Marcelo Hernán Cassini from Individual Habitat Use to Species

Home , Ideal free distribution

Marcelo Hernán Cassini Distribution Ecology From Individual Habitat Use to Species Biogeographical Range Distribution Ecology

Marcelo Hernán Cassini

Distribution Ecology

From Individual Habitat Use to Species Biogeographical Range Marcelo Hernán Cassini National Scientiﬁ c and Technical Research Council & Luján University Argentina

ISBN 978-1-4614-6414-3 ISBN 978-1-4614-6415-0 (eBook) DOI 10.1007/978-1-4614-6415-0 Springer New York Heidelberg Dordrecht London

Library of Congress Control Number: 2013931179

© Springer Science+Business Media New York 2013 This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is concerned, specifi cally the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfi lms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. Exempted from this legal reservation are brief excerpts in connection with reviews or scholarly analysis or material supplied speci fi cally for the purpose of being entered and executed on a computer system, for exclusive use by the purchaser of the work. Duplication of this publication or parts thereof is permitted only under the provisions of the Copyright Law of the Publisher’s location, in its current version, and permission for use must always be obtained from Springer. Permissions for use may be obtained through RightsLink at the Copyright Clearance Center. Violations are liable to prosecution under the respective Copyright Law. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a speci fi c statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. While the advice and information in this book are believed to be true and accurate at the date of publication, neither the authors nor the editors nor the publisher can accept any legal responsibility for any errors or omissions that may be made. The publisher makes no warranty, express or implied, with respect to the material contained herein.

Printed on acid-free paper

Springer is part of Springer Science+Business Media (www.springer.com) To my father

Preface

A well-accepted de fi nition of Ecology is that of Charles Krebs (1985), who said that it represents the scienti fi c study of the interactions that determine the distribution and abundance of organisms. During a number of decades, this discipline has emphasised the research on the second of these two components, investigating the process that explains changes of populations and communities over time, with less consideration of the spatial context of these changes. In the 1970s, it was possible to recognise at least two new directions in ecological research: an emphasis on experimentation as a way to inject greater explanatory power to the discipline and a decision to make more explicit consideration of the spatial context in which ecological processes take place. This book is a compendium of the modelling products that result from the second of these conceptual reforms. This book combines a set of approaches to enable the analysis of the distribution of organisms between environmental sites, which structure is based on levels of organisation. There are chapters for the distribution of individuals, aggregations, societies, subpopulations, populations, and species. Two chapters analyse the distribution at additional organisational levels: communities and genes. The analysis of the distribution of genes crosses the boundaries of the fi eld of ecology, but their inclusion in this book illustrates how ecology distribution methods can be applied to other areas of knowledge. Two further chapters deal with the applications of distribution ecology to biodiversity conservation and animal production. The fi nal chapter is for synthesis and conclusions. This tiered organisation respects the way biology in general and ecology in particular have developed historically. The book aims to take an overarching view of all of the branches of ecology that embrace a spatial approach. Given that the scope of this book is very broad, it is not specialised in any particular ecological discipline or taxon. This book is intended for advanced undergraduate and graduate students. Researchers of ecological science should not seek in this book to get deeper in their areas of expertise, but to fi nd an occasion to frame their knowledge in a wide spatial-oriented approach. It can be also useful for researchers in other disciplines such

vii viii Preface as population genetics, epidemiology, conservation biology, wildlife management, and range management. Daniela Centrón García and Juan Ignacio Túnez are main authors of Chap. 9 . J.I. Túnez carefully commented on the other chapters of the manuscript. I acknowledge the Consejo Nacional de Investigaciones Cientíﬁ cas y Técnicas and the Universidad Nacional de Luján (Argentina) for their institutional support. The idea of this book emerged 10 years ago during the internal seminars of the GEMA group, for what I acknowledge to all the attendance to these seminars. My thanks also go to Melissa Higgs and the rest of the staff of Springer who help me through production. Finally, my greatest debt goes to all my family: Daniela for her unceasing and loving support; Francisco and Jerónimo for being there; Sofía, who kindly designed ﬁ gures and organised references; María; my parents Hernán and María Rosa; and my brother Fernando.

Buenos Aires, Argentina Marcelo Hernán Cassini Contents

Part I Introduction

1 Concepts and Deﬁ nitions ...... 3 1.1 Introduction ...... 3 1.2 Scales ...... 4 1.3 Levels ...... 8 1.4 Habitat Selection ...... 9 1.5 Movement, Dispersal, and Migration ...... 11

Part II Levels within Species

2 Distribution of Individuals ...... 17 2.1 Introduction ...... 17 2.2 Movements ...... 17 2.2.1 Ethology of Movement ...... 18 2.2.2 Cognitive Aspects of Movement ...... 20 2.2.3 Movement as an Adaptation...... 22 2.2.4 Statistical Description of Movement ...... 24 2.3 Home Ranges ...... 25 2.4 Site Suitability Models ...... 27 2.5 Patch Models of Foraging Theory ...... 30 2.5.1 Introduction: Foraging Theory and Foraging Behaviour ...... 30 2.5.2 Types of Patches ...... 31 2.5.3 Marginal Value Theorem ...... 31 2.5.4 Patch Selection ...... 35 2.5.5 Patch and Prey Selection ...... 37 2.6 Foraging in Heterogeneous Habitats by Plants ...... 38 2.7 Balancing Demands and Use of Refuges ...... 39 2.8 Distribution at Equilibrium: The Matching Rule ...... 41

ix x Contents

3 Distribution of Aggregations ...... 43 3.1 Introduction ...... 43 3.2 Site Suitability Models ...... 44 3.3 Individual-Based Models ...... 45 3.4 Ideal Free Distribution ...... 48 3.4.1 Habitat Selection and Intraspeciﬁ c Competition ...... 48 3.4.2 Fretwell and Lucas’ Model ...... 49 3.4.3 Ideal Free Distribution with Interference ...... 52 3.4.4 Ideal Free Distribution with Unequal Competitors ...... 52 3.4.5 Other Ideal Free Models ...... 53 3.5 Distribution at Equilibrium: The Habitat-Matching Rule ...... 54 4 Distribution of Societies ...... 57 4.1 Introduction ...... 57 4.2 Dispersion Patterns ...... 58 4.3 Grouping ...... 59 4.3.1 Protection Against Predators ...... 60 4.3.2 Finding Resources ...... 62 4.3.3 Changing Local Environment ...... 63 4.4 Defending Territories ...... 63 4.5 Habitat Use and Reproductive Systems ...... 64 4.6 Distribution of Groups: Ideal Free Approach with Allée Effect ...... 67 4.7 Distribution of Territories: Ideal Despotic Distribution ...... 71 4.8 Mating Systems and the Ideal Free Approach ...... 72 4.8.1 Distribution of Reproductive Birds: The Polygyny-Threshold Model ...... 72 4.8.2 Distribution of Mammals: The Effect of Sexual Harassment ...... 73 5 Distribution of Subpopulations ...... 77 5.1 Introduction ...... 77 5.2 Metapopulations ...... 78 5.3 Source–Sinks ...... 81 5.4 Fitness Maximisers, Foraging Behaviour, and Landscape Distribution ...... 83 5.5 Individual-Based Models and Spatially Structured Populations ...... 86 5.6 Pattern-Oriented Approach of Landscape Ecology ...... 87 5.7 Cognitive Constraints on Landscape Utilisation ...... 90 6 Distribution of Populations ...... 93 6.1 Introduction ...... 93 6.2 Distribution of Close Populations ...... 95 Contents xi

6.3 Distribution of Spatially Structured Populations ...... 97 6.4 Role of Behavioural Mechanisms in Distribution of Populations ...... 98 7 Distribution of Species ...... 101 7.1 Introduction ...... 101 7.2 Species Distribution Models ...... 102 7.3 Internal Structure of Biogeographical Ranges ...... 108 7.4 Individuals, Physiological Mechanisms, and Species Distribution ...... 110 7.5 Behavioural Adaptations and Species Distribution ...... 112

Part III Levels Outside Species

8 Distribution of Species Assemblages ...... 117 8.1 Introduction ...... 117 8.2 Isoleg Model: Distribution of Guilds Composed of Fitness Maximisers ...... 118 8.3 Macroecology ...... 121 8.4 Species Distribution Models Applied to Species Ensembles...... 124 9 Distribution of Genes ...... 127 9.1 Introduction ...... 127 9.2 Biological and Environmental Features Leading to Population Genetic Structure ...... 128 9.3 Geographical Genetic Patterns Observed in Eukaryotes ...... 129 9.3.1 Random Patterns ...... 130 9.3.2 Clines ...... 130 9.3.3 Ecotypes ...... 131 9.3.4 Isolation by Distance ...... 133 9.3.5 Stepping Stone ...... 133 9.3.6 Discontinuities to Gene Flow ...... 135 9.3.7 Metapopulations ...... 136 9.3.8 Source–Sinks ...... 137 9.3.9 The Central–Marginal Model ...... 137 9.4 Distribution of Genes in Bacterial Communities ...... 138 9.4.1 Mechanisms of Lateral Genetic Transfer in the Bacterial World ...... 138 9.4.2 Role of Class 1 Integron Genes in Antibiotic Resistance...... 138 9.4.3 Pattern of Distribution of Class 1 Integron Genes at Different Ecological Scales ...... 140 xii Contents

Part IV Applications

10 Distribution Ecology in Conservation Biology ...... 145 10.1 Introduction ...... 145 10.2 Environmental Challenges ...... 145 10.3 Individual (and Gene) Distribution and Conservation Biology ...... 147 10.4 Aggregation Distributions and Conservation Biology ...... 149 10.5 Metapopulations, Source–Sinks, and Conservation Biology ...... 156 10.6 Landscape Ecology, Pattern-Based Models, and Conservation Biology ...... 156 10.7 Species Distribution Models and Conservation Biology ...... 158 10.8 Ensemble Distributions and Conservation Biology ...... 160 11 Distribution Ecology in Animal Production ...... 163 11.1 Introduction ...... 163 11.2 Foraging Behaviour and Distribution of Livestock ...... 163 11.3 Landscape Predictors of Livestock Distribution ...... 166 11.4 Quantitative Analyses of Livestock Distribution ...... 167 11.5 Species Distribution Models Applied to Livestock Management ...... 168 11.6 Species Assemblage Distribution and Livestock Management ...... 172

Part V Conclusions and Prospects

12 Final Remarks ...... 177 12.1 Introduction ...... 177 12.2 Criticisms of Models ...... 178 12.3 Molecular Versus Molar Models and Processes Versus Associations ...... 179 12.4 Amount of Resources Versus Patch Arrangement ...... 180 12.5 Organisation Levels and Ecological Scales Revisited ...... 181 12.6 Equilibrium and the Generalised Matching Law ...... 182 12.7 Holistic or Reductionist Future of Distribution Ecology ...... 186

References ...... 189

Index ...... 211 Part I Introduction Chapter 1 Concepts and Deﬁ nitions

1.1 Introduction

This chapter has two objectives. In this introductory section, distribution ecology is defi ned and the structure of the book is outlined. In the remaining sections, some terms that will be frequently used in the course of the book are defi ned. Which interactions determine that organisms of a particular species distribute heterogeneously in space? This question is what attempts to answer distribution ecology. In this book, the term distribution refers to differences between places either in individual residence times, population abundances, species occurrences, or probabilities of use, depending on ecological scales and levels of organisation. This application of the term differs from the implicit or explicit sense of traditional ecology, which frequently confi nes the distribution problem to the question of presence versus absence of species. In this book, the concepts of distribution and abundance are mixed. Thus, the comparisons of numbers of organisms between places can provide information of determinants of abundance within a site. The term place also has a broad meaning, as it includes the heterogeneity in the distribution of the environmental characteristics at different ecological scales, including foraging patches, habitats, landscapes, regions, and biogeographical zones. There is consensus among ecologists to accept that much of the history of the discipline has developed without a special concern for the spatial dimension of ecological problems. The typical example is that of population dynamics models. These models are designed to predict changes over time in clusters of organisms that are primarily defi ned by their intrinsic properties, with a vague consideration for the site in which they fi nd themselves. Another feature of ecology before the 1980s was the emphasis on internal processes, especially the phenomena of intra- and interspecifi c competition. Since then, a signifi cant change has occurred in the importance given to the spatial dimension of ecological problems. New theories have appeared, like those applied to meta-populations, foraging, habitat selection, island biogeography, and new disciplines, such as landscape ecology, macroecology, movement ecology, aerography, and geoecology. These theories and disciplines appeared with

M.H. Cassini, Distribution Ecology: From Individual Habitat Use 3 to Species Biogeographical Range, DOI 10.1007/978-1-4614-6415-0_1, © Springer Science+Business Media New York 2013 4 1 Concepts and Defi nitions a completely new approach regarding the role of the environment in modulating the distribution of the abundances of organisms. This ‘spatial revolution’ in ecology is now more than 30 years old and is still a growing process. This historic change in ecology seems to have accompanied a phenomenon of awareness of the regional and global environmental problems caused by humans. Ecology suffered a crisis when it was faced with the need to produce scientifi c models to explain the ecological effects of human degradation of nature and to predict the consequences of these impacts at the level of nations and regions. As the scale of anthropogenic impact increased, the traditional models of ecology seemed to become less effi cient, making ecology produce new theoretical solutions. One example is the strong link between the development of the meta-population theory and the increasingly frequent occurrence of processes of anthropogenic habitat fragmentation. Another suggestion is that the search for new paradigms emerged from the recognition that there are important environmental variables that have a chaotic behaviour and, therefore, conventional models were unable to predict changes in abundance over time. Whatever the historical reason, the fact is that ecology suffered a spatial revolution that still continues today. This book has four other parts. The next one introduces the different models that have been used to answer the question of why organisms of a species use certain sites more than others. The different approaches will be divided according to levels of organisation, because that was the implicit form in which they developed historically: distribution of individuals, groups, aggregations, subpopulations, populations, and species. The third part expands spatial analysis to two additional levels of organisation: species assemblages at one end and genes at the other. While distribution ecology applies primarily to individual species, this third section describes spatial analyses in the fi eld of community ecology, macroecology, and landscape genetics. The fourth part is devoted to the application of the ecology of distributions to biodiversity conservation and animal production. The last part is intended to attempt a synthesis. Because the book covers a variety of topics, it was impossible to go into the details of all topics. The greatest value of this review is that it brings together in one text all of the most important theories and models, using common terminology and a hierarchical organisation.

1.2 Scales

Scale is the physical dimension of entities or phenomena (O’Neill and King 1998 ). Although several types of scales exist, the cartographic scale has the most precise deﬁ nition, which is the ratio between the distance on a map and the distance on the ground, which is normally expressed as 1:100,000, for example. When the ratio is small (small cartographic scale), the map covers a large real surface. For example, a map of the entire planet has an approximate cartographic scale of 1:40,000,000 m, i.e. a very small scale. When the term scale is used, dimensions and units of 1.2 Scales 5

Fig. 1.1 Ecological scales. Example for a coastal species (e.g. otters) living in the Beagle Channel, Argentina, South America

measurement should be assigned or identifi ed. It is possible to talk about landscape scale as it has an implicit reference to large spatial scales of the order of kilometres. However, it is not possible to use the term population scale, because it is impossible to assign a dimension to a population. This book uses an operational defi nition of ecological scale as a level of organisation of the environment in which we can distinguish heterogeneities in biotic and abiotic conditions of the Earth’s surface. As opposed to the cartographic scale, a large ecological scale corresponds to a large geographic area. Five ecological scales or spatial units are considered: patches, habitats, landscapes, ecoregions, and biogeographical regions or climatic zones (Fig. 1.1 ). They are not strictly defi ned by exact measurements, but they provide an approach of the resolution at which an ecological pattern should be measured or represented. This is a categorisation of many available in the literature and can also distinguish several subcategories depending on the species considered. Site , spatial unit, and location are general terms that will be used when referring to any spatial scale. Two terms commonly used by ecologists to refer to scales of observation are grain and extent. Grain refers to the smallest temporal or spatial resolvable unit of study. On a digital map, a grain is one cell or pixel. Extent is the total area or the length of time over which observations of a particular grain are made. 6 1 Concepts and Defi nitions

Wiens ( 1976 ) defi nes patches as a surface area differing from its surroundings in nature and appearance. Within this defi nition, patchiness exists along different scales, including heterogeneities within an individual’s home range, habitat patches containing subpopulations of a meta-population, and a highly isolated island of habitat containing an independent population. This book uses the term patch (without any adjective), only in the case of the smallest of scales. Thus, the patch may be considered the smallest unit of environmental heterogeneity. In general, it refers to a site with a single resource type, usually food, which can be depleted by the activity of the individual who uses it. When the term patch is used for large scales, it will always contain an adjective that refers to the scale, e.g. habitat patch. A resource can be defi ned as a substance (water), object (shelter, potential mate), or energy source (food) required for normal body maintenance, growth, and reproduction (Ricklefs 1979 ). Sometimes it is called a microhabitat . Examples of patches are a tree with fruits for frugivorous vertebrates, a sheet for a phytophagous insect, a herd of antelope for a large feline, a patch of sunlight for a thermoregulating insect, or a coastal rookery that provides shelter for marine vertebrates. The maximum size of a patch would be less than ten hectares for a medium vertebrate. The term habitat is ambiguous and its use depends on the author and the particu- larities of the study. This term typically refers to a subset of physical and biotic conditions that can be spatially delimited from other adjacent subsets. In functional terms, this subset of conditions represents a certain combination of resources needed by a focal species, so that an individual or a population belonging to that species will respond differently to each habitat type. Overall, the same habitat includes various types of resources so that an individual could remain within the habitat for its entire life. This differs from a patch, which is smaller than a habitat and is usually defi ned in terms of a single resource. An individual may have its home range in one habitat. A habitat can also be represented as a set of patches in a matrix of unused resources. For animals, the habitat is often defi ned or delimited by a plant community. A typical example would be a forest as the terrestrial habitat, with trees as patches. The dimension of a habitat ranges from tens to hundreds of hectares. The concept of ‘landscape’ is consistent with the intuitive or dictionary defi nition. It has to do with a perspective or scenario that can be seen from a high view- point. It covers a mosaic of habitats and their size varies from hundreds to thousands of hectares. The landscape scale is usually studied from satellite images and maps. Landscapes are composed of habitats, which are arranged in two basic structures: mosaic-like or matrix–patch structures (Fig. 1.2 ). In the jargon of landscape ecology, habitats are just called patches, but they will be referred to here as habitat patches or elements. An element is a recognisable area on a landscape that contrasts with adjacent areas and has defi nable boundaries (Kotliar and Wiens 1990 ). The original notion of landscape was applied primarily to continental environments at a scale of tens or hundreds of kilometres, since this is where heterogeneities can be recognised at fi rst glance. The elements of a typical landscape include meadows, forests, crops, urbanisations, streams, and lagoons. However, landscape ecology is applicable to any type of environment, including oceans, and to a wide range of scales, depending mainly on the processes relevant to the species under study. 1.2 Scales 7

Fig. 1.2 Types of landscapes

Landscape elements can be formed by human and biotic and abiotic agents. The presence, form, and metabolism of plants often create patches. For example, a group of trees differs from the surrounded by grassland in soil organic matter, nutrient content, water holding capacity, and other important features. Animals can also operate on landscape elements. Typical examples are burrowing animals and beavers that increase landscape heterogeneity. The ecoregion is the next ecological scale. It can be defi ned as a large portion of geographically defi ned territory in which certain geomorphologic and climatic conditions dominate and determine physiognomic vegetation characteristics, with typical dominant species of plants and animals. Ecological regions can be identifi ed through the analysis of the patterns and the composition of biotic and abiotic phenomena that affect or refl ect differences in ecosystem quality and integrity (Omernik 1987 ). These phenomena include geology, physiography, vegetation, climate, soils, land use, wildlife, and hydrology. The relative importance of each characteristic varies from one ecological region to another, regardless of the hierarchical level. In 1999–2000, the World Wildlife Fund conducted an extensive review of the ecoregions of the world (http://wwf.panda.org/about_our_earth/ecoregions). They defi ned a total of 238 ecoregions: 142 terrestrial, 53 freshwater, and 43 marine. The WWF Global Ecoregions are the results of regional analyses of biodiversity across the continents and oceans of the world, completed in collaboration with hundreds of regional experts worldwide and by conducting extensive literature reviews. They distinguished 26 major ecotypes that describe different areas of the world that share similar environmental conditions, habitat structures, and patterns of biological complexity and that contain similar communities and species adaptations. They classifi ed ecoregions within the seven biogeographical realms or regions based on these ecotypes. 8 1 Concepts and Defi nitions

The following and last scale implies dividing the Earth into its largest portions. In the nineteenth century, six biogeographical regions in the world were recognised, using global faunal distributions: Nearctic, Neotropical, Palearctic, Ethiopian, Oriental, and Australian; although occasionally with different names and minor amendments, this nomenclature survives today (Cox and Moore 2005 ). This clas- siﬁ cation corresponds only to terrestrial ecosystems. At the largest scale, the Earth can also be divided into fringes corresponding to climatic zones: tropical, subtropical, temperate, sub-polar, and polar.

1.3 Levels

Scales are more related to patterns, while levels are more related to processes. There are levels of organisation in a hierarchical system. Levels are the study object of hierarchy theory, which investigates fl ow, interaction, and rates within and between levels (O’Neill et al. 1989 ). Within a hierarchy, higher levels have slower rates and react more slowly. Levels are relative characterisations of the organisation of a system. In summary, a level of organisation can be associated with a certain spatial scale, but they are not the same concept. Hierarchy theory is part of the conceptual foundation of landscape ecology and is applied to the principles of system organisation (O’Neill and King 1998 ). It is pertinent to any system with a medium number of component parts, too many to be described as a small-number system and too few to be described by the stable statistical properties of a large-number system (King 1997 ). Provided the environment is organised into scales that have these properties, then the hierarchical theory is appropriate for the study of relationships between scales. The biology in general is often explained in levels (Curtis et al. 2008 ), as well as ecology (Begon et al. 2006 ). While these divisions have been criticised (O’Neill and King 1998 ), the fact remains that the history of ecology is organised around these divisions. Theories included within the ecology of distributions, as the unit of analysis, have some level of organisation. Therefore, the chapters are organised according to the following levels of organisation, or units that distribute themselves: individuals, aggregations, societies, subpopulations, populations, and species (Fig. 1.3 ). This classifi cation is mainly operational; however, it is possible to characterise each level by its own internal processes that, in accordance with different theories, governs how organisms distribute in space. Individuals follow behavioural decision rules that are designed by natural selection to maximise resource acquisition. Most approaches agree that habitat selection is the main process controlling the distribution of aggregates. The problem lies in the defi nition of habitat selection (as it will be discussed in the following section), although there is general agreement that interference in resource acquisition due to intraspecifi c competition is a main factor affecting habitat selection and, therefore, aggregation distribution. The main internal processes that are incorporated at the group level are associated with the costs 1.4 Habitat Selection 9

Fig. 1.3 Levels of organisation and some of the models and theories that are described in this book

and beneﬁ ts of living with conspeciﬁ cs. Subspecies distribution within a population is modulated by processes of dispersal, extinction, and colonisation. Groups of populations reach equilibrium densities at different sites that are determined by birth, death, immigration, and emigration rates. Species are distributed in accordance to their environmental niche.

1.4 Habitat Selection

Habitat selection is the behavioural mechanism that determines where an organism goes and stays. Habitat selection is an individual process and a behavioural phenomenon. Its product or result is individual habitat use, which can be measured as the relative probability of utilisation or the amount of time that an individual spends in a habitat. As with any behaviour, habitat selection may be studied in terms of proximate or ultimate causes (Tinbergen 1963 ). Habitat selection is the sum or result of several spatial behaviours that are organised in a hierarchy (Fig. 1.4 ). 10 1 Concepts and Deﬁ nitions

Fig. 1.4 Habitat structure and decision process during an activity cycle

Most animals show daily rhythms of activity, being more active in a particular part of the day. Within this active period, animals usually alternate foraging bouts with resting periods, usually in a shelter. If more than one type of habitat is inside an individual habitat, habitat selection is usually performed when starting one of these search bouts. Once in the habitat, the individual must decide upon the searching trajectory, whether to use a patch when it fi nds it, the searching strategy to use once in the patch, whether to use a resource once it is encountered, and fi nally when to leave the patch. Foraging theory analyses each of these decisions or choices between behavioural alternatives with models based on the optimisation principle, which uses a surrogate of fi tness maximisation as a currency, such as maximising energy intake rate or minimising searching time (Sect. 2.4 ). Maximising resource energetic payoff is not the only selective pressure that suffers habitat selection: individual behaviour is affected by predation risk (Sect. 2.7 ), intraspecifi c competition (Sect. 3.4 ), and other biotic and abiotic factors, such as diseases or climate (Sect. 7.4 ). The result of all these behaviours that determine the movement and length of stay in different sectors of space is what determines the habitat selection made by an individual. Site suitability models (Sects. 2.4 and 3.2 ) analyse this output of the habitat selection process. The amount of use of a type of habitat by an individual is 1.5 Movement, Dispersal, and Migration 11 measured and then compared with the amount of habitat available, using an index or a probability function (Manly et al. 1992 ). This type of analysis can be classifi ed as a molar kind, while the analysis made by foraging theory should be defi ned as a molecular type (Sects. 2.5 and 12.3 ). There is a large amount of confusion around this term, especially after the development of the resource selection functions (Sect. 2.4 ). Habitat selection is an individual process that may be affected by intraspecifi c competition but it is not a population attribute. Therefore, it is incorrect to estimate habitat selection using a pool of individuals, and the term resource selection function should not be used in these cases. A clear example of this mistake is provided in Sect. 3.4 , which describes a model that predicts how organisms’ abundances can be distributed in perfect proportion to habitat availability, i.e. lack of habitat selection in accordance to a ‘population’ use of the term, when all individuals are selecting the habitat to stay in a way that maximises their individual payoff. Habitat selection implies the choice of an alternative among different behavioural options available to solve an environmental problem. An individual is said to be selective when it systematically uses one or a few options, while it is said to be opportunistic when it uses a wide range of options. It is important to distinguish the selective–opportunistic dichotomy from the specialist–generalist alternative. The former is a plastic character related to decision-making, while the latter are fi xed strategies that are determined by the phylogenetic history of the species. This distinction becomes critical when it is applied to habitat selection in plants. Plants have several mechanisms that improve their ability to reach good habitats despite their immobility, especially by means of propagules. However, most responses of plants to habitat heterogeneity expressed evolutionary adjustment (adaptations or constraints), rather than decision processes associated with the alternatives available for an individual. As Bazzaz (1991 ) said, it is the habitat that makes the choice rather than the plant. However, there are some cases that will be discussed in Sect. 2.6 , in which plants appear to select resources in patches based on payoff maximisation rules. Habitat or patch selection in plants occurs by means of differential growth between substrates and especially by actual movement by clonal spread and fragmentation.

1.5 Movement, Dispersal, and Migration

Movement , dispersal , and migration are all terms that are related to spatial dynamics and changes in loc.tions that determine the distribution patterns. As for other terms in ecology, there are no universal deﬁ nitions for them. Movement is mainly applied to individuals. Dispersal applies to individuals but is also a population property. Migration involves group changes of location and involves a large proportion of a population. Movements can be in any direction, dispersal means a spreading of individuals away from others, while migration is a mass directional movement that necessarily implies changing habitats. Plants do not move or migrate; they only disperse. 12 1 Concepts and Deﬁ nitions

Fig. 1.5 A conceptual framework to investigate individual variation in dispersal, proposed by Clobert et al. ( 2009 ). Dotted lines represent unexpected feedbacks between dispersal stages caused by transfers of information through individual movements (modiﬁ ed from Clobert et al. 2009 )

Dispersal can be active or passive; the latter depends on active animals, wind, gravity, or current. Passive dispersal is generally density independent, while active dispersal is normally density dependent. Another category is between close- and long-distance dispersal; the latter is a rare phenomenon both in frequency of occurrence and in the number of individuals that are involved. In terms of frequency of occurrence, movements occur often, many times per day. Dispersal occurs normally once in life, and there are two common types: natal and breeding dispersal. Migration is normally cyclical and occurs seasonally or annually, although diurnal changes of habitats made with tide that involve mass movements should also be considered as migration. Dispersal is diffi cult to study; however, a new impetus has been given to its research, mainly because it became a key process for most of the new approaches of population ecology and landscape ecology (described in Chap. 5 ). The importance of dispersal and migration is becoming increasingly apparent as more populations 1.5 Movement, Dispersal, and Migration 13 face the major threats posed by global climate change and the fragmentation of their habitat. The development of new techniques has favoured research on these two processes. Global Positional System (usually named with the abbreviation GPS ) devices, radio-trackers, bio-loggers, and other devices for following and locating individuals in the fi eld have facilitated this research. Molecular techniques such as polymerase chain reaction (usually named with the abbreviation c) and the use of multiple genetic markers have allowed the recording of dispersal directions and the identifi cation of individual dispersers. Clobert et al. (2009 ) recently proposed a conceptual framework to investigate individual variation in dispersal (Fig. 1.5 ). Individual decisions to leave a population and settle in a new habitat patch rely on a set of external cues and on internal factors that increase dispersal propensity. They proposed three main processes shaping individual variation in dispersal: fi rst, phenotypic differences between residents and dispersers should depend on external factors that cause dispersal (Fig. 1.5 ); second, individuals may vary in their sensitivity to conditions encountered during transience and at settlement, given their phenotype and dispersal motivation; and third, a transfer of information through individual movements across the landscape might cause unexpected feedbacks between dispersal stages (Fig. 1.5 ) . Part II Levels within Species Chapter 2 Distribution of Individuals

2.1 Introduction

The distribution of individuals is a field mainly studied by behavioural sciences. It is largely concerned with animals, although some models can be applied to plants. Behavioural ecologists study decision rules that follow non-sessile individuals on when to start a movement; speed and direction of a movement; sites at which to stop, forage, be vigilant, or rest; and time spent for each activity. The result of these behaviours is an individual pattern of habitat use. Measured along the life of an animal, this pattern demarcates a home range. This chapter begins with a description of movement strategies in animals as mechanisms that result in an individual pattern of distribution. It is followed by two sections that discuss two types of descriptive models of individual habitat use: home-range estima- tions and site suitability models. The analytical approach is presented in the following two sections, and it is based on foraging theory, which developed models to predict residence times in foraging areas and to resolve the trade-off between forage and defence against predators. The chapter concludes with an analysis of the expected distributions at equilibrium.

2.2 Movements

Movement is a change in the spatial location of a whole organism in time (Nathan et al. 2010). Most living creatures move to a greater or lesser degree. Sessile species such as plants have a stage of development where they can move. The movement may be passive or active. Most species without a nervous system passively move due to the direct effect of external factors such as wind or pollinators for pollen and seeds, or a single active response to environmental conditions, as is the case with micro-organisms.

M.H. Cassini, Distribution Ecology: From Individual Habitat Use 17 to Species Biogeographical Range, DOI 10.1007/978-1-4614-6415-0_2, © Springer Science+Business Media New York 2013 18 2 Distribution of Individuals

The study of movement has a long tradition in the field of animal behaviour research. It was studied in the first half of the twentieth century, when it was argued that animals moved based on simple orientation mechanisms triggered in response to environmental stimuli: tropisms, taxes, and kineses (Gould 1982). During the second half of the twentieth century, the study of movement followed three well-defined lines: (1) neuro-ethology, especially of invertebrates, focused on the underlying sensory and neural mechanisms; (2) cognitive psychology focused on the information-processing mechanisms that underlie navigation; and (3) behavioural ecology focused on the adaptive importance of different movement strategies. At the beginning of the twenty-first century the study of movement was revolu- tionised, and three simultaneous phenomena were promoted: the new emphasis on spatial dimension as a global concern of ecology, the development of new tech- nologies to track animals accurately anywhere on the planet, and the progress of new methods of statistical analysis and new computational tools for processing movement patterns. This new approach defines itself as movement ecology and has a clear emphasis on the statistical description of the movement, the relationship between external conditions and the spatial response, and the effects on distribution patterns (Getz and Saltz 2008; Holyoak et al. 2008; Nathan 2008; Nathan et al. 2008; Giuggioli and Bartumeus 2010). The capacity to collect high-resolution spatio-temporal movement data was due to recent technological advances encompassing numerous techniques, such as miniaturised radio transmitters, global positioning systems, cellular and satellite networks, acoustic transmitters, global positioning systems, and light-level geo-locators (Viswanathan et al. 1999; Bartumeus et al. 2005; Bartumeus and Levin 2008; Smouse et al. 2010; Urbano et al. 2010). In summary, there are four paradigms in the study of movement. Some have shown a successful integration, as is the case of studies of information processing in food searching (Stephens et al. 2007; Dukas and Ratcliffe 2009). Others have been developed in relative isolation. For example, movement ecology has little connection with previous studies in other disciplines, despite the call for greater integration, probably due to its recent development (Nathan et al. 2008; Giuggioli and Bartumeus 2010).

2.2.1 Ethology of Movement

Insects show adaptations to follow cues that are specific for a type of patch or prey. Olfactory and visual cues are common mechanisms. A well-known example is that of male moths, which search for suitable female mates by sensing sex pheromones in the air. The species-specific odour is blown downwind, and males search for these trails by casting back and forth across the wind. When they detect the olfactory sign stimulus, the males fly upwind to the source (Gould 1982). Classical experiments of Nobel Prize winner Niko Tinbergen showed how a digger wasp finds its tunnel home. 2.2 Movements 19

Fig. 2.1 Classical Tinbergen’s experiment on insect cognition (modified from Tinbergen 1963)

He hypothesised that once back in the vicinity, wasps used local landmarks to guide them to their nests. To test this idea, he placed rings of pine cones around the nests, waited until each wasp flew out on a hunting trip, and then moved the circles a metre or so away (Fig. 2.1). Invariably, the wasps would return to the centre of the pine cone ring and search for their nests (Tinbergen 1963). Searching for distant targets can be made when environmental cues are either available or not. Many terrestrial predators use regular roads when searching for prey patches in their territories. Studies using signs or camera traps, for example, show that felines and canines tend to use trails and roads built by people. The other famous case is flower recognition by bees. Another Nobel Prize winner Karl von Frisch showed experimentally that the navigation system of bees is based on the sun-centred pattern of polarised ultraviolet light in the sky. Adult white butterfliesPieris rapae follow natural and man-made topographical structures as they range from one area to another (Baker 1978). These animals follow roads, hedges, and lines of telephone poles, and they orient towards distant objects. The butterflies use different cues depending upon wind conditions and probably other abiotic factors. 20 2 Distribution of Individuals

Recently, Lohmann et al. (2008) proposed a new hypothesis to explain long-distance natal homing in salmon and sea turtles, called geomagnetic imprinting. Several marine vertebrates move across vast distances of ocean before returning as adults to their natal areas to reproduce. Salmon are known to use chemical cues to identify their home rivers at the end of spawning migrations, but this is not a possible explanation of long-distance orientation. Both salmon and sea turtles detect the magnetic field of the Earth and use it as a directional cue. Turtles also derive positional information from two magnetic elements (inclination angle and intensity) that vary predictably across the globe. Lohmann et al. (2008) argued that salmon and sea turtles imprint on the magnetic field of their natal areas and later use this information to direct natal homing. The simplest variant of this hypothesis involves imprinting on a single element of the field (e.g. inclination angle or intensity), and to locate the area later in life; the animal would need only to find the coastline and then swim north or south along it to reach the target region. Even though Lohmann et al. (2008) showed that some field data support their hypothesis, they did not provide a possible proximal (neurophysiological or cognitive) mechanism that might support it.

