, 306, 309–10, 313–14, 315arrival and change 177, 178, 181, 182, 188biodiversity ‘hotspots’ 47, 50, 66, 72, 74community assembly and dynamics 126conservation of ecosystems 331–2, 337emergent models of evolution 238island theory and conservation 257scale and ecological theory 158speciation and island condition 206species numbers 88

niche:breadth 173–4expansion 179space 215see also empty niche space; niche shifts and

syndromesniche shifts and syndromes 87, 177, 179, 181–94

dispersal powers, loss of 182–4fecundity and behaviour, changes in 190–2island syndrome in rodents 192–4size shifts in species and island rule 186–90woodiness in herbaceous plant lineages 184–6

Niuafo’ou 19non-adaptive differences 200non-adaptive radiation 208, 218, 230, 235, 239, 243, 245,

247non-equilibrium 150–6, 158–9, 283–5

static 151, 155Norfolk Island 70, 262Normania 62North America 92, 214, 224, 290

island theory and conservation 277–8, 284see also Canada; United States

northern spotted owl 286Notechis ater serventyi (tiger snake) 189–90Novocariensis 62nuées ardentes 42null hypothesis 113–18null models 117, 118, 122, 126, 144, 176nutrient fluxes 268

Oahu (Hawaii) 30, 160, 222, 225, 226, 320oceanic islands 10, 14–15, 36, 47

arrival and change 167biodiversity ‘hotspots’ 52, 53–4, 56, 57, 67emergent models of evolution 220scale and ecological theory 152–3, 159speciation and island condition 203

oceanic plate islands 12Oecanthinae 225Oeno atoll 312Okino-Tori-Shima: strategic economic importance 324

Olea (olive tree) 234Oligosoma grande (skink) 181Opuntia (cactus) 300, 317, 327orb-weaving spiders 102, 105, 261orchids 50, 135, 142Oriental zoogeographic regions 52–3, 54origin, modes of 12–22

intraplate locations 20–2plate boundary islands 16–20

orioles 256Orkney isles 190–1Oryctolagus cuniculus (rabbit) 299outcrossing 177–8outliers 276over-compensation 179

Pachycephala (flycatcher) 111Pacific Ocean and islands 16

anthropogenic losses and threats to ecosystems 294,302, 322

ants 209arrival and change 170–1, 173, 183biodiversity ‘hotspots’ 70, 73, 74birds 79, 84, 307–12conservation of ecosystems 338–9speciation and island condition 201, 202

palaeoendemism 63–5palaeoinhabited islands 305palaeoislands, fusion of 237Palma, La (Canary islands) 21–2, 28, 29, 43, 44

conservation of ecosystems 333, 335emergent models of evolution 231, 232Taburiente National Park 333

Pandanus (screw-pines) 130, 323Pandion haliaetus (osprey) 267Panjang island 132, 133Papasula abbotti (booby) 310parallel convergent morphology 228parapatric phase 248paraphyletic lineages 203Paroreomyza (Molokai creeper) 223parrots 255–6, 260–1, 339–40parthenogenesis 178–9Partula (land snail) 198, 299, 315Parulidae 122Parus (tits) 174–5, 237, 258, 275Paschalococos disperta (palm) 310passive sampling 88, 127patch dynamics 274–5Pearl Reef 177Pelargonium 238Peléan eruptions 42Penguin Bank seamount (Hawaii) 30peninsula effect 273

I N D E X 395

Page 409: Whittaker. Island Biogeography. 2007

Pennisetum setaceum (grass) 327Pentaschistis 238Pericallis 62, 67, 185, 229, 235Peromyscus 119Perth islets 93pes-caprae formation 131, 133petrels 312Petroica macrosephala dannefaerdi (black tit) 159–60Pezophaps solitaria (solitaire) 182, 313Phaeornis obscurus (thrush) 319Phalacrocorax harrisi (cormorant) 183Phelsuma (gecko) 180, 181, 314Philesturnus carunculatus (saddle–back) 331Philippines 47, 69, 71, 304–5, 337Phoenix (palm) 300phosphate mining 324–5, 340Phigys solitarius (parrot) 332Phyllis 62, 178Phylloscopus collybita (chiffchaff) 258phylogenesis 243, 245phylogenetic analyses 214, 217phylogeny (phylogenetic tree) 195–9, 207, 208,