2.2.2 Cognitive Aspects of Movement

Non-sessile animals move within their home ranges and occasionally out of them. Most controlled movements that involve changing position can be defined asgoal orientation, or movement towards a different place where the animal can find better conditions. Dyer (1998) classified goal orientation movements based on scales, which in turn related to previous experience with the goal, in three types: (1) heading for a nearby object in plain view, (2) heading for an unseen patch of resources within a familiar home range, and (3) heading for home from a distant, unfamiliar location. At all scales, goal orientation movement necessitates that the animal obtain two general sorts of information about the environment. First, the animal must be able to discriminate between different directions (body orientations) relative to some external reference (a celestial body, a landmark, or a chemical cue). Second, the animal must be able to determine its position in space relative to its goal. The animal can use a simple measure of location or it may be able to measure angular position and distance in relation to some reference (Dyer 1998). To characterise the specific cognitive challenge faced by an animal that is orient- ing to a goal, Dyer (1998) identified three parameters that affect in some way the procurement of positional and directional information and therefore the goal: (1) the geometry of the space containing the goal, (2) the distance of the goal from the starting point, and (3) the variability of the environmental cues that provide navigational information. Cognitive maps are mental representations of space, independent of immediate contingencies (Tolman 1948). An animal with a cognitive map should not need to 2.2 Movements 21

Fig. 2.2 Mazes used in classical Tolman’s experiment with rats, testing spatial memory. Meaning of letters in text (modified from Tolman 1948) follow gradients or landmarks and should take novel routes across unknown terrain. According to Tolman (1948), a feature of a cognitive map is the ability to make novel short-cuts between two points. He conducted an experiment to test this idea. In the first part, he used the maze represented in Fig. 2.2a. The animals ran from A across the open circular table through CD (which had alley walls) and finally toG , the food box. H was a light which shone directly down the path from G to F. After four nights, three trials per night, in which the rats learned to run directly and without hesitation from A to G, the apparatus was changed to the maze shown in Fig. 2.2b. The starting path remained the same but a series of radiating paths was added. The animals again started at A and ran across the circular table into the alley and found themselves blocked. They then returned and began exploring practically all the radiating paths. Rats showed a significant tendency to choose the patch that ran the closest to a point where the entrance to the food box had been. Some authors are sceptic about the conclusions of this and similar experiments and suggest that no animal has been conclusively shown to have a cognitive map (e.g. Bennett 1996). Recently, Jacobs and Schenk (2003) proposed that a cognitive map is composed of two sub-maps, one derived from directional cues and the other derived from positional cues. Food hoarding has been one of the preferred behaviours used to investigate the role of memory in animal movements in the last 20 years. Nicky Clayton and her colleagues have been pioneers of this type of study. A research line involved food- storing behaviour of western scrub jays Aphelocoma californica in the laboratory. They have investigated several aspects of behaviour that indicated complex mental capabilities of this species. One study demonstrated that scrub jays remember what they have stored and use episodic-like memory to remember when and where they 22 2 Distribution of Individuals have stored food (Emery and Clayton 2001). Another experiment was conducted by Clayton and Krebs (1994) using two storer/non-storer pairs of species, marsh tit Parus palustris/blue tit P. caeruleus and jay Garrulus glandarius/jackdaw Corvus monedula. These pairs were compared on a one-trial associative memory task with results that suggest food-storing birds respond preferentially to spatial information than non-storers.

2.2.3 Movement as an Adaptation

Behavioural ecology defines behaviour as a design that results from natural selection, and foraging theory conceives behaviour as decision rules modelled by this evolutionary process (Stephens and Krebs 1986). The function of movement is varied, the main one being to search for resources. The movement can fulfil other purposes such as escape from predators, patrolling an area, dispersing, or exploring to learn or evaluate changes in the environment. The rules of movement are usually different depending on these motivations. In most cases movement is an adaptation for resource searching, so its components are expected to be affected by the distribution of resources. Because of the hierarchical nature of spatial distributions of resources, we would expect to find animals initially using gross cues, indicative of certain habitats; then homing in further using patch cues; and finally employing cues from individual resources. Bell (1991) classified searching behaviour based on ecological scales. He firstly analysed mechanisms related to patch location and then how animals restrict search to a patch. He assumed that, when searching for a resource patch, an animal attempts to localise an assemblage of resources, whereas when searching within patches, it avoids leaving the assemblage until it becomes unprofitable to remain. The following terms are used depending on scale. For locomotory movements and scanning within a patch, the term is local search (Jander 1975). Immediately after leaving a patch or a resource and seeking others, regardless of its orientation mechanism, it is designated as ranging or travelling. Ranging refers mainly to the internal constraints, i.e. search using orientation mechanisms until finding external sensory information (Jander 1975), while travelling refers to external constraints, i.e. moving along a matrix without resources (Charnov 1976). Movements between habitats are called dispersal or migration (see Sect. 1.5). An example of an adaptive approach to the study of movement is provided. Cassini and Krebs (1994) conducted a field experiment with hedgehogs Erinaceus europaeus, in which they recorded movement behaviour in response to food manipulations. Hedgehogs are nocturnal solitary insectivores with overlapping home ranges (Morris 1991a, b). When active, they spend most of the time searching for food (Wroot 1984). The study was conducted in a 4-ha ground in Oxford, United Kingdom. In order to analyse the distribution of hedgehogs, the study area was divided into four sectors—NW, NE, SE, and SW. A ‘prey item’ 2.2 Movements 23

Fig. 2.3 Results of a field experiment on foraging hedgehogsErinaceus europaeus. Explanation of references in the text (modified from Cassini and Krebs 1994)

consisted of two chunks of cat food. A ‘patch’ consisted of 36 prey located in a sector of the study area. The five experimental phases were (1) ‘No food 1’; (2) ‘Food NW’, with 36 prey distributed in a 10 × 10-m grid and flags marking the prey sites; (3) ‘No food 2’; (4) ‘Food NE’, prey randomly distributed in a sector with no flags; and (5) ‘No food 3’. On the last day of the ‘No food 2’ phase, 25 flags without food were located in SW sector in a 10 × 10-m grid (the ‘Flag’ test). During the phases without food addition, hedgehogs expressed a natural preference for NW sector, and they responded to the experimental manipulation by increasing the use of those patches that received additional food (Fig. 2.3). During the experiment, hedgehogs were individually observed, counting the number of times they changed 90° in the direction of their movement (turns/10 min). Hedgehogs responded differently to the two experimental manipulations (Fig. 2.3). When they had inf.rmation on food location (regular distribution and visual signals), they showed more straight movements, but when they did not have environmental cues (random distribution of no signalled food), they increased the frequency of turns. Ad libitum observations allowed getting a description of the actual behaviour of hedgehogs: in ‘Food NW’ experimental phase, they moved in a straight line for several metres after they initiated movement following a capture, while in ‘Food NE’ phase, their movement resembled area-restricted searching behaviour. In summary, hedgehogs were able to learn about the properties of the environment and responded efficiently to improve foraging success. 24 2 Distribution of Individuals

2.2.4 Statistical Description of Movement

Traditional studies that investigate the adaptive value of movement were mainly based on experimental tests conducted in small scales. These studies were limited by the lack of statistical tools to describe the movement patterns. Recently, modern methods to interpret behavioural changes in movement trajectories or to identify the role of resource distribution on animal movement patterns have been developed (Giuggioli and Bartumeus 2010). The most important contribution that has been made to movement ecology approach is the use of statistical tools borrowed from physics, such as Lévy statistics and anomalous diffusion, and optimal stochastic search (Giuggioli and Bartumeus 2010). Several studies have suggested that food searching in several species of vertebrates follows what is called a Lévy flight (Mandelbrot 1982). When animals spend time without finding sparsely and randomly distributed resources, they appear to start a random search trajectory, which is composed of jumps with lengths that follow a heavy-tailed distribution, corresponding to a power law (Viswanathan et al. 1999). In modern movement ecology, individual trajectories are followed with GPS and bio-loggers, which allow recording details of animal movements (Giuggioli and Bartumeus 2010). Trajectories are often analysed by considering the collection of fixes as a series of random events (e.g. displacements, turns), whose spatial and temporal distributions are assumed to produce patterns that can be interpreted statistically. These methodologies have been applied to numerous species, and the number of publications has grown exponentially. A few examples published in a period of 3 years are in bacteria (Jarrell and McBride 2008; Zhang et al. 2010), in plants (Damschen et al. 2008), in insects (Ovaskainen et al. 2008), in fish (Patterson et al. 2009), in reptiles (Godley et al. 2008), in birds (Mandel et al. 2008), and in terrestrial (Dalziel et al. 2008) and marine mammals (Gurarie et al. 2009). Also, there is a large number of reviews on modelling techniques and concepts. An example of a study in movement ecology is provided. Ramos-Fernández et al. (2004) have determined the distribution of the sojourns made by fruit-eating spider monkeys Ateles geoffroyi, who forage in a semievergreen tropical forest in Yucatán, Mexico. The results show power-law behaviour P(l) ~ l−α, with an exponent α = 2.2, which suggests that these foraging movements may be described by Lévy walks. Boyer et al. (2006) suggested that this Lévy foraging behaviour may be the outcome of the distribution of resources. They observed in the field that spider monkeys can follow regular routes, so they hypothesised that their foraging movements rely on processing information on resource distribution. They developed a foraging model in which (1) monkey territory is composed of many foraging patches with varying sizes and in which the spatial structure can be varied and (2) foragers know the location and size of the targets and follow a simple foraging strategy of maximising food intake in a minimum travel distance. They used this model to explore the conditions that lead to Lévy foraging patterns. They found agreement between the field exponents (for step length, tree size, and waiting time distributions) and their theoretical values at a special parameter βc = 3, suggesting that the model 2.3 Home Ranges 25

Fig. 2.4 Movement ecology of primates, based on a study by Boyer et al. (2006). Once a monkey has stopped at a tree, it stays for a time τ before moving to another site. This waiting time is plotted against its fit by an inverse power lawΨ (τ) ~ τ−w, with w ≈ 2.0 (one time unit represents a 5-min interval)

correctly captures the interactions between spider monkeys and the environment (Fig. 2.4). This study by Boyer et al. (2006) is an attempt to link the statistical approach of movement ecology to other areas of behavioural research, such as cognitive ecology and foraging theory.

2.3 Home Ranges

The home range is a concept applicable only to mobile animals. There are descriptions of the home ranges of birds, reptiles, amphibians, fish, and some invertebrates. However, it has been studied mainly in terrestrial mammals, carnivores being the favourite taxonomic group. Burt (1943, p. 351) provides the first definition of home range as ‘that area traversed by the individual in its normal activities of food gather- ing, mating, and caring for young. Occasional sallies outside the area, perhaps exploratory in nature, should not be considered part of the home range’. The study of home ranges followed the inductive direction from data to theory. Its evolution is the consequence of the advent of radiotelemetry in the late 1950s. The animal car- ries a transmitter implanted or attached to a collar or harness. The transmitter emits a signal that is captured by a receiver from distance. This method was adopted as the main tool for fieldworks by wildlife researchers. Hundreds of studies using radiotelemetry have been published since its release. Satellites have improved enormously the use of radiotelemetry in two directions. On the one hand, the interpretation of satellite images and the development of geographic information systems substantially improve the acquisition of environmental information that can be compared 26 2 Distribution of Individuals with home-range locations. On the other hand, some transmitters implanted in the animals can now be used through satellites, which allow the following from almost any researcher location to any animal position in the world. This new technique has been used for oceanic vertebrates, migrant birds, and many other species that live in hard environments or move extensively. Many reasons have been advocated for estimating home ranges, in both research and management fields. Several studies deduce social organisation and breeding system for the overlapping of home ranges. Knowing home ranges also provides information on foraging and food choices, limiting results, among other important components of habitat. However, it is important to note that the study of home ranges was the result of a technological opportunity. In other words, once the technique was released and the data obtained, then came the scientific questions. Therefore, the study of home ranges is characterised by a strong empiricism, so it is often difficult to use this information to test hypotheses arising from a theoretical framework. Data used to estimate home ranges are a collection of points on a map for each radio-tracked individual. The boundaries of a home range are difficult to measure because the number of points normally decreases to the centre of the area, so decreasing the precision of the measurements. Several methods exist to transform this set of locations in an area that can be considered an animal’s home range. The most common statistical methods to estimate home ranges are grids, minimum convex polygon, circle and ellipse approaches, Fourier series (Anderson 1982), harmonic mean distribution (Dixon and Chapman 1980), fractal estimators (Bascompte and Vilà 1997), and Kernel estimators (Powell 2000). Brief descriptions of these methods follow. A first method consists of a grid superimposed on the study area and represents a home range as the cells in the grid having an animal’s locations (e.g. Doncaster and Macdonald 1991). This simple approach has several advantages. It avoids assuming that data fit some underlying distribution. Occasional sallies are easily identified as cells visited only once. It allows straightforward comparisons between samplings taken at different moments in the same study area. A traditional and spread method is to estimate the minimum convex polygon. It is also easy to imple- ment and consists of a draw that results from joining known or estimated locations that produced the smallest convex polygon (Hayne 1949). Again, the advantage is that it is not necessary to assume any statistical distribution of the spatial distribution of the individual. However, it has several problems, such as high sensitivity to extreme data points, an outline only with large sample sizes, and incorporation of large areas that are never used (Powell 2000). Bivariate normal models use circles or ellipses to estimate home ranges; animals are assumed to move randomly about their home range, with the most probable location being the very centre; then a 95% ellipse is calculated around the mean location. The area of this ellipse is an estimate of the animal’s home range (Jennrich and Turner 1969). In recent years, Kernel density estimators are probably the most common method used for individual use of space. This produces an unbiased density estimate directly from data and is not influenced by grid size or placement. Therefore, it is considered the best estimator available (Powell 2000). A typical study on home ranges extracted from wildlife literature is described. The giant anteater Myrmecophaga tridactyla is a member of an emblematic group 2.4 Site Suitability Models 27

Fig. 2.5 Home ranges of giant anteaters Myrmecophaga tridactyla (modified from Medri and Mourão (2005)) of Neotropical mammals. It used to distribute in a wide biogeographical range from Guatemala to Argentina, but now is a locally endangered species due to human persecution and habitat destruction. In spite of its conspicuousness and importance, it is still a poorly known species of Neotropical fauna (Shaw et al. 1987). Medri and Mourão (2005) captured, radio-collared, and monitored seven anteaters of the Pantanal wetland, Brazil. Males covered areas of 4.0–7.5 km2 (5.7 ± 1.7 km2), and one of the two females monitored occupied a larger area (11.9 km2) than the males, but none of the curves of cumulative area unequivocally reached the asymptote. The home-range estimates were calculated using the 100% minimum convex polygon and 95% adaptive kernel methods. There was considerable overlap among individual areas used (Fig. 2.5). Giant anteaters rested mainly in forest patches and savanna and frequently used grasslands and scrub savanna for foraging.

2.4 Site Suitability Models

Site suitability models are descriptive models that relate, with the availability of habitat or components of the habitat, normally resources, the probability of utilisation by an individual or set of individuals. Preference indexes and resources selection functions (Manly et al. 1992) are site suitability models. As it was explained in Sect. 1.4, the term selection is not used in this context; instead the general term site suitability models is used. Site suitability models have been applied to all levels of organisation, from individuals to species, so I return to them in subsequent chapters. In Sect. 3.2, we will see site suitability models applied to aggregations (at habitat and landscape scales), while in Sect. 7.2, to species distribution (at regional scale). 28 2 Distribution of Individuals

Site suitability models describe the environment in terms of some attribute of resource categorisation or units. These discrete units of resources (in our case habitats) are often simply a grid cell which divides the home range of an animal. Studies normally involve classifying resource units into qualitative entities or categories or measuring specific variables or properties that are characteristic of those units. Site suitability models have a clear applied approach. The purpose is to use a rapid method for field studies to establish habitat requirements of individuals or populations. Site suitability models can be used with three types of sampling designs, or three different procedures to estimate availability: random, case–control, and use– availability (Keating and Cherry 2004). In a random sampling design, sampling units are selected randomly and the habitat/resource characteristics of units are evaluated according to the presence/abundance or the absence of the target species (at the proper level). In a case–control sampling, two data sets on resource/habitat traits are collected, a pool of sampling units with the species and another pool without it. In the use–availability sampling method, habitat/resource variables are measured in units where the species is present and, then from a pool of all units in which the study area was divided, selected randomly. Statistical modelling procedures used for site suitability models are of two types: simple sample comparisons and generalised linear models. Indices of selection and simple sample comparisons fit into the first category, while linear regression, logistic regression, log-linear, and proportional hazards models are special cases of the second type of statistical methods. They are applicable to different data sets, but always with the same objective of establishing the statistical relationship between use and availability. One example of the use of site suitability models to describe habitat distribution of individual organisms is provided. Moose or elk Alces alces is a well-known cer- vid species that inhabits temperate to subarctic regions of the northern hemisphere. Predation, food availability, climate, parasites, and disease are the most important natural factors that can potentially limit moose populations across North America (Van Ballenberghe and Ballard 1994). Hunting is also a major limiting factor of moose populations in areas accessible to humans. Dussault et al. (2005) studied a moose population from a protected area of Québec, Canada. Hunting was prohib- ited, but there were two potential predators, the wolf Canis lupus and the bear Ursus americanus, so the prediction was that habitat selection at larger spatial scales should be oriented towards avoiding exposure to predation risk. At a smaller ecological scale, the prediction was that moose habitat selection should be primarily influenced by food availability and secondarily by snow. Forty moose were monitored with GPS telemetry collars. A total of 186 forest stands were surveyed for availability of food, concealment cover, and winter cover. Dussault et al. (2005) expected that, if moose traded off food availability with protection from predation risk, they would be attracted to edge between habitats providing abundant food and sheltering against predation risk. Also, moose that traded off food availability with cost of locomotion in deep snow were expected to be attracted by the edge between habitats providing abundant food and those sheltering from snow. At the landscape scale, the extent of edge density (both edge types) was used as a measure of food 2.4 Site Suitability Models 29

Fig. 2.6 Habitat selection by moose Alces alces (modified from Dussault et al. 2005). The comparison between moose home ranges and the random permutations provides an estimation of the variables that are relevant for moose distribution. Variables analysed: (I) overlap with wolf territories (%), (II) density of edge between stands offering high food abundance and stands offering protection against snow (m/ha), (III) area with low snowfall (%), (IV) area with moderate snowfall (%), (V) area with high snowfall (%), (VI) stands offering protection against protection risk (%), (VII and VIII) area offering high food abundance, (IX) stands offering protection against snow, (X) other habitats (non-regenerated and non-forested areas, lakes), (XI) density of habitats providing abundant food and those sheltering against predation risk (m/ha). Asterisk represents statistical differences between bars and cover interspersion. Habitat selection at the landscape scale was determined by comparing habitat composition and edge density within individual moose home ranges with those available within the study area. At the landscape scale, the authors used a stepwise logistic regression to determine which habitat and edge variables discriminated individual moose home ranges from random permutations. At a home-range scale, habitat use and availability were measured at telemetry locations and within the home range, respectively. Telemetry locations were pooled by individual and period to calculate standardised habitat selection ratios. Selection ratios constituted the resource selection function and were used as the basic unit in all subsequent statistical analyses of habitat preference. The authors used these selection indices to create independent variables by subtracting adjacent pairs of values and then applied a repeated measure MANOVA using individual moose as sampling. At the landscape scale, three variables discriminated the composition of moose home ranges from random permutations (Fig. 2.6). The overlap between moose home ranges and wolf territories was much lower than expected. Moose selected 30 2 Distribution of Individuals areas where habitats providing high food abundance were interspersed with habitats providing shelter against snow. Additionally, moose home ranges contained fewer areas with low snowfall compared with random permutations. At a home-range scale, patterns of habitat selection varied between time periods but the tendencies were usually the same. Neither predation risk nor food abundance alone determined habitat preference of moose since stands offering the highest food abundance and those offering the best concealment cover were not the most preferred by moose. In summary, at both scales, moose traded off food availability with exposure to deep snow and predation risk, as expected by the authors.

2.5 Patch Models of Foraging Theory

2.5.1 Introduction: Foraging Theory and Foraging Behaviour

Foraging theory was born in 1966 from two studies published by John Emlen and by Robert MacArthur and Eric Pianka. They were interested in foraging behaviour as a mechanism to explain the processes in populations and communities. The principles of their work were adopted by behavioural researchers, and the theory went on to have a huge influence in the next two decades, dominating the field of behavioural ecology. Dave Stephens and John Krebs published their book Foraging Theory in 1986, at a time when—paradoxically—the influence of this theory began to decline. However, several of its principles remained central in behavioural ecology: the need to refine the hypothesis so as to generate accurate quantitative predictions, the value of experimental work combined with field observations, and the concept of behaviour as a decision-making process. Foraging behaviour studies expanded in two directions: looking at the neurophysiological and cognitive mechanisms involved in foraging decisions and analysing the ecological consequences at the level of populations and communities. The book ‘Foraging: Behavior and Ecology’, written by David W. Stephens, Joel S. Brown, and Ronald C. Ydenberg in 2007, shows how foraging theory went from being a body of simple but robust models to a rich and diverse field, extending into diverse disciplines such as neuro-ethology, animal psychology, ecology, population biology, and conservation. The classic theory of foraging emphasises the use of optimisation models as a tool for studying behaviour. Optimisation models are applied to the interactions of the animals with the environment and have the following characteristics: • They treat the behaviour as a decision-making process. Decisions are not ‘con- scious’ in the human sense, but are behavioural alternatives that are subject to natural selection and, therefore, can be analysed with cost–benefit models. • They analyse the adaptation as a balance between costs and benefits. This balance is achieved through natural selection operating on biological traits. 2.5 Patch Models of Foraging Theory 31

• They generate hypotheses about the adaptive value and the ecological consequences of foraging, anti-predatory behaviour, and habitat selection. • Habitat selection is divided in a hierarchy of behaviours that are analysed separately (Fig. 1.4).

2.5.2 Types of Patches

The ‘classic’ habitat for foraging theory is a set of patches where the resource, normally food, is concentrated and a matrix through which an individual ‘travels’ without using resources. A patch defines in terms of the availability of one type of resource, mainly food. A patch has the dimensions for the exploitation of one individual. It is affected by the foraging activity so that the resource becomes scarcer as the individual remains in the patch. Foraging patches have two basic properties: the initial density of food and the so-called gain curve, i.e. how the resource is depleted as an animal stays. There are three main types of gain curves (Fig. 2.7): without depletion, gradual depletion, and sudden depletion (Sutherland 1996). These gain curves depend on the rate of replenishment of the resource and searching behaviour of the organism. In gain curves without depletion, resources do not diminish their density in the animal’s foraging period. This occurs, for example, in the case of prey that is super-abundant, but has a high anti-predatory response. This behaviour of the prey determines resource availability decreases in the short term, but the abundance stays approximately constant. In sudden-depleted patches, the probability of obtaining resource is constant until all the resource is consumed. This occurs when animals search systematically within a site so that the encounter rate with resource items remains constant until the patch is depleted. The progressive depletion can occur for two main reasons: because the animals search randomly and every time it is harder to find food or because the food must be transported to a central location. Examples of this are birds feeding their chicks or social insects (bees) carrying the food to a central place. In this case there is a cost associated with transportation of the food that determines a gradual decrease of the gain with increasing load to be transported.

2.5.3 Marginal Value Theorem

In the case of gradual depletion, the classic model called the theorem of the marginal value of capture is applied. This model was published in its first version in 1976 by Erich Charnov. The model predicts an optimum time of departure that occurs when the instantaneous rate of acquisition of resources on the site equals the overall rate environment. In this case, the prediction changes depending on the average travel time between sites or the quality of the patches (Fig. 2.8). The characteristics of the model are: 32 2 Distribution of Individuals

Fig. 2.7 Type of patches in relation to resource gain functions

Fig. 2.8 Graphical representation of the marginal value theorem of capture. The maximal slope of the function relating gain to time (travel between patches + patch residence) is reached when n* = 5

Decision: How long to stay in a patch. Context: Patches separated by areas without resources, slow recovery of the offer, and random search.

Variables involved: ni = resources obtained at site i, bi = searching time within patch i, and v = travel time between patches. 2.5 Patch Models of Foraging Theory 33

Decision criterion: Maximise the rate of acquisition of resources:

Tn=+()vb (2.1) Decision rule: Leave a patch when the marginal rate of capture equals the average rate environment:

n n i = ∑ i b ()vb+ i ∑ i (2.2) An experiment that was conducted on a South American mammal (Cassini et al. 1990), the screaming hairy armadillo Chaetophractus vellerosus, is described as an example of many tests of the marginal value theorem (Nonacs 2001). These animals with a funny name are found in central Argentina and occupy arid and semi- arid environments where the soil is not hard so they can dig its burrows (Abba and Cassini 2008). They are solitary foragers with an omnivorous diet and a preference for insects (Abba and Cassini 2010). One of the few field studies on this species was conducted by Greegor (1985). He used a small roll of thread attached to the animal and was able to describe preliminary animal movements (Fig. 2.9). Armadillos followed a typical pattern with relatively long straight lines interspersed with looping and spiralling movements. This area-restricted pattern was normally observed when armadillos reached a bush. In the semi-desert environment, where these armadillos live, food is concentrated mainly under bushes that are scattered in a matrix of nude soil. This is a typical context of foraging patches and a travelling matrix between them. Cassini et al. (1990) trained armadillos in a U-shaped alley with each arm being almost 4 m long (Fig. 2.9). The ‘patches’ were the ends of the arms, where two small containers received food items (commercial dog food pellets) through a vertical pipe. Food was delivered in accordance with a patch gain function with a power form, so the feeding schedule simulated a depleting patch. The authors used a discrete version of the marginal value theorem applied to an environment with two equally common patch types, a ‘poor’ and a ‘good’ type, which were encountered in alternation and differed in the rate at which resource depression occurred. The model predicts the number of prey per visit to take from each patch type (p* and g*) that maximises capture rate (R) in the cycle of two successive patch visits. Rate of gain in such a cycle is

()pg+ R = pg, ()rq++2t (2.3) pg where p and g are numbers of prey items per visit to poor and good patches, respectively; r and q are poor and good patch residence times; and t is the travel time between patches. Using (2.3), they calculated by iteration the pair (p*, g*) that yields the maximum profitability (R*). The marginal value theorem predicts that optimal exploitation of each patch type should depend on the average rate of gain in 34 2 Distribution of Individuals

Fig. 2.9 Foraging by individual armadillos Chaetophractus vellerosus. In nature: example of searching trajectory of one night of foraging activity by an armadillo in the Argentinean step (modified from Greegor 1985). It travelled from shrub to shrub, where most insect prey were available. In the lab: Apparatus used in a laboratory experiment with the same species of armadillos (modified from Cassini et al. 1990). Armadillos ‘travelled’ from one dispenser to the other, when they entered the ‘patch’; food item was provided through a pipe

the environment, and thus changing this average should affect the number of prey collected from both patch types. Cassini et al. (1990) manipulated the average rate of gain for the environment by differing sp. There were two treatments or ‘environments’, which differed only in the gain curve in the poor patch (sg = 0.5, sp = 0.2 in treatment A and sp = 0.4 in treatment B). Qualitative trends observed in the experiment were as predicted by the marginal value theorem. Within treatments more prey were taken from good than from poor patches. Between treatments exploitation of good patches decreased and of poor patches increased when the quality of the poor patches was increased. Prey per visit consumed by armadillos in poor and good patches of both treatments closely agreed with the expectation of the patch use model (Fig. 2.10). 2.5 Patch Models of Foraging Theory 35

Fig. 2.10 Observed versus predicted consumptions in the experiment with armadillos Ch. vellerosus (modified from Cassini et al. 1990, experimental set described in Fig. 2.9)

2.5.4 Patch Selection

An experiment with armadillos was just described in which predictions of a discrete version of the marginal value theorem were tested. Another experiment in a similar set-up was conducted, which differed in the way that armadillos got food in patches (Cassini 1993). The ‘environment’ consisted of a V-shaped runway. Each ‘patch’ was made up of a square plastic grid divided into 400 square compartments of 1.3- cm side. Ten or 40 dog food pellets were randomly distributed, representing ‘poor’ and ‘good’ patches. There were two treatments, ‘long travel’ and ‘short travel’, located at the distal and proximal ends of the runway arms, respectively. In this experiment, armadillos were free to move within patches and decided a searching strategy. The predictions on patch use of two searching strategies were tested. A random forager is an individual that has a certain probability of exploring an already visited section of the patch (Fig. 2.11). Its mean probability of finding food is a decreasing function of time in the patch, and the functions of patch gain are negatively accelerated (Fig. 2.7). Under these conditions, the marginal value theorem predicts that patch residence time and prey consumed per visit increase as travel time increases (Fig. 2.11). A systematic forager is an individual that does not visit a section of the patch already visited. Its mean probability of finding food is constant within each patch time and, therefore, the functions of patch gain are linear. Because systematic foragers are always expected to deplete patches fully, no changes are predicted in prey per visit and residence time for each patch type (Fig. 2.11). 36 2 Distribution of Individuals

Fig. 2.11 Schematic representation of the searching paths of systematic and random foragers in a circular patch, food randomly distributed prey of two types, and shape of gain functions under both foraging strategies. Foraging theory predicts that random foragers will stay longer in patches when environments have long travel times, while systematic foragers will use the same departure time with long and short travel times (modified from Cassini 1993)

Figure 2.12 illustrates the results of this experiment. Armadillos tended to depleted patches in all conditions, as was expected for a systematic forager. The actual behaviour of the armadillos observed during the training sessions also invalidates random searching. The path pattern commonly observed in the animals while they visited a patch had a spiral configuration: individuals often began foraging at the periphery of the patch and then moved progressively towards the interior, leaving when they reached the centre. This spiral route clearly minimised the probability of revisiting the same compartments. This example shows how a systematic type of movement within a patch with randomly distributed food produced a linear gain function with sudden depletion. Each patch type may be characterised by the slope of the gain function. Under these conditions, a gain-maximising strategy will predict that a set of poor patches will be abandoned without exploitation because gain is under the marginal value. This is a 2.5 Patch Models of Foraging Theory 37

Fig. 2.12 Results of the second experiment with armadillos Ch. vellerosus. In this experiment, armadillos developed a systematic search (Fig. 2.11). Under these conditions, the optimal strategy of a rate maximiser is to stay until patches are depleted, independently of patch quality and travel time. Maximal food items were 10 and 40 in poor and good patches, respectively (modified from Cassini 1993) prediction analogous to the one produced by another famous foraging theory product: the prey choice model (Stephens and Krebs 1986). Foragers are expected to rank patches in terms of their gain function slopes as if they were prey profitabilities in the prey choice case. In summary, systematic searching allows animals not only to use patches more efficiently but to select patches in a way that poor patches are never visited.

2.5.5 Patch and Prey Selection

The classical prey choice model is applied to a homogeneous environment where prey types differ in profitability, while the classical patch use model is applied to patchy environment where patches contain the same type of prey. In nature, resource types frequently concentrate in different areas of the habitat, so resource patches differ in the relative abundances of prey types. Figure 2.13 represents an example of two prey types with two types of distribution. There are 10 highly profitable prey and 18 low profitable prey in both cases. In the homogeneous habitat, the rate of encounter with large prey is high enough and the prediction is that the forager will specialise in this type of prey. In the heterogeneous condition, patch type B has large proportion of large prey, but the overall prey density is low. The opposite occurs for patch type A. In this environment, the expected behaviour depends on the distance between patches. When travel time is large enough between patches, a systematic forager will select patch A exclusively, because prey density in B is low, and will be prey opportunistic when in patch A, because the density of high profitable prey in there is low. 38 2 Distribution of Individuals

Fig. 2.13 Schematic representation of two environments with same amount of highly (X) and low (x) profitable prey, one with food distributed homogeneously and the other with distribution in patches. Foraging theory predicts prey selectivity in the homogeneous but not in the heterogeneous habitat. See text for details

2.6 Foraging in Heterogeneous Habitats by Plants

The growth pattern of stoloniferous and rhizomatous plants may be considered analogous to the search path of a foraging animal (Bell 1984). Hutchings and de Kroon (1994) defined foraging behaviour in plants as the plastic physiological and morphological alterations that directly or indirectly enhance the capture of essential resources. Aboveground, such responses include the shade-induced elongation of stems and petioles by which laminas are projected upwards, along gradients of increasing photon flux density, into locations with higher light availability and the light-fleck-induced activation of the photosynthetic apparatus that increases the capture of ephemeral light pulses (de Kroon et al. 2009). Belowground foraging responses include the initiation and activation of lateral root primordia and the subsequent growth of lateral roots in patches of substrate with high nutrient concentration, as well as the activation of nitrate transporters in response to high nitrate availability (de Kroon et al. 2009). Plants often grow in patchy environments and modifications of the growth pattern in response to patch quality may alter the distribution of ramets. Plasticity of some clonal species allows them to change their growth form so that they become more compact in favourable habitat. In a patchy environment, this could enhance the placement of ramets in relatively resource-rich microhabitats. Accordingly, morphological plasticity has been viewed as a behavioural phenomenon, which allows 2.7 Balancing Demands and Use of Refuges 39

Fig. 2.14 Individual foraging in plants. Experiment conducted with the herb Glechoma hederacea, bred in different types of environments. The figure represents the ‘patchy’ environment, with the ‘good patch’ located at IV (the central circle contained half of the potting compost). Roosts were harvested in four fractions of soil: I the small pot containing the original parent ramet; II outside the central circle in the proximal half of the box; III outside the central circle in the distal half; IV inside the central circle (modified from Birch and Hutchings 1994) clonal plants to arrange their ramets within their environment in a selective manner. This capability has been regarded as a manifestation of foraging behaviour (Grime 1979). Many experiments have been conducted with clonal herbs that showed this plant’s ability to select the best patch where to concentrate growth (Slade and Hutchings 1987; Birch and Hutchings 1994; Cain 1994; De Kroons et al. 2009). Birch and Hutchings (1994) conducted an experiment to test whether plants may exploit patches of resources following optimal foraging rules. They grew the herb Glechoma hederacea in three artificial soil environments: ‘uniform’, ‘patchy’, and ‘poor’. In the uniform condition, potting compost was evenly mixed with sand (Fig. 2.14). The ‘patchy’ treatment contained the same quantity of potting compost, but half was concentrated in a central resource-rich circle. The poor environment had half of the compost of the other two treatments. This experiment showed that G. hederacea was able to produce over two and a half times biomass in the patchy environment. This improvement was achieved by concentrated 80% of root biomass within the resource-rich central circle.

2.7 Balancing Demands and Use of Refuges

There is abundant evidence regarding insects, fish, birds, and mammals that prey are able to respond spatially to the abundance or presence of predators. When the environment with more abundant food resources is also the most risky, individuals that are prey use patches less than expected on the basis of the food supply. A patch in 40 2 Distribution of Individuals which food has a homogeneous distribution can be used in heterogeneous form by the effect of predation risk. This is typically the case when the patch is associated with a shelter, such as a den or a border with plant cover for mammals, woodland for birds, or rocks for fish. Prey typically uses with greater intensity portions of a foraging site that are closer to the shelter. However, plant-covered environments are not always a source of refuge, since some predators use them to hide and stalk their prey. Therefore, in some species such as ungulates, the spatial response to the presence of coverage is the opposite, i.e. avoid feeding in areas close to high and dense vegetation. Brown (1988) extended the marginal value theorem to include the effects of predation risk and alternative activities on patch use. A fitness-maximising forager should cease harvesting a resource patch when the value of its harvest rate (H) no longer exceeds the sum of its energetic cost of foraging (C), predation risk (P), and the missed opportunity cost (MOC): (2.4) HC=+PM+ OC This type of modelling the trade-off between foraging and danger avoidance remained elusive during the first developments of foraging theory because the costs and benefits were in different units. More recent developments used a life history perspective that allowed translating both foraging gain and danger into a common measure of fitness (Stephens et al. 2007). These models have proved to be useful in predicting theoretically the response of organisms to changes in predation risk at different scales. However, quantitative tests of these predictions can be difficult, especially when state-dependent costs are incorporated. Brown (1988) developed a simple method to measure risk while foraging, by means of measuring giving-up densities. Foragers at depletable patches do not consume all of the contents. Brown called the amount of food that a forager leaves behind the giving-up density. It is a surrogate of the quitting harvest rate predicted by the marginal value theorem. This model predicts, for example, that a forager should have a higher giving-up density as travel time among patches declines. Giving-up densities have been used to estimate the predation cost of foraging of several species, e.g. rodents (Jacob and Brown 2000), fish (Petty and Grossman 2010), and birds (Oyugi and Brown 2003). A study conducted by Oyugi and Brown (2003) is described. Giving-up densities of robins Turdus migratorius and starlings Sturnus vulgaris were estimated to test for (1) the role of cover as safety in microhabitat preferences, (2) temporal habitat preferences, and (3) the habitat quality of mowed grass. The habitat comprised clumps of shrubs and tall trees within open mowed lawn. Experimental patches consisted of trays filled with vermiculate and 30 mealworms Tenebrio molitor bur- ied in this substrate. The birds had to search within the patch, and the longer a forager spent in the patch, the lower was the rate at which it found new worms. The number of mealworms left in the resource patch after the birds have quit the exploited patch is the giving-up density. The location of patches was lumped into three distances categories of 0–4, 4–8, and 8–12 m from the cover. Giving-up 2.8 Distribution at Equilibrium: The Matching Rule 41

Fig. 2.15 Giving-up densities of birds. Mean ± SE giving-up densities of American robins Turdus migratorius and European starlings Sturnus vulgaris foraging on mealworms increased with distance from cover. The type of cover also influenced patch use, with lower giving-up densities near dense shrubs than near tree-canopy cover. Giving-up density was measured as the number of mealworms remaining per tray at the end of each day of observation (modified from Oyugi and Brown 2003)

densities were lowest near cover (independent of cover type) and increased steadily with distance from cover (Fig. 2.15). By using the giving-up densities as a measure of quitting harvest rate and foraging efficiencies (Brown 1988), robins and starlings revealed that microhabitats near cover provide safer feeding areas than those away from cover. Giving-up densities increased with distance from the edge of shrub or tree canopy. Association of risk with closed versus open habitats, or with distance from cover, varies among bird species (Lima and Dill 1990). Other species of birds perceived cover as a source of attacks and stay foraging for shorter periods.