215–16physical changes of fragment systems 268–9physical environment of islands 32–7

climatic characteristics 32–6topographic characteristics 32tracks in the ocean 37water resources 36–7

pigeons 118, 122pigs 302, 313, 314, 317, 318, 328, 330Pimelia (beetle) 235, 237Pinaroloxias inornata (finch) 174, 204, 220Pinta (Galápagos islands) 220Pinus (tree) 317, 318Pinzón (Galápagos islands) 233pioneers of interior habitats 141–2Pisonia grandis 317Pitcairn Island 24, 312Plagithmysus beetles 93plant climbers 297plants 74, 177, 183–4, 240–1

adaptive radiation 228–30anthropogenic losses and threats to ecosystems 293,

294–5, 316–20, 322endemic 65–7herbaceous plant lineages, woodiness in 184–6hermaphroditic 177invasive 327scale and ecological theory 153

Plasmodium (parasite) 301plates see tectonic platesPlatydesma 235Platypsaris niger (becard) 213

Platyspiza (finch) 219Plaza Sur 225Pleasant Island see NauruPlebejus argus (butterfly) 260Plinian eruptions 42Plocama 178Plumbaginaceae 67Poaceae 50Podarcis (lizard/skink) 181, 299pollination 179–81, 322

changes 299–300networks 194

Polynesian islands 47, 84anthropogenic losses and threats to ecosystems

307, 308 arrival and change 183–4biodiversity ‘hotspots’ 72, 73community assembly and dynamics 118emergent models of evolution 219, 245, 247

polyploidy 205–6, 207, 230Ponape island 23population:

differentiation scale 164dynamics scale 147–8, 164size, effective 254trajectories 256viability analyses 253see also metapopulation; minimum viable

populationpopulation dynamic scale 147–8Porto Santo island (Madeira) 230Porzanula palmeri (rail) 299power function model 81predation 105, 298–9

by humans 295, 322community assembly and dynamics 115, 121–2island theory and conservation 278scale and ecological theory 160

previously uninhabited islands 305prickly pears 300, 327Principe 337Pritchardia (palm) 50Procyon lotor (racoon) 278Proechimys semispinosus (spiny rat) 119Prognathogryllus (cricket) 235, 236progression rule 226, 228, 234–5, 236, 248promotion of sexual out–crossing hypothesis 185propagules 85, 100, 102protected area 333–7Psammodromus (lizard) 199Pseudonestor (parrotbill) 223pseudoturnover 100–1, 103Psidium (guava) 300, 319, 328Psittirostra 223

396 I N D E X

Page 410: Whittaker. Island Biogeography. 2007

pteridophytes (fern) 65, 240Pterodroma (petrel) 312, 318Pteropus (flying fox) 69, 148–9, 160–1, 173, 246Ptilinopus (dove) 312Puerto Rico 39, 40, 157, 216, 217, 234, 314

island theory and conservation 255–6, 260–1puffbirds 122pygmy hippo 313Pyrrhocorax pyrrhocorax (chough) 231Psychotria 318

Quercus (oak) 196

r-selected species 87, 110, 130rabbits 299, 317radiation 203

archipelagic 245explosive 238fluxes 268intraisland 235model, three-stage 231zone 71, 87, 162, 219, 245, 247see also adaptive; non–adaptive

Raiatea island 23, 318Raillardiopsis 228rails 182–3, 294

see also flightless railsraised reef islands 24Rakata island 132, 133, 134, 137, 138, 139, 140, 154Rallus owstoni (rail) 303–4random placement 87–8Raphus cucullatus (dodo) 182, 313Rattus (rat) 169, 278, 328, 331, 332

losses and threats to ecosystems 302, 307, 312, 313,314, 317

ravens 300recolonization 134–6, 144, 259–60, 288reduced predation and parasitism hypothesis 191reefs 22–4regional scale, comparisons between taxa at 69–71regionalism 52–60Regulus (goldcrest) 275Reinwardtoena 112relaxation 159, 264, 269–74, 289release programmes 331–3Remya (plant) 234reproduction 177–9reproductive isolation 196reptiles 140, 292, 293, 296, 313–14rescue effect 102, 120, 261reserve configuration 263–8reserve systems 281–2, 283, 286reserves 285, 287, 289

see also Single Large Or Several Small reserves

residency 148–9resource hypothesis 192resource predictability (Ashmole) hypothesis 191Réunion 68, 313Revillagigedo island 48Rhetinodendron 178Rhynchophorus ferrugineus (weevil) 300rhinos 254–5, 266Rhodacanthis 224Rhyncogonus (weevils) 93Rhynochetos jubatus (kagu) 182richness 276