2.8 Distribution at Equilibrium: The Matching Rule

The pattern of distribution of an individual in its home range depends on a decision process. The individuals select between alternative behaviours such as stay, move, forage, escape, or rest. Foraging theory is based on the fundamental principle that this decision process is modulated by natural selection. Optimisation models are 42 2 Distribution of Individuals applied, not because animals are expected to behave as perfect machines but as an instrument to organise hypothesis around this principle. Marginal value theorem is an optimisation model that describes behavioural decisions while foraging in a patchy environment. It predicts the amount of time that the animal should invest in each patch depending on resource availability. But which is the expected distribution in the equilibrium? Staddon (1983) conducted a simple deduction to obtain the overall or molar expectation of patch rules that maximise fitness. He supposed that the environment is composed of two types of depleting patches—‘good’ and ‘poor’—as illustrated in Fig. 2.8, with a power function form:

s (2.5) CAt = r where A and s are constants and 0 < s < 1. Taking the first derivative to find the marginal value gives

dC sC ==Astr s−1 t dt t (2.6) i.e. the derivative of the cumulative food intake function; Ct can be written as a function of the ratio Ct/t. From the marginal value theorem, dC/dt should be the same for all patches; thus, for two patches g and p, dCg/dtg = dCp/dtp. If s is the same in both patches, equating the marginal value yields Cg(tg)/tg = Cp(tp)/tp, or

Ct() t gg= g Ct() t pp p (2.7) Equation (2.7) states that the ratio of the times the animal spends in both types of patches is equal to the ratio of the amounts of food obtained. This is termed the matching rule. If it is not the same for both patches, then choice proportions should be proportional (not equal) to payoff proportions, the so-called biased matching. Matching law was firstly described by animal psychologists who have a long tradition of studying behaviour experimentally. They have been observed matching in many experiments where subjects choose between alternatives that reinforce probabilistically (Herrnstein 1961). The ‘psychological’ matching law is applicable in a type of experimental set-up called concurrent schedules, in which subjects can distribute their behaviour between two or more simultaneously available schedules of reinforcement. More specifically, individuals of a diverse range of species distribute their responses or time between in proportion to alternative qualities, in experiments with concurrent variable-interval schedules (Baum 1981). In these experiments, reinforcement is based on the passage of time and the requirements for reinforcement change randomly around a mean. Therefore, the animal can learn that the delay in an alternative is greater on average than in the other, but cannot exactly determine the exact moment when the device is ready for delivery of the reinforcement. Chapter 3 Distribution of Aggregations

3.1 Introduction

This chapter discusses models that, by being applied to sets of individuals, do not demand any specific collective properties for these sets, as group or population models do. Some models have been explicitly applied to aggregations that are the result of the conglomeration of individuals in patches because of the concentration of resources on those sites. Other models simply emphasise environmental properties and use measurements of abundance or occurrences of sets of individuals without any specification of the properties of these sets. Although these models may be applied to groups and populations, they are preferably included in the category of aggregations. Finally, some models included in this chapter have the ability to incorporate levels of complexity in a way that can be applied to aggregations, groups, and populations. The chapter begins with a section of the descriptive models that were already presented in Sect. 2.4 and which were called site suitability models. The following section describes individually based models, which are widely used simulation models. Some of them can be applied to sets of individuals which have an internal dynamic as groups or populations but the basic principle of individual-based models is that they have properties that result from the processes occurring at the individual level within an aggregation. In its simplest form, these models explain heterogeneous distribution patterns of aggregations from random movements. The following section introduces a new mechanism that operates at the level of aggregation: competition. It describes the types of intraspecific competitions in a context of habitat selection. The habitat selection mechanisms and intraspecific competition are based on models of ideal free distribution, which are described in the next section. Ideal free distribution is a theory of equilibrium. It predicts the equilibrium distribution of aggregations between habitats, when individuals reach the maximal payoff within a certain environmental and competitive context. This chapter closes with an analysis of the expected pattern of distribution of aggregations at equilibrium.

M.H. Cassini, Distribution Ecology: From Individual Habitat Use 43 to Species Biogeographical Range, DOI 10.1007/978-1-4614-6415-0_3, © Springer Science+Business Media New York 2013 44 3 Distribution of Aggregations

3.2 Site Suitability Models

This category of models includes habitat suitability indexes (US Fish and Wildlife Service 1981), resource selection functions, and some preference indices (Manly et al. 1992). In Sect. 2.4, the use of site suitability models was described as applied to habitat selection by individual organisms. This type of model is most commonly used for sets of individuals which are only defined by their presence or abundance in a site. Use of habitat is frequently measured as number of individuals associated with a habitat or a component of the habitat, or by indirect evidence of the aggregation abundance, such as the density of dens or faeces. The units of use are normally arbitrary portions of land (pixels on a satellite image), although sometimes real areas occupied by a type of habitat area are measured. Availability is measured in two ways: (1) the amount of portions or the area occupied by a type of habitat in the study area or (2) the mean value adopted by components of the habitat in (1). Habitat components have several other names: resource variables, predictors, and covariates. Examples are plant cover, elevation, and distance to roads. The prevailing statistical model is a binomial generalised linear model, usually focused on logistic regression, although in the case of presence/available sampling designs, logistic regression is used as an estimating function and not for statistical inference (Boyce et al. 2002). When linked to a geographical information system, site suitability models can be useful tools in natural resource management, with applications for cumulative effects assessment, land-management planning, and population viability analysis (Boyce and McDonald 1999). Information criteria such as the Akaike information criteria or Bayesian information criteria are tools that can be useful in selecting a model from a set of biologically plausible candi- dates. Statistical inference procedures, such as the likelihood-ratio test, can be used to assess whether models deviate from random null models. However, for most applications of site suitability models, usefulness is evaluated by how well the model predicts the location of organisms in a landscape (Boyce et al. 2002). Site suitability models have the problems of all descriptive models: their prediction cannot be extrapolated to a large set of new conditions because (1) the estimates are a posteriori of data collection and (2) do not identify the decision criteria used by individuals in the expression of spatial preferences. In other words, the results of a study in a geographical area in a given period of time must be used with caution to explain the distribution in other areas and other periods. Site suitability models are still useful because they provide a rapid estimate of the distribution that can be used in applied problem solving or as a basis for more complex studies. Site suitability models have also been criticised because they assume that the preference for habitat types or qualities remains constant, regardless of the density of habitat or resources, i.e. they assume a linear functional response when actually this response may take different forms (Nolet and Klaassen 2009). Another frequent criticism is in relation to the difficulties in measuring availability (Garshelis 2000). 3.3 Individual-Based Models 45

Despite these criticisms, site suitability models are widely used, especially in relation to the management and conservation of wildlife. An example of this type of use is provided. Arzamendia et al. (2006) studied habitat selection by vicuñas Vicugna vicugna in a protected area of South America. Vicuña is an emblematic species of one of the major arid ecosystems of the neotropics: the Puna or Altiplano. In the past, excessive commercial hunting of vicuña for their valuable fleeces caused a severe decline in the population, with the vicuña almost becoming extinct by the mid-twentieth century. Effective protection resulted in the recovery of some of the population. The study area of the study (Laguna Pozuelos Reserve) was divided in 11 habitat types. Habitat use was determined by the frequency with which groups of vicuña were observed in each habitat type, and habitat availability was the proportion of the area occupied by each vegetation community. Habitat selection was analysed using the methodology of Manly et al. (1992) in which the relationship between the use of a resource and its availability was combined in a selection index that was compared by using a log-likelihood χ2 test. The selection index a was calculated as

  r  1  i   (3.1) a = ni  rj   ∑   nj 

where ri and rj are proportions of habitat type i or j in total habitat use and ni and nj are proportions of habitat type i or j in the environment. The a values are normalised so that the sum of all values equals 1. When selective habitat use does not occur, a is equal to the inverse of the total number of habitat types. During 2 years of study in the Toqueros area of the Laguna Pozuelos Reserve, vicuñas did not use the study area homogeneously (Fig. 3.1). They spent more time in habitats dominated by grasses (pajonal and esporal) and with high overall plant cover than in habitats with low cover dominated by bushes such as Parastrephia lepidophylla and Boutelowa incarum (with poor digestibility) or Tetraglochin cristatum. This information was used to improve management criteria of the protected area.

3.3 Individual-Based Models

Site suitability models use a molar approach, assuming an equilibrium distribution and identifying the predictors of the pattern of distribution. Individual-based models are more concerned with the process that originates the pattern. They are applied to an environment in which local interactions occur and the behaviour of individuals, following certain procedural rules and characteristic parameters, is tracked through time (Huston et al. 1988). When individuals are located in a geometrical space, it is said that the model is spatially explicit. This type of model can use either continuous or grid-like space. Individuals can be mobile, so that searching rules can be established in the virtual environment. 46 3 Distribution of Aggregations

Fig. 3.1 Site suitability model applied to study habitat use by vicuñas Vicugna vicugna in Puna region, South America. Habitats were classified in accordance to dominant plant species.Asterisk represents statistical differences between expected use obtained with a site suitability index and observed used measured as a frequency of vicuñas’ foraging groups (modified from Arzamendia et al. 2006)

Judson (1994) found three reasons for the rise of individual-based models in ecology: (1) the difficulty of prediction as implied by the theory of chaos, (2) the importance of local interaction between individuals, and (3) improvements in computer power. Lomnicki (1999) added a fourth reason, progress in evolutionary ecology studies, especially with regard to the study of the adaptive value of individual behaviour. The development of high-powered computers allows the exploration of these bottom-up models, which have been applied to a wide range of species and theoretical and applied problems. They have the advantages that it is easier than in analytical models to incorporate realistic spatial information about the environment and to explore the implication of different scenarios with regard to what can be achieved with traditional population models. Uchmánski and Grimm (1996) proposed four characteristics of individual-based models that differ from other similar approaches in terms of (1) the degree to which the complexity of the individual’s life cycle is reflected in the model, (2) whether or not the dynamics of resources used by individuals are explicitly represented, (3) whether real or integer numbers are used to represent the size of a population and (4) the extent to which variability among individuals of the same age is considered. Berec (2002) simplified the discussion about what should be considered a typical individual-based model, by giving an operational definition: a model in which a discrete individual that has at least one unique feature is the fundamental modelling unit. This unique feature is simply (at least) its spatial location. 3.3 Individual-Based Models 47

Fig. 3.2 Schematic representation of the process involved in an individual-based model applied to a host–parasite system. Flowchart of main model processes involved in the interaction between an ectoparasitic tick (Aponomma hydrosauri) and its lizard host (Tiliqua rugosa) at Mt. Mary, South Australia (modified from Tyre et al. 2006). Processes that are attributes of lizard population dynamics and behaviour which indirectly affect the ticks are placed in ovals, while processes directly affecting ticks are in rectangles

An example of the use of an individually based model is provided, applied to an arachnid parasite. Tyre et al. (2006) studied the interaction between an ectoparasitic tick Aponomma hydrosauri and its lizard host Tiliqua rugosa at Mt. Mary, South Australia. The authors developed an individually based, spatially explicit model with 28 parameters. It is based primarily on Chilton (1989), who conducted extensive measurements of the response of life history parameters of A. hydrosauri, from development time and survival to the effect of temperature and humidity. From a tick’s point of view, the landscape consists of the lizard hosts and their nocturnal refuge sites (Fig. 3.2). The model is structured on randomly distributed refuges in a 1 × 1-km2, used by lizards whose home ranges are centred on a randomly chosen refuge. All refuges within 100 m of the centre refuge are included in the home range. There are two timescales in the model. On the short timescale, movement of 48 3 Distribution of Aggregations lizards, birth, development, and the death of ticks are modelled each day. A series of days is then aggregated into a season, which is 210 days long, under the assumption that autumn/winter temperatures are too low for tick activity. Ticks experience over- wintering mortality, and a simplified form of population dynamics among lizard hosts also takes place between seasons. Within a single day, the model goes through several steps in the following order: ticks board lizards, lizards move between refuges, engorged ticks disembark from lizards, and ticks develop (Fig. 3.2). Tyre et al. (2006) conducted the sensitivity analysis of the model on the 28 parameters by forming a Latin hypercube. They successfully found some parame- terisations that replicated a landscape-level pattern of an increase in the aggregation coefficient with an increasing mean abundance of ticks. This pattern is independent of particular initial conditions on the landscape, which allows for analysing this system, despite limited empirical data of landscape distributions. Most importantly, they generated these patterns based only on biological knowledge of individual- level life history and interactions.

3.4 Ideal Free Distribution

An individual-based approach is based on simulation models that are used to explain the distribution of aggregations between habitat patches, based on strategies of movement and other individual traits. It is thus possible to track the process or mechanism that produces the pattern of distribution. Ideal free models follow a different strategy than individual-based models. They are analytical models that deal with the distribution in the equilibrium instead of simulating the process which is responsible for the equilibrium pattern. This is conceptually analogous to an evolutionary stable strategy (Maynard Smith 1982) because the equilibrium is reached when no individual can do any better by moving from its habitat patch to another patch (Tregenza 1995). In this sense, ideal free distribution models may be considered a special case of game theory models (Krivan et al. 2008).

3.4.1 Habitat Selection and Intraspecific Competition

In Sect. 2.5, we analysed the distribution patterns between foraging sites of solitary individuals. We predicted which patches avoid and in what sites individuals remain longer if the decision criterion is to maximise the rate of obtaining resources. In the models that will be described in this section, individuals forming aggregations behave in the same way, i.e. the models assume that individuals behave adaptively, by remaining in the patch or habitat that provides maximal suitability. Thus, mechanisms for resource and habitat selection are the same as in the case of individual consumers (Chap. 2). Now we can incorporate the problem of intraspecific competition or the use of density-dependent foraging. Overall, when analysing the distribution pattern of the 3.4 Ideal Free Distribution 49 organisms of a species in the range of environments where survival is possible, it can be observed that the distribution is rarely random, but certain areas of the range tend to concentrate greater population density. This heterogeneity in the distribution of aggregations is usually associated with the existence of different amounts of available resources between types of environments. In such a context in which several individuals are using the same resources in a habitat, competition appears inevitable. The notion of competition has been instrumental in the development of the science of ecology. An individual’s competitive success is determined both by their physical strength and by the behavioural strategies they use when competing with others. The reduction of an individual’s resource gain as a result of competition is called interference. Several types of competitive mechanisms produce interference. Simple exploitation competition or pseudo-interference of co-specifics produces a decrease in the individual consumption rate, without differences or interactions between competitors. This is the result of the aggregative response in patches of resources. Consumers do not need to meet each other and use similar foraging rules to decide whether and for how long to exploit a resource, irrespective of the behaviour of other competitors. Milinski and Parker (1991) prefer to distinguish between exploitation and scramble competition. They use the latter term only when each competitor can assess that others exploit the same resource and can increase its foraging effort to get as much as possible the resource only by trying to exploit it. They may increase their efforts, dependent on what their competitors do. An example would be to increase the speed of travelling between patches above the optimal value in energetic terms, in order to get faster access than its competitors to the best patches. Most studies use exploitation and scramble competition interchangeably, and the same criterium is followed here. A second type of competition occurs when the population can be divided into categories of individuals or phenotypes with different competitive abilities. Sex, age, and dominant hierarchies within a sex are common forms of this type of competition. A scramble competition with different ‘phenotypes’ occurs, for example, when sexual dimorphism in size allows males to reach the best patches of resources faster than females. There is still no direct interaction between competitors. Contest (direct or aggressive) competition occurs when there are direct negative interactions between the members of a population. This includes kleptoparasitism and territorialism. Contest interactions may occur with or without different competitive abilities between categories of competitors. This chapter discusses only the effect of the scramble competition, while contest competition will be discussed in the next chapter.

3.4.2 Fretwell and Lucas’ Model

The a priori method of studying habitat selection generates predictions about the distribution of organisms that select their resources with rules that maximise payoff. The first a priori model was developed by Fretwell and Lucas in 1970 and is known 50 3 Distribution of Aggregations

Table 3.1 Ideal free distribution approach No. of consumers Expected gain Obtained gain A B C A B C A B C 0 0 0 6 3 1 6/1 = 6 3/1 = 3 1/1 = 1 1 0 0 6 3 1 6/2 = 3 3/1 = 3 1/1 = 1 2 0 0 3 3 1 6/3 = 2 3/1 = 3 1/1 = 1 2 1 0 3 3 1 6/3 = 2 3/2 = 1.5 1/1 = 1 3 1 0 2 3 1 6/4 = 1.5 3/2 = 1.5 1/1 = 1 4 1 0 1.5 3 1 6/5 = 1.25 3/2 = 1.5 1/1 = 1 4 2 0 1.5 1.5 1 6/5 = 1.25 3/3 = 1 1/1 = 1 5 2 0 1.25 1.5 1 6/6 = 1 3/3 = 1 1/1 = 1 5 2 1 1.25 1.5 1 6/6 = 1 3/3 = 1 1/2 = 0.5 5 3 1 1.25 1 1 6/6 = 1 3/4 = 0.75 1/2 = 0.5 6 3 1 1 1 1 Ten consumers distribute themselves one by one between three habitats (A, B, and C) with different initial suitabilities (6, 3, and 1 food items per minute). Each consumer goes to the habitat that provides maximal expected gain. When all consumers are distributed (last file), 6 occupy habitat A, 3 occupy habitat B, and occupy habitat C, each one gaining 1 item per minute as the ideal free distribution model. It was intended to be applied to the use of feeding sites, but as we will see, it has also been applied in the use of other resources. It is a simple model that applies to the following situation: • Resources have a patchy distribution, and patches have a constant rate of flow of resources. • Animals are ‘ideal’ in the sense that they ‘know’ the offer of resources available everywhere in the environment. • Animals are ‘free’ to move between sites since no travel costs between sites or social spacing behaviour. • Each animal remains at the site that offers the highest rate of gain of resources. Table 3.1 shows an example for deriving the predictions of this model. Each row represents the changes that occur when a new consumer enters the environment. There are three sites with different initial quality (expected profit) (A > B > C). In this example, consumers wait for resources that arrive into patches and are consumed immediately by any individual with equal probability. Patches differ in the input rate of resources. Finally, when all consumers already entered the environment, the qualities 3.4 Ideal Free Distribution 51

Fig. 3.3 Graphical representation of ideal free distribution model. The first consumers that arrive at the area enter into habitat I, the site with the initially highest density of resources. Foraging activity decreases good habitat suitability to point A, when suitability is equalised between habitats I and II. Foragers use both habitat types until they reach suitability B that is the same as the initial quality of habitat III. New consumers will occupy any of the three habitats of the sites are balanced and the distribution of consumers is proportional to the initial quality of the sites. Fretwell and Lucas (1970) formalised this process by using the following method. Patches are organised so that the first patch has a higher intrinsic suitability (measured as individual payoffs) than the second patch and so on. Consider

H habitat patches with qualities Vi:

VQ=-hn() (3.2) iiii

Here Qi is the initial suitability, ni is the population density in patch i, and hi is an increasing function that is 0 when patch i is unoccupied. In particular, Q1 >…> QH. As the density of individuals in patch 1 increases, the individual payoff there will decrease and, at a certain critical density, patch 2 will have the same payoff. Then, for a higher population density, both patches will be occupied and the payoff in either patch will be the same. As density continues to increase, the third patch will give the same payoff and it will also be occupied. Thus, all occupied patches will provide the same payoff and no unoccupied patch will have a higher payoff. This pattern is the ideal free distribution. Fretwell and Lucas (1970) showed that, for each positive total population size M, there will be a unique ideal free distribution. Figure 3.3 describes this inference graphically. The example shows three sites that, in the absence of competitors, differ in quality. As the number of consumers using a site increases, the quality of this site diminishes. The first individuals to use the environment will be established in the best place until it reaches point A, when the quality of the best place equals the intermediate site that has no competitors. Since A, both sites will be used. As more competitors arrive, they are 52 3 Distribution of Aggregations established in both sites to point B, moment in which the quality of both sites equals the worst site. The new individuals arriving also start using the worst place. The model is very simple and therefore does not capture the complexity of the natural situation. Nevertheless, it has been a cornerstone for the development of the theory of habitat selection. The model was tested repeatedly with the general results: (1) at equilibrium, populations were distributed according to the law of matching; (2) however, the individual gains were heterogeneous, with some individuals getting resources at a greater rate than others (Milinski and Parker 1991). Key insights into the ideal free distribution approach include the fact that it explains animal distributions in terms of individual decisions and that, as the density of foragers on a patch increases, the suitability of that patch decreases. Other researchers used similar forms of rationale when investigating resource competition and animal distribution in different ecological contexts (Brown 1969; Orians 1969; Parker 1970; Rosenzweig 1974). This convergence is an interesting historical moment in which ecology meets an adaptationist programme in the search for general principles to answer the big questions of ecology.

3.4.3 Ideal Free Distribution with Interference

Sutherland (1983) introduced an important modification to the original model by incorporating the interference constant and making it applicable to a wide spectrum of consumer–resource relations. The interference represents the decrease in individual gain rates by obtaining resources in areas with increasing densities of conspecifics.

In a patch i with ni competitors, let the gain rate Gi of a consumer be Q GQ()==i C (3.3) ii m ni

Qi is the intrinsic suitability measured as individual payoff of patch i, m is the interference constant, and C is a constant if the assumption that the payoffs are constant for all competitors and patch is met.

3.4.4 Ideal Free Distribution with Unequal Competitors

Individuals in a population can be divided in categories (‘phenotypes’) depending on their abilities to scramble competition. For example, old individuals may forage more efficiently than young ones due to their wider experience of searching and capturing food, and large males can move faster between and within patches than small females. Note that these different competitive abilities between phenotypes (denoted with letters A, B, C, etc.) are not related to aggression or direct 3.4 Ideal Free Distribution 53

Fig. 3.4 Interference when phenotypes differ in searching efficiency (phenotype scales intercept) or in direct interactions (phenotype scales slope) encounters between individuals. Differences between phenotypes may be of two types: competitors may differ in search efficiency or they may suffer different levels of interference. In the first case, phenotypes differ in the intercept of the interference function, while in the second case, they differ in slope (Fig. 3.4). Sutherland and Parker (1985) described the expected distribution in the equilibrium of aggregations of different phenotypes. Let KAi be the competitive weight of phenotype A in patch i. With differences in intercept between phenotypes, the payoff G for an individual of phenotype A in patch i will be

æ K ö G = Ai Qn-m (3.4) Ai ç ÷ ii è Ki ø With difference in slope within phenotypes, the payoff will be

æ K ö G = Ai Qn-mK(/iAK i ) (3.5) Ai ç ÷ ii è Ki ø In (3.4), differences in phenotype affect the intake rate while in (3.5) they scale the effect of m.

3.4.5 Other Ideal Free Models

In addition to this general change, numerous further developments have been incorporated into the basic idea of Fretwell and Lucas, other variables not considered by the original model: the risk by the variance in the availability, limitations in animals 54 3 Distribution of Aggregations

Fig. 3.5 Schematic representation of the increase in the number of habitats used with increasing depletion

in the estimation of the quality of the sites and the need to learn, sample or sensory limitations, and the costs associated with relocation, such as the energy cost of travel between sites and the risk of predation by moving. Several population distribution models take into account the effect of resource depletion, for example, those developed by Royama (1971), Bernstein et al. (1991a, b), and Sutherland and Anderson (1993). One of the main predictions of these models is that consumers initially concentrated on the set of better places but, as these are exhausted, the range of sites used is higher (Fig. 3.5).

3.5 Distribution at Equilibrium: The Habitat-Matching Rule

So far in this chapter, three types of models that are applicable to aggregations were briefly described. Site suitability models are statistical descriptions of the relationships between two data sets, so they can only generate predictions of distribution within the environmental and competitive contexts in which the models were built. Individual-based models are mainly concerned with the process rather than the pattern, so the distribution at equilibrium is critically dependent on the movement rules of individuals that have been incorporated into the model. Thus, there is not a general theory from which to derive an expectation of the distribution under certain environmental or competitive context. Ideal free models are specially designed to 3.5 Distribution at Equilibrium: The Habitat-Matching Rule 55

Fig. 3.6 Predictions of the ideal free distribution model, with different values of interference m (modified from Sutherland 1996)

produce predictions about distribution at equilibrium. For Sutherland and Parker’s (1985) model the equilibrium distribution is summarised in the following equation:

Em/ æ Qi ö Ni = ç ÷ (3.6) è C ø where E is the mean exploitative competitive ability of all individuals divided by the ability of individual i. When all individuals have the same competitive abilities and the degree of interference is maximal, then 1 N = Q (3.7) iiC This is the habitat-matching law that states that individuals distribute in aggregations the sizes which are proportional to the suitability of patches. This pattern of distribution is expected to occur when all individuals get the same payoff, independently of location.

There are other types of equilibrium, depending on values of m and Ki (Fig. 3.6). When m = 0, there is no competition, so all consumers should be concentrated in the best patches. With intermediate values of interference and no difference in competitive abilities, the use-suitability function has different shapes, but there is always a trend towards a smaller amount of users of the poor patches, i.e. overmatching. Sutherland and Parker (1985) proposed four different scenarios: (1) individuals can differ in the effect of interference on intrinsic quality of patches, with ratios of competitive abilities constant for all habitat types; (2) foragers can also differ in foraging efficiencies, but some phenotypes are more efficient in a type of habitat, which will usually be the best kind of site; (3) the third case is when interference affects m; and (4) foragers differ simultaneously in both the ordinate and the slope of the interference function. Two types of outputs are expected at equilibrium. Scenarios (2) and 56 3 Distribution of Aggregations

(3) predict one pattern of distribution, with best patch types occupied by the most competitive phenotype. In this case, animals are spatially separated in different habitat types. Scenarios (1) and (4) produce a range of possible outputs: one of them is the habitat niche segregation pattern. The others are different proportions of good and poor competitors distributed in good and poor habitats. Although the number of individuals that correspond to each phenotype can differ considerably between different solutions, foraging success of good exploitative competitors will be always higher than that of poor competitors and in the same proportion. In this case, animals of different types can share habitat but are separated by different levels of success in the use of habitats. Chapter 4 Distribution of Societies

4.1 Introduction

So far we have seen how the distribution of environmental resources, mainly food and shelter, determines the distribution of individuals and aggregations. Assuming that individuals use resources based on decision rules that maximise their fitness, it is expected that these exploiters distribute their time between patches of resources within the habitat, in direct relation to the quality of those patches. In this chapter we will see how social behaviour affects the distribution of animals, both favouring a contagious distribution of individuals and, the opposite effect, imposing social spacing. A society is defined operationally as an aggregation of individuals with some kind of social organisation that can include grouping and territoriality. In social groups, individuals benefit by being surrounded by conspecifics, while in territoriality, social spacing promotes a uniform distribution. This chapter links distribution ecology with evolutionary ecology, more specifically with social evolution. The chapter begins with a description of the most common dispersion patterns. Two sections follow that describe the mechanisms that induce the formation of social groups and spacing. More specifically, they analyse the ecological conditions that favour the expression of social behaviours in animals. Then two sections describe analytical models on the distribution of groups and territories. Both approaches are based on the seminal work of Fretwell and Lucas (1970), along with other authors who in the late 1960s and early 1970s began the integration of the principles of population ecology and evolutionary ecology in the context of habitat selection studies. The chapter ends with a section that analyses the distribution of organisms during the breeding season when reproductive behaviour predominates.

M.H. Cassini, Distribution Ecology: From Individual Habitat Use 57 to Species Biogeographical Range, DOI 10.1007/978-1-4614-6415-0_4, © Springer Science+Business Media New York 2013 58 4 Distribution of Societies

This chapter focuses on animals; however, many ideas on the adaptive value of social behaviour can be applied to species without a nervous system, such as plants and microorganisms. Ciszak et al. (2012) have proposed that, to exploit soil resources optimally, plants have developed intricate root systems that are characterised by complex patterns and based on the coordinated group behaviour of the growing root apices. It has been shown that plants can distinguish between self and nonself roots (Gruntman and Novoplansky 2004) and that the sensory information collected by one plant is shared with neighbouring plants to optimise territorial activities, including competitive behaviours and symbioses with fungi and bacteria (Heil and Karban 2010; Novoplansky 2009). Similarly, there is a growing literature on the social behaviour of microorganisms. For example, West et al. (2006) propose that microorganisms can communicate and cooperate to perform a wide range of behaviours, such as dispersal, cooperative nutrient acquisition, biofilm formation, and quorum sensing. Microbiologists are rapidly gaining a greater understanding of the molecular mechanism involved in these behaviours and the underlying genetic regulation. Biofilms, which are aggregates of microorganisms that are adhered to a surface and frequently embedded in a matrix, have rapidly become the paradigm for the study of grouping in this taxon. For example, Kreft (2004) studied the conflicts of interest in biofilms, under the assumption that the key problem for the evolution of grouping is the fitness cost for those individuals investing in cooperative behaviour.

4.2 Dispersion Patterns

In a homogeneous environment where resources are distributed at random, if the organisms are distributed strictly in terms of decision rules as we saw in previous chapters, then the dispersion of aggregations of organisms in a given time would have to show a random pattern. This is because, according to those rules, each individual moves according to its own encounter rate with resources. However, when analysing the dispersion pattern of an aggregation of individuals within a homogeneous habitat, it is rarely random, but is often clustered or uniform (Fig. 4.1). Most organisms are not distributed in space independently of other organisms of the same species. This type of distribution cannot be explained only by the distribution of the resources; it has to do with the characteristics of the social organisation of the species involved. Animals have developed social behaviours that maintain group cohesion or promote social spacing somewhat independently of short-term environmental conditions, while environmental conditions affect the intensity of the expression of social behaviour and therefore the dispersion patterns. It is necessary to introduce some concepts developed by evolutionary ecology about social behaviour in order to understand certain patterns of distribution which cannot be explained solely as an individual response to environmental conditions. 4.3 Grouping 59

Fig. 4.1 Schematic representation of the distribution of individuals within and between sites of ‘good’ and ‘poor’ quality when there is social behaviour involved

4.3 Grouping

A group exists when the animals stay close through some form of social attraction. This definition distinguishes groups from aggregations, in which the proximity of individuals is triggered by some environmental factor, such as a food source. The social attraction can be expressed by some kind of active social interaction or simply provided by the proximity of a conspecific. It can recognise at least five benefits associated with the formation of groups: anti-predation defence, improved feeding, temperature control, reduced transportation costs, and encounter of a breeding pair. This section will discuss the first four while the latter will be discussed in Sect. 4.5. The most common disadvantages of forming groups are competing for resources and contagious diseases. There are also costs related to the formation of dominance hierarchies and predation. 60 4 Distribution of Societies

4.3.1 Protection Against Predators

Foraging behaviour can be divided into eight components: • Search: Navigate in the environment in order to find food • Encounter: Reach a distance from a prey at which it is possible to detect it • Detection: Distinguish the prey from the context • Choice: Identify prey and its productivity • Persecution: Approach or attack the prey • Capture: Reach and grasp the prey • Handling: Divided into edible and chewing • Digestion: Ingest and process food Being part of a group can reduce the effects of all components prior to capture, in ways that will be briefly explained below. Thus there is a clustering effect for each of these five components: effects of avoidance, encounter, vigilance, dilution, and confusion. The effect of avoidance occurs when the predator spends some time searching for prey without being able to detect it. If prey is grouped, then the predator will take longer to find it. If the predator consumes a single prey in each group, then this effect will lead to a decrease in the so-called productivity (energy/time of encounter). According to the classical diet model, this drop in productivity may determine that the predator decides to feed on another type of prey (Stephens and Krebs 1986). An animal that is part of a social group may have a lower likelihood of being found by a predator by the encounter effect. This effect is because the individual likelihood of being detected by a predator decreases with increasing group size. This effect is fulfilled when the probability of detecting a prey group is independent of group size. There is little evidence of the existence of this effect, although it is well established that the psychophysics of sensory detection generates an effect for which the probability of detection of an object is independent of size, after exceeding a certain size. If there is no encounter effect, then the probability of detection is higher in a group that is alone. Specifically, if a group of size N is N times more likely to be detected, then there is no advantage in clustering for this effect. And, as may occur in many cases, if the detectability increases faster than the linearity with the size of the group, then forming groups will be a disadvantage from the standpoint of the encounter effect. Much prey that forage in open areas often scan for predators, leading to the vigilance effect. An animal that is part of one of these groups gets the benefit of this vigilance shared with other group members. A study case on this effect is provided. Wild guinea pigs Cavia aperea are Neotropical herbivorous rodents (Lacher and Cassini 2001). Typically, their environment in the pampas grasslands of South America has a cover zone with high and dense vegetation, which they use as a protection from predator attacks (Lacher and Cassini 2001), and a foraging zone where predation risk would depend, among other environmental factors, on vegetation 4.3 Grouping 61

Fig. 4.2 Dilution effect in wild guinea pigs. Mean foraging bout duration and mean scanning rate when cavies are foraging alone or in a group. Asterisks show statistical differences (modified from Cassini 1991)

height (Cassini and Galante 1992). In short vegetation, cavies make a series of foraging bouts, each one consisting of a relatively brief visit to the foraging zone close to the cover. While foraging, cavies periodically stop grazing and elevate their heads from over the line projected by their backs, during which moment they probably scan for predators (Rood 1972). Cassini (1991) studied the effects of distance to cover and of socialisation in cavies’ foraging behaviour and found support for the hypothesis that foraging in groups improves foraging efficiency. As expected from this hypothesis, shorter residence times and greater scanning rates were observed when cavies forage alone than when in a group (Fig. 4.2). The so-called dilution effect determines that the probability of an individual being chosen by the predator in a group decreases with group size. This is because predators are usually satisfied after consuming low amounts of prey. If the predator consumes only one prey on each attack on a group, then the probability for a prey of being chosen by the predator decreases at the ratio 1/N, where N is the size of the group. This effect is probably a common ecological factor to explain the formation of groups in vertebrates. Hamilton (1971) developed a model called the selfish herd, which is a simple explanation of the evolution of group formation based on the dilution effect. A prey that is away from his group is more likely to be attacked by a predator than an individual within the group. Therefore, the probability of being attacked is diminished by simply reducing the distance to conspecifics, as long as the predator consumes only one or a few prey per attack. 62 4 Distribution of Societies

The confusion effect is produced when for a predator it is easier to pursue a single moving object than deciding among several that are moving. This idea was formalised in a neural network model (Krakauer 1995). This model not only predicts the expected effect but has other ramifications, such as, for example, the benefit increases. First, the advantage increases with the size of the group according to a negative exponential function so that increased benefit reaches an asymptote. Second, the more compact the group, the more protection there is. Third, the effect is more effective when prey are more similar to each other. There is partial evidence for these predictions. Group formation may decrease the likelihood of being caught by a predator because defence cooperation exists among group members. The most obvious example of this defence effect is that of social insects. A wasp, a bee, or an ant alone is hardly able to avoid the attack of a vertebrate predator. However, the coordinated defence of a colony of social insects can force a predator away from the colony and even cause death. Among vertebrates, there are numerous examples of cooperation in defence, such as the group attack performed by colonial birds towards aerial or terrestrial predators and defensive formations of elephants.