–area relationships 278–ordered incidence functions 108

Ritter Island 105, 108, 111, 130, 131Robinsonia 240rodents 69, 71, 74, 188, 192–4, 314Rodrigues (Mascarenes) 68, 313, 314Rota 317–18Rubia fruticosa (shrub) 299Rubiaceae 50Rutaceae 50, 235

Saccharum spontaneum (grass) 133St Croix island 189St Helena 22, 184, 201

anthropogenic losses and threats to ecosystems 305,307, 316–17, 320

biodiversity ‘hotspots’ 49–50St Kitts 212, 215, 216, 256St Lucia 212, 298St Maarten 216St Paul Island 313St Vincent 333, 340Salvage Islands 60, 61Samoa 40, 160–1, 210, 332San Clemente island 156Sandwich arc 20Santa Barbara island 100Santa Cruz (Galápagos islands) 219–20, 233Santa Fe (Galápagos islands) 233Santiago (Galápagos islands) 233, 328Santorini (Aegean Sea) 41São Paulo 275São Tomé 337SAR (species–area relationship) 271Sardinia 189, 276Sarona 67, 201, 236saturated/basal water zone 36saturation curve 84Saxicola dacotiae (chat) 231scale 266–8

effects 93–7see also scale and ecological theory

I N D E X 397

Page 411: Whittaker. Island Biogeography. 2007

scale and ecological theory 145–64equilibria and non-equilibria 150–6equilibrium model of island biogeography (EMIB)

145–9future directions 161–4scale and dynamics of biotas 147–9see also carrying capacities, temporal variation in

Scalopus 119Scaptomyza 226Schiedea 236Schizachyrium (bunchgrass) 320Sclerurus 274Scilly Isles 34scorpions 227Scrophulariaceae 206sea level change 22–4

eustatic 22, 24–6rises 324, 337–8

seamounts 12, 16seas, heavy dependence on 338Sebesi island 132sedentary species 87Sedum 62Semele 62Serianthes nelsonii 317Sertung island 132, 133, 140sexual selection 230, 231

differential 168Seychelles 17, 22

arrival and change 186biodiversity ‘hotspots’ 47, 56, 58conservation of ecosystems 337emergent models of evolution 219species numbers 93

sheep 307shield volcanoes 41shrews 118–19, 121–2shrubs 135, 297–8Sicily 189, 314Sideritis 67, 234, 237Silene (plant) 169silversword alliance 66, 184, 226, 228, 234, 235simulation models 117, 127Singapore 321Single Large Or Several Small reserves (SLOSS) debate

252, 263–8, 275, 276, 284, 288–9single-island endemics 230–1, 232, 233, 239, 242, 244, 245sink habitats 262, 281, 288

see also source and sinkSitta (nuthatches) 275size shifts in species 186–90skinks 70–1, 74, 304Skokholm island 101small-island effect 88, 90, 96, 119, 267, 277

snails 227, 296, 299, 315–16see also land snails

snakes 189–90, 219see also Boiga irregularis

Snares Islands (New Zealand) 159–60Society Islands (Polynesia) 21, 69, 318Solanum 62Solifugae 227Solomon islands 84, 116

anthropogenic losses and threats to ecosystems 305, 307arrival and change 173, 183biodiversity ‘hotspots’ 68, 69, 71conservation of ecosystems 337, 339emergent models of evolution 209, 212, 219

Sonchus 62, 67, 184, 228, 234, 235Sophora toromiro (tree) 310Sorex (shrew) 118–19source and sink populations 147, 261, 278, 288South America and islands 125, 186, 196, 214, 217, 290South-East Asia 303Spain 275sparrows 100Spartina alterniflora (grass) 103spatial and size-frequency

distribution 280speciation:

allopatric 200, 218, 247archipelago 200, 205interisland 245intraisland 245with little or no radiation 208–9parapatric 200sympatric (competitive) 199, 200, 205, 222, 247within an archipelago 230–3see also speciation and island condition

speciation and island condition 195–207allopatric or geographical speciation 202–4competitive speciation 204–5distributional context 199–201lineage structure 206–7locational and historical context 201–2polyploidy 205–6species concept and place in phylogeny 195–9

species 199abundance distributions 79, 80, 81, 82–3, 84, 85, 89accumulation curves 79–80, 94–5-based approach 163concept and place in phylogeny 195–9-energy theory 88, 97–9-isolation 162lists, dynamics of 136–41numbers games: macroecology ofisland biotas 77–106species-area relationships,

398 I N D E X

Page 412: Whittaker. Island Biogeography. 2007

systematic variation in 87–9see also equilibrium theory; island species-area

relationship; turnovernumbers variation in short and medium term 157–8poverty 49–50, 320protection systems 333–7swarms 230see also species-area

species-area 162approach 283curves 80, 158–9equations 265models 271, 272, 284plot 83relationships 77, 81, 82, 87–9, 92–3, 304type 286see also island species-area relationship (ISAR)

spermatophytes 141Sphenodontia 314spiders 219, 227, 234, 236, 261

see also orb-weavingSpitsbergen island 97Sri Lanka 47starlings 100static equilibrium 150, 155static non-equilibrium 151, 155Steganecarus (mite) 237Stephens Island wren 309stepping-stone islands 19, 29, 61–2, 84, 247, 278

see also corridorsSterculiaceae 201Stewart Island 332stochastic demographic effects 288, 289stochasticity 235storm vulnerability 338strandlines 135, 141strategic economic importance 324stratovolcano 42stream-continuum concept 279Strigops habroptilus (kakapo) 182, 332Strix occidentalis caurina (owl) 260, 286strombolian eruptions 42Sturnus vulgaris 327subspecies 196, 197, 199succession 131–4

in fragmented landscapes 274–6successional island ecology 129–31successional model of island assembly 130succulents 298Sumba (Indonesia) 97Sunda islands 19, 52, 54, 84, 179Sundaland 47sunken continental shelf 12super-generalists 194

endemic 180–1supersaturation 159, 264supertramps 87, 276

community assembly and dynamics 109, 110, 111, 118,128, 130–1, 143

Sus scofra (pig) 307sustainable development 337–40sweepstake dispersal 52, 73Sylvilagus nuttallii (cottontail) 92sympatry 199, 201, 207, 217, 224–5, 228, 248system extent 266–8

Tahiti 126, 240, 307, 318, 320, 321, 322Takapoto atoll 178–9Tamias 119Tarentola (geckos) 237Tasmania 19, 181, 305Tatoosh island 158taxon cycle 5, 186, 209–17

anthropogenic losses and threats to ecosystems 291Caribbean lizards 215–17emergent models of evolution 208, 211–15, 217, 230–1,

245, 247–8evaluation 217Melanesian ants 209–11species numbers 87

taxonomy 196tectonic plates 12–13, 22

boundary islands 16–20convergent plate boundaries 19–20divergent plate boundaries 16–19transverse plate boundaries 20

control model 21, 22damage vulnerability 338

Telespiza cantans (finch) 160, 177Teline 237Tenerife (Canary islands) 21, 28, 29, 35–7, 42, 43–4

arrival and change 180biodiversity ‘hotspots’ 62, 71conservation of ecosystems 326–7, 337emergent models of evolution 227, 228–9, 231, 232,

233, 237, 243, 244island theory and conservation 278National Park 329, 333

Teno (Canary islands) 28, 29Tenrecidae 18tens rule 321, 327terminal resolution 236Terminalia catappa (tree) 131, 136territoriality, relaxation in 191–2Tetramolopium 235, 237theory and conservation 251–89

climate change and reserve systems 286corridors 278–9, 281

I N D E X 399

Page 413: Whittaker. Island Biogeography. 2007

theory and conservation (cont.)edge effects 276–8habitat islands 252–3hierarchies and fragmented landscapes 285–6islands and conservation 251–2isolation 279–81metapopulation dynamics 259–63minimum viable populations and minimum viable

areas 253–9mobile species 282–3nestedness 275–6non-equilibrium world 283–5physical changes and hyperdynamism of fragment

systems 268–9relaxation and turnover 269–74reserve configuration: Single Large or Several Small