4.3.2 Finding Resources

Among the benefits of group formation related to foraging is the increased likelihood of finding places with food. When food is highly abundant in concentrated patches, but these patches are randomly distributed in time or space, information provided by conspecifics can be of great benefit. Two typical examples are colonial seabirds and scavenger birds. In the first case, the fish on which they feed normally form large schools that are hard to find, but when determining their location, they may provide food for a large number of birds. The carcasses used by scavengers also are randomly distributed. In these species, members of a group use the information provided by other members to decide where to search for food. This theory proposes that the communal resting places of scavengers and seabird colonies function as food information centres (Ward and Zahavi 1973). At these sites, hungry animals would be able to find food by following other individuals that start to feed and show signs of having been successful in encounter- ing food. The theory of information centres explains the advantages of clustering in social insects like bees, in rats, and in some birds. In the case of colonial birds, this has generated controversy because it is rare that birds in the colony follow individuals returning to feeding areas (Mock et al. 1988; Richner and Heeb 1995). Another advantage linked to resource information is in the case that members of a group can use the information from the others to determine whether a foraging site is exhausted and must change sites. In other words, an individual can assess the quality of a site without having to explore the site itself (Boulinier and Danchin 1997; Valone 1989). There have been numerous laboratory experiments with social birds like starlings in which creative experimental designs have identified 4.4 Defending Territories 63 that social learning skills are very important in the use of this public information (Smith et al. 1999; Templeton and Giraldeau 1996). Group feeding also allows the capturing of large prey or prey that are difficult to capture. The typical case is that of many social carnivores, especially canids. These species can capture prey of a body size significantly greater than their own, such as elk Cervus canadensis and American buffalo Bison bison, thanks to the coordinated use of group hunting techniques. This is also common in fish and sea mammals that feed on shoals. In this case, the fish are herded into shallow areas or are surrounded until they form a compact group that is easier to attack. Another advantage of grouping that is related to foraging is when it increases the chances of effectively defending a foraging site. This phenomenon has been observed in several species of birds. It has also been proposed that a spatially coordinated foraging allows resource recovery in those places already visited, particularly for herbivores. This hypothesis has been postulated for both mammals and plant parasitic insects.

4.3.3 Changing Local Environment

Cooperation within groups also allows some aspects of the local environment to be modified, which is especially noticeable in colonial insects that built nests (bees, ants, termites). One problem with endotherm animals is the maintenance of the temperature in cold regions or periods. Some species, such as bats and penguins, save energy by forming compact groups. It has also been proposed that in the case of fossorial rodents, the presence of many individuals not only reduces the surface exposed to the heat loss but can increase the environmental temperature. Another advantage related to the modification of the micro-environment around the group has been observed in large concentrations of larvae of some insects, for which there seems to be a benefit in reducing the surface exposed to evapotranspiration. Another way to change the local environment occurs during travel in groups. Experiments have shown that staying behind another group member reduces energy costs by reducing friction with the water or the air, especially through the use of the vortices created by the movement of others. This benefit has been described in birds, schools of fish, and even lobsters moving in groups. For birds, it has been postulated that this benefit operates only in relatively large birds and is best when the birds follow a V-shaped formation. It has been shown that this type of arrangement reduces the metabolic cost of flight.

4.4 Defending Territories

We have seen how the formation of groups, i.e. the active search for the proximity of other conspecifics, may increase the chances of survival of the members of the group. However, under certain environmental conditions, it is more beneficial to 64 4 Distribution of Societies follow the opposite strategy, i.e. to actively move away from conspecifics. A territory is defined as an area of exclusive use that is defended by fighting or is delimited using marks that other individuals detect and which deter them from entering. Normally there is a degree of overlap between territories. A distinction should be the territory of the so-called home range (Sect. 2.3). Animals use different techniques to defend their territory from the invasion of intruders. The most obvious is attacking conspecifics that cross the boundaries of their territory. But many species use strategies that do not necessarily involve territorial fights. Many mammals use scent marks: everyone knows the example of cats and dogs that use urine as a territorial delimiter. Other mammals have special glands that secrete substances with the specific function of marking the boundaries. Among birds, vocalisations are a widely used method of territorial defence. In fish, visual displays are common. A notable example of demarcation is the electrical signal of eels. The resource defended in a territory varies, but in most cases it has to do with food or shelter, including nesting sites for birds. During the breeding season, it is common for males to defend territories with resources needed for breeding females. Resources should be distributed in a heterogeneous and predictable way to be economically defensible. A concentrated and stable distribution enables the defence of the resource without excessive cost. Territorial defence also appears when animals use fixed shel- ters and territory extends around the shelter. The use of a restricted area also has the advantage of familiarity with the characteristics of the territory. This information facilitates both the exploitation of food resources and the search for protection from predators. But defending a territory inevitably leads to expenses associated with the time and energy costs of defence. Patrolling, singing, and producing marks are expensive. Territorial fights have the additional cost of receiving injuries.

4.5 Habitat Use and Reproductive Systems

A reproductive system is the way in which a population is organised to reproduce (mate and breed). It includes forms of courtship and competition, duration of couples, and form and duration of parental care. There are four basic types of reproductive systems, distinguished primarily by the number of breeding pairs that individuals of each sex have on average during a breeding season: • Monogamy: Each male breeds with a female. • Polygyny: Some males mate with more than one female. • Polyandry: Some females reproduce with more than one male. • Promiscuity: Males and females have multiple mating. Traditionally, the reproductive systems were classified according to the social relations established during reproduction. For example, over 90% of birds form monogamous pairs and in many cases these pairs persist throughout the entire reproductive life of the male and female. Nowadays, it is known that these relationships do not always correlate with genetic ties. This is seen clearly in the case of birds. In genetic 4.5 Habitat Use and Reproductive Systems 65

Fig. 4.3 Schematic representations of the factors that affect breeding organisation in mammals. Because female reproductive success tends to be limited by resources, whereas the male reproductive success tends to be limited by access to females, it is expected that female dispersion depends primarily on the distribution of the resources, whereas male distribution depends primarily on female dispersion studies on the paternity of nestlings, the occurrence of multiple mating was frequently noted, so that these couples often care for eggs and chicks that are from other parents. The mating system of a population depends on the phylogenetic history of the species and the environment (Emlen and Oring 1977). The phylogenetic history determines the amount of parental investment that both sexes can make. There is considerable variation among taxa in the relative amount of investment of the sexes. In mammals, long gestation and lactation require a large investment from females. In most species, males only invest in sperm. In birds, males often invest as much as females and feeding the offspring is the responsibility of both sexes. In teleost fish almost 80% have no parental care. Where there is care, one sex is responsible and it is common for the male to care for the offspring. When females are phylogenetically conditioned to perform almost all parental investment, as in the case of mammals, the reproductive system depends first on the distribution of resources and then on the distribution of females (Fig. 4.3). When resources are distributed in stable clusters, males may defend territories containing those resources and attract females. Under these conditions, a breeding system called resource defence polygyny evolves. In contrast, when resources have a random temporal distribution, the females are not set in fixed locations but move in stable groups. Under these conditions, males may accompany the movements of females and defend groups or subgroups of females called harems. One system that is rare but well studied is called lek. In this system, males congregate in arenas that are visited by breeding females for the sole purpose of copulation. In the leks, males defend minimal territories and only a few of them get most of the copulations. The degree of clumping can be substantially larger during than outside the reproductive period, so the models described in Chap. 2 are frequently inappropriate to explain the high gregariousness observed during reproduction. Figure 4.3 indicates that, in order to understand the distribution of social groups or territories during mating periods, it is necessary to investigate both the environmental determinants of 66 4 Distribution of Societies

Table 4.1 Life history traits of two families of pinnipeds. Two phocids (grey seals and elephant seals) are presented in a separated column because the diverge from the rest of phocids H. grypus Trait Otariids Most phocids and Mirounga Number of species 15 15 3 Female density Gregarious Solitary/moderately gregarious Gregarious Mating system High polygyny Slight polygyny High polygyny Male parental care No No No Sexual dimorphism High Low High Breeding substrate Land Ice or land Land Mating substrate Land Water Land Lactation Long Short Short Breeding site availability Low High Low Coloniality Yes No Yes

female distribution and male and female reproductive strategies. An example of the effect of reproductive behaviour on group distribution is provided. The order Pinnipedia (sea lions, seals, and walruses) is a mammalian group with some special traits. They show a combination of marine feeding and terrestrial breeding, while parturition and mating are seasonal and highly synchronised (Bartholomew 1970). The two main families, Otariidae (sea lions and fur seals) and Phocidae (seals), show contrasting distributions during the reproductive period (Table 4.1). While most otariids form reproductive colonies with animals forming compact clusters of hundreds and even thousands of individuals, most female seals reproduce in isolation. The distribution of coast with suitable breeding sites cannot entirely account for the extreme clustering observed in some species. It is normal to find empty shorelines with similar characteristics to those of the rookeries that support high densities of pinnipeds, even in close proximity (e.g. Boness 1991; Le Boeuf 1991). Three main hypotheses have been proposed with regard to the ecological factors that influence female clustering during reproductive periods in this species: (1) reduce predation risk, (2) find mates, and (3) avoid male harassment. Cassini (1999) discussed these hypotheses and found support for the latter. Interspecific comparisons indicate that there is a strong relationship between the substrate (on land or in the water) on which copulation takes place, the intensity of male harassment, the degree of polygyny and sexual dimorphism, and population distribution. Mating harassment is a common phenomenon in nature, although its ecological and evolutionary consequences have not always been sufficiently appreciated. Male coercion has been described in several groups of mammals including pinnipeds, cetaceans, ungulates, canids, and primates (Cappozzo et al. 2008). Reproduction of social pinnipeds shows characteristics that predispose females to being affected by sexual harassment, including high sexual size dimorphism and low mobility on land, and that copulations occur when pups are only a few days old. If the newborn pup 4.6 Distribution of Groups: Ideal Free Approach with Allée Effect 67

Fig. 4.4 Behaviour in relation to female density in sea lions Otaria flavescens in Patagonia, Argentina. Male harassment that is received by a female in a reproductive group decreased when the number of females surrounding her increased (modified from Cassini and Fernández-Juricic 2003) becomes separated from the mother, the probability of reunion is expected to be low, and death is likely from starvation, aggression, or crushing (Vilá and Cassini 1990). Dominant males normally defend a group of females or a territory with females, while satellite males use alternative reproductive behaviours on the periphery of the large patch of females. A breeding group is composed of a resident, dominant male and the females surrounding it (Cassini and Fernández-Juricic 2003). For females, staying in a breeding group under the influence of a resident male can reduce the reproductive cost of harassment by satellite males since resident males defend their harems against intruders (Cassini 1999). However, males holding female groups can also harass the females within their groups. Female gregariousness can reduce this harassment by dilution effect, which reduces the probability of female interactions with the resident male as breeding group size increases (Fig. 4.4). If harassment by dominant males is a relevant force in modulating female gregariousness in this species, it is expected that males holding larger breeding groups would interact less frequently with females than males with small groups, and the frequency of interactions with the resident males experienced by individual females is lower in large than in small groups.

4.6 Distribution of Groups: Ideal Free Approach with Allée Effect

The benefit of group formation is one of the mechanisms that explain the Allée effect and usually refers to a decrease in the population growth rate at low densities (Courchamp et al. 2008). In populations without the Allée effect, growth rates decrease monotonically with density and there is one equilibrium point when the 68 4 Distribution of Societies

Fig. 4.5 Relationship between population growth rate and population density. (a) Without the Allée effect, there is a continuous decrease in growth with density and only one stable equilibrium point when r = 0. (b) With the Allée effect, the function adopts an inverted U shape. As density increases, social benefits and other mechanisms improve growth until a maximal point is reached and growth begins to decrease due to intraspecific competition (modified from Courchamp et al. 1999)

Fig. 4.6 Relationships between density and costs, benefits, and individual fitness in populations without (a, b) and with (c, d) Allée effects (modified from Stephens et al. 1999) population reaches carrying capacity (Fig. 4.5a). In populations with the Allée effect, growth rates increase at low densities, then decline due to competitive interactions (Fig. 4.5b). The result is a dynamic system with two equilibrium points, the first one unstable (Courchamp et al. 1999). Figure 4.6 shows a schematic representation of the costs and benefits of population aggregations and how the classical shape of the interference function is altered by the Allée effect (Stephens et al. 1999). The benefits of group formation usually 4.6 Distribution of Groups: Ideal Free Approach with Allée Effect 69

Fig. 4.7 Relationship between habitat quality and local population density (modified from Bernstein et al. 1991a, b). (a) Two habitats are represented; habitat A has higher quality than B. (b) The process of colonisation is initiated when the first individuals found a new population in the original best habitat; when abundance reaches 12 individuals, the quality of the best habitat matches the maximal possible quality of the poor one. (c) When the population reaches 12 individuals, the quality of the original good habitat matches the initial quality of the poor one. (d, e) A migration process occurs from habitat A to B until the quality of the original poor habitat reaches its maximum. (f) Both colonies are equally occupied are a decelerating function of density. For example, by approaching a conspecific, a solitary prey reduces its chances of being attacked by a predator by half due to the dilution effect; however, a prey in a group of hundreds will receive a low payoff if a new member joins. Several studies have investigated the effect of the Allée effect on habitat distribution with an ideal free approach (Greene and Stamps 2001; Morris 2002; Cassini 2011a) after the seminal paper of Fretwell and Lucas (1970) introduced the problem (Sect. 3.4). Figure 4.7 represents a hypothetical example of the formation process of 70 4 Distribution of Societies two populations in an environment that contains two patch types in a matrix of unsuitable habitats (Bernstein et al. 1991a, b). Patch of habitat A has a higher initial quality than habitat B (Fig. 4.7a). According to a principle of maximising payoff, the first individuals occupy the best patch (Fig. 4.7b). When the first population reaches 18 individuals (Fig. 4.7c), patch types match qualities, so the next animal that enters the system will be the founder of patch A. Due to the Allée effect, the entrance of that animal slightly improves patch quality so that it will be advantageous for another animal from the original population to migrate to the new one. A positive feedback loop results in rapid migrations from A to B (Fig. 4.7d). This phenomenon lasts until patch qualities are matched again (Fig. 4.7e). From this point, both patches are equally occupied (Fig. 4.7f). A key feature of Fretwell and Lucas’s model is what Cassini (2011a) called the Allée paradox: there is a range of local population densities at which local individual fitness is less than the potential mean gain that could be obtained in the environment; however, individuals cannot disperse. They must ‘wait’ until the density drops to a value that is less than the value for one individual in the new patch. In the example of Fig. 4.7, when density in the originally good patch reaches 18 individuals, mean fitness is lower than if they distribute more homogeneously, e.g. 13 individuals in the good patch and five individuals in the poor one. The ideal free distribution model with the Allée effect predicts the following ecological consequences: Instability of local densities: When the spatial dimension is incorporated to models of population dynamics with Allée effect, the most important result is that local densities in patches shift suddenly as a function of overall population size. When patch quality in an initially good patch approaches the quality in an empty, initially bad one, a very small increase in population size may result in a very large change in the distribution of abundances in populations with Allée effect. Thus, a smooth asymptote in population density is not reached. This prediction differs from the prediction obtained for single populations, in which population growth under the Allée effect shows only two equilibrium points. Here, the equilibrium point varies depending on the overall population density and the number of patches. This is a typical result for game theory models, for which the equilibrium depends on an optimal balance between alternative strategies in different social contexts. Initial period of latency and very fast initial population growth: When overall population density is high, colonisation of empty patches follows a particular pattern under the Allée effect. Solitary immigrants are expected to reach empty patches; however, they fail to produce a population due to the susceptibility to extinction produced by the Allée effect. These occasional occupants create a baseline number of individuals prior to population formation. When a nearby source population reaches its threshold density, the emigration process starts and a rapid occupation of the empty patch occurs. The new population grows fast due to this high immigration rate. Sink populations are formed when individual fitness in the source population is low due to density-dependent effects (probably on fecundity and mortality, while birth rate can be high due to large population size). Initial fitness in new populations 4.7 Distribution of Territories: Ideal Despotic Distribution 71 is low but it should rapidly increase when density increases due to the arrival of new immigrants. Clumping of populations: When a meta-population originates from a founder population, the spatial arrangement of sink populations is expected to follow a Poisson distribution. The slope of the relationship between the probability of a population formation and the distance to the source population depends on the ability of the species to move across the environment. When the Allée effect operates, the shape of the relationship between probability of population formation and distance changes. New populations depend on the probability of several dispersers meeting in the same patch (solitary dispersers are not expected to succeed), so at close distance the probability is relatively high but it rapidly decreases. The expected consequence of this process is that sink populations form near the founder populations and show a cluster distribution. At a geographical level, social species are expected to show a tendency to form clusters of populations within their range.

4.7 Distribution of Territories: Ideal Despotic Distribution

Fretwell and Lucas (1970) distinguished between three hypotheses on the effect of territorial behaviour on animal distribution: the density assessment, the density limiting, and the spacing hypothesis. The density assessment hypothesis proposed that territorial behaviour is used as a density cue by unsettled individuals so they can avoid highly populated patches. Under this hypothesis, consumers distribute following ideal free principles, i.e. with equal payoffs in every patch. This hypothesis implicitly assumes that the size of the territories can vary without costs to meet ideal free distribution requirements. The density limiting hypothesis proposed that territories can be compressed but with a high cost to the newly settling individuals, so ideal consumers maximising their own success would settle in unoccupied patches where habitat suitability is not the highest but the cost of territorial defence is low. Under this hypothesis, the ‘free’ assumption is violated. The spacing hypothesis states that territorial behaviour is not related to resource defence within patches, but individuals separate as much as possible for other reasons, such as to strengthen pair bonds or to prevent the spread of disease. Under this hypothesis, there is not a consistent change in population distribution in the face of variation in population size. Fretwell and Lucas (1970) did not provide with an analytical approach for the spacing hypothesis, using it as a kind of null hypothesis, while the density assessment hypothesis falls into ideal free category. Fretwell (1972) proposed a model for the density limiting hypothesis called the ideal dominance distribution. Let Si be the profitability of patch i for territory holders, Ti the profitability for newcomers, and t a density-dependent factor which calibrates the advantage of holding a territory, then

=- (4.1) TSii()1 t 72 4 Distribution of Societies

Fig. 4.8 Graphical representations of the ideal despotic distribution models. In A, new, subordinate arrivals start using the poor patch (modified from Sutherland 1996)

The value t ranges between 0 and 1 and increases with density since the advantage of exclusive resources rises accordingly, as competition increases. If all competitors entering the habitat maximise Ti, then equilibrium will be produced in which the values of Ti are the same in all patches. The ideal despotic distribution model provides that in territorial animals, the relevant variable is the order of establishment of territories in different sites. The quality of sites for each individual decreases with the order established. The presence of others does not diminish the quality for those who are already established (Fig. 4.8). Individuals who are arriving will be established in the best place until it reaches point A in Fig. 4.8. At this point, a newcomer will gain the same in this site and in the worst site. The following individuals that are introduced to the system will be established at both sites in turn so that the gain will remain the same. This model predicts that the average gain of the individuals in the best site will be greater than at the worst site. However, an individual who is about to establish a new territory will gain the same by using either of the two types of patches.

4.8 Mating Systems and the Ideal Free Approach

4.8.1 Distribution of Reproductive Birds: The Polygyny-Threshold Model

In some birds, it was found that the same population can show both monogamous couples and polygynous groups. This led to the development of one of the first cost–benefit models to explain the environmental conditions in which the monogamous and polygamous systems are adaptive. This is known as the polygyny- threshold model, which is designed with the same approach as ideal free distribution 4.8 Mating Systems and the Ideal Free Approach 73

Fig. 4.9 Orians’ polygyny-threshold model. The distance δf represents the difference in fitness between females mated monogamously and females mated with other females on the territory of the same male. PH is the polygyny threshold; when differences in environmental quality between the territories of a male with a female and a male without a female are larger than PH, then it is advantageous for a new female to breed polygynously (modified from Orians 1969)

(Orians 1969). When a small proportion of males can monopolise all available sites for breeding, females are forced to accept polygyny. When most males defend breeding sites, females can choose between breeding monogamously and polygyni- cally (Fig. 4.9).

4.8.2 Distribution of Mammals: The Effect of Sexual Harassment

Because most mammalian systems of social organisation ultimately depend on female dispersion (Davies 1991), the study of habitat selection by breeding females can help us to understand the ecology of mammalian mating systems. Fretwell and Lucas’s model can be used to explain female aggregations (harems, territories, or leks) in mammals (Cassini 2000). Male harassment has been postulated as a major social factor influencing female movements and mate selection, ultimately affecting the evolution of mammalian mating systems (Bartholomew 1970; Clutton-Brock et al. 1992). Fretwell and Lucas’s model with the Allée effect described in Sect. 4.6 can be applied to female habitat selection during breeding seasons, with (1) the probability of offspring survival as currency, (2) the increase in female–female competition for resources required for breeding as the cost of increasing density, and (3) a reduction in male coercion as the benefit (Fig. 4.10). The model predicts that male coercion forces females to form denser aggregations than expected based on the distribution of resources alone, because a reduction in male harassment compensates for the increase in female competition. The formal 74 4 Distribution of Societies

Fig. 4.10 Model of female habitat distribution. Female aggression increases linearly with female density (a), while the benefit of male harassment avoidance increases at a decelerated rate (b). These different responses of costs and benefits produce an Allée effect

development of a model of habitat selection by breeding females under male coercion is as follows (Cassini 2000). When female density in a breeding site increases, female competition increases and female reproductive success decreases. The model describes this effect as a linear function of the form

fq=- F (4.2) where f is the cost of female competition, F is female density, and q is the female interference coefficient. Male–male competition prevents the number of males within a female group increasing in proportion to the number of females. As a consequence, when female density increases in a group, male harassment decreases due to dilution effects. Thus, male density can be described as a decreasing function of female density. The model uses a power function of the form

s Md= Fr (4.3) where M and F are male and female densities and d and s are ordinate and slope, respectively (0 < s < 1). The equations that describe the cost–benefit functions are

-r æ F ö m =-1 ç ÷ (4.4) è MF+ ø

pg=+mf+ (4.5) where m is the cost of male harassment, p is the probability of offspring survival, g is the maximal value of p, and r is the male interference coefficient. The parameter g represents the suitability of a site in terms of the availability of resources. If the 4.8 Mating Systems and the Ideal Free Approach 75 environment is comprised of sites with different g, the equilibrium distribution for a given population size can be easily obtained by iterating across the sites until the maximal p is obtained in each site. The main predictions of the model are as follows: (1) without male interference, the maximum probability of pup survival is obtained by solitary females; (2) when male interference increases, there is a change in the maximal value of p towards high values of female group density; (3) species with a similar intensity of male harassment can show different degrees of gregariousness depending on the intensity of female–female competition (q). Chapter 5 Distribution of Subpopulations

5.1 Introduction

This chapter deals with the distribution of local populations or subpopulations. A subpopulation is composed by aggregations and societies and is a component or unit of spatially structured or patchy populations. It is largely characterised by its connection with other subpopulations through dispersal. Other properties of subpopulations are extinction rate and birth/death rates, depending on modelling approaches. This chapter describes four types of models, which differ in theoretical background: metapopulation, source–sink, isodar, and pattern-based models. Metapopulation theory is deeply linked to traditional population ecology. The original source–sink model was based on an ideal-free distribution approach. Isodar models also combine the principles of maximising individual fitness with those of population dynamics. The pattern-based models originated in landscape ecology and emphasise landscape properties rather than the quality of individuals or local populations. The latter type of models is included in this section because landscape ecology frequently assumes to work at this level of organisation, when analysing the distribution of organisms of a species. The concepts of metapopulation and source–sink have become very influential in population ecology. This discipline had developed over several decades without any concern for the issue of the spatial structure of the environment, focusing instead on temporal change processes within defined groups of individuals in a more or less arbitrary way (Chap. 1). These models are the primary way in which the ecology of populations was able to incorporate a spatial dimension into its theoretical body.

M.H. Cassini, Distribution Ecology: From Individual Habitat Use 77 to Species Biogeographical Range, DOI 10.1007/978-1-4614-6415-0_5, © Springer Science+Business Media New York 2013 78 5 Distribution of Subpopulations

5.2 Metapopulations

Traditional population ecology was developed on the implicit assumption that habitats are homogeneous. The discipline incorporated the spatial axis as an explicit variable model in the 1970s. Metapopulation theory emerged as the solution found for environmental heterogeneity. It is applicable to matrix-embedded structures, where the matrix occupies substantially more space than does the suitable habitat. Under this condition, populations are expected to be divided into subpopulations, each one occupying a suitable element of the landscape. This theory is built on the idea that subpopulations are internally unstable and that stability is reached through an extinction–colonisation process. Most species show some kind of specialisation or preference for a type of habitat, so they are expected to respond numerically to habitat heterogeneity. Suitable habitat elements are unstable enough to produce local extinctions of species that will remain absent immediately after habitat recovery. In a heterogeneous landscape it should be possible to find both occupied and empty areas of suitable habitat. Levins (1969, 1970) original model was applied to an infinite number of identi- cal habitat patches. Every subpopulation has a fixed extinction ratee , while any empty habitat patch has a colonisation rate that is proportional to the number of occupied patches and a colonisation parameter c. Let p be the fraction of patches occupied at a given time, so the rate of change of occupied habitat is given by

dp =-cp()1 pe- p (5.1) dt

The critical parameter of the model is c/e because the metapopulation persists forever if c/e > 1, and otherwise becomes extinct. Metapopulation models emphasise two properties of landscapes that are not included in most models that are based on habitat quality, such as the ideal-free approach: geometry and stochasticity. Habitat area and inter-habitat distance are key factors in metapopulation geometry. Small habitat patches are predicted to support smaller populations, each having a higher extinction risk than larger populations residing in large patches of habitat (Hanski 1994), while populations in distant patches of habitat are expected to be prone to extinction because they receive fewer immigrants or colonists than do those in habitats that are closer together, due to a general decline in the frequency of dispersers with increasing distance (Turchin 1998). On the other hand, the probability of the persistence of a local population is calculated by adding stochastic variation to birth and death, either separately or combined (Thomas 1994). Mathematical ecologists have enjoyed producing variants of this original metapopulation model, so that the literature is full of these, of which an example is described. Harding and McNamara (2002) proposed a unifying framework for metapopulation dynamics. They suggested that two preset functions of the original 5.2 Metapopulations 79

Fig. 5.1 Dynamic of Levins model. Colonisation rate (C) and extinction rate (E) depend on the proportion of occupied patches (N/T). The proportion of occupied patches is in equilibrium when C and E, i.e. when the colonisation and extinction functions, intersect. There is only one stable equilibrium point, and it is reached only when m is sufficiently large. A, C, and E illustrated for m = 0.5 (modified from Harding and McNamara 2002)

Levins model omit important biological processes that act at the metapopulation scale. The two functions are total colonisation rate C and total extinction rate E:

æ N ö Cc=-N ç1 ÷ (5.2) è T ø Ee= N (5.3) The dynamics of the Levins model is visualised in Fig. 5.1. An equilibrium point occurs where the colonisation rate equals the extinction rate. A higher migration rate always implies higher patch occupancy (N/T) at equilibrium in Levins metapopulation scenario. In order to capture the dynamics of metapopulations, Harding and McNamara (2002) suggest that we should allow for different types of dependencies between migration and colonisation rates and between migration and extinction rates. They define two general functions, the colonisation rate per patch

Cpatch(α) and the extinction rate per patch Epatch(β), where α denotes the rate of arrival of colonisers onto an unoccupied patch and β is the rate at which immigrants arrive at an occupied patch. These arrival rates for immigrants are the functions αN and βN of the number of occupied patches, N. The rate at which empty patches are colonised in a metapopulation according to this new generalised framework is given by =- µ (5.4) CNtotalp() ()TNC atch ()N

The rate at which occupied patches switch to empty patches Etotal(N) in the new framework is

= b (5.5) ENtotalp() NE atch ()N 80 5 Distribution of Subpopulations

The rate of change in patch occupancy is given by

dN =-CN()EN() (5.6) dt totaltotal By allowing for various dependencies between extinction and colonisation or migration, the model of Harding and McNamara is able to account for most metapopulation-level dynamics. They provided a series of examples explaining how migration has a different influence on subpopulation extinction and establishment, including the rescue and Allée effects, and the consequences of territoriality and conspecific attraction. There are hundreds of predictive tests derived from the metapopulation model (with variations in how strictly a metapopulation is defined) and its variants for different taxa, including plants, amphibians, insects, birds, and mammals. A case for plants is provided. Purves and Dushoff (2005) studied metapopulations of Eichhornia paniculata, an aquatic plant restricted to ephemeral pools. Seed dispersal is by waterfowl and so is directed to suitable habitat. Because the original Levins model assumes random dispersal, these authors developed a modified version of that model that was adapted to plant species with directed dispersal. They reformu- lated the original model in the following terms:

dy =-Spa ()hp- ep (5.7) dx where S is the seed output from each occupied population and α is the probability that a seed (or other propagule) landing in a suitable unoccupied patch will establish a new population. Purves and Dushoff (2005) introduced two parameters that set the nature and strength of the directed dispersal, one of which measured dispersal towards unsuitable habitat (ϕ) and the other towards occupied patches (γ). This is implemented as simply as possible: the total seed rain of the metapopulation is divided between the three patch types (unsuitable, suitable occupied, suitable unoccupied) in proportion to their fractional covers, but weighted by a preference for each patch type (weighting 1 for suitable unoccupied patches, φ for unsuitable patches, and γ for occupied suitable patches). If γ = ϕ = 1, dispersal is random. The authors expected that ϕ < 1 and γ > 1 because plants are expected to be dispersed by animals that share the same habitat preferences. With directional dispersal, the equation for the metapopulation dynamics becomes

dp  hp−  = Spa − ep (5.8) dt  φ()1−+hpg +−hp Purves and Dushoff (2005) applied their model to the data set for the E. paniculata metapopulation provided by Freckleton and Watkinson (2002), who surveyed patch occupancy along sections of road with different densities of suitable patches. Figure 5.2 illustrates the prediction produced by the original Levins model and the expectations of the directional dispersal model. The latter shows a considerable quantitative agreement with observation. 5.3 Source–Sinks 81

Fig. 5.2 Test of a metapopulation model developed by Purves and Dushoff (2005) with data surveyed by Husband and Barrett (1998) on the aquatic plant Eichhornia paniculata. The dotted line is the Levins model (φ = γ = 1) with the parameter e/Sα fit to the data via least squares regression. The solid line was given by the model with directional dispersal, fit to the data using least squares regression, with γ fixed at 1 andφ free to vary (modified from Purves and Dushoff 2005)

5.3 Source–Sinks

The notion of the source–sink dynamic is another famous achievement of population ecology. Similar to what occurred for metapopulations and their potential ability to predict the effects of habitat fragmentation, source–sink models were also rapidly absorbed by conservation biology. The major appeal of these models is that they could explain the apparent paradox that degraded environments could support stable, local populations. Source–sink models have also been used as the main argument to show that the internal processes within populations may be more important in predicting the distribution of abundances than the effect of habitat quality. Even though several authors had earlier discussed the need to distinguish between source and sink habitats in population regulation (e.g. Lidicker 1975), it was Pulliam (1988) who first developed a formal model. Pulliam divided habitat into adjacent patches or compartments, such that local populations living in source patches produce a surplus of descendants (finite rate of population increase λ > 1), whereas subpopulations in sinks produce a deficit (λ < 1). The subpopulation of the sink is maintained by emigration from the source. The model predicts that, at equilibrium, the overall growth rate of the whole population is zero (λ = 1), with individuals continuing to colonise sink patches from the source. Pulliam’s (1988) model has some restrictive requirements in terms of habitat suitability and social behaviour: (1) species must be territorial, so the model is a special case of an ideal despotic distribution; (2) source habitats are of high suitability, and territory owners always produce a surplus of descendants; and (3) sink habitats are abundant and provide suitable resources for survival (migrants have an equal probability of survival to residents) but are sufficiently poor during 82 5 Distribution of Subpopulations the reproductive season, so that migrants fail to reproduce. Pulliam’s model applied the general principle of ‘using always the habitat that provides the maximal payoff’. In this case, because it assumes equal survival, the rule can be translated to ‘always occupying the best available breeding site’. This source–sink model must be considered a special case of despotic distribution in which habitats have the peculiarity of being able to provide excellent resources for survival but poor resources for reproduction, so λ < 1 and individuals from the source habitats continuously produce surplus offspring, and individuals in sink patches never return to their natal habitat. This paper by Pulliam has been cited by at least 3,363 publications in Google Scholar (28/7/2012). Many of these cita- tions probably studied subpopulations with different environmental and social contexts to the background required by the original model. For example, many studies lack sufficient data to separate a source λ( > 1) and a sink (λ < 1) from simply a good or bad quality habitats (Diffendorfer 1998; Runge et al. 2006). Another problem with this model is that the unidirectional dispersal rule is probably evolutionarily unstable. Although Pulliam (1988) argues that it is an evolutionary stable strategy, other authors insist that a certain degree of reverse migration back to the source is necessary to provide stability in the long term (Morris 1991a, b). Long-term field researches and ecological experiments appear to have been the best methodological approaches for testing source–sink population structures (e.g. in insects, Caudill 2003; in birds, Zanette 2000; in mammals, Fryxell 2001). An example in microorganisms is provided. Amezcua and Holyoak (2000) tested the predictions of a source–sink system in laboratory microcosms using a ciliated protozoan Tetrahymena pyriformis as the target species. They investigate source–sink dynamics in a predator–prey context, so they used the heliozoan protist Actinosphaerium nucleofilum as a predator of T. pyriformis. Holt (1985, 1993) firstly proposed that a source–sink dynamic could be established when predators are confined to a single patch type, which would then become sinks, while patches free of predation are the sources. Holt showed that the dynamic is unstable when these two patch types are isolated from each other, but that when dispersal is allowed from sources to sinks, the predator–prey system tends to stabilise. Amezcua and Holyoak (2000) conducted three experiments to test Holt’s ideas. In the first experiment they evaluated if spatial subdivision of microcosms was capable of enhancing the persistence of two protist species that do not coexist in undivided microcosms (Fig. 5.3). The purpose of the second experiment was to examine the dynamics operating in spatially subdivided microcosms to determine if a classic metapopulation dynamics was sufficient to promote persistence. The third experiment served to estimate the rate of migration of prey and the influence of prey emigration on the local dynamics of the prey-only patches of the first experiment (Fig. 5.3). The results were consistent with Holt’s model predictions. In Experiment I, both T. pyriformis and A. nucleofilum persisted in spatial microcosms more than three times as long as they did in non-spatial microcosms, showing that spatially structured predator–prey systems extended persistence. In Experiment II, there were no differences in persistence in undivided microcosms of 30 and 60 mL, suggesting that enhanced persistence in the divided artificial environment was not due to an 5.4 Fitness Maximisers, Foraging Behaviour, and Landscape Distribution 83

Fig. 5.3 Experimental design of a microcosm experiment conducted to test the source–sink model. An oval indicates the presence of bacterivorous Tetrahymena pyriformis, and a star indicates the presence of predator Actinosphaerium nucleofilum (modified from Amezcua and Holyoak 2000) effect of population or patch size. In Experiment III, the minimum density to which prey decreased in predator–prey bottles of spatial microcosms was much larger than that reached by non-spatial microcosms. This indicates that prey dispersed from prey-only bottles to predator–prey bottles, augmenting small prey densities in the latter. The dispersal rate was also consistent with a source–sink dynamic, because there was no significant relationship between initial prey density and the proportion that emigrated.

5.4 Fitness Maximisers, Foraging Behaviour, and Landscape Distribution

In this section, a different approach to the study of the distribution of spatially structured populations is presented. Similarly to the model of Pulliam (1988), it is based on the principle that individuals migrate between subpopulations in accordance with decision rules that maximise payoff. Douglas Morris has produced a long- standing theoretical and empirical body of work on the interaction between evolutionary ecology and population and landscape ecology. He emphasised the importance of scale and expanded ideal-free reasoning to landscape use by local populations, connected by colonisation processes. He proposed that migration can 84 5 Distribution of Subpopulations

Fig. 5.4 Design of the isodar model (modified from Morris 1992) with and without cost of dispersal (modified from Morris 1992). Dispersal cost produces a devaluation of alternative home ranges, and the resulting isodar has a higher intercept and a greater slope than the cost-free alternative

also be modelled by its benefits and costs, as was previously applied to patch utilisation by foraging theory. Morris (1987, 1996) developed his isodar theory, which serves to formalise this link between foraging behaviour and landscape ecology. A synthesis of this theory is provided. Consider two similar habitat mosaics, which differ only in productivity, in a landscape. The effect of increased density on fitness should be similar in both habitats such that the feedback of density on fitness (the fitness–density function) can be assumed to be represented by a parallel negatively sloped straight line in each case (Fig. 5.4). If individuals select habitats on the basis of fitness gains, the distribution of individuals should be adjusted such that

WW=-C (5.9) ijijto where Wi is the fitness in habitati and C i to j is the cost of dispersal from i to j (which was assumed to be zero in the original Fretwell and Lucas’ (1970) model). Fitness– density functions can be modelled as

WA=-bN (5.10) gg g 5.4 Fitness Maximisers, Foraging Behaviour, and Landscape Distribution 85

where Wg is the fitness in good habitat, Ag is the basic fitness in good habitat at zero density, and bNg is the linear relationship between fitness and density in good habitat. Similarly, for poor habitat, =- (5.11) WAppbNp If habitat selection obeys an ideal-free distribution (Fretwell and Lucas 1970), then individuals will adjust their densities in each habitat, so that

= (5.12) WWgp that is, -=- (5.13) AbggNAppbN Solving for N in the habitat with the greater density (assumed to be good habitat),

()aa- æ b ö N = gp+ ç p ÷ (5.14) g ç ÷ bp è bg ø Equation (5.14) is the cost-free isodar of habitat selection between good and poor habitats. With emigration cost, however, individuals should leave good habitat in favour of poor habitat only when the fitness in the latter exceeds that in the former by a value that is at least equal to the cost of emigration. The simplest version of this model is when that cost is independent of density. ³+ (5.15) WWpgCgtop The effect of the dispersal cost represented in the general inequality of (5.15) can be approximated by an isodar in terms of N~ by

()AA-+C æ b ö N = gpgtop + ç p ÷ N (5.16) g ç ÷ p bg è bg ø Morris (1992) provided methods to test his isodar theory in the field. He and others developed his ideas in several directions, including competition Morris (1989), community structure Morris (1990), dispersal Morris (2004), predator–prey interactions Morris (2004), parasitism (Krasnov et al. 2003), and population regulation (Jonzén et al. 2004). Another approach to the study of adaptive behaviour in the use of landscapes is the effect of predation risk. Instead of measuring the predation rate as a direct effect on population abundance, Laundré et al. (2001) coined the term landscape of fear, which is the effect produced by the spatial variation in predation risk. It describes how the foraging cost of predation varies in space for an individual. Position with respect to refuges, escape substrata, sight lines, and other landscape properties is considered explicitly to build a special representation of the foraging cost of predation. 86 5 Distribution of Subpopulations

5.5 Individual-Based Models and Spatially Structured Populations

There are several approaches to the integration between the individual and population levels. The combination of individual-based models with metapopulation theory has been one of the most prolific interactions within the discipline of population ecology in the past two decades. It is well known that the original Levins model contained simplistic assumptions regarding metapopulation structure, migration, and landscape configuration. Simulation models provided a methodological tool ideally suited for representing dispersion among local populations with more realism. Connectivity refers to a property of landscape matrices that facilitates or impedes movements among habitat patches (Taylor et al. 1993). The importance of behaviour for connectivity has been repeatedly stressed (Lima and Zollner 1996; Bélisle 2005; Baguette and Van Dyck 2007). Figure 5.5 represents the links between the behavioural and population mechanisms and landscape properties that are involved in the connection between local populations. The dynamics of spatially structured populations are the result of processes at the level of individuals. As described in Sect. 5.1, each patch in a metapopulation can be characterised by two external and two internal processes: emigration and immigration rates and birth and death rates. Individual-based models allow the behavioural responses of organisms to be associated to the two latter processes in a relatively simple way. An example is described, in which Revilla and Wiegand (2008) explore how to integrate movement ecology mechanisms into models for spatially structured populations. They schematically represent the modelled process as shown in Fig. 5.6. Individuals are characterised by an internal state (W) that is directly related to the reasons governing the decision of the individual to move (goals), by a motion capacity (Ω, the ability to use information related to the goals), and by a navigation capacity (Φ, an inherent ability to move with some properties). Individual motion capacity is modulated by the navigation capacity, affecting the movement path as a function of the information available to the individual. The dynamic interaction of these components with the environment (R) through time generates movement paths (Fig. 5.6). These authors built and parameterised the model by using field data relating to the demography and individual movement behaviour of Iberian lynxes Lynx pardinus living in Doñana National Park, Spain. Virtual movements in a grid were previously modelled based on actual field data (Revilla et al. 2004). Population dynamics were modelled by adding survival and reproduction to the individual dispersal model. Dispersers move until they either die or settle, suffering a higher mortality when outside the park. As expected from a typical metapopulation, local processes (B and D) were more important on a per capita basis than between population processes (E and I), with the local populations found inside the park acting as sources (B > D) and net exporters (E > I), while sinks outside the park are net importers. The model showed that inter-patch processes depend on the heterogeneity of the 5.6 Pattern-Oriented Approach of Landscape Ecology 87

Fig. 5.5 Schematic representation of the interaction between biological and landscape processes at each of the three stages of dispersal (modified from Bowler and Benton2005)

Fig. 5.6 Relationship between movement properties and the dynamic of spatially structured populations (symbols described in text, modified from Revilla and Wiegand 2008)

matrix where patches are embedded and also on the parameters defining individual movement behaviour. They were also very sensitive to the dynamic of demographic variables limiting the time spent moving, the within-patch dynamics of available settlement sites (both spatio-temporally heterogeneous), and the response of individuals to the perceived risk while moving.

5.6 Pattern-Oriented Approach of Landscape Ecology

Much of landscape ecology is concerned with understanding the processes that explain the causes and consequences of spatial heterogeneity. In other words, the landscape and the arrangement of its elements are considered as the dependent variable. Landscape ecology investigates complex systems. It has developed many techniques to describe this complexity and sophisticated statistical analyses to simulate system behaviour and has produced hypotheses for the causes of the patterns. 88 5 Distribution of Subpopulations

Landscape ecology must study simultaneously the role of historical factors and the relative importance of the presence of multiple factors at multiple scales (Turner 2005). The ultimate aim is to understand the ecological, geological, and anthropogenic processes that lead to the presently observed landscape patterns. Landscape ecology describes environmental compositions by means of a wide range of metrics, including a series of simple data measured directly from fields or maps (e.g. patch size, edge, inter-patch distance, proportion) to more complicated indices (evenness, contagion, fractal dimensions). A statistical programme commonly used to estimate landscape measures is FRAGSTATS (McGarigal and Marks 1995). Some common measures are (1) edge density, which is the total length of patch edge per unit area; (2) contagion, which is an index designed to quantify the degree of aggregation found within landscape classes; (3) the mean nearest-neighbour distance, which defines the average edge-to-edge distance between a patch and its nearest neighbour of the same class; (4) the mean proximity index, which is designed to measure the isolation of a patch within a complex of patches, given a specified search radius; (5) the perimeter–area fractal dimension which provides information on the irregularity of patch configurations and can be calculated in a number of ways; and (6) the mass fractal dimension, which can be used to quantify the complexity of an area containing internal heterogeneity rather than the complexity of the area’s border (Hargis et al. 1997). When studying individual species, landscape ecologists frequently assume a metapopulation approach, so their pattern-based approach is included into the level of analysis of subpopulation distributions. A field of landscape ecology which is concerned with the distribution of individual species, and which is especially relevant for this book, corresponds to a set of models that use landscape metrics as environmental descriptors that influence species distributions. In this case, the interest of landscape ecologists is reversed in the sense that it is focused on the influence of landscape patterns on the spatial heterogeneity of species abundance. Li and Wu (2004) conducted a conceptual review of these metrics when they are applied in this context, and found a number of difficulties. They suggested that an effort be made to measure metrics and indices with ecological relevance. For example, evaluating wildlife habitats requires the development of indices that can be related to food, cover, reproduction, and other population processes such as dispersal. To ensure ecological relevance, these variables of species-habitat requirements should be incorporated into the map data used in spatial pattern analysis. The qualitative analyses on the response of organisms to landscape complexity have proof to provide new insights on distribution ecology. Their main contribution is to show that local habitat conditions may be inadequate to explain species presence or abundance. The size of a landscape element has a strong effect on species that are specialised to live on edge or interior but is negligible for generalist species (Turner 2005). A significant effect of boundary shape or characteristics of the surrounding matrix may be present. Many examples demonstrate the importance of corridors that connect habitat patches. In summary, landscape models have incorporated the notion that the configuration of landscape elements may be 5.6 Pattern-Oriented Approach of Landscape Ecology 89 critical to explain variation in the presence or abundance of organisms in a wide range of taxa. A good example of the use of the pattern-oriented approach is to estimate the effect of matrix resistance on species distribution. Analytical models such as metapopulation approach or island theory often measure isolation of habitat patches in a matrix, solely based on distance between nearest occupied habitat patch and to source areas, or a combination of distance and patch size (e.g. Hanski 1994). These measures assume that the matrix environment is homogeneous with respect to animal movement. Sometimes landscape elements like roads, rivers, or corridors facilitate or complicate dispersal, changing the dispersal path or speed or survival during the process. Several authors preferred to use a landscape pattern approach to estimate the effect of this heterogeneous matrix resistance on species distribution. In the following paragraphs, some examples are provided of the use of the pattern-oriented approach of landscape ecology. Ricketts (2001) conducted a fieldwork with butterflies in Colorado, USA. Butterflies typically live in matrix–patch landscapes. In this study, butterfly community inhabited meadows of a naturally patchy landscape. Ricketts (2001) used maximum likelihood to estimate the relative resistances of two major matrix types (willow thicket and conifer forest) to butterfly movement between meadow patches. For four of the six butterfly taxa studied, conifer was 3–12 times more resistant than willow. For the two remaining taxa, resistance estimates for both zones of the matrix were not significantly different. These results indicate that responses to matrix differ even among closely related species and suggest that inter-patch distance alone is not an appropriate measure of patch isolation. Fractals are shapes or geometric forms made of parts similar to the whole and whose irregular details recur at different scales (Mandelbrot 1982). Fractals can be thought of as objects or patterns that have non-integer dimensions. While Euclidean dimensions such as points, lines, and planes have integer dimensions 0, 1, and 2, respectively, fractals adopt values between them. For example, a coastline can have a fractal dimension between 1 and 2, indicating that the bends in the coastline tend to not ‘fill’ the plane in which it is embedded. Coastlines are also good examples to illustrate the other two properties of fractals: self-similarity and scale dependence. In a jagged coastline, every small peninsula is but a knob on the side of a yet larger one (Milne 1997). They also relate to the classic question ‘how long is the coast of Britain’, which does not have a unique answer because the length depends upon the resolution at which it is measured. The classic way of measuring a coastline is to apply a caliper set at some arbitrary gap width and to step along the line as rendered on a map or aerial photo- graph (Milne 1997). It is possible to get a function that relates the number of steps

NL in relation to the caliper length L. This function will be negative and will follow a power law:

- = Dc (5.17) NbL L where Dc is a fractal dimension estimated by the caliper method and b is a constant. 90 5 Distribution of Subpopulations

Milne (1997) provides an example of the use of fractals applied to estimate potential habitat for red-cockaded woodpeckers in the United States. It is based on a map provided by Evans and Zhu (1993) on the pine forest of the south of this country. He found the box dimension of the potential habitat by first counting the number of 1-km-wide grains that contained southern pine forest. Next he overlaid this map with a grid of 2-km-wide boxes in which he counted the number of boxes that contained at least one 1-km grain of forest. Then, he increased the boxes to 4-km length and so on up to 128 km to provide two orders of magnitude range in box size. Logarithms were taken of both the number of boxes NL that contained forest and the corresponding box length L. He used linear regression to estimate the slope of the function that relates the logarithms of the two variables:

=- (5.18) NKLbloglDLog which is equivalent to the fractal power law described in (5.17). The slope provided a first approximation of the box fractal dimension. Without the use of fractal geometry, Milne (1997) showed that there is a danger that mapping habitat at an arbitrary scale could either underestimate or overestimate the habitat available for the birds. In recent years, a long list of sophisticated methods of analysis of the heterogeneity of landscape has been developed, many of them sound intractable for non-mathematical ecologists, such as Fourier (Lundquist and Sommerfeld 2002) and wavelet analysis, percolation theory (Keitt et al. 1997), lacunarity analysis (Plotnick et al. 1993; McIntyre and Wiens 2000), multi-fractal models (Gamarra 2005; Laurie and Perrier 2009), and spectral analysis (Platt and Denman 1975). Wavelet analysis is applicable to many studies that require temporal or spatial patterns, including landscape analysis (Schröder and Seppelt 2006). Wavelets have shapes that resemble a wave, i.e. a sine function with a period of 2π. For example, there is a wavelet that is based upon the first derivative of the standard Gaussian probability density function (Percival et al. 2004). Wavelet analysis uses these finite templates applied over the length of a data sequence, to analyse spatial data. The wavelet template, over a range of sizes and positions, is compared to the data. Sizes and positions that match well produce a positive score (Dale et al. 2002). Examples of the use of wavelet analysis in ecology can be found in Harper and Macdonald (2001), Klvana et al. (2004), and Rosenberg (2001).

5.7 Cognitive Constraints on Landscape Utilisation

The content of this section is related to Sect. 2.2, but specifically thought for individual movements between landscape elements. Ecologists working on animal responses to landscape structure became interested in the cognitive abilities shown by different species to recognise features of the environment and move in accordance to this information. Thus, the perceptual range became a topic in landscape ecology. It can be defined as the ability of animals to perceive habitat at a distance, and it is expected that it determines the ease with which animals can locate 5.7 Cognitive Constraints on Landscape Utilisation 91

Fig. 5.7 Angular orientations of grey squirrels Sciurus carolinensis in a field experiment (modified from Zollner 2000). Grey squirrels were released 300, 400, and 500 m from the woods. The solid squares represent the release point and the dotted line the direction of the woods. In the large circle, there are small ones that represent the angular orientation of each released squirrel. Only in the shortest distance, there was a statistically significant orientation towards the woods, which is represented with an arrow habitat patches, so the time spent searching in a hostile matrix (Zollner and Lima 1997; Zollner 2000). Even when the actual psychological abilities that are involved in perceptual range are not clearly established, it is assumed to be a species-specific attribute that affects the persistence of spatially structured populations. Zollner (2000) conducted a series of ingenious experiments with eastern chipmunks Tamias striatus, grey squirrels Sciurus carolinensis, and fox squirrels Sciurus niger in fragmented landscapes composed of a mosaic of forested patches and agricultural lands in eastern North America. Animals were captured in remote woodlots and translo- cated to unfamiliar agricultural fields. There they were released in a hostile, open habitat at different distances from a woodlot and their movements towards or away from the woodlot were used to assess their ability to perceive forested habitat. Figure 5.7 illustrates results obtained with grey squirrels. These squirrels were capable of being orientated towards the woodland when they were at 300 m, but they failed to do so at larger distances. There were also interspecific differences in perceptual range: it was of 120 m for chipmunks and of 400 m for fox squirrels, suggesting that differences in landscape-level perceptual abilities may influence the occurrence of these species in isolated habitat patches. Chapter 6 Distribution of Populations

6.1 Introduction

This chapter deals with the distribution of groups of populations that may cover part of or the whole range of a species. The concept of population remains central in ecological thought and practice. As for many other important terms in ecology, its definition remains elusive (Schaefer 2006). At an organisation level, a population is composed of aggregations, groups, or subpopulations and is a component of communities. A population is formed by individuals of the same species which have little or no contact with other individuals of the same species, so any numerical changes are largely determined by birth and death processes (Caughley 1977; Krebs 1985; Berryman 2002). The concept of population is strongly associated with these internal processes. Due to the effect of scramble competition in close populations, the interplay between the concepts of carrying capacity and intrinsic growth rate, as described by the logistic equation, has become the standard general theory of single-species population growth (Pianka 1974; McNaughton and Wolf 1979). One of the most fundamental assumptions of the logistic equation is that the level of carrying capacity is set by the availability of resources (see review by Soberón 1986). According to this assumption, the answer to the starting question is that the distribution of abundances among a group of populations (closed and in equilibrium) will be a positive accelerating function of habitat suitability. Populations can be of various types: open and closed populations, source–sinks, metapopulations, and patchy populations. The internal, spatial organisation of spatially structured populations was analysed in the previous chapter. Thomas and Kunin (1999) produced a classification for spatial structures of populations. A spatially structured population can typically be divided into a series of interacting subsets which may represent discrete habitat patches, population clusters, or arbitrary sampling areas (population units). Population changes within any such unit can be described in terms of inputs such as birth (B) and immigration (I) and outputs such as death (D) and emigration (E). The net inputs and outputs for a population unit can be separated into internal processes (B-D) on the one hand and external processes

M.H. Cassini, Distribution Ecology: From Individual Habitat Use 93 to Species Biogeographical Range, DOI 10.1007/978-1-4614-6415-0_6, © Springer Science+Business Media New York 2013 94 6 Distribution of Populations

Fig. 6.1 Types of spatially structured populations. The Compensation Axis is represented with the dashed line. B birth; D death; I immigration; E emigration; per capita rates. More explanation in the text (modified from Thomas and Kunin(1999))

(I-E) on the other (Fig. 6.1). The balance of these four processes differentiates many of the types of population units that have been described. Thomas and Kunin (1999) used net internal and external processes to define ademographic space into which population categories can be arranged. At equilibrium, classical populations fall on the intersection of the B-D and I-E axes, with source populations placed on the lower right and sink and pseudosink population units situated on the upper left. Categories lie along a diagonal in demographic space, along the line defined by (B + I) − (D + E) = 0. This occurs because, if a population is to be at equilibrium, the factors that are increasing population size must be balanced by the forces decreasing it. Thomas and Kunin (1999) also defined amobility axis, that is, (I + E) − (B + D), that describes the involvement of a local population in regional (I + E) rather than local (B + D) processes, running from separate populations, through metapopulations (population units connected by the dispersal or occasional flow of individuals), to patchy populations (regular flow of individuals in and out of each unit). This chapter is divided into three sections with different approaches. In the first section, we analyse what should be the distribution of abundances for sets of close populations that show internal equilibrium or a growth rate equal to zero. This section includes three models: a classical model developed by Slobodkin (1953), from which it is possible to derive the habitat-matching rule; a model by Brown (1984) based on the distribution of populations within a species range; and a model called NICHE that explicitly incorporates landscapes into a model of multiple population distributions, developed by Pulliam (2000). In the second section, we analyse the equilibrium of spatially structured populations 6.2 Distribution of Close Populations 95 and the distribution of many of this type of population within a species geographical range. In the final section we briefly evaluate the role of behavioural mechanisms on the distribution of multiple populations.

6.2 Distribution of Close Populations

Slobodkin (1953) was the first to formalise the relationship between population abundance and habitat quality. Measuring the environment in terms of the number of organisms in the population implicitly assumes that there are resources (r) in the environment, such that each animal at equilibrium requires a/K of r, being K the carrying capacity (Slobodkin 1953). If the total available quantity of r increases in the environment, the equilibrium number of organisms in the population increases proportionally. In effect, the amount of r required by each organism in the population is independent of the other organisms in the population, while the amount of r available to each organism is dependent both on the total amount available in the environment and on the number of organisms competing for it. Slobodkin (1953) wrote the equation of population growth as dN ()Ka- N = lN (6.1) dt K where N is the number of organisms in the population, l is the intrinsic growth rate, and a is a proportionality constant. The relationship between the size of the environment (measured as the amount of resources, R) and the number of animals at the upper asymptote can be written as

Ka= N (6.2)

The expectations of this relationship at equilibrium will be that l1 = l2 = … = li = 0 and that the population abundance in rich environments will be higher than in poor ones. Slobodkin (1953) incorporated the effect of contest competition into the model of closed environments:

dN ()Kb- N 2 = rN (6.3) dt K and thus the relationship between quality of the environment and population abundance becomes

2 Ka=+NbN (6.4) It is also possible to incorporate the third component of a power series that may represent the effect of differences in competitive abilities, such that

23 Ka=+NbNc+ N (6.5) 96 6 Distribution of Populations

Slobodkin (1953) defined the efficiency of an asymptotic population as K/N, that is, as the number of organisms that can be maintained by a unit of environment. For the same value of K, the efficiency is higher in populations without social behaviour. This means that the population abundance (probability of occurrence in the species distribution model) will be relatively lower when species exhibit aggressive behaviour compared to when only scramble competition operates. Hutchinson (1947) modelled the outcome of competition between two species subsisting on a common resource and maintained in a homogeneous closed universe. In summary, it is theoretically possible to describe the relationship between population abundance and habitat suitability by basing it on the traditional logistic regression function for population growth. The capacity of these models to describe the real world depends on many factors; a major factor is whether a population may remain within its carrying capacity or if this parameter even exists. Brown (1984) was the first to propose a link between population dynamics at the local scale and the overall pattern of distribution at the geographical scale. He assumed that the abundance at different locations within the distribution of a species is the result of the interplay of many different biotic and abiotic factors that define a species’ niche. Spatial variation in population density is assumed to reflect the probability density function of the required combination of niche factors along the spatial dimension. Brown also assumed that the environmental variables that characterise a niche are in fact composed of groups of co-varying factors. Brown and Maurer (1989) formalised these ideas. They proposed that the dynamics of a single local population may be viewed as a change in density over time:

dn =+bN(,Xi)(NX,)--dN(,Xe)(NX,)=-gN(,Xl)(NX,) (6.6) dt where b, i, d, e, g, and l are birth, immigration, death, emigration, gain, and loss rates, respectively, and X represents space. When X is constant, the gain and loss functions will intersect to produce one or more equilibrium points. Changes in the g and l functions in space can be attributed to spatial variation in environmental factors that are independent of population size. Over their geographic ranges, most species not only experience significant variation in abiotic factors, such as temperature and moisture, but also encounter varying biotic conditions. These environmental variations determine the equilibrium points reached by each local population, and the sum of these local equilibrium densities builds the shape of the geographical distribution of abundances of the species. Pulliam (2000) constructed a model with a similar conceptual framework, linking niche, demography, and distribution. The model called NICHE is more specific in defining the shape of the demographic response function. The landscape consists of a grid of k cells; each cell c is occupied by n individuals and is characterised by particular values of environmental variables e. Pulliam provided an example (e.g. annual plant species) with two niche dimensions: one environmental factor 6.3 Distribution of Spatially Structured Populations 97

e1 (e.g. pH) that influences juvenile survival Pj and another factor e2 (e.g. moisture) that influences reproductive success in accordance with parabolic functions:

dn  n  cc=−11Pj maxm{(−−ae opt )}2 Pj ax {(1−−ee opt )}2   11ce1 22c e2 (6.7) dt  k 

 nc  where 1−  is a density-dependent factor. The model also specifies the landscape  k  structure by assigning environmental conditions to all cells, as well as dispersal rules, which govern movements between cells. The model has the virtue that allows the niche space to be transformed into a geographical space. However, the function that describes how environmental factors influence demographic variables should be determined by empirical evidence, which is hard to achieve in many real species.

6.3 Distribution of Spatially Structured Populations

Section 5.2 described how subpopulations should be distributed in a landscape composed of habitat patches separated by a matrix of unsuitable environments. This section scales up one level and asks which would be the expected distribution of whole spatially structured populations between landscapes. It is divided into two parts. In the first, an explanation is given for how the classical models of metapopulations and source–sinks predict equilibrium distributions of individual populations composed of multiple, unstable units. In other words, if environmental conditions remain stable in a landscape, the total population will equilibrate when λ = 1. In this way, it is possible to estimate the distribution of several spatially structured populations among landscapes. In the second part of this section a model developed by Holt and Keitt (2000) that formalised this approach is presented. Models of metapopulation and source–sink have often been used as an argument to show that nature is essentially dynamic and that it is therefore unrealistic to suggest that environments can sustain stable populations based on the provision of resources. Metapopulation models show that a suitable environment may not contain individuals as a result of local extinctions, and source–sink models show that an unsuitable environment may contain a local population by directional migration. These patterns would oppose the tendency towards the habitat matching between abundance and quality models that is proposed for site suitability and the ideal free distribution approaches. While metapopulation and source–sink models demonstrate that, at the subpopulation level, instability can be the rule for certain species, the same models provide a solution for equilibrium distribution when a larger level is taken into account. In Levins model, the equilibrium proportion of occupied patches is

e p* =-1 (6.8) c 98 6 Distribution of Populations

Equation (6.8), which gives the rate of change in the proportion of patches that is occupied in a metapopulation, is equivalent to the logistic model with a carrying capacity K = p* and a growth rate r = c − e. The critical parameter of the model is c/e; the metapopulation persists forever if c/e > 1; otherwise, it becomes extinct. In Pulliam’s model, the total population equilibrates when λ(N) = 1, so population density at equilibrium can be estimated as

P nˆ ()bb- * j 12 (6.9) N = ()1--PPAJb2 where Pj and Pa are the probabilities of survival of juveniles and adults and β1 and

β2 are offspring produced in patches 1 and 2. If there are many habitats, the total population reaches equilibrium when the total surplus in all source habitats equals the total deficit in all sink habitats. In summary, both meta- and source–sink populations reach an equilibrium with the environment at the proper scale. Holt and Keitt (2000) preferred to use a metapopulation perspective to draw geographical ranges. They expanded the ecological scale to a set of landscapes in which a group of metapopulations develops uncoupled dynamics. Their model assumes that the basic Levins model is a good descriptor of the behaviour of each metapopulation in its landscape and that landscapes are organised along a smooth environmental gradient, characterised by a single spatial dimension x. The fraction of patches occupied (n) changes with time t as

dn ()xn=-()xk[(xn)(xc)] ()xe- ()()xnx (6.10) dt where k(x) describes the maximum fraction of suitable patches at point x along the gradient, n(x) is the actual fraction of patches occupied at x, e(x) is the extinction rate (per occupied patch) at point x, and c(x) scales the colonisation rate of suitable, empty patches, per occupied patch.

6.4 Role of Behavioural Mechanisms in Distribution of Populations

Ecologists working on the distribution of populations in and between landscapes have been concerned about behavioural mechanisms that can constrain the use of certain portions of the landscape. In particular, perceptual ranges and philopatry have been considered as internal processes that modulate dispersal in vertebrates. Philopatry is common in migrant animals and, as was described in Sect. 2.2, can emerge through different cognitive mechanisms, including social learning. Site fidelity is a result of adult philopatry and implies absence of dispersal. This behaviour is expected to evolve in stable environments and when costs of dispersal are very high. The important consequence of this inflexible behaviour, in contrast to 6.4 Role of Behavioural Mechanisms in Distribution of Populations 99 payoff-maximising strategies, is that animals can become concentrated in some locations but are absent from others, independently of habitat suitability. González Suárez and Gerber (2008) constructed a demographic stage-structured model that includes the costs and benefits of movements based on two possible cognitive mechanisms: philopatry and direct assessment. Those individuals that take part in habitat selection and move between sites may receive the benefit of higher fecundity when the best quality sites are found, but also incur a sampling cost d that reduces survival. In contrast, philopatric individuals that remain at the natal site do not incur sampling costs but may have a lower fecundity if the natal site becomes a low-quality habitat. They applied the model to a life table for California sea lions Zalophus californianus from Los Islotes Island in the Gulf of California including information on behaviour that dramatically changed predicted population size, model dynamics, and the expected distribution of individuals among sites. Estimated population sizes projected for 100 years diverged by up to one order of magnitude between scenarios that assumed different movement behaviours. The scenarios also exhibited different model dynamics that ranged from stable equilibria to cycles or even extinction. Chapter 7 Distribution of Species

7.1 Introduction

Ecological distribution models at the species level are applied to the whole range of species, although they are also frequently used at regional or national scales. In the first case, the subject of distribution ecology approaches that of ecological biogeography and niche ecology, depending on whether the main interest of the researcher is in producing distribution maps or reconstructing species’ ecological requirements. Section 7.2 forms the bulk of this chapter and describes the so-called species distribution models, which can be considered equivalent to the site suitability models described in Sects. 2.4 and 3.2, with the main differences being that they are applied on a coarser scale and use a wider range of statistical tools and some of them can be used to produce distribution maps. Species distribution models have three characteristics: (1) they apply to groups of populations at a coarse scale; (2) they emphasise the effect of environmental variables on internal processes; and (3) they do not use populations as the unit of study (with associated measurements of birth, death, and migration parameters), but the set of individuals is defined by categories determined by environmental heterogeneity or by arbitrary divisions of space (in most cases, cells on a grid superimposed on the landscape). Section 7.3 deals with the hypotheses that try to explain the shape of the function that best fits the distribution of species abundance within a biogeographic range and is almost entirely based on the analysis of Gaston (2003). Section 7.4 describes models at the level of species that are based on the physiological mechanisms involved in determining the causes of limits in species’ ranges, while the last section describes how species distribution models can be improved by incorporating information on behavioural traits of the target species.

M.H. Cassini, Distribution Ecology: From Individual Habitat Use 101 to Species Biogeographical Range, DOI 10.1007/978-1-4614-6415-0_7, © Springer Science+Business Media New York 2013 102 7 Distribution of Species

7.2 Species Distribution Models

Species distribution models are associative models relating occurrence or abundance data at known locations of individual species (distribution data) with information on the environmental characteristics of those locations (modified from Elith and Leathwick 2009). Of the terms used to describe them, species distribution models is used here, although there are many other terms, some with a more restrictive definition: habitat models, ecological niche models, bioclimatic models, climatic enve- lopes, resource selection functions, species-habitat modelling, and gradient analysis. Species distribution models can be used to cover the whole biogeographical distribution of a species or can be applied to a portion of the range. Strictly speaking, in the latter case models are applied to sets of populations rather than to the species. In any case, most developers of these kinds of models implicitly or explicitly follow the principle that what they are modelling represents components of the niche. Species distribution models have two main uses: identifying predictors or key factors of the environment that affect species distributions or predicting distributions in new scenarios, assuming that the variables included in the model are relevant factors. When species distribution models are used for the latter purpose, their output is normally a habitat suitability map. The first approach has a more theoretical focus and considers the causal drivers of species distributions. The second approach is influenced by the strong demand for mapped products for use in conservation and land management (Elith and Leathwick 2009). In the first category, it is possible to include spatial regression models that use information–theoretical methods for model selection (Burnham and Anderson 2002; Johnson and Omland 2004), quantile regression analysis (Cade et al. 2005; Vaz et al. 2008), resource selection functions (Manly et al. 1992), and many other statistical methods that have been used to relate distributions of species abundances or occurrences with multiple environmental variables (e.g. Utzinger et al. 1998; Roobitaille and Laurance 2002). These models have already been analysed in Sect. 3.2, with the only difference being that the present chapter deals with discussions at the level of a set of several populations distributed at a coarse scale, while Chap. 3 is mainly related to aggregations of individuals within a population. The predictive role of species distribution models can be subdivided according to its scope (Elith and Leathwick 2009). Firstly, predictions are made for new sites within the range of environments sampled by the training data and within the same general time frame as that in which the sampling occurred. Typical applications include global analyses of species distributions, mapping within a region for conservation planning or resource management, and identifying suitable habitat for rare species. Secondly, models are also used to make predictions concerning new and unsampled geographic domains and/or future or past climates. The environments in these new times and places need to be assessed carefully, particularly for new combinations of predictor values or for predictor values outside their original ranges in the training data. Most species distribution models are based on correlation statistics, 7.2 Species Distribution Models 103 from which causation cannot strictly be inferred (e.g. Austin 2002), but summing correlative results based on ecologically meaningful predictors can provide support to a hypothesis. The consequence is that species distribution models should only be used with care, to predict potential ranges or to extrapolate from the current to alternative conditions (e.g. climates; Elith et al. 2006), for instance by ensuring that theoretically well-supported predictors are used (Austin 2002). A strong performance by a particular method in one set of conditions does not guarantee a similar performance outside the range of the environments on which the original model was based (Araújo and Pearson 2005). Only models based on a fundamental knowledge of the actual processes determining species distributions can be extrapolated reliably to new environments and to future or past conditions (Soberón and Peterson 2005; Elith et al. 2006). Recently, extensive reviews have found that species distribution models have been used with a good measure of success to characterise the natural distributions of species and to apply this information to investigate a variety of scientific and applied issues (Guisan and Thuiller 2005; Elith et al. 2006; Elith and Leathwick 2009). This ecological tool has enjoyed an explosive success among wildlife and land managers because it allows them to obtain decision criteria in a relatively short time. With presence–absence and GIS-based descriptions of habitats, models that predict species’ responses to changes in environmental conditions can be generated. Species distribution models have numerous applications: to assess the potential threat of pests or invasive species, to identify hotspots of endangered species or biodiversity, to prioritise areas for conservation, and to restore ecosystems, among many uses (Hirzel et al. 2002; Elith et al. 2006). The development and increased success of these models has been accompanied by criticism about their validity and reliability. There is an extensive discussion on the theoretical framework underlying the models and on the assumptions on which models are built (Van Horne 1983; Rotenberry 1986; Terrell et al. 1996; Thomson et al. 1996; Fielding and Bell 1997; Garshelis 2000; Railsback et al. 2003; Thuiller 2003; Gibson et al. 2004; Cade et al. 2005; Soberón and Peterson 2005; Barry and Elith 2006; Jiménez Valverde and Lobo 2006; Kearny 2006; Real et al. 2006; Austin 2007; Soberón 2007; Jiménez Valverde et al. 2008; Raes and ter Steege 2008). There are several publications that review the available species distribution models (Guisan and Zimmermann 2000; Austin 2002; Anderson et al. 2003; Guisan and Thuiller 2005; Heikkinen et al. 2006; Elith et al. 2006; Elith and Leathwick 2009). Environmental envelope models are the simplest of these. An envelope is defined by minimum and maximum values of the habitat variables and is calculated so that the envelope encompasses a predetermined percentage of the observed species’ occurrences. Regression is the most commonly used technique to model the relationship between species distribution and one or more habitat variables. Tree-based models provide a non-parametric alternative to linear and additive regression models. Other more recent modelling approaches are artificial neural networks, genetic algorithms, and Bayesian analyses. Associative models based on a variety of descriptive statistical techniques (goodness-of-fit metrics, analysis of variance, ordination), which 104 7 Distribution of Species have been used to identify important habitat variables, are also included as species distribution models. Some models are close to the theory of niche, such as Canonical Correspondence Analyses (Austin 2002) and ENFA (Hirzel et al. 2002), and share the assumption that the suitability has a bell shape. Most models provide relative indices of suitability to be interpreted as probabilities of occurrence and then to be translated (or not) onto habitat suitability maps (Elith et al. 2006). Some authors have searched for means of transforming these probabilities into measures of absolute abundance (Boyce and McDonald 1999; Pearce and Boyce 2006). Araújo and Guisan (2006) and Elith and Leathwick (2009) have suggested dis- tinguishing between niche and geographical models. The former, which are based on environmental predictors, only yield projections of potential habitat for species. The latter combine environmental variables with spatially explicit factors. This classification is linked to the purpose of the models. In the majority of cases, this purpose is the prediction of species distributions, while the detection of functional relationships between species and environment and the testing of ecological theory tend to be secondary considerations (Austin 2002). The selection of candidate variables that will be incorporated to the species distribution model often relies on expert knowledge (Guisan and Zimmermann 2000; Manly et al. 1992). Then, most methods look for the smallest combination of variables that can produce the best fit to the data. Stepwise algorithms automate this selection/rejection process but risk rejecting variables more because of spurious correlations with other variables than for ecological reasons (Hirzel et al. 2002). Another approach is the information–theoretic model comparison, which is used when multiple hypotheses are plausible or multiple predictors are considered in combination (Stephens et al. 2005). Two criteria commonly used in ecology and evolution are the Akaike information criterion and the Schwarz criterion (Johnson and Omland 2004). Manly et al. (1992) provide a detailed description of model selection using Akaike information criterion. Guichón and Cassini (1999) developed a simple method that consists of using a Principal Component Analysis to synthesise environmental variables into composite, uncorrelated variables—called factors—that summarises significant ecological information and can be used as new variables in species distribution models. Two studies are described as examples of the use of species distribution models, one in trees and the other in mammals. Thuiller (2003) investigated the predictive ability of three statistical methods (generalised linear models, generalised additive models, and classification tree analysis). They applied these models to the species distribution data of four Mediterranean species (Juniperus oxycedrus, Pinus pinas- ter, P. pinea, and Quercus suber) at three scales: fine (Catalonia), intermediate (Portugal), and coarse (Europe). The choice of common species took into consideration the number of records per species in the databases, as both very rare and very common species are both bound to be more difficult to model. Table 7.1 lists the environmental variables used to build the models. To provide good predictions of current species distributions, the authors proposed that caution must be taken in the selection of environmental variables for analysis in plants. Firstly, relevant climatic variables that have a direct impact on physiological 7.2 Species Distribution Models 105

Table 7.1 Environmental variables used to construct three species distribution models applied to tree species at three scales in Europe (modified from Thuiller 2003) Variable Catalonia Portugal Europe Precipitation Mean annual X X X Mean summer X Mean winter X Mean autumn X Mean spring X Temperature Mean annual X X Mean, coldest month X X Mean, warmest month X X Mean annual maximal X Mean maximal, hottest month X Mean annual minimal X Mean minimal, coldest month X Radiation Mean annual X Potential evapotranspiration Annual X Coldest month X Warmest month X Elevation Mean X X X Maximal X Minimal X Standard deviation X Slope Mean X X Standard deviation X Variation coefficient X Geology X parameters should be selected, including temperature, precipitation, and radiation. Secondly, soil type, slope, and elevation are also considered as good indirect variables to predict plant distributions, even if their link with physiological mechanisms is more complex. In the study, the authors used generalised linear models in two ways: a version with linear terms and interactions and a more complex version that included quadratic and polynomial terms but not interactions. As in most recent uses of these models, they used the Akaike information criterion to select for the most parsimonious model, in the form of an automatic stepwise model. They applied three models. Generalised linear models are probably the most common type of model used to describe the relationship between species and the environment in recent years. The well-known advantage of these models is that they provide error distributions for the dependent variable that are not normal and non- constant variance functions. They also used generalised additive models that are a 106 7 Distribution of Species non-parametric extension of generalised linear models, in which the linear function of the predictor variable is replaced by a smoothing function, useful when the variables adopt a complex form (Goodall 1990). Thuiller (2003) specifically used a cubic spline smoother. The third type of model that the authors used involved classification and regression trees, which provided a completely non-parametric alternative to regression models (Breiman et al. 1984). The first type of tree is used when the response variable is categorical, and the second type when it is numeric. The goal of a tree-based model is to resolve relationships within a complex data set by producing the best empirical classifier or binary tree. Thuiller (2003) found that variables selected by models were relatively consistent across scales and the predictive accuracy of models varied only slightly. However, there were differences in how the methods performed. Classification tree analysis had a lower accuracy than the generalised methods, especially at finer scales. The performance of generalised linear models also increased with scale. At a fine scale, generalised linear models with linear terms showed better accuracy than generalised linear models with quadratic and polynomial terms. This is probably because distributions at finer scales represent a linear subsample of the entire realised niches of species. In contrast to generalised linear models, the performance of generalised additive models was constant across scales, being more data-oriented. The predictive accuracy of generalised additive models was always at least equal to the other techniques, suggesting that this modelling approach is more robust when managing variations of scale because it can deal with any response shape. Categorical trees never performed better than generalised models at any of the scales, even if they yielded the lowest deviance, probably due to the lack of interactive effects among environmental variables. Species distribution models have been extensively used with marine mammals, probably due to the large extent of the distributions of these animals and to their dependency on climatic and oceanographic factors. Gomez and Cassini (unpublished data) conducted an exploratory analysis on the climatic correlates of sexual segregation in elephant seals Mirounga leonina at a global scale. The spatial segregation of sexes outside the mating season has been documented in a wide variety of mammals (Ruckstuhl and Neuhaus 2000). The differential use of space by the sexes is intimately linked to sexual size dimorphism, being more common among more dimorphic species (Mysterud 2000). Elephant seals are among the most sexually dimorphic and polygynous species of all mammals (McCann 1981). Elephant seals forage exclusively at sea and dive continuously and deeply throughout their foraging trips (Bailleul et al. 2007). Gomez and Cassini’s study had the aim of giving a preliminary assessment of the role of climatic factors on ocean distributions of male and female elephant seals and exploring the possibility that different thermal requirements between sexes could eventually explain differences in distribution. Gomez and Cassini used website and bibliographic sources to obtain information on the ocean distribution of adult elephant seals breeding in four colonies: King George Island, Macquarie Island, and Valdés Peninsula in the southern hemisphere (Fig. 7.1), including Campagna et al. (1998), Bornemann et al. (2000), and Tosh et al. (2008). The predictor variables used to calibrate the model included sea surface temperature, bathymetry, and productivity. 7.2 Species Distribution Models 107

Fig. 7.1 Global distribution of female and male elephant seals Mirounga leonina. Males use colder water in higher latitudes than females (Gómez and Cassini, unpublished data)

Gomez and Cassini used two species distribution models, maximum entropy and ecological niche factor analysis, to predict the present and future suitability of ocean regions. Maximum entropy was developed by Phillips et al. (2006) and is based on the principles whereby a target probability distribution is estimated by finding the probability distribution of maximum entropy, i.e. the distribution that is most spread out or closest to uniform, subject to a set of constraints that represent incomplete information about the target distribution. Ecological niche factor analysis is based on ordination of data in a multivariate space of environmental variables (Hirzel et al. 2001). This technique is based on the computation of the factors explaining the major part of species environmental distribution. Extracted factors are uncorrelated and have biological significance: the first factor is the marginality factor, which describes how far the species optimum is from the mean environmental profile in the study area; the second is the tolerance factor, which is sorted by a decreasing amount of explained variance and describes how specialised the species is by reference to the available range of environments in the study area. The suitability maps derived from maximum entropy predicted that females use lower latitudes than males in all populations (Fig. 7.1). Males from King George Island were predicted to occur mainly in the Weddell Sea while females showed the highest probability of occurrence towards the south-western sectors of the colony, around the Antarctic shelf break. Similarly, males of Macquarie Island are expected to occur along the shelf and shelf break, as the ice expands 108 7 Distribution of Species

Table 7.2 Ocean distribution of southern elephant seals from three colonies in Southern Atlantics King George Macquarie Valdés F1 F2 F1 F2 F1 F2 Males Temperature −0.66 −0.74 −0.92 −0.37 −0.32 −0.78 Bathymetry −0.05 0.21 0.38 −0.93 0.86 −0.47 Productivity 0.75 −0.64 0.11 0.08 0.39 0.40 Variance (%) 35 25 53 37 36 33 Females Temperature 0.67 0.14 0.8 0.37 0.85 0.45 Bathymetry 0.15 0.94 −0.55 0.77 −0.52 0.73 Productivity −0.73 0.31 −0.23 −0.52 0.02 −0.52 Variance (%) 78 13 45 29 58 24 Scores for the first two factors (F1 and F2) and explained variance of the ENFA in the autumn and winter. The highest probabilities for females were also found in the latter area and to the north. Males from Valdés Peninsula are distributed along the Argentine continental shelf while females occur mainly at the shelf break and in deep ocean waters. Analysis of the contribution of each variable to the ecological niche factor analysis model revealed that temperature was the variable with the highest positive scores for the most relevant factor for females of all colonies. Males also showed a relatively strong relationship with temperature, but with the opposite sign (Table 7.2). All analyses conducted to determine which was the main environmental predictor of these differences agreed that ocean temperature is the principal factor, with females being distributed in warmer waters than males. Males showed a tendency to be found in more productive waters than females, but these trends were not statistically significant. These results indicated that temperature can play a role in sexual segregation in these species. One possibility is that sexual segregation of elephant seals in the sea is influenced by the different thermal requirements of males and females. Due to their relatively small body size, females should be metabolically constrained and have a thermal limit for water temperature while males should use the coldest waters of the polar regions due to their large size and thick blubber layer. Data on marine mammal distribution in the oceans is growing enormously and Gomez and Cassini used only a limited set of this data, so their analysis should be seen as a methodological and theoretical exercise that illustrates the uses of two types of species distribution models.

7.3 Internal Structure of Biogeographical Ranges

This section deals with the form of the full pattern of variation in abundance that a species exhibits over its geographic range, the so-called abundance surface (Gaston 2003). Many studies have been conducted that describe how the abundance of a species 7.3 Internal Structure of Biogeographical Ranges 109

Fig. 7.2 Hump-shaped model of biogeographical range

changes towards its range limit or periphery. Most studies support a hump-shaped structure, with a decline in abundance towards the edges (Fig. 7.2). This pattern has been described in different taxa (plants, Huntley et al. 1989; fish, Machpherson 1989; molluscs, Kiflawi et al. 2000; insects, Svensson 1992; birds, Telleria and Santos 1993; mammals, Kauhala 1995). These studies may use different methodologies, including visual inspection of maps, analyses of relationships between local abundances and position along transects across geographic ranges, analyses of spatial autocorrelation functions, and interspecific relationships between local abundance and proximity to range edge (Gaston 2003). Gaston (2003) has reviewed five hypotheses on the mechanism responsible for the internal pattern of geographical ranges. They constitute different levels of explanation rather than competing mechanisms. A brief description of each hypothesis follows. Optimum response surface model: Difference in abundance reflects spatial variation in environmental conditions. A classical interpretation of the hump-shaped structure is that environmental conditions deteriorate from the centre towards the periphery of a range, in relation to species-specific requirements (Brown 1984). Local demographic (basin) model: MacCall (1990) developed a model based on the assumption of density-dependent habitat selection. Per capita population growth rate is depicted graphically as increasing downward, and habitats can be described as a continuous topographic surface of geographic suitability that has the appearance of an irregular basin. This model proposed that species distribute following the habitat-matching rule (Sect. 3.5), in a way that the carrying capacity K in any location is proportional to habitat quality in that location. Vital rates model: This model assumes that the intrinsic population growth rate r declines across an environmental gradient from a central peak towards the extremes and that there is a spatially invariant density-independent death rate. Equilibrium local density varies directly with r, peaking towards the centre of the environmental gradient. Gain and loss surface model: This is another model that tries to explain the effect of environmental factors on demographic parameters. Since a large number of biotic 110 7 Distribution of Species and abiotic factors may act simultaneously to limit local densities of species populations, rather than trying to identify this, Brown and Maurer (1989) focus on the four general processes that determine the rate of population change: birth, death, immigration, and emigration. Birth and immigration sum to a net rate of population gain and death and emigration sum to a net rate of population loss. These two functions intersect to give a stable equilibrium point. The set of equilibrium points of populations in space determines the general pattern of population abundance in space. The model assumes that either population gain rate or population loss rate must be a function of distance from the geographic centre. Stochastic temporal variation model: Ives and Klopfer (1997) developed a spatially explicit simulation model that predicts variation in the local abundance of a species in the absence of any fixed spatial pattern of environmental conditions. This occurs when population dynamics are characterised by very weak density dependence, so that population densities exhibit near-random-walk behaviour. Their simulations show that log-rank abundance curves, similar to those obtained assuming fixed spatial variation, can be easily created by temporal environmental variation. The authors suggest that this is presumably a result of temporal environmental variability acting sequentially and, hence, multiplicatively, on changes in population densities.

7.4 Individuals, Physiological Mechanisms, and Species Distribution

In Sect. 7.3, species distribution models were introduced, which statistically associ- ate patterns in distribution of occurrence or abundances of a target species with environmental (mainly geographical and climatic) variables. These types of models have the advantage that they can be deployed quickly and easily, but have several limitations that have been widely discussed in the literature (Pearson and Dawson 2003; Thuiller 2003; Araújo and Guisan 2006). The main concern is related to the true explanatory and predictive ability of these models. Species distribution models use prediction in two ways: (1) within the range of environments sampled by the input data (interpolation) and (2) for new geographic conditions or simulated future scenarios. A main problem with these associative models is that they ignore the ecological mechanisms that mediate the relationship between species distribution and environmental variables, restricting confidence in model predictions (Oswald et al. 2011). Strong performance of a particular method in the present conditions does not guarantee a similar performance outside the geographical range on which the original model was based (Araújo and Pearson 2005). In contrast, models based on fundamental knowledge of the actual processes operating on the pattern of species distribution can be extrapolated to unsampled sites, new environments, and future or past conditions (Soberón and Peterson 2005; Elith et al. 2006). In this and the following section, a sample is presented, of numerous attempts to integrate physiological, behavioural, and population mechanisms into models that predict distribution of species at coarse scales. 7.4 Individuals, Physiological Mechanisms, and Species Distribution 111

There is a group of mechanistic models that rely on physiological traits and are applied at large scales, in most cases to determine species’ geographic ranges. They typically assess the fundamental niche, modelling the potential response of a species to environmental conditions by explicitly incorporating physiological processes calibrated using observations of individuals in natural populations (MacNally 2000; Morin and Lechowicz 2008; Kearney and Porter 2009). Although interest in these types of models has grown exponentially in recent years, they were conceived in the 1970s with the development of plant and animal ecophysiology (Botkin et al. 1972; McNab 1974, 2002; Evans 1976). In plants, a typical example is forest gap models, which predict the distribution of tree species at continental scales, based on the ability of plants to resist frost and drought events (Bugmann 2001). Chuine and Beaubien (2001) developed a model called PHENOFIT that can serve as an example of this approach. PHENOFIT focuses on the stress limits for survival and on the impact of temperature on pheno- logical processes and the success of producing viable offspring. This model estimates the probability of the presence of an adult tree over several years by its fitness on a given site. Fitness is defined as the product of the individual probability of survival and the individual probability of reproductive success. Probabilities of presence predicted over several sites or over a grid produce the potential distribution of the tree species. PHENOFIT comprises several process-based models: pheno- logical models, a frost injury model, a survival model, and a reproductive success model. The parameters of the model are fitted using physiological data obtained from natural populations, so that the observed distributions are not needed in the model calibration. The frost injury, survival, and reproductive success models are based on the principle of the adequacy of the development cycle of an individual within the seasonal climatic conditions of a site (Fig. 7.3). For instance, survival is reduced when a severe drought event occurs between budburst and leaf senescence, or reproductive success decreases when frost events occur during flowering. Animal ecophysiologists and ecologists have been largely concerned about the biogeographical consequences of physiological adaptations and constraints, especially in relation to energetic balance (e.g. McNab 2002; Oswald et al. 2011). Kearney and Porter (2009) recently reviewed a promising line of modelling based on the biophysical properties of organisms, mainly thermal tolerances. Porter and Kearney (2009) established one of the most sophisticated biophysical models that can be applied to endotherms by deriving a geometric model of the thermal dependence of metabolic rate for ellipsoidal, furred endothermic organisms as a function of the radiative and convective environment. The model is mathematically complex, but can be simplified in the following equation:

()tt- Q = cf (7.1) gen ()II++I III where Qgen is the heat generation requirement, (c − f) is the gradient between core and fluid temperatures, and I, II, and III are three thermal resistances. Resistance I is for a body with a distributed heat generation, II is for furry insulation, and III is the 112 7 Distribution of Species

Fig. 7.3 Description of the model PHENOFIT. Ta daily mean temperature; Ti daily minimum temperature; Pm monthly precipitations; Dl leafing date;Df flowering date;Dr ripening date; Dc leaf colouring date; Il frost injury of leaves index; Er thermal energy available since flowering; If frost injury of flowers index;Ir fruit ripening index; Sf probability of frost survival; St probability of drought survival (modified from Chuine and Beaubien 2001) equivalent resistance for convective and thermal infrared radiant heat exchange with the environment. This formula gives the critical minimum temperature at which the thermoneutral zone starts, i.e. the range of temperatures at which endotherms maintain their basal metabolic rate and can therefore survive.

7.5 Behavioural Adaptations and Species Distribution

Correlative approaches to species distribution identify the environmental conditions at locations where a species is known to occur and then expand the expected distribution to other locations with similar conditions. Physiology-based models identify the tolerance limits of species for key environmental conditions and then draw the distribution of the combination of the extreme values of these environmental variables to construct the expected biogeographical range of the species. Neither of these two types of model is able to capture the effect of variance caused by biotic interactions and behavioural adaptations along the range of suitable environment for a species, in other words to describe the distributional consequences of the realised niche of species. This section introduces several examples of recent studies that incorporate behavioural traits of target species into traditional species distribution 7.5 Behavioural Adaptations and Species Distribution 113 models. The following behaviours are taken into account in different publications that improve the prediction of the species’ realised distribution: foraging (Lundy et al. 2012), benefits of group formation (Osborne et al. 2007), time budget (Willems and Hill 2009), territoriality (Sergio et al. 2004), dispersal (Allouche et al. 2008), and antipredation avoidance (e.g. Willems and Hill 2009). The first three studies are briefly described. Lundy et al. (2012) studied the geographical distribution of the bat Myotis nattereri in Ireland. This bat is a foraging generalist, like most other bats, and uses a roost as a central place where it rests during the day and from where it searches for food. The authors developed a multi-scale species distribution model to examine habitat associations of M. nattereri maternity roosts at a landscape scale. The foraging areas of M. nattereri can be up to 4 km from the roosts; Lundy et al. (2012) used this distance as the upper limit of spatial scales to model roost occurrence, delineat- ing available habitat into seven sequential tori starting at roosts with limits at 0.25, 0.5, 0.75, 1, 2, 3, and 4 km. Presence data were obtained from all known roosts spread across the study region (n = 22), and 22 pseudo-absences were generated at random. Eight types of habitats were identified in each torus. Fifty-six generalised linear models were applied. These analyses provided parameter estimates which could be compared to habitat selection ratios (Manly et al. 1992) derived from radio-tracking collected in one independent roost. The results of the study showed that habitat selection of radio-tracked bats mirrored the species distribution model, with bats selecting for woodland in the immediate vicinity of individual roosts but avoiding this habitat in foraging areas, while pasture was significantly positively selected for in foraging areas. Using habitat selection derived from radio-tracking enables a multi-scale species distribution model to be interpreted in a behavioural context that, in the specific case of M. nattereri, appeared to describe a trade-off between the central roosting location and foraging habitat. Osborne et al. (2007) studied the distribution of great bustards Otis tarda in Spain. Previous studies have applied classical species distribution models with the result that, while models were very successful at predicting known occupied locations, they failed because they predicted use of suitable habitat patches that were actually empty. Osborne et al. (2007) tested two hypotheses for the cause of this patchiness in bustard distribution: (1) a high site-fidelity hypothesis that predicts no average difference in habitat quality in either space or time, or in patch size, between occupied and unoccupied patches and (2) a benefits of group formation hypothesis (information transfer on patch quality through the presence of others) that predicts occupied patches to be the more environmentally stable over time. They compared suitable habitat patches that were used and not used by bustards. Occupied patches were larger, over-used in proportion to their size, and possessed different characteristics over time, than unoccupied patches. Osborne et al. (2007) interpreted these results as reflecting a tendency to form groups and as a benefit associated with the use of information about the predictability of the habitat patches. They postulated that sites where great bustards choose to aggregate have a greater temporal stability of climatic factors and that there is a logical link between this stability and the expected breeding performance. Although great bustards have a tendency to return 114 7 Distribution of Species

Fig. 7.4 Comparison of the maximum ecologically tolerable group size as predicted by the time- budget model and the observed maximum group sizes in the 36 populations of vervet monkey (Cercopithecus aethiops) throughout Africa for which accurate group counts were available. The solid line serves as a visual aid for interpretation: in the area below the line, predicted maximum group size exceeds the observed maximum group size (modified from Willems and Hill 2009) to their birthplace, philopatry alone does not explain the difference in size they found between occupied and unoccupied patches. Willems and Hill (2009) studied the distribution of the vervet monkey Cercopithecus aethiops in Africa. They started from the premise that time is an important resource affecting animal activity and, as is true for most resources, is usually limited. To maximise individual fitness, animals must thus adaptively allo- cate their time over functionally distinct behaviours (see Chaps. 2 and 4). For primates living in social groups this entails time being divided over four main time-budget components: feeding, moving, social interactions, and resting (Dunbar 1992). Willems and Hill (2009) compared two models: a ‘classical’ associative model and the maximum entropy distribution model (see Sect. 7.2) with a time- budget distribution model that resulted from integrating a time-budget model developed by Dunbar (1996) into a geographic information system. They produced the maximum entropy model taking 13 environmental variables and presence-only data from 174 sites throughout Africa as input, with an additional 58 sites retained to test the model. The time-budget model considered the same environmental variables but was constructed from detailed behavioural data from 20 groups representing 14 populations, with presence-only data from the remaining 218 sites reserved to test model predictions. They also validated both models against a reference species distribution map that used the African Mammals Databank. The time-budget and maximum entropy models produced accurate and remarkably similar species distribution maps, despite fundamental differences in their conceptual and methodological approaches (Fig. 7.4). Such a strong convergence not only provides support for the credibility of current results, but also relieves concerns about the validity of the two modelling approaches. Part III Levels Outside Species Chapter 8 Distribution of Species Assemblages

8.1 Introduction

One of the central goals of ecology in recent decades has been to seek an explanation of the factors determining biodiversity, surely motivated by the global biodiversity crisis. The amount of work published on this subject is enormous, as are the different approaches and number of schools who study it, so it is obviously impossible to cover completely in this chapter. We will concentrate on three approaches that study diversity in space: isoleg models, macroecology, and species distribution models applied to assemblages. These different approaches depend on the emphasis given to the processes involved in generating biological diversity. They can be divided into biotic, abiotic, and historical processes. Biotic mechanisms relate mainly to interspecifi c competition and predation and normally occur at fi ne ecological scales and in bottom - up paths , while abiotic mechanisms can be divided into ‘prevailing’ processes vs. disturbances and are mainly studied at coarse scales and as top - down paths. Historical processes imply that species richness depends on the length of time available for a species to evolve to fi ll habitats and niches in a region. The isoleg model is an ideal-free type of analytical approach that is based on the effect of biotic factors at the local scale, in particular interspecifi c competition effects on habitat selection. Macroecology is a whole discipline that is concerned with understanding the abundance and distribution of species at large spatial and temporal scales. It emphasises the importance of a regional perspective for understanding the structure and dynamics of local species assemblages, together with an interest in regional-scale issues per se. Thus, it deals mainly with abiotic and historical processes. Species distribution models are also applied at large scales, but their approach mainly emphasises the effect of abiotic interactions and, in some way, assumes that the effect of catastrophes and historical events is relatively unimportant.

M.H. Cassini, Distribution Ecology: From Individual Habitat Use 117 to Species Biogeographical Range, DOI 10.1007/978-1-4614-6415-0_8, © Springer Science+Business Media New York 2013 118 8 Distribution of Species Assemblages

8.2 Isoleg Model: Distribution of Guilds Composed of Fitness Maximisers

Michael Rosenzweig published his theory on isolegs in 1981. In this work, he described several models and called the fi rst one the model of density - independent habitat selection. This model is essentially equivalent to the marginal value theorem (Charnov 1976 ), which was described in Sect. 2.5.3. It applies to individuals who are distributed between foraging sites when no intraspecifi c competition exists or, similarly, when alone in the environment. Rosenzweig developed a second model incorporating ‘density dependence’ into a population. It is similar to the model of ideal-free distribution developed by Fretwell and Lucas ( 1970 ), described in Sect. 3.4.2, with the only difference being that Rosenzweig incorporates the cost associated with switching between foraging sites. What interested Rosenzweig in this second model was to fi nd the threshold of population density M o , at which a population M passes from using a single type of site to occupying two (always modelled for a situation with two types of sites). In the terminology we have used in the previous sections, where we talked about food acquisition rates R , this threshold is reached when

RR= pg ( 8.1 ) where p and g are patch types. In Rosenzweig’s terminology, this threshold is reached when lnWW= ln pg ( 8.2 ) where W is individual ﬁ tness, including number of offspring or reproductive rate and survival. The model includes the costs of movement between patches:

æ t ö lnWW=+ ln 1 m ( 8.3 ) pgç ÷ è tp ø where tm is travelling time and tp is residence time in patch p. The third model described by Rosenzweig is what is now known by the name of isoleg theory . This model incorporates the problem of interspecifi c competition to previous ones. It analyses the distribution of two populations M and N in two environments P and G , with both species using the same resources in both environments. In the fi rst statement of this theory, both environments differ in the ability of each species to use them. That is, M is specialist, suppose in P , and N in G . The central argument of isoleg theory can be stated as follows (Fig. 8.1 ). (1) At low densities of both species, they are segregated into their preferred environments. (2) Suppose the density of M is equal to 0, then N will use G until the density reaches the threshold Ng and also begins using P. (3) Suppose now that the density of M begins to increase but there is no interspecifi c competition. The only thing that changes is how the changes of strategy N are represented because now, instead of a vector, there is an area where there is a threshold line that demarcates the change in strategy. 8.2 Isoleg Model: Distribution of Guilds Composed of Fitness Maximisers 119

Fig. 8.1 Steps in obtaining selective–opportunistic thresholds. The population N is specialist in habitat P , while population M is specialist in G . When population N is alone, at low densities it uses only P , while at high densities, it uses both P and G . With both species but without competition, N becomes opportunistic when it reaches a certain density threshold in P . This threshold is constant regardless of the density of M . Similarly, M is selective in G at low densities and uses both habitats when it reaches a density threshold, which is independent of the density of N . When species compete, when M density increases, the selective– opportunistic threshold (isoleg) increases

This threshold is an isoleg. (4) Now we incorporate interspeciﬁ c competition. By increasing the density of M in environment P , the availability of resources in P decays against the possible use of P by G . That is, the increase of M in B produces a decrease of quality in P for the other population. Therefore, there is a decrease in the advantage for N in using P , so the specialist–generalist threshold for N must change. When patch B becomes poorer, more individuals consuming population N in G will be needed for the qualities of both sites to become equal. This is represented in the graph with a positive slope line. (5) Using the same rationale for M , the isoleg for the other population is obtained. Figure 8.2 shows graphically how to obtain the density threshold at which a species happens to be selective to be opportunistic. The overall result for the case of two competing species, each specialising in a different type of site, is that there are three regions of habitat use: in one region, M specialises in G and N is opportunistic; in another, N specialises in G and M is opportunistic; and in the third, both species are selective in their preferred environments. One of the conclusions relevant to the community ecology of this model is that it demonstrates that habitat selection is an ecological force to achieve competitive coexistence. Pimm et al. ( 1985 ) modiﬁ ed this model to include differences in competitive abilities between the two species (Fig. 8.3 ). In this case, both species are specialised in the same habitat type but one species is a better competitor. This model is similar 120 8 Distribution of Species Assemblages

Fig. 8.2 Isoleg model. Selectivity regions for populations N and M in habitats P and G , after steps described in Fig. 8.1

Fig. 8.3 Isoleg model. Selectivity regions for N and M populations in habitats P and G. In this version of the model, both species are specialised in G , but N dominates over M 8.3 Macroecology 121 to the ideal distribution model with competitive differences developed by Sutherland and Parker 1985 and described in Sect. 3.5 , but is applied here to two species instead of one. For the subordinate species, the isoleg initially has a negative slope because the increased density of the dominant species diminishes the quality of their preferred habitat. When the isoleg touches the axis of the dominant species, all members of the subordinate species have an opportunistic strategy. If the density of dominant individuals occupying the best initial habitat continues to increase, the isoleg begins to separate two strategies: opportunistic on the left side and selective in the initially rejected habitat (P ). The isoleg has a positive slope for the dominant species, due to the effect of the interference imposed by subordinate species in the poor quality site. If this effect did not exist, the isoleg would be completely vertical. Pimm et al. (1985 ) applied the model of interspeciﬁ c competition to a community of hummingbirds from the south-east of Arizona, USA. The dominant species of blue-throated hummingbirds Lapornis clemenciae was dominant over two other species, Rivoli’s hummingbirds Eugenes fulgens and black-chinned hummingbirds Archilochus alexandri . They used artiﬁ cial feeders with sweetened water solutions, representing a good habitat (1–2-M sucrose) and a poor habitat (0.3-M sucrose). These authors investigated habitat use at different densities that were due to natural variations and experimental perturbations and found support of the predictions of the model.

8.3 Macroecology

Macroecology is a young discipline that investigates at a coarse-scale approach. It represents another response of ecologists to the lack of spatial considerations found in traditional ecology. Macroecology starts describing large-scale patterns and then ﬁ nds explanations to them. Gaston and Blackburn (2000 ) enumerated 26 different patterns of principal concern to macroecologists: the species–area relationship, species richness–isolation relationship, peninsular effect, local–regional richness relationship, latitudinal gradient in species richness, species richness–energy relationship, longitudinal gradient in species richness, altitudinal gradient in species richness, species–range size distribution, geographical range structure, range size– niche breadth relationship, extinction–range size relationship, speciation–range size relationship, nestedness of species occurrence, spatial turnover in species identities, latitudinal gradient in geographical range size (Rapoport’s rule ), abundance–range size relationship, abundance–niche breadth relationship, latitudinal gradient in abundance, species–abundance distribution, species–body size distribution, extinction–body size relationship, speciation–body size relationship, range size–body size relationship, latitudinal gradient in body size (Bergmann’s rule), and abundance– body size relationship. Books by Brown (1995 ) and Gaston and Blackburn (2000 ) are excellent reviews on the subject. We will take one example of the way in which macroecology analyses these coarse-scale patterns. One of the most important generalisations found by 122 8 Distribution of Species Assemblages

Fig. 8.4 Example of the abundance–range relationship (modiﬁ ed from Brown 1984 ). Data on abundance and distribution of zooplankton species in 45 lakes in north-western Ontario (from Patalas 1971 )

Fig. 8.5 Apparent correlation ( solid line ) is obtained in an abundance–range graph due to differences between two taxonomic groups, represented by open and closed circles . The correct analyses (dotted lines ) show no correlations

macroecologists is that locally abundant species tend to be widespread and locally rare species tend to be narrowly distributed (Fig. 8.4 ), i.e. a positive interspecifi c abundance–range size relationship (Brown 1984 ). Gaston et al. ( 1997 ) reviewed the evidence for this pattern and the hypotheses that have been proposed to explain it. They described seven different hypotheses. Sampling effort artefact: This is a necessary, and frequently demanded, null hypothesis. It can occur when sampling effort is insuffi cient, because by chance rare species will tend to be unrecorded in some locations. In other words, the greater the average abundance of a species, the more likely would it be to appear in samples and the greater would be its apparent range. Phylogenetic non - independence: This is a general problem with the interspecifi c comparative method that has been extensively analysed by Harvey and Pagel ( 1991 ). Phylogenetic relatedness violates a main assumption of this type of analysis, which is that of independency between data points. Figure 8.5 illustrates an example in which an apparent correlation is obtained in an abundance–range graph due to differences between two taxonomic groups. Harvey and Pagel (1991 ) have provided methodological tools to overcome this undesired effect. Sampling area artefact: This is another type of sampling artefact due to inequalities in range positions. Section 7.3 described how species abundance normally follows a 8.3 Macroecology 123 hump shape within its biogeographical range (Fig. 7.2 ). If the study area is selected when some species (but not others) are closer to the edges of their ranges, this effect of range position will produce a positive relationship between abundance and range size. Niche breadth hypothesis: In Sect. 7.3 we described Brown’s niche hypothesis (Brown 1984 ) on the origin of the normal distributions of abundance over space. This hypothesis predicts that generalists should have larger ranges than specialists. If the range of a set of species is compared, a different pattern is expected between species with similar niches than between species with different requirements. If niches are multidimensional, and if spatial variation in the environment tends to be autocorrelated, then there should be a positive correlation between abundance and distribution for those species that differ in only a very few niche dimensions. Under this assumption, tolerant species with wide niche breadths will share many sites over a relative large area, while species that are otherwise similar, but have narrow niche breadths, will necessarily be restricted to the few sites within the limited geographic region where their needs can be satisfi ed. The differential processes associated with species niche demands will eventually produce an abundance–range size relationship. Niche position hypothesis : This hypothesis postulates that wide range species use resources that are both locally and widely abundant while the opposite occurs for species that are locally rare. An example of this hypothesis corresponds to host–parasite and prey–predator systems in which parasites or predators that are specialised for widespread and abundant host/prey will be locally abundant and widespread. Habitat selection hypothesis : This hypothesis incorporates a local mechanism, which was analysed in Sect. 3.4 . A prediction of ideal-free distribution models is that, driven through intraspecifi c competition, locally more abundant species will tend to occupy more habitats. The consequence of this density-dependent habitat selection mechanism is that abundant species will be more widespread than rare ones. Metapopulation hypothesis : In accordance with Gaston et al. ( 1997 ), another possible mechanism involved in the abundance–range size relationship is metapopulation dynamics. Although several metapopulation models have been developed to understand the determinants of biogeographical borders, it is not straightforward to understand how these predictions can be extrapolated to obtain a positive trend between abundance and range. Gaston et al. ( 1997 ) found that a number of predictions are common to all of the biological (as opposed to artefactual) mechanisms, but the combination of predictions and assumptions made by each is unique, suggesting that, in principle, conclu- sive tests of all of the mechanisms are possible. Based on present evidence, they concluded that there is no single mechanism that has unequivocal support and argued that there are at least four main reasons: the models have not been adequately tested, it is not possible to discriminate among hypotheses, none of the mechanisms are appropriate, and some or all of the mechanisms are complementary. The latter is probably the most plausible explanation, and most of the present research in macroecology tends to integrate not only hypothesis over one pattern but also patterns between them. 124 8 Distribution of Species Assemblages

8.4 Species Distribution Models Applied to Species Ensembles

Section 7.2 presented species distribution models, the coarse-scale version of site suitability models. The use of these models has grown signifi cantly in recent years and the range of applications has expanded both in the spectrum of applications and in the range of theoretical topics they cover. We have seen how most of these models are built with the assumption that interspecifi c interactions (competition, predation, and parasitism) are not relevant factors in modulating species geographical distribution or that their effect is homogeneous, both spatially and temporally. One of the more recent theoretical efforts has been an attempt to model not only the distribution of individual species but of species assemblages (Baselga and Araujo 2009 ; Lucifora et al. 2011 ; Trebilco et al. 2011 ; Broennimann et al. 2012 ; Kissling et al. 2011 ). This section will describe some of these approaches. Ferrier and Guisan ( 2006 ) fi rstly reviewed ‘community - level ’ distribution models . As is the case for typical species distribution models, the community-level approach has the following characteristics: (1) it fractionates the study area with a grid, to which both species data and environmental predictors refer, and (2) it relates both types of data to extrapolate patterns across an entire study area. There are three modelling strategies. In the classifi cation - then - modelling strategy, the model is applied to the relationship between environmental information and assemblages of species, which were previously obtained from species data using some criteria of classifi cation. The predict fi rst , assemble later strategy classifi es species distributions that result from individual species distribution models. The assemble and predict together strategy allows single multi-response models to be fi tted to data for several species. Some of these community-based models are canonical quadratic ordination, canonical additive ordination, generalised dissimilarity modelling, multiple adaptive regression splines adapted for community modelling, and multi- response neural networks (Ferrier and Guisan 2006 ). Baselga and Araújo (2009 ) compared the performance of generalised linear modelling and canonical quadratic ordination for projecting distributions of species under climate change scenarios. They found that projections from these two methods varied both in accuracy and in the diversity of patterns yielded. Kissling et al. ( 2011 ) performed another review of published attempts at incorporating species interactions into the classical species distribution model framework, and they recognised three types of methods. The fi rst approach involves incorporating an interacting species into the model as an additional predictor. This is clearly the simplest method but presents potential statistical problems: false absences and collinearity within regression models when two species respond similarly to ecological factors. A second approach implies modelling the distribution of interacting species separately and then representing interactions by restricting the distribution of one species to the modelled distribution of the other. This approach should be a special case of the predict fi rst , assemble later strategy described by Ferrier and Guisan (2006 ). The third approach is to incorporate interaction processes into the model. This is the most diffi cult approach: it has only been formalised with plants, 8.4 Species Distribution Models Applied to Species Ensembles 125

Fig. 8.6 Schematic representations of two methods for modelling species assemblages using interaction matrices. (i) Static distributions use spatial presence/absence data and can be applied to model species’ co-occurrences in the error matrices of multivariate logistic regressions. (ii) Temporal dynamics use time series of multi-species abundances and can be modelled with multivariate versions of population dynamics models (modifi ed from Kissling et al. 2011 ) by representing competition for resources under climatic stress in forest gap models (Bugmann 2001 ). Kissling et al. ( 2011 ) conducted an innovative and inclusive analysis of the conceptual and methodological strategies that can be followed to model biotic interactions in multi-species assemblages over large spatial extents (Fig. 8.6 ). They found that the main limitations are (1) a poor knowledge of spatio-temporal variation in the existence and strength of species interactions and (2) model complexity. They proposed that new approaches that incorporate multispecies complex interactions into the projection of species distributions and community structure should include interaction matrices for multispecies co-occurrence data sets across large-scale environmental gradients. An example is described of the use of the species distribution approach to ensembles. Leathwick et al. (2005 ) implemented multivariate adaptive regression splines that allow a model to be fi tted that describes relationships between multiple species and their environment. This statistical method is capable of fi tting complex, non- linear relationships between species and predictors and can be used in one of its implementations. They use these models to analyse the environmental relationships of fi fteen diadromous fi sh species using distributional data from New Zealand rivers and streams. In a previous study, the same authors tested the same data using individual multivariate adaptive regression splines analysis. However, multi-species models may offer advantages by their ability to identify a set of environmental predictors that best recover overall variation in species composition. 126 8 Distribution of Species Assemblages

Leathwick et al. ( 2005 ) studied diadromous fi sh, which comprise a highly distinctive component of New Zealand’s indigenous freshwater fauna. The majority of the 15 species included in the analysis are from the families Galaxiidae and Eleotridae. Most of these species spend the majority of their lifespan in freshwater, rather than in the sea. Most are widely distributed in New Zealand, including on its offshore islands, and a number also have a wider regional distribution. The predictors they used can be divided into factors describing the character of the river segment within which the sampling site was located, downstream factors affecting the ability of diadromous fi sh to migrate from the sea to that river segment, and upstream/catchment-scale factors affecting environmental conditions at the sampling site. Segment scale predictors were summer air temperature, winter air temperature, segment fl ow, riparian shade, and segment slope. Downstream predictors were distance to coast, downstream average slope, and maximum downstream slope. Upstream/catchment-scale predictors were average temperature, days with rain greater than 25 mm, average slope in the upstream catchment, area with indigenous forest, average phosphorus and calcium concentrations of underlying rocks, average hardness of underlying rocks, area of peat, and area of lake. Multivariate adaptive regression splines is a technique in which non-linear responses between a species and an environmental predictor are described by a series of linear segments of differing slope, each of which is fi tted using a basic function (Hastie et al. 2001 ). Breaks between segments are defi ned by a knot in a model that initially over-fi ts the data and which is then simplifi ed using a backward/ forward stepwise cross-validation procedure to identify terms to be retained in the fi nal model. Leathwick et al. ( 2005 ) obtained two main types of results with their analysis: (1) relative importance of predictors independently of fi sh species and (2) interspecifi c comparison of individual distributions. The fi rst analysis indicated that a relatively small set of predictors plays a dominant role in explaining variation in the probability of occurrence for most fi sh species. These included key functional aspects of stream character or their correlates, such as summer temperature, distance from the sea, and stream size, together with catchment-scale drivers of variation in water temperature and fl ow variability. An examination of probabilities fi tted for individual species by the multivariate adaptive regression splines analysis indicated that there were marked differences among the environments where they most frequently occurred. For example, species varied widely in the distances that they penetrate upstream from the coast. This study is an excellent example of the advantages of applying these new statistical methodologies to investigate the distribution of species ensembles. Firstly, Leathwick et al. ( 2005 ) discussed how the environmental relationships that emerge from their modelling approach accord strongly with previously published accounts of diadromous fi sh ecology at various scales of study. This result shows that this kind of method can capture signifi cant traits of the process involved in community distributions. Chapter 9 Distribution of Genes

Tunez JI , Cassini MH , and Centrón D

9.1 Introduction

Technological innovations in spatial analyses coupled with the increased availability of spatial data and hypervariable genetic markers have resulted in great advances in our ability to study the infl uence of landscape on genetic variation in space (Storfer et al. 2007 ). Landscape genetics and phylogeography are the two research fi elds that are primarily concerned with understanding the distribution of genetic variation across natural environments (Avise et al. 1987 ; Manel et al. 2003 ). Both fi elds overlap in their goals and methods; however, several important distinctions exist between them (Wang 2010 ). The most important is that phylogeography investigates the historical processes generating patterns of genetic variation (Manel et al. 2003 ; Storfer et al. 2007 ), while landscape genetics concentrates on the contemporary processes affecting genetic variation in space (Knowles 2009 ). This chapter deals with landscape genetics and the ecological process associated with the distribution of genes rather than with the historical processes that are studied by phylogeography. Landscape genetics is a rapidly growing research fi eld that integrates methods from landscape ecology, spatial statistics, geography, and population genetics to understand the spatial distribution of genetic variation (Holderegger and Wagner 2006 ; Storfer et al. 2007 ). It aims to provide information about the interaction between landscape features and microevolutionary processes, such as gene fl ow, genetic drift, and selection (Manel et al. 2003 ) (Fig. 9.1 ). This chapter is divided into another three sections. The following section will be dedicated to reviewing the microevolutionary processes than can affect the distribution of genes in space, the concept of genetic structure, and the role that functional connectivity plays in avoiding genetic differentiation. In the second section, the most common spatial patterns of genetic variation observed in eukaryotes will be revised, including some exemplifying studies in different taxa. The third section will be dedicated to the distribution of bacterial genes. Integrons, which are genes associated with horizontal gene transference, will be used as an example.

M.H. Cassini, Distribution Ecology: From Individual Habitat Use 127 to Species Biogeographical Range, DOI 10.1007/978-1-4614-6415-0_9, © Springer Science+Business Media New York 2013 128 9 Distribution of Genes

Fig. 9.1 The landscape genetics approach

9.2 Biological and Environmental Features Leading to Population Genetic Structure

The amount and type of genetic diversity within a species varies across its natural range due to different microevolutionary processes that can affect the distribution of genes in space. The resulting genetic differentiation can be a consequence of different processes including (1) natural selection favouring different genotypes in different environments, (2) random processes in the transmission of alleles from one generation to the next (genetic drift), or (3) limited gene fl ow between populations. The latter process is the most relevant from the point of view of spatial genetics. Gene fl ow is defi ned as the transfer of genes from one population to another of the same species, either by migration or by the dispersal of seeds and pollen. A general consequence of a high rate of gene fl ow is the lack of genetic differentiation between populations interchanging genes. Two important features of gene fl ow are the dispersal capacity of migrants and the direction in which dispersal occurs (Garant et al. 2007 ). However, gene fl ow can be affected by other factors. The spatial arrangement of populations itself can affect the degree of genetic interchange between populations. Moreover, physical or ecological barriers can prevent migration between geographically close populations giving rise to unexpected genetic differentiation if only the individual capacity of dispersal and the distance between populations is taken into account (Epperson 2003 ). The three microevolutionary processes mentioned above can lead in different ways to a genetic differentiation between populations and, consequently, to a genetic population structure. In the remainder of this section we will focus on the biological features of species as well as the environmental features that can lead to a genetic population structure. Population substructure is almost universal among organisms. In Part II, we have seen distribution patterns at the level of individuals, aggregations, groups, and 9.3 Geographical Genetic Patterns Observed in Eukaryotes 129

subpopulations. When there is population subdivision, there is almost always some genetic differentiation among the subpopulations and an unequal distribution of genes (Hartl and Clark 1997 ). The structuring of genetic variation arises when the individuals in a natural population do not breed at random. This departure from panmixia can be related to the mating system and social structure (Storz 1999 ), dispersal barriers (Ernest et al. 2003 ), and spatial distribution of individuals or populations (Peakall et al. 2003 ). All of these processes lead, directly or indirectly, to reproductive isolation and a decrease in effective population size, which was defi ned by Wright (1938 ) as the number of individuals that would result in the same inbreeding, or genetic drift, if they behaved in the manner of an idealised population. Dispersal barriers and the spatial distribution of individuals and populations also act to reduce the probability that an individual will mate . Geographical space will infl uence genetic structure through an interaction between spatial heterogeneity (Li and Reynolds 1995 ) and dispersal (Olden et al. 2004 ). More specifi cally, spatial heterogeneity modifi es how effectively distant (Verbeylen et al. 2003 ) individuals are from one another (Schooley and Wiens 2003 ). As such, it changes the degree of functional connectivity, the species ability to disperse within the limits imposed by the interaction between its natural history traits and the surrounding landscape, among populations and individuals (McIntyre and Wiens 1999 ). Following Wright’s island model (1931 ), functional connectivity is proportional to the effective number of dispersers. Thus, high levels of connectivity would be associated with low levels of genetic differentiation and a random spatial distribution of genetic variation. On the other hand, low levels of connectivity would be correlated with high levels of genetic differentiation and a structured spatial distribution of genetic variation. To apply Wright’s model to natural populations and landscapes, it is fi rst necessary to estimate the levels of functional connectivity within the studied landscape, and to do this, we fi rst have to take into account the spatial scale. Habitat heterogeneity, and therefore functional connectivity, is scale-dependent (Calabrese and Fagan 2004 ). In contrast to traditional population genetics studies, which were limited in spatial inference (Storfer et al. 2007 ), the integrative approach of landscape genetics has addressed a broad array of questions, including identifying specifi c barriers to dispersal, quantifying diversity, inferring the effect of landscape change, identifying migrants in relation to landscape conditions, estimating source–sink and metapopulation dynamics, predicting the spread of disease or invasive species, and comparing the observed genetic patterns between contemporary and historic landscapes (reviewed by Storfer et al. 2010 ).

9.3 Geographical Genetic Patterns Observed in Eukaryotes

As introduced in previous sections, patchily distributed populations can become genetically distinct over time as a consequence of random genetic drift and in response to locally varying selection. Gene ﬂ ow counters these two processes and 130 9 Distribution of Genes acts as a homogenising force that opposes differentiation (Manier and Arnold 2005 ). The net result of all three processes is some spatial pattern of population genetic structure.

9.3.1 Random Patterns

Random patterns are the simplest spatial genetic patterns in which there is a random distribution of genetic variation among individuals within populations. It implies that the presence of one individual has no infl uence on the distribution of others (Manel et al. 2003 ) and that landscape features do not impose any restriction to gene fl ow. This random pattern of genetic variation has been described, for example, in the North Atlantic eel, Anguilla rostrata. Juveniles of this species inhabit coastal and island waters of the Americas. During sexual maturation, they migrate to the western tropical mid-Atlantic Ocean, where spawning takes place. Leptocephalus larvae disperse to coastal regions, partly or largely through passive transport by ocean cur- rents. Williams et al. (1984 ) suggested that spawning in this species is essentially panmictic, which implies that collections of juveniles from any locality along the species distribution are all samples of the same breeding population. Two years later, Avise et al. (1986 ) used restriction site data on mitochondrial DNA to evaluate the geographic differentiation in Anguilla . Using 14 informative restriction endo- nucleases and 138 eels, they found no evidence of genetic divergence among samples from a 4,000 km stretch of the North American coastline, suggesting that barriers to gene fl ow do not exist in the study area. The results obtained emphasise the importance of life history in shaping population genetic structure. Random patterns of genetic variation have also been described in Ornithogalum montanum , a Mediterranean plant species belonging to the Liliaceae family, which dwells in arid and stony grassland. Piglucci and Barbujani (1991 ) studied the geographic variation of the species in Italian populations. Using 15 electrophoretic loci, they found that the allele frequencies of only 19 of the 40 alleles studied appear heterogeneously distributed, but only three of them show signifi cant spatial structure. The authors stated that these fi ndings support a model of population differentiation in which the genetic relationships between isolates do not depend on their spatial distances.

9.3.2 Clines

Clinal variation is deﬁ ned as a geographical trend in the frequency of a given allele. Clines can result from natural selection, by migration, or by a combination of both processes (Frankham et al. 2002 ). Natural selection can produce a clinal pattern of genetic variation when the relative ﬁ tness of one genotype varies geographically across an environmental gradient almost continuously, for example, according to latitude, altitude, aridity, or salinity 9.3 Geographical Genetic Patterns Observed in Eukaryotes 131

(Hartl and Clark 1997 ). If sufﬁ ciently stable in time, this selection gradient along a region can result in an allelic frequencies gradient along the region. For example, latitudinal clines of malate dehydrogenase-1 (Mdh-1) allozymes have been described within the honey bee Apis mellifera ’s populations in Europe, North America, and South America (Del Lama et al. 2004 ). The clinal pattern observed is characterised by a ‘medium’ electrophoretic allele that increases in frequency with increasing latitude and a ‘fast’ allele that decreases with latitude on all the three continents (Nielsen et al. 1994 ). The best environmental variable that explains this clinal pattern observed is the average daily high temperature for July on all continents, which is associated with unequal thermostability of allelic forms of malate dehydrogenase in honey bees (Cornuet et al. 1995 ). Taking into account that A. mellifera is a species that was introduced to the American continent, the currently observed Mdh allelic clines observed in America would have been established within the last 150 years (Nielsen et al. 1994 ), demonstrating how quickly natural selection can produce this kind of clinal pattern of genetic variation. Clinal patterns can also result when differences in allele frequencies in each local population at the extremes of the range result from chance processes (i.e. different founding populations) and migration from the extremes into the intermediate zone produces the cline. This is the case illustrated by B blood group allele frequencies in human populations across Eurasia (Mourant et al. 1976 ). Prior to about ad 500, the B blood group allele was absent in Western Europe, but it existed in high frequencies in the east. Between ad 500 and 1500, Mongols and Tartans invaded Europe and the invaders mated with local people, interchanging their alleles. As a consequence, a cline showing a gradual decrease in the frequency of the B blood group allele can be observed in Eurasia from east to west. The B blood group allele is currently still absent in native Basques in Spain and in other isolated populations that did not have contact with invaders and their descendants. Finally, clines can also form when there is a balance between natural selection for different alleles in different habitats, which is termed local adaptation, and there is migration across habitats (Frankham et al. 2002 ). This is the case for the cline observed in heavy-metal tolerance in colonial bent grass passing from polluted mine wastes to nearby unpolluted pastures in Wales (Bradshaw and McNeilly 1981 ). Selection favours heavy-metal tolerant plants on the mine waste, but acts against them in the unpolluted pasture. Pollen ﬂ ow moves alleles among populations, such that there is a gradation in the frequency of heavy-metal tolerant plants across the transition zone between the two habitats. Heavy-metal tolerance declines with distance in the downwind direction, as pollen reaches more distant pastures.

9.3.3 Ecotypes

Large differences in many loci between populations can be produced as a result of natural selection and/or genetic drift. When selection appears to be the primary factor leading to differentiation, as suggested by major differences in environment predictably being associated with particular combinations of genetic variation, the 132 9 Distribution of Genes

Fig. 9.2 Cluster analysis of silicicolous and calcicolous (grey , italics) populations of Silene nutans (modiﬁ ed from Van Rossum et al. 1997 ) observed pattern is called ecotypic. This geographical pattern of genetic variation has been described in a great variety of taxa, from bacteria to vertebrates and plants (e.g. Moore et al. 2005 ; Keeley et al. 2007 ; Climent et al. 2008 ). Silene nutans is a predominantly outcrossed, herbaceous plant that belongs to the Caryophyllaceae family. It is a steppe species with a widespread Eurosiberian distribution. In Belgium, at the north-western margin of its geographical range, S. nutans is a rare species, which has evolved a silicicolous and a calcicolous eco- type, with contrasting morphometric traits. Van Rossum et al. ( 1997 ) examined the genetic diversity and population genetic structure of the species for seven allozyme loci in 16 silicicolous and 18 calcicolous marginal and disjunct populations and related its ﬁ ndings to ecotypic differentiation. The results obtained from genetic distance measures and cluster analysis revealed that the populations were differenti- ated according to their ecotypic property in two distinct gene pools, despite the fact that some silicicolous populations were geographically intermixed with calcicolous populations (Fig. 9.2). As the two ecotypes have evolved genetic adaptations to the 9.3 Geographical Genetic Patterns Observed in Eukaryotes 133 mineral composition of their respective soils, the authors found evidences for correlated differences in allozymes and ecological traits.

9.3.4 Isolation by Distance

The term isolation by distance was fi rst used by Wright (1943a , b ) to describe patterns of population genetic variation that derive from spatially limited gene fl ow. Isolation by distance is defi ned as a decrease in the genetic similarity among populations as the geographic distance between them increases. Malécot ( 1968 ) formalised the relationship between genetic differentiation and geographic distance, showing that individuals that are closer in space are also more genetically similar. The author demonstrated that there is a strong relationship between geographic distance and genetic distance, which implies that gene fl ow and dispersal are in some way related. Genetic patterns of isolation by distance have been extensively described in nearly all major groups of organisms, including plants, insects, marine invertebrates, freshwater and marine fi shes, amphibians, birds, and mammals (Epperson 2003 ). Pogson et al. (2001 ) examined the effect of geographic scale on the pattern of isolation by distance in the Atlantic cod Gadus morhua . The authors reported signifi cant relationships between the inferred levels of gene fl ow and geographic distance in this species at ten nuclear restriction fragment length polymorphism loci at a regional scale of 1,600 km in the western north Atlantic region. This relationship was in accordance with that previously detected over the entire geographic range of the species (7,300 km). Despite the authors fi nding a weak population structure, the distance separating populations explained between 54 and 62% of the variation in gene fl ow depending on whether nine or ten loci were used to estimate Nm. The authors suggested that the correlation found between gene fl ow and distance at small and large spatial scales shows that dispersal distances and effective population size are much smaller than previously predicted and that the recent age of populations, rather than extensive gene fl ow, may be responsible for its weak population structure. As a consequence, interpreting limited genetic differentiation among populations refl ecting high levels of ongoing gene fl ow should be made with caution.

9.3.5 Stepping Stone

In natural populations, individuals are often distributed more or less discontinu- ously to form numerous colonies and may be exchanged between adjacent and nearby sites. To analyse such a situation, Kimura ( 1953 ) proposed a model which he termed the stepping-stone model of population structure. His model describes a large population subdivided into subpopulations aligned one after the other in one 134 9 Distribution of Genes

Fig. 9.3 Stepping-stone pattern in South American sea lions, Otaria ﬂ avescens (modiﬁ ed from Túnez et al. 2007 )

dimension, on nodes of a grid in two dimensions or on the nodes of a cube in three dimensions. Subpopulations are equally spaced, and migration occurs only among adjacent subpopulations. A decade later, Kimura and Weiss (1964 , 1965) analysed the effects of isolation by distance within Kimura’s model of population structure. They derived expectations as to the correlation in allele frequencies between populations that were at any distance (in this case, steps) apart, in one, two, and three dimensions. Their results show that the correlation in allele frequencies among populations falls exponentially with distance. In addition, they show that the rate of differentiation increases with the number of dimensions. Stepping-stone models have been described, for example, in different species of marine mammals (Lehman et al. 1993 ; Lamont et al. 1996 ; Túnez et al. 2007 ). The South American sea lion Otaria fl avescens is distributed along approximately 10,000 km of the southern coast of South America (King 1983 ). The distribution of breeding colonies of the species shows heterogeneity at three geographic scales. Sea lions congregate in 37 breeding colonies, which are clumped in two sectors of the Atlantic coast separated by 100 km, and these colonies are separated from those in the Pacifi c Ocean by thousands of kilometres. Túnez et al. ( 2007 ) found a strong correlation between this spatial pattern of distribution of individuals and the distribution of mitochondrial DNA cytochrome b haplotypes. Populations of sea lions from the same cluster of colonies do not show signifi cant differences in haplotype frequencies, while there are signifi cant differences between clusters of colonies. This pattern of haplotype frequency distribution resembles a stepping-stone model of gene fl ow among colonies (Fig. 9.3 ), where most gene fl ow occurs between colonies within a cluster and fl ow between clusters is absent or very low (Kimura and Weiss 1964 ). Stepping-stone models have also been described in benthic marine invertebrates, where larvae movements are thought to determine dispersal capability. The solitary scleractinian Balanophyllia elegans possesses crawling larvae capable of only limited dispersal. Using the frequencies of eight polymorphic allozyme loci, Hellberg 9.3 Geographical Genetic Patterns Observed in Eukaryotes 135

(1995 ) examined the pattern of gene fl ow in B. elegans at a 1–50 km spatial scale to determine (1) whether gene fl ow conformed to the expectations of the stepping- stone model and (2) whether continuing long- distance gene fl ow or historical changes in gene fl ow were responsible for the weak relationship between gene fl ow and distance observed previously at the range-wide spatial scale (4,000 km). The author found that populations were signifi cantly subdivided among localities separated by 1–50 km and that the observed results conformed to equilibrium expectations for a linear stepping- stone model. However, those from the range-wide spatial scale did not, which implies that the mechanisms conferring patterns of genetic differentiation between localities in B. elegans differ with spatial scale. At a scale of 1–50 km, continuing gene fl ow and drift have equilibrated and the process of isolation by distance may facilitate local adaptive change. At a broader spatial scale, historical changes in gene fl ow disrupt the equilibration of gene fl ow and genetic drift, so that genetic differentiation may not increase continuously with separation between populations.

9.3.6 Discontinuities to Gene Flow

As was mentioned in the fi rst section of this chapter, the rates of migration between populations may be affected by factors other than distance and the direction of dispersal. The spatial arrangement of populations may affect which populations exchange migrants. Moreover, physical or ecological barriers may prevent migration among populations that are geographically close (Epperson 2003 ). A genetic discontinuity is defi ned as a geographic zone of sharp genetic change (Manel et al. 2003 ), where breaks in gene fl ow occur without any obvious cause (Guillot et al. 2005 ). The fl ightless ground beetle Carabus violaceus is one example of how physical barriers may prevent migration and produce genetic differentiation between close geographically populations. Keller and Largiader (2003 ) used six microsatellite loci to assess the genetic population structure of this species in a Swiss forest, which is divided into several fragments by a highway and two main roads. The authors found that the number of roads between sites explained 44% of the variance in pairwise FST estimates and that the largest genetic differentiation was observed between samples separated by the highway. Furthermore, a comparison of allelic richness showed that the genetic variability in a small forest fragment isolated by the highway was signifi cantly lower than in the rest of the study area. The authors state that these fi ndings strongly support the hypothesis that large roads produce a genetic discontinuity to gene fl ow in C. violaceus . Migration among populations can also be prevented by ecological barriers. The South American sea lion O. fl avescens is a good example of how the lack of suitable habitats for reproduction produces discontinuities to gene fl ow. As previously explained, O. fl avescens shows a patchy distribution of breeding activity. In the Atlantic coast, breeding colonies aggregate in three areas, Uruguay, north- central Patagonia, 136 9 Distribution of Genes and southern Tierra del Fuego (Túnez et al. 2008 ). Uruguay and north-central Patagonia breeding areas are separated by a 1,200 km segment of coast without breeding activity, the coast of Buenos Aires province. Genetic studies carried out in these breeding areas have shown that they only shared an ancestral mitochondrial haplotype, which suggests that current gene fl ow between breeding areas is almost absent (Túnez et al. 2007 ). On the coast of Buenos Aires, breeding colonies were abundant until the second half of the nineteenth century (Rodríguez and Bastida 1998 ). The disappearance of these colonies appears to be related to the large-scale pattern of human settlement that occurred at the end of that century, when the coastal zones were rapidly colonised by humans. Breeding colonies in Buenos Aires and gene fl ow between Uruguay and north-central Patagonia were not re-established, probably due to a combination of factors; the most important is an anthropogenic barrier. Most of the Buenos Aires coast has been used since the beginning of the twentieth century as summer holiday centres during the peak of breeding season of sea lions, making the area unsuitable for the settlement of breeding colonies.

9.3.7 Metapopulations

Section 5.2 describes the distribution of subpopulations within spatially structured populations. Wright ( 1931 ) was the ﬁ rst geneticist who discussed the effect of metapopulation dynamics on population genetics. However, in spite of the large amount of literature about this topic, it remains unclear what patterns in genetic variation are associated with metapopulation structures in nature (Harrison and Hastings 2005 ). Harrison and Hastings ( 2005 ) attempted to clarify seemingly conﬂ icting ideas on how metapopulation processes affect, or are related to, the partitioning of neutral genetic variation within and among populations. The authors showed that population turnover is generally associated with low levels of among population variation, in both the classical metapopulation model (Levins 1970 ) and other types of metapopulations (Boorman and Levitt 1973 ). Therefore, adaptive evolution is unlikely to be promoted by selection among populations. The black-tailed prairie dog Cynomys ludovicianus is a good example that illustrates the genetic consequences of metapopulation structure in nature. Several human activities that modify its habitat contributed to local extinction and a steady decline of prairie dog numbers throughout its range. As a consequence, the species currently live in metapopulations (Hoogland 1995 ). Roach et al. ( 2001 ) analysed the pattern of genetic variation associated with this metapopulation structure. Analysing seven microsatellite loci in 155 individuals, the authors found moderate levels of genetic differentiation in 13 colonies of the species from northern Colorado, USA. These results suggest that extinction and recolonisation of colonies over the last 10 years have not increased genetic differentiation among colonies, which is only expected in metapopulations when initial colonisation is very different to subsequent dispersal (Barton and Whitlock 1997 ). 9.3 Geographical Genetic Patterns Observed in Eukaryotes 137

9.3.8 Source–Sinks

A particular case of metapopulation dynamics, called source–sink, is given when one or more permanent populations (sources) supply individuals to restart one or more transient (sink) populations (Sect. 5.3 ). If there are frequent extinctions in the sink populations and recolonisation occurs only from one source population, less genetic diversity is retained than in a single large population of the same total size (Gilpin 1991 ). Low levels of genetic differentiation between source and sink populations would also be expected due to high rates of gene fl ow. Manier and Arnold (2005 ) used 11 microsatellite markers to examine the population structures of two coexisting species of garter snake, Thamnophis elegans and Thamnophis sirtalis, in order to determine if shared landscape and biology imposed similar population genetic structures. These snakes inhabit a series of ponds, lakes, and fl ooded meadows in northern California and tend to converge on prey type. The authors found that both garter snakes had comparable effective population sizes and bidirectional migration rates, with low but signifi cant levels of genetic differentiation. Asymmetrical gene fl ow revealed large source populations for both species as well as potential sinks, suggesting a frequent extinction–recolonisation dynamic.

9.3.9 The Central–Marginal Model

Peripheral or marginal populations are those on the boundaries of a species geographic range. They exhibit unique properties that are not evident in populations in the centre of the range (Brussard 1984 ), which are embodied by the central–marginal model developed by Lewontin ( 1974 ). The model proposes that marginal populations contain a lower density of individuals, lower levels of genetic variation, and higher levels of genetic differentiation and are more isolated than populations in the centre of a species range. In a recent revision, Eckert et al. ( 2008 ) analysed the results of 134 studies representing 115 species of plants and animals that tested for declines in within- population genetic diversity and/or increases in among population differentiation towards range margins using nuclear molecular genetic markers. The authors found that, on average, 64.2% of studies detected the expected decline in diversity and 70.2% showed increased differentiation and that these trends were positively associated. However, the authors noticed strong taxonomic and biogeographical biases in the revised literature. Twenty-one percent of the animal species studies were frogs, 22% of plants were pines, and 40% of the studies focused on the northern limit of species in the temperate zone of the northern hemisphere, which calls for a cautious generalisation of the results. Moreover, these geographical trends were found in putatively neutral variation, but little effort has been put into analysing this pattern in selective variation, which is likely to inﬂ uence the adaptive potential of populations across the geographical range. 138 9 Distribution of Genes

9.4 Distribution of Genes in Bacterial Communities

9.4.1 Mechanisms of Lateral Genetic Transfer in the Bacterial World

In bacteria, reproduction is not linked to sexuality, as occurs in eukaryotes. Bacteria divide by binary ﬁ ssion, which generates individual clones. However, there are parasexual processes of lateral genetic transfer that, coupled with mutations, generate variability in populations. This mechanism of gene exchange takes place independently of reproduction. It is a driving force in the adaptation to novel niches and the evolution of genomes based on the mobilisation of DNA among different strains, genera, and even phyla ( Touchon et al. 2009 ). With the availability of these mechanisms, it is likely that any piece of DNA from a bacterial source may be subjected to processes involving lateral genetic transfer. Furthermore, bacteria are exposed to a constant provision of DNA most of the time, from which they can acquire large blocks that can generate remarkable quantities of biodiversity, which may in turn be adaptive under selection pressure. The mechanisms that allow this exchange of information at a DNA level are transformation, conjugation, and transduction. The process of transformation is the genetic alteration of a bacterial cell resulting from the uptake and expression of foreign DNA or RNA. Transduction occurs when the bacterial DNA is moved from one bacterium to another by a virus (or bacteriophage). Finally, conjugation is a process in which a bacterial cell transfers genetic material to another cell by cell-to-cell contact. Despite the intense gene ﬂ ow in the eubacterial world, microbial biogeography analyses suggest that genetic population structure in the environment is non-random, implying that there are ecological processes that are responsible at least in part for this spatial variation (Martiny et al. 2006 ).

9.4.2 Role of Class 1 Integron Genes in Antibiotic Resistance

The most manifested and documented example of the changes produced by a continuous pressure/selection over a bacterial community is the antimicrobial stress exerted over the nosocomial bacterial environment. Bacteria respond to this burden in complex and successful processes that most often are due to mechanisms of lateral genetic transfer resulting in extensively drug-resistant and even pandrug-resistant isolates ( Arduino et al. 2012 ). This is one of the issues where this genetic mechanism has been shown to have huge consequences for human health. Among the vast amount of processes identiﬁ ed in gram-negative bacteria that confer antimicrobial resistance that is subject to lateral genetic transfer, the most successful and dissemi- nated mechanisms are the class 1 integrons that can render a single cell resistant to almost all families of antibiotics used in the clinic currently. 9.4 Distribution of Genes in Bacterial Communities 139

Fig. 9.4 Genetic structure of class 1 integrons. The typical class 1 integron with the variable region where the gene cassettes are inserted by activity of the IntI1. The gene cassettes are also shown as circularised DNA with its attC site as an oval . The integrase inserts the cassette in the variable region

Class 1 integrons are genetic units that are made up of three functional components: an integrase gene, intI1; an adjacent attI site, which is recognised by the integrase and the antimicrobial resistance gene cassettes by a site-specifi c recombination event mediated by IntI1; and a common promoter region, which allows expression of antimicrobial resistance gene cassettes inserted in the variable region of integrons (Fig. 9.4 ) (Hall and Stokes 2004 ). The gene cassettes are the mobile elements of the system integrin/cassette, composed by an open reading frame with an attC site. This attC site is recognised as well as the attI1 site by the IntI1 which integrates or excises from the variable region of the integron by a site-specifi c recombination mechanism. More than 130 gene cassettes conferring resistance to antibiotics in almost all antibiotic families have been described including beta-lactams, aminoglycosides, trim- ethoprim, chloramphenicol, fosfomycin, macrolides, lincosamides, rifampicin, and quinolones (Partridge et al. 2009 ). Single isolates harbouring up to seven class 1 integrons have been found, as have class 1 integrons with eight antimicrobial resistance gene cassettes conferring resistance to four families of antibiotics. These results illustrate the enormous fl exibility of both class 1 integrons and bacterial genomes that possess them. To understand the mobility of the class 1 integron/cassette system, it is necessary to know that class 1 integrons are embedded in the transposon Tn 402 which is a mobile DNA element that allows spreading over a wide range of different bacterial lineages. This is how class 1 integrons can be located directly in the chromosome, or as part of genomic islands like the Salmonella genomic island 1 (SGI1), or embedded in larger transposons of the Tn21 family when found in plasmids (Fig. 9.4 ). This genetic backbone functions like clockwork for the mobilisation of antimicrobial resistance genes between a wide range of gram-negative bacteria by very specifi c and effi cient mechanisms, allowing for rapid adaptation to changing antimicrobial pressure environments. The progressive and global increase of multidrug resistance in all geographical regions has been identifi ed as a public health priority according to the World Health Organization, 2011 ( http://www.who.int/drugresistance). In recent years, research 140 9 Distribution of Genes on the function of antibiotic resistance in non-clinical environments has begun to receive attention. This interest is based on the idea that a better understanding of the diversity of patterns and biological functions of antibiotic-resistance mechanisms may eventually help to control the threats towards human and animal health. Nowadays, it is assumed that the environment plays a role in the spread of antimicrobial resistance determinants, including class 1 integrons, among pathogenic species (Stokes and Gillings 2011 ; Nardelli et al. 2012 ). Recently, the relevance and widespread nature of class 1 integrons has been shown in strains isolated in non- clinical environments (Stokes et al. 2006 ; Nardelli et al. 2012 ). This data, coupled with the enormous amount of genes with a huge variety of functions that have the structure of gene cassettes in environmental samples, shows how the class 1 integrons confer a benefi t to the host cell due to their ability to acquire these cassettes, which could provide advantages for survival in hostile environments. This new scenario has opened doors to the understanding of the ecological role of the open environment as a reservoir of genes among bacterial communities: from nature to the hospital and vice versa. Simultaneous analyses at ecological and molecular levels are showing to be a successful strategy for elucidating the role of each component of the processes associated with antimicrobial resistance.

9.4.3 Pattern of Distribution of Class 1 Integron Genes at Different Ecological Scales

Global scale . The study of class 1 integrons in natural environments has only been initiated recently. In these few years of research, they have been discovered in bacteria collected in diverse ecosystems in a wide range of ecoregions. Noteworthy, class 1 integrons are spread in different natural ecosystems, both aquatic and terrestrial, as well as in environments impacted by industrial waste, agriculture, and urban sites. Figure 9.5 shows the distribution of some of these discoveries, illustrat- ing the distribution of class 1 integrons at a global scale. Landscape scale . Singer et al. ( 2006 ) have proposed that landscape ecology can be a useful tool for studies of antibiotic resistance. By addressing the complex array of selection pressures, routes of transmission, and background levels of resistance that exist in ecosystems, it can help to design studies to improve our understanding of the dynamics of resistance. The following is an example of this approach. It is a study of the distribution of bacterial genes related to lateral gene transference, conducted in several landscapes of Tierra del Fuego Island, in the southern pick of South America. The frequency of class 1 integrons in environments with different degrees of urbanisation, including urban sites and sites with no history of human disturbances, was studied on this Patagonian island, which is recognised as being one of the last regions containing wild areas (Fig. 9.6 ) in a model of a culture-based method without antibiotic selection (Nardelli et al. 2012 ). The class 1 integrons that were found more commonly than previously described throughout natural communities were, unexpectedly, not associated with urbanisation 9.4 Distribution of Genes in Bacterial Communities 141

Fig. 9.5 Global distribution of class 1 integrons in non-urbanised sites, based on the few recent studies conducted around the world

Fig. 9.6 Study area for the Nardelli et al.’s (2012 ) study. The geographic sites where the sampling was performed are the following: 1 Ushuaia (Pipo river), 2 road no. 26 and Turbio river, 3 Turbera Maucasen, 4 Escondido Lake (hotel), 5 Escondido Lake (stream), 6 National Park (Ovando river), 7 National Park (Ensenada Bay stream), 8 National Park (Pipo river), 9 Moat (stream at Baliza Davidson), and 10 Ushuaia river. The circles represent low (white ), medium ( dark grey ), and high ( black ) degrees of urbanisation (modifi ed from Nardelli et al. 2012 ) 142 9 Distribution of Genes

Fig. 9.7 Mean occurrence (+SD) of genes and pseudogenes in sites with different degrees of urbanisation. Only sul1 showed statistical signiﬁ cance in this trend (rs = 0.74, p = 0.01) (modiﬁ ed from Nardelli et al. 2012 )

(Fig. 9.7 ). These fi ndings, at a landscape scale, permitted the identifi cation of sites without urbanisation, also named clean sites, as sources and reservoirs of class 1 integrons. High rates of class 1 integrons in clean sites allow higher availability of class 1 integrons that may be then selected under antibiotic pressure and follow their way to the pathogenic species through the successful mechanisms of the lateral gene transfer. Nardelli et al. (2012 ) have suggested that high rates of class 1 integrons in clean sites could be one factor to consider in regions where the nosocomial isolation of multidrug-resistant isolates is more common. The fl ow of class 1 integrons between natural and clinical communities cannot be explained with a single general process at a global scale, but is a direct consequence of the interaction of multiple factors operating at molecular, ecological, phylogenetic, and historical levels, which can be untangled by using a landscape ecology approach. Habitat scale. The analysis of class 1 integrons from one landscape with different sediment compositions can answer additional questions concerning the role of the environment as the source of these genetic elements. Sediments from freshwater catchments differing in both the type and extent of pollution from 30 spatially distinct locations within Victoria, Australia, were analysed for their abundance of class 1 integrons. The metals have heterogeneous distributions in the environment. A strong positive correlation with the heavy metals zinc, mercury, lead, and copper was identifi ed (Rosewarne et al. 2010 ). This fi nding can be explained by the fact that co-selection of metal and antibiotic resistance has been documented, since class 1 integrons are usually embedded in Tn21 which is a transposon that harbours a mercury-resistance operon. Part IV Applications Chapter 10 Distribution Ecology in Conservation Biology

10.1 Introduction

Conservation biology uses ecology as one of the main sciences that provide theoretical paradigms and methodologies to further its aim of developing solutions to the biodiversity crisis. Some of the most important areas of concern in conservation biology are sustainable management, protection of endangered species, design and management of protected areas, the preservation of ecosystems, and global climate change. Before analysing the contribution of distribution ecology to conservation biology, we will see a brief description of each of these areas. The contributions of distribution ecology to biodiversity conservation are arranged according to levels of organisation, in the following sections: individual (and gene) distribution, aggregated distributions, metapopulations and source–sinks, landscape ecology and pattern-based models, species distribution models, and ensemble distribution.

10.2 Environmental Challenges

Endangered species: One goal of conservation biology is to recover populations or species threatened by extinction. Normally, the species under study are called umbrella or key species. They are selected because they (1) play a vital role in main- taining the structure of a biotic community (e.g. prairie dogs Cynomys spp. in North American grasslands), (2) have an aesthetic or symbolic value (cetaceans and large carnivores), or (3) represent a potential economic resource for local people (Neotropical parrots). The most common methods used in the recovery of threatened species are the use of population viability models, programmes for captive breeding, and reintroduction and evaluation of habitat requirements. Other more specific applications are the avoidance of hybridisation and control of diseases. Population viability analysis is a method consisting of design models based on the use of population and environmental information to generate projections on the

M.H. Cassini, Distribution Ecology: From Individual Habitat Use 145 to Species Biogeographical Range, DOI 10.1007/978-1-4614-6415-0_10, © Springer Science+Business Media New York 2013 146 10 Distribution Ecology in Conservation Biology probability of the extinction of a species under natural conditions. These models can incorporate explicit information about the structure or metapopulation genetic structure of populations. Genetic models consider random effects that occur in small populations such as genetic drift, inbreeding, and bottlenecks. These models predict the minimum effective population size that can guarantee a population’s survival. Assessing habitat requirements may result in the discovery of key information about an endangered species that can be used to generate a specific method of recovery. This analysis is also able to detect particularly vulnerable species that are specialised in their requirements. Another aspect of research on endangered species is the avoidance of interspecific interactions that are detrimental to the conservation of these species. The formation of hybrids was recently discovered as a critical problem for certain taxonomic groups, as is the case for carnivorous mammals. Disease transmission is another key factor in the conservation of certain species, for which contact with domestic species represents a great danger of contagion that can decimate populations. Biodiversity and protected areas: The recovery of endangered species to prevent loss of biodiversity is one of the fundamental goals of conservation biology. Due to the high rate of extinction of species that has been occurring in recent decades, a policy solely focused on the protection of individual species may be inefficient under certain circumstances. That is why, since its conception, conservation biology has been concerned with identifying areas of high biodiversity and understanding the ecological processes involved in their maintenance, as well as generating criteria for the design and management of protected areas. One of the central themes in the conservation of landscapes is their permanent loss and fragmentation. Understanding the effect of habitat fragmentation on wildlife communities is essential for conservation biology. The creation of protected areas is usually based on the principle of preserving biodiversity hotspots. Conservation biology has provided conceptual principles for the design of these areas, derived from theories such as island biogeographic models or metapopulation dynamics. In some cases, the areas are selected to protect a group or a key species. As the rate of creation of new national parks and reserves is declining, the emphasis of conservation biology is shifting towards getting better effectiveness of these areas in preserving biodiversity, by improving their design and defining general management principles. This redesign of protected areas includes resizing them or the creation of corridors to improve connectivity between reserves. The new protected areas are usually zoned according to different uses. Both the redesign and management require a deeper an. more detailed knowledge of the landscape and plant and animal communities that conservation practitio- ners are trying to preserve. Ecosystem preservation: Much of biodiversity lies outside of protected areas on land with different forms of human use. Managing these altered ecosystems requires a holistic approach where protection of fauna and flora is harmonised with the rational use of abiotic resources, such as water and soil, and sustained regional development policies. In some places, the anthropogenic ecosystem decline is so great that techniques for the restoration of the original communities that populated these sites have been developed. Another central problem of ecosystem conservation 10.3 Individual (and Gene) Distribution and Conservation Biology 147 is environmental pollution. To assess the impact of pollution on biotic communities, indicator species are used frequently to predict the disruptive effect of pollutants before they eventually impact the entire system. Climate change: The global climate is changing. There has been an increase of temperature in recent decades, probably caused by increased carbon dioxide emis- sions. Temperature increase is accompanied by other climatic, physical, and bio- chemical changes in the planet surface. The effects of climate change on the composition of plant and animal communities and the geographical distribution of wild species have begun to be registered. For example, in marine environments at least 11 processes of change due to the effects of global warming have been described: in temperature, sea level, ocean circulation, ice extent, salinity, carbon dioxide concentrations and pH, rainfall patterns, storm frequency, wind speed and wave conditions, and in climate patterns.

10.3 Individual (and Gene) Distribution and Conservation Biology

The contents of several sections of Chaps. 2, 3, and 4 are based on the premise that the individual strategies for use of space are the result of a process of natural selection and are therefore designs whose currency is oriented towards maximising fitness. The importance of behavioural ecology in conservation biology has been stressed in several reviews, which found that the greatest influence has been in the following areas: in the establishment of sustained use strategies, captive breeding programmes and rehabilitation of endangered species, the estimation of habitat requirements, reducing the impact of human presence, avoiding harmful interspecific interactions, and in reducing the random effects on genetic structure (Cassini 1999). These reviews indicate that distribution ecology at the level of individuals and sets of individuals has had a major influence on the identification of habitat requirements. Several examples of these applications are provided. Valenzuela et al. (2009) conducted a fascinating study on the annual movements of southern right whales Eubalaena australis. These whales make long annual migrations between mid-latitude coastal winter nursery grounds and high-latitude offshore summer feeding grounds (International Whaling Commission 2001). These whales used to have six known feeding grounds in the South Atlantic. The intensive catching by whalers conducted during nineteenth and twentieth centuries reduced this number to only one in South Georgia (International Whaling Commission 2001). At present, they still use only one feeding ground, despite the species’ sustained recovery from near extinction in the early twentieth century to a population that was close to 20,000 in 2008 (International Whaling Commission 2001). This situation surely reduces the availability of better foraging opportunities elsewhere. The reproductive success of southern right whales breeding at Península Valdés, Argentina, is affected by sea surface temperature (correlated with periods of low krill abundance) anomalies off South Georgia (Leaper et al. 2006). This reproductive failure resulting from food constraints is a cause for conservation concern. 148 10 Distribution Ecology in Conservation Biology

Valenzuela et al. (2009) hypothesised that, if calves learn these routes from their mothers and then follow them faithfully for life, matrilines will continue to use the same feeding grounds for generations, jeopardising the utilisation of eventually better foraging grounds. A method to test this hypothesis is to search for genetic differentiation of mitochondrial DNA markers among whales using different feeding and nursery grounds, because this type of marker is transmitted through females, so genetic differences are consistent with female-directed fidelity. Valenzuela et al. (2009) describe the population genetic structure of southern right whales in their feeding grounds by using a novel combination of genetic and stable-isotopic analyses of skin samples collected from live whales at Península Valdés, Argentina. Each skin sample provides the maternal lineage of the whale and information on its feeding location several months before sampling. They found that individuals from a given maternal lineage tend to have similar carbon and nitrogen isotopic values. These associations suggest that, throughout life, individuals tend to follow the migratory routes that they learned from their mothers during their first year. In the following example, the authors explored changes in the home-range sizes of animals that were reintroduced in a natural environment after being bred in captivity. Captive breeding and reintroductions are one of the conservation strategies used for endangered species. Saltz et al. (2000) studied space-use patterns in an introduced population of Asiatic wild ass Equus hemionus. Home-range distributions may allow inferences of social behaviour that can be critical in the success of reintroduction programmes. Between 1983 and 1987, 28 asses were released in Makhtesh Ramon, nature reserve in Israel. Home ranges of dominant males over- lapped only slightly, suggesting that in this population males are territorial. After the first release, only one male home range was established and covered most of the study area. After 6 years, the number of male home ranges increased with a conse- quent decrease in territory size. A corresponding temporal trend was also found in the level of association between males and females; while in the first years the home range of a male was able to include a whole female home range, after several years, females ranged through several male territories. These results indicate that male density is important in shaping the spatial organisation of males and the social system of the population. Territoriality and male dominance is expected to have an effect on effective population size. In the case of the Asiatic wild ass in Makhtesh Ramon, the presence of a single territorial male in the reintroduced population during the 6 years of the reintroduction programme greatly reduced the measure of the population’s risk of extinction (Saltz et al. 2000). The following example highlights the impact of human perturbation on patch use by a specialist species. There is general agreement that specialist species are more vulnerable to habitat degradation than generalists. The argument is that species that exhibit flexible behavioural or physiological traits will be more able to cope with environmental changes than species that are restricted to only a narrow set of resources. If these resources are depleted or lost, populations will decline because individuals are unable to exploit alternative resources or habitats. Nevertheless, this hypothesis is hard to test because it is not always possible to disentangle the numerous effects produced by habitat loss. Chaverri and Kunz (2011) presented the results of a field study that examined the short-term effects of 10.4 Aggregation Distributions and Conservation Biology 149 resource loss in the disc-winged bat Thyroptera tricolor, a species that is highly morphologically specialised for roosting in the developing furled leaves of trees of the order Zingiberales. Because these bats use a discrete, easily identifiable habitat feature that occurs over limited spatial scales and can be temporarily removable, they are an excellent biological model to test experimentally the hypothesis of the link between specialism and vulnerability. The authors experimentally mimicked the loss of roosting habitat (by removing some developing furled leaves) in a natural setting located in south-western Costa Rica. They monitored individuals before and after the perturbation to determine patterns of resource use, spatial behaviour, and group stability. They hypothesised that removing roosting habitat would decrease roost selectivity, increase mobility of individuals while attempting to locate suitable habitat, and increase mortality. To mimic habitat loss, they cut all plants that could potentially be used as roosts in the experimental plots. This includes all four species of Heliconia and two Calathea, which were the exclusive genera used by bats before the experiment. The experiment produced a significant perturbation on habitat use, with home- range size increased fivefold and group stability severely affected. In contrast, roost site selection varied only slightly, in that bats incorporated only three new species of tree and varied the proportion of utilisation of different trees, but never used roost sites other than those with developing furled leaves. These results show that bats maintained selectivity and responded to habitat loss by increasing mobility, which reduced social cohesion and potentially increased energetic expenditure and predator exposure. Chaverri and Kunz (2011) concluded that extreme specialisation of T. tricolor could ultimately jeopardise the long-term persistence of local populations under conditions of habitat degradation. In a similar kind of experiment, Patriquin and Barclay (2003) investigated the effect of habitat fragmentation on bat foraging behaviour. They measured activity in three forest types and at four tree densities, ranging from unharvested forests to clear-cuts, in the boreal mixed-wood forest of Alberta, Canada. Results suggest that smaller and larger bats use harvested forests differently and that smaller bats (Myotis spp.) were less influenced by clutter than were larger bats. The largest species Lasionycteris noctivagans preferred clear-cuts and avoided intact patches. The authors concluded that management for forest-dwelling bats must take such species-specific effects into consideration. Harvesting that creates a mosaic of patches with different tree densities is likely to satisfy the requirements of more species than a system with less diverse harvesting styles.

10.4 Aggregation Distributions and Conservation Biology

In this section, we will see examples of the use of three approaches: site suitability, individual-based, and ideal-free approaches, although it is worth taking into account that some of these approaches can be used at different levels of organisation. In the first example in this section, we see a typical study that applies site suitability models to explore the environmental demands of species of conservation interest. 150 10 Distribution Ecology in Conservation Biology

Many of the world’s most troublesome mammalian pests are far from being pests in their region of origin, where their numbers may be in decline or they may be integral to the food chains of native species (e.g. muskrat Ondatra zibethicus in North America, Hall 1981; European rabbit Oryctolagus cuniculus in the Iberian peninsula, Moreno and Villafuerte 1995). The coypu Myocastor coypus is a large caviomorph rodent indigenous to aquatic habitats in southern South America. Since the 1920s, wild populations have become established in many parts of Europe, Asia, and North America, as a result of escapes and releases from fur farms (Carter and Leonard 2002). Coypus are considered to be a serious economic pest in several of these regions of introduction, principally because they eat a variety of agricultural crop and pasture plants and because their burrows damage drainage systems (Carter and Leonard 2002). Among the indigenous coypus of South America, in contrast, populations in several areas of Argentina are severely suppressed by hunting. The coypu is com- mercially the most important wildlife mammal species in Argentina’s furbearer market, yet there is no effective control over coypu hunting permits and there exists a substantial illegal market at a national level (Bertonatti and Corcuera 2000). These pressures on indigenous coypus are exacerbated by degradation of its pampas habitat caused by intensified agricultural production and urbanisation. From 1994 to 2004, Cassini and collaborators conducted a research programme on native coypus in order to provide basic information on the species that can simultaneously (1) preserve native local populations that are at risk, (2) advance commercial hunting sustainability, and (3) improve control strategies in countries where coypus have been introduced (Guichón and Cassini 1999, 2005; Borgnia et al. 2000; D’adamo et al. 2000; Guichón et al. 2003a, b, c; Túnez et al. 2009; Leggieri et al. 2011). In one of these studies, Guichón and Cassini (1999) surveyed the riparian habitat along the Luján river, which crosses the pampas plain in Argentina for 130 km (Fig. 10.1). They determined the presence/absence of coypus based on a methodology standardised for sampling riparian mammals (Mason and Macdonald 1986), and the following habitat variables in each transect were recorded: (1) the river variables of width and depth; (2) riverbank variables of slope, height, width, shallow shores, and vegetation cover (% of herbs, shrubs, and trees); (3) border variables of width and vegetation cover; and (4) upland variables of vegetation cover, land use (% land with agriculture, livestock, timber production, urban–industrial activity), and local perturbations (presence of docks, houses, roads, bridges, railways, and recreational centres near the river). A correspondence analysis was performed to identify the principal features of the environment and associated them with coypu distribution (Fig. 10.2). A stepwise logistic regression analysis was conducted to determine the effect of environmental variables on coypu distribution. The first three factors of the correspondence analysis (which explained most data variability) were used as independent variables, and the presence or absence of coypu as a binary dependent variable. The presence of coypu was documented in 66% of transects sampled along the river. Logistic regression analysis retained only the second factor of the correspondence analyses as a predictor of distribution of coypu. Variables associated with human disturbance were the most important negative correlates of coypu distribution. 10.4 Aggregation Distributions and Conservation Biology 151

Fig. 10.1 Map of the study area of the study conducted with coypus Myocastor coypus in the Pampas, South America (modified from Guichón and Cassini 1999)

In other studies of the same research programme, Guichón and Cassini (2005) showed that hunting activity is a major influence in demographic and genetic structures in loc.l populations of the Luján river, so heterogeneity in coypu distribution along this river appeared to be related to hunting activity. As a management strategy, protecting corridors that connect suitable habitats, favouring the dispersal that may be needed for population recovery, were proposed. Individual-based models are built mainly on a case-by-case approach. They have a large potential for application in conservation biology when information exists on the species under consideration. An area of frequent application for individual-based models is invasion ecology. Biological invasions are considered to be a major threat to biodiversity (Clavero and Garcia-Berthou 2005). Thus, it is important to discover the processes underlying invasion success and rates of spread (Hastings et al. 2005). Another area of application of this type of model is for following the spatial response of endangered species to human perturbations. An example of this type of application is described. Regan et al. (2003) constructed a stochastic, spatially explicit, individual-based model applied to Grevillea caleyi, an endangered plant species with a restricted range lying partly within Ku-ring-gai Chase and Garigal National Parks in Australia. The principal threatening processes affecting G. caleyi are habitat destruction and adverse fire regimes combined with high levels of seed predation. In this model the age, position, height, and radius of the crown of each plant in a rectangular patch were recorded at each time step. Eleven age classes were assumed, and plants were 152 10 Distribution Ecology in Conservation Biology

Fig. 10.2 Site suitability analysis of the distribution of coypus M. coypus. (a) Arrangement of the 26 variables sampled along the Luján river. Variables of high contribution to the factors: A riverbank width; B riverbank shrub cover; C agricultural activity; D livestock production; E roads; F houses; G upland shrub cover; H urban activity, I timber production; J docks; K recreational areas. (b) Arrangement of the 82 transects sampled along the river. Most positive samples have positive values of Factor 2 (modified from Guichón and Cassini 1999) categorised into two groups to model the effects of intraspecific competition: those that are below other plants and those that are not. The change in height of each plant over one time step was assumed to be a function of its age, and the survival coefficient for a plant was assumed to depend on its age and its competitive status. The effects of competition were further incorporated by ensuring that the growth of a plant was halted if the edge of its crown coincided with the stem of an adjacent taller plant. Seed predation rates and viability were estimated from empirical observations. Fire is incorporated into the model by the use of a probability function: the probability of a fire in yeart + 1 depends on the number of years since the last fire. 10.4 Aggregation Distributions and Conservation Biology 153

When a fire occurs, the model assumes that all burnt plants die and only seeds are left in the seed bank. Environmental stochasticity is incorporated into the model by assigning a probability distribution for each of the survival and plant growth parameters. The model is initialised at t = 0 by randomly assigning positions and heights to a specified number of plants in each age class in a rectangular patch of pre- assigned dimensions. The heights of plants in each age class were randomly selected from a normal distribution with a mean calculated from a similar height function to those above with a maximum height of 2.5 m and a coefficient of variation of 0.1. The Monte Carlo method was used throughout to select random deviates from the specified probability distributions. The model was configured to reflect an actual population of the species. The dynamics of G. caleyi populations were simulated by generating 50-year trajectories, 1,000 times. The model of Regan et al. (2003) predicted the future of the endangered plant based on empirical data from the unmanaged situation. Under these conditions, most population trajectories declined to zero. Simulations with average predation rates of 80%, in combination with managed fire frequency and fire size, indicated a marked improvement in both extinction and decline. Regan et al.’s model (2003) highlights just how difficult it is to maintain small populations of G. caleyi in a fire- prone habitat. The only circumstance in which a small population has a reasonable chance of not being lost completely is if there is an imposed reduction in predation rates and if fire events are managed at regular intervals. The optimal fire management strategy depends on the extent to which seed predation is reduced. The following is a good example of the application of the ideal-free distribution approach to a conservation challenge. Visiting areas of conservation interest has turned into an increasingly popular and profitable activity, particularly when tourists have the opportunity to watch in situ flagship, charismatic species. This is the case in coastal areas, where the patchy distribution of some species (e.g. marine bird and mammal colonies) allows easy access and the development of an infrastructure for visitors (Cassini 2001; Cassini et al. 2004). However, this may generate a conflict of interests: ecotourism may be a source of income for local communities but uncon- trolled visitor access can have a negative impact on the target species through the effects of human disturbance at the individual and population levels. Velando and Munilla (2011) analysed the interaction between recreational boating and a nearshore marine bird, the European shag (Phalacrocorax aristotelis), in the marine protected area around the Cíes archipelago, which is part of the National Park of the Atlantic islands of Galicia (north-western Spain). This protected area holds about two-thirds of the European shags breeding in Atlantic Iberia, a population that is under a risk of extinction. European shags form conspicuous feeding groups in the vicinity of the islands where the breeding colonies are located, which attract visits by numerous recreational boats. Daunt et al. (2007) have shown that reduced foraging time after independence results in high juvenile mortality, and therefore foraging disturbance by tourism could represent a serious risk for this species. Velando and Munilla (2011) used an ideal-free distribution framework to model the consequences of different management options aimed at reducing the impact of recreational boating on the foraging efficiency of shags. Their model simulated a population of shags foraging in a two-dimensional area divided into a grid of 500 m2 154 10 Distribution Ecology in Conservation Biology

Fig. 10.3 Observed effects of tourist boat presence on behaviour of European shags Phalacrocorax aristotelis in Galicia, Spain. (a) Increase in the spatial aggregation and (b) decrease in foraging activity with boat abundance (modified from Velando and Munilla 2011)

cells. They represented disturbance as a competitive interference for space between boats and shags. Thus, the number of grid cells available to foraging shags decreased with an increase in cells used by boats, according to a depleting exponential function with a parameter k that represented the rate of cell depletion by boats. The model also included a parameter m = rH/rL, where H and L were cells of high and low quality according to food resource supply r, respectively. To simulate the effect of habitat protection (i.e. the establishment of set-aside areas free of boat traffic), they varied the conditions of boat disturbance (k parameter) and the difference in quality between the two types of cells (m parameter). Three management scenarios were considered: (1) no habitat protection, (2) protection of the 14 higher quality (H) cells, and (3) protection of 14 lower quality (L) cells. Shags foraged mainly in shallow waters over sandy bottoms close to the colonies. Recreational boats made frequent visits to these areas. Shags showed two types of behavioural response to boat presence: they formed dense and clumped groups (Fig. 10.3a) and they reduced foraging activity by ten times (Fig. 10.3b). The 10.4 Aggregation Distributions and Conservation Biology 155

Fig. 10.4 Predicted effect of tourist boat presence on the food intake rate of shags P. aristotelis according to an ideal-free distribution model. High and low boat disturbance and two levels in the difference between the high- and low-quality habitat patches m are modelled. The black line represents the change in food intake in relation to the number of boats using the marine reserve. The grey line H represents the change in food intake rate when all boats are banned from the high- quality habitat patches; the light line L represents the change in food intake rate when all boat traffic is banned from low-quality habitat patches (modified from Velando and Munilla 2011)

model of Velando and Munilla (2011) predicted stronger effects on food intake when boat interference (k) is high, especially as the number of boats increased (Fig. 10.4). At low levels of boat interference (e.g. k = 0.01), the protection of the higher quality grid cells (H) would only have a strong positive effect on food intake if the difference in food supply from the lower quality grid cells is high (m = 3). When boat interference is high (e.g. k = 0.03), the protection of habitat (irrespective of its quality) had mainly beneficial effects on food intake and was stronger when the high-quality grid cells were protected. When a higher preference of boats for the higher quality grid cells (e.g. k > 0.03) was modelled, similar results were achieved, though the patterns observed became more exaggerated (Fig. 10.4). The conclusion drawn from the application of this model is that reduction of tourist visits in key foraging areas is a required conservation tool. However, protection of lower quality grid cells may have negative effects on food intake and therefore on conservation of this endangered species. 156 10 Distribution Ecology in Conservation Biology

10.5 Metapopulations, Source–Sinks, and Conservation Biology

Spatial population ecology models were those most readily absorbed by conservation biology. The metapopulation structure was incorporated into population viability models and the concepts of sources and sinks were integrated into the theoretical background of this applied science. A search of the word metapopulation in the journal Conservation Biology obtained 644 results, and the result for sink was 525. This strong influence is expected, considering that conservation biology and wildlife management traditionally use the principles of population ecology as a major source of their theoretical framework. In this section one example is described, but many others can be found in journals such as Conservation Biology, Biological Conservation, and Biodiversity and Conservation. Although there is some debate about the ability of population viability analysis to predict extinction risk, it remains an important tool in conservation biology. In fact, it is probably the only tool with the potential to estimate long-term persistence, if not as an accurate predictor of absolute measures of extinction risk, at least as a reliable tool for ranking relative persistence among different scenarios. Gutierrez (2005) studied a metapopulation of the dingy skipper butterfly Erynnis tages, a regionally rare declining species of Great Britain. In the Creuddyn Peninsula (North Wales) it inhabits patches of its host plant, Lotus corniculatus, growing in lightly grazed areas in sheltered microhabitats. In this area, 16 of 34 patches of habitat (out of 26.20 ha) are located in reserves. The remaining unprotected habitat is mostly located in wasteland areas vulnerable to urban development. Using a metapopulation model, the incidence function model, the author evaluated the extent to which the persistence of the dingy skipper in the landscape could be guaranteed within the existing reserve network. According to model projections, the dynamics within the reserve system were relatively stable when the unprotected habitat remained. When unprotected habitat was completely removed, however, the dynamics in the reserve network became markedly unstable, with an increased extinction risk (19 patches, Fig. 10.5). The result of this study showed that, in the case of dingy skipper butterfly of North Wales, the present availability of protected areas is insufficient to guarantee the persistence of this endangered species.

10.6 Landscape Ecology, Pattern-Based Models, and Conservation Biology

Landscape ecology has been of fundamental importance for the design of protected areas and other conservation strategies associated with land management. Notions such as connectivity and fragmentation (Wiens 2009), accessible habitat (Eigenbrod et al. 2008), or land of fear (Laundré et al. 2010) have emerged from this discipline 10.6 Landscape Ecology, Pattern-Based Models, and Conservation Biology 157

Fig. 10.5 Proportion of replicates that went extinct of time series for 1,000 simulations of 100 years generated for four scenarios of the dingy skipper butterfly in Erynnis tages in the Creuddyn Peninsula (North Wales). Scenario with 34 patches considered the observed habitat distribution in 1997–1998. Scenario 33 patches simulated the observed habitat destructions in 1999, in which a large core populated patch was destroyed. Scenario with 19 patches considered the habitat including only those sites located in reserves. In scenario ‘19+14 patches’, the mean average predicted proportion of patches occupied, the proportion of replicates that went extinct, and the time for ≤95% chance persistence were calculated for the reserves only (19 patches) (modified from Gutierrez 2005) and some of them have been very influential in the conservation of biodiversity. There are many examples in the literature on the interaction between landscape ecology and conservation biology. An example was selected of the infrequent application of fractal theory to evaluate the impact of habitat loss on extinction risk. The Florida panther Puma concolor coryi has large home-range requirements, so is susceptible to the largest threats to wildlife conservation, which are habitat loss and hunting. Extensive forested landscapes are critical to the survival of this species. Kerkhoff et al. (2000) determined the minimal amount of forest required by panthers to survive in South Florida, USA. They considered a hypothetical panther accumu- lating habitat resources subject to the metabolic costs of travelling. They divided the study area into 15 plots of 1,000 × 1,000 pixels, in which they estimated the percentage of forest cover and the number of panther locations. They suggested that fractal geometry allows proper estimation of the amount of additional forest gained for an increase in area searched as the panther moves around a potential home range. 158 10 Distribution Ecology in Conservation Biology

Fig. 10.6 Number of panther Puma concolor locations per plot as a function of forest fractal dimension D, in South Florida, USA (modified from Kerkhoff et al. 2000)

The question is as follows: how much forest cover a panther will gain for a given increase in the area searched? To estimate this, they used the following formula:

= D (10.1) FkL L where L is the width of a square cell, k is a proportionality constant, and D is the mass fractal dimension. Mass refers to the mean amount of the forest that is found in a window of width L. The authors calculated FL on a map at different values of L, and from then they estimated D as the slope of a least squares regression of log FL on log L. Figure 10.6 shows the relationship between the number of panther locations per plot as a function of forest fractal dimension D. When D drops below the threshold 1.8 (corresponding to about 25% forest cover on the scale of the plot), the likelihood of intensive panther use declines dramatically. The fact that several plots exhibited D > 1.8 but contained few, if any, panther locations suggests that, above the threshold, panthers respond to other environmental influences in addition to forest cover.

10.7 Species Distribution Models and Conservation Biology

As we saw in Sect. 7.2, species distribution models are used for two purposes: to identify relevant factors or habitat requirements and for spatial prediction, especially from mapping distribution. These two applications have significant appli- cability for conservation biology. There is an enormous number of studies using relatively simple statistical methods to find relationships between the abundance or occurrence of a species and environmental variables or types. This technique is used to obtain information quickly about the environmental needs of 10.7 Species Distribution Models and Conservation Biology 159 species of conservation concern, especially in Third World countries, which have great natural richness but little biological information on species to support the application of more sophisticated models of population viability. An example is provided of this type of study conducted in the South Cone of South America. Cassini et al. (2010) evaluated the conservation status of Patagonian otters Lontra provocax using limited information. Argentinean Patagonia encompasses the southern part of the country between parallels 36°S and 55°S, between the Andean mountain chain and the Atlantic Ocean. It is dominated by two main types of ecosystems: the steppe that occupies most of the region and the temperate forest that forms a narrow fringe lying almost entirely in the mountains. This fringe of forest includes large numbers of lakes, lagoons, and rivers. In its southern portion it runs adjacent to the Beagle Channel, which connects the waters of the Atlantic and Pacific oceans. The length of this fringe is similar to the distance between London and Rome. Three teams surveyed 435 locations in lakes, rivers, and sites on the marine coast. The authors used two spatial criteria provided by the International Union for the Conservation of Nature to decide on the conservation status of individual species. To assess the extent of occurrence (IUCN criterion B; IUCN 2001, 2003), they measured the area of minimum convex polygons containing all sites with evidence of otters, grouping those that were less than 100 km apart and always connected by water systems, bearing in mind that the greatest distance known to have been travelled by a dispersing huillin is less than 50 km. The estimation of area of occupancy is defined by the IUCN as the area within the extent of occurrence that is effectively used by the taxon (e.g. avoiding the inclusion of unsuitable habitat) and it is generally recommended to estimate this using grids. The authors used grids that comprise cells with a length corresponding to the mean home range (and half of this mean) estimated for the species. This work found that, at a regional level, this South American otter showed a geographic range that is close to the limit expected of a species at risk of extinction. In a simultaneous survey to that of otters, the same authors estimated the availability of their main prey: the macro-crustacean genera Aegla and Sammastacus. Cassini et al. (2009) used the results of both surveys to analyse the role of food availability on otter distribution using a scale-dependent analysis of prey and predator distributions. They compared the distributions of otters and macro-crustaceans along a north–south regional gradient between the river basins of northern Patagonia, at an altitudinal gradient within a river basin, and between habitat types within a lake. There were heterogeneities in the distribution of macro-crustaceans at four scales, which were reflected in the distribution of freshwater otters. The only exception occurs at a sub-regional level, where the occurrence of basins with crustaceans but without otters appeared to be due to dispersal limitation after human overexploi- tation. The agreement between distributions of prey and predator at different ecological scales suggested that food availability is an important factor regulating otter distribution. Species distribution models have a broad range of applications and have been used to assess the potential threat of pests or invasive species, to obtain insights into the biology and biogeography of species, to identify hotspots of endangered species 160 10 Distribution Ecology in Conservation Biology or to predict biodiversity, to prioritise areas for conservation, and to establish suitable locations for species translocations or cultivation (reviewed among others by Beaumont et al. 2005). Importantly, species distribution models are currently the only means by which we can assess the potential magnitude of changes in the distribution of multiple species in response to climate change (e.g. Brereton et al. 1995; Eeley et al. 1999; Berry et al. 2002; Erasmus et al. 2002; Midgley et al. 2002; Peterson et al. 2002; Williams et al. 2003; Meynecke 2004; Thomas et al. 2004). Recently, distribution models have been used to assess the feasibility of current conservation strategies and the value of existing reserves in Great Britain under future climate scenarios and to examine the effects that different climate regimes may have on biodiversity within existing South African National Parks. The output of these models has also been used to estimate the extinction probabilities of species in response to global warming (Thomas et al. 2004). Predicting the current or future distributions of species has principally been conducted using bioclimatic models that assume that climate ultimately restricts species distributions. These models summarise a number of climatic variables within the known range of a species, thus generating a bioclimatic envelope. The models can then be used to (a) identify the species current potential distribution, that is, all areas with climatic values within the species bioclimatic envelope, and (b) assess whether these areas will remain climati- cally suitable under future climate scenarios. These models play a vital role in assessing the potential distribution of species and are useful ‘first filters’ for identifying locations and species that may be most at risk from a changing climate.

10.8 Ensemble Distributions and Conservation Biology

Patterns of diversity determined by using macroecological tools have been used extensively in conservation biology. For example, Laurance (1991) tested the effi- cacy of seven ecological traits (body size, longevity, fecundity, trophic level, dietary specialisation, natural abundance in rainforest, and abundance in the surrounding habitat matrix) for predicting the responses of 16 non-flying mammal species to rainforest fragmentation in tropical Queensland, Australia. In turn, Kattan (1992) analysed the rarity and vulnerability of the avifauna of the Central Cordillera of Colombia using a macroecological approach, which defined eight forms of rarity in three dimensions. We will see in more detail an example of the use of indicator species for estimating biodiversity patterns. Pearson and Cassola (1992) suggested that risk analysis should be a major consideration in designing programmes that use indicator taxa. They recognised that the preponderant strategy of conservation biology has relied on charismatic vertebrates that operate as umbrella species for the ecosystems that they inhabit. They criticised this approach of using vertebrates as indicator taxa because these animals tend to be relatively long-lived and have low rates of population increase, long generation times, and comparatively low habitat specificity, all of which tax time and finances in proper investigation. They proposed—as do other researchers—to use 10.8 Ensemble Distributions and Conservation Biology 161 insects instead of vertebrates as indicators. Insects represent more than 80% of the several million animal species now extant in the world, and extinction is probably more significant in terms of absolute numbers for insects than for any other global group of organisms. They were in fact more specific, suggesting that the family of tiger beetles (Coleoptera:Cicindelidae) is a potentially ideal indicator taxon, for the following reasons: (1) their taxonomy is well known, (2) all are predaceous on small arthropods, (3) they share similar larval and adult body forms around the world, (4) adults are generally diurnal and found primarily on soil surfaces, (5) larvae are easy to raise in the laboratory and are relatively easy to find and observe in the field, and (6) there is plenty of information available about them. As a test of the hypothesis that tiger beetles are an excellent indicator of biodiversity, they compared the characteristics of tiger beetles to each of the seven criteria proposed for an ideal indicator taxon. In addition to establishing the strengths and weaknesses of tiger beetles as indicators, they conducted an extensive bibliographical survey of worldwide species richness biogeographical patterns, using diverse macroecological approaches comparing beetles with butterflies and birds. Logistical advantages provide some of the strongest arguments for selecting tiger beetles as an appropriate indicator taxon. Species numbers of tiger beetles are relatively well known for 129 countries. Eight countries alone account for more than half the world total of the 2,028 known species. Species numbers were also indicated for 11 biogeographical zones of the world. For gridded squares across North America, the Indian subconti- nent, and Australia, there were significant positive correlations in species richness of tiger beetle, birds, and butterflies. However, Kattan (1992) proposed that tiger beetles are a better indicator of biodiversity than other taxa because their species numbers can be reliably determined within 50 h on a single site, compared to months or even years for birds or butterflies. Thus, the advantage of using tiger beetles in conservation biology is evident. Another example of the use of assemblage distribution in conservation biology is provided by the work of Boris Worm and collaborators, who have developed a rapid method for generating richness patterns based on the overlaying of predicted distributions using habitat suitability models for all relevant species (Worm and Duffy 2003; Kaschner et al. 2006; Lucifora et al. 2011, 2012; Trebilco et al. 2011). This has replaced more traditional methods based on expert knowledge or regional observations. The individual model delineates the environmental tolerance of each species with respect to basic parameters known to determine the distribution of the taxon directly or indirectly. It does so by combining available data on species occurrence and habitat usage, supplemented by expert knowledge. The relative environmental suitability of different habitats for a given species can then be computed and used to predict long-term mean annual species distributions. The authors superimpose individual species predictions to generate global patterns of species richness, defined as the number of species present in a given area, which they sub- sequently validate using independent survey data. We will see in more detail one of these studies. Lucifora et al. (2012) analysed the spatial pattern of chondrichthyan (sharks, rays, and chimaeras) diversity in the south-west Atlantic. At present, fishing effort 162 10 Distribution Ecology in Conservation Biology

Fig. 10.7 Distribution of diversity of chondrichthyan fish in the south-west Atlantic (modified from Lucifora et al. 2012) has increased dramatically in this region, and overfishing has become the most important threat to fish communities. These authors used a data set that was collected between 1966 and 1978, i.e. prior to the major development of industrial fishing, in particular of demersal trawling, one of the most damaging and unselective fishing practices (Crowder et al. 2008). They also explored the relationship between richness and reported catches of chondrichthyans from commercial, demersal fisheries. Environmental data were ocean depth, mean annual bottom temperature, mean annual bottom salinity, mean annual bottom dissolved oxygen content, mean annual surface nitrate, phosphate and silicate concentration, and bottom rugosity. Generalised linear models were used to assess the relationship of species richness with environmental variables, marine fronts, and with commercial catches. Species richness increased towards the north, particularly close inshore and on the outer shelf. Dissolved oxygen was related negatively with richness; depth, latitude, and longitude were also related with richness, but to a lesser extent (Fig. 10.7). Chondrichthyan diversity hotspots, i.e. cells with richness higher than the 90th percentile of the richness distribution, were more likely to occur on marine fronts than elsewhere. High chondrichthyan catches were significantly associated with hotspots, but low-catch areas were not related to coldspots, i.e. cells with less than a 10th percentile of the richness distribution. Areas of high richness of chondrichthyans coincided with high levels of catches and are used by other species of high conservation concern. The conclusion of this study was that, given the importance of these areas to multiple species in the south-west Atlantic, any conservation strategy based on a network of protected areas should include marine fronts. Chapter 11 Distribution Ecology in Animal Production

11.1 Introduction

Theoretical advances, made in the 1970s in relation to habitat selection behaviour and its ecological consequences in terms of spatial distribution patterns, also had an influence on animal production. Ideas from foraging theory, community ecology, and hierarchy theory were absorbed by researchers in the field of range management. This chapter deals only with applications to domestic species involved in animal production. There are other studies applicable to management of wild and game species that will not be considered (e.g. Bristow and Ockenfels 2004; Larsen et al. 2007). This chapter is divided into five sections related to foraging behaviour, landscape predictors, quantitative analyses, species models, and species assemblages’ approaches, all applied to the link between livestock distribution and animal production.

11.2 Foraging Behaviour and Distribution of Livestock