(SLOSS) debate 263–8reserve systems 281–2succession in fragmented landscapes 274–6

thistles 102Thousand Islands region 120, 121threats to ecosystems see anthropogenic losses and

threats to ecosystemsthrushes 301tiger snakes 189–90Tilia cordata (lime) 266time-space scales 5Timonius compressicaulis (tree) 133, 142tit species 174–5Tobago 122Tolpis 62Tonga 332topographic characteristics 32tourism 325–8, 329, 339, 340Trachydosaurus rugosus (sleepy lizard) 169–70tracks in the ocean 37trade winds 35, 39tramps 87, 108–10, 130–1, 143

see also supertrampstransition probabilities 113translocation 331–3tree crickets 67tree lettuces 184tree sunflowers 184treecreeper 258trees 135, 298Trinidad 122tripartite model of biogeography 162–3Tristan da Cuhna 42, 50, 52, 182, 305Trochetiopsis (tree) 201, 202Troglodytes troglodytes (wren) 148, 158trophic cascades 270, 322trophic diversification 231trophic level 266–8

true islands 10Truk island 23Tsavo National Park (Kenya) 286tsunamis 43–4, 158Tuamotus 69Turdus merula (blackbird) 319turnover 99–106, 269–74

community assembly and dynamics 144equilibrium 101–2, 103–4and equilibrium model of island biogeography (EMIB)

84–7extinctions, causes of 104–6heterogeneous 103, 104homogeneous 103island theory and conservation 255, 289Krakatau 136–41pseudo- and cryptoturnover 100–1rate 85, 102rescue effect and island area effect on immigration rate

102scale and ecological theory 150–1, 154, 157, 158species numbers 103, 106

types of islands 10–12Tyrannidae 122

UNESCO designation systems 333United States and islands 92, 257, 265, 275, 281, 286, 331

vadose zone 36vagility, high 235Vanuatu 84, 209, 339

community assembly and dynamics 115, 116, 117, 118veil line 82Venezuela 214Vespula (wasp) 299Vestiaria coccinea (i’iwi) 224vicariance/dispersalism debate 4, 52–60Victoria, Lake 240, 293Vini australis (parrot) 332Vipera berus (adder) 189Virgin Islands 189Visnea 62Viverridae 18volcanic eruptions 41–2volcanic islands 3, 13, 32, 42volcanism 22vulcanian eruptions 42Vulcano 42

Wallacea 47, 52–3Wallacean shortfall 290Wallace’s line 52, 54, 55water:

fluxes 268

400 I N D E X

Page 414: Whittaker. Island Biogeography. 2007

lack of 337resources 36–7

wattsi series 216, 217weather:

actual evapotranspiration 97cyclonic storms 246freezes 158hurricanes 39–41, 156–7, 255–6storm vulnerability 338tsunamis 43–4, 158

Weber’s line 52weevils 201, 217West Indies 72, 115, 200

conservation of ecosystems 339–40emergent models of evolution 213, 214species numbers 79, 81

Western Ghats 47Wilkesia 66, 228, 235

wind 268pollination see anemophily

Windward Islands 67woodiness 184–6, 202woodlands 135woolly mammoth 189World Heritage Site 328, 333World Wildlife Fund 47, 48, 284, 340worms 227, 296Wrangel Island 189wren 258Wurmbea 238

Xylocopa darwini (bee) 180

Zapus 119Zosterops japonica (white–eye) 318, 319Zosterops lateralis (silvereye) 170–1, 318

I N D E X 401

Page 415: Whittaker. Island Biogeography. 2007

This page intentionally left blank

Page 416: Whittaker. Island Biogeography. 2007

New Guinea 792 500

Madagascar 587 041

Great Britain 218 100

Java 126 700

Iceland 103 001

Tasmania 68 330

Sardinia 24 090

New Caledonia 16 890

Hawaii 10 400

Cyprus 5 897

Isabela 4 588

Socotra 3 625

Tenerife 2 034

Tahiti 1 042

Guam 541

Easter 166

Amsterdam 61

Norfolk 36

Lord Howe 11

Surtsey 1.3

New Guinea 5 029

Hawaii 4 205

Taiwan 3 997

Tenerife 3 718

Hispaniola 3 175

Réunion 3 069

South Georgia 2 934

Corsica 2 710

La Palma 2 425

Jamaica 2 256

Tristan da Cunha 2 060

Isabela 1 707

Fiji 1 323

Ireland 1 041

Mauritius 826

Easter 510

Rodrigues 396

Bermuda 79

Aldabra 8

Maldives 2

Hawaii 3 650

Amsterdam 3 370

Tristan da Cunha 2 820

Gough 2 580

St. Helena 1 840

New Zealand 1 641

Heard 1 560

Azores 1 356

Macquarie 1 070

Mauritius 858

Cape Verde 568

Madagascar 410

Sao Tomé 235

Cuba 152

Canaries 96

Sumatra 40

Sri Lanka 27

Newfoundland 17

Sakhalin 6

Chiloé 2

Madagascar 180.0 My

New Zealand 70.0 My

New Caledonia 65.0 My

Jamaica c. 50.0 My

Fuerteventura 20.0 My

Iceland 15.0 My

La Gomera 12.0 My

Christmas Island

(Indian Ocean) 10.0 My

Santa María, Azores

8.0 My

Kauai 5.0 My

S. Cristóbal, Galápagos

c. 4.5 My

Ascension 1.5 My

Tahiti 1.2 My

Tristan da Cunha

1.0 My

Hawaii c. 0.5 My

Newfoundland 14 000 y

Tasmania 13 000 y

Barro Colorado 103 y

Anak Krakatau 75 y

Surtsey 40 y

Spitsbergen 77–80°N

Wrangel 71°N

Iceland 63–67°N

Aleutian 52–54°N

Hokkaido 42–46°N

Azores 37–38°N

Canaries 27–29°N

Hispaniola 18–20°N

Socotra 12°N

Sri Lanka 6–10°N

Galápagos 1°S–1°N

Ascension 8°S

Tahiti 18°S

Rodrigues 20°S

Pitcairn 25°S

Lord Howe 32°S

Amsterdam 38°S

Tasmania 42°S

Kerguélen 50–48°S

Macquarie 54°S

Ascension AD 1 815

Tristan da Cunha

AD 1812

Bermuda AD 1 609

Juan Fernández AD 1 574

St. Helena AD 1 501

Azores AD 1 432

New Zealand AD 900

Hawaii AD 800

Tahiti c. AD 700

Easter AD 400

Marquesas AD 300

Maldives 400 BC

Canaries c. 500 BC

Fiji–Samoa 1 500 BC

Majorca c. 2 040 BC

Crete c. 6 250 BC

Corsica c. 6700–8 600 BC

Hispaniola 7000 BC

Cyprus c. 8 320 BC

Tasmania

20 000–30 000 BC

New Guinea

40 000–50 000 BC

Malta 1 202

Maldives 933

Taiwan 611

Mauritius 574

Gran Canaria 550

Puerto Rico 416

Comoros 301

Guam 277

Cape Verde 102

Ireland 66

Fiji 44

Madagascar 19

Easter 16.4

Socotra 12

Tasmania 7.3

Iceland 2.7

Galápagos 1.9

Spitsbergen 0.1

Kerguélen 0.01

Bermuda 30 000

Hawaii 26 030

Great Britain 21 200

Newfoundland 19 360

Taiwan 14 000

Cyprus 13 500

Malta 12 900

Mauritius 10 300

Nauru 10 000

Seychelles 7 000

Réunion 4 300

Sri Lanka 3 800

Jamaica 3 660

Solomon 3 000

Tonga 2 250

Maldives 1 800

Cape Verde 1 360

Sao Tomé 1 000

Madagascar 730

Comoros 685

Data for selected islands

Area Altitude Isolation Age Latitude Human Population Income (km2) (m) (km) (y) settlement density per capita

(date) (inhabitants/km2) (US$)

Data drawn from the National Geographical Society’s National GeographicAtlas of the World, 7th edn, 1999 and numerous other sources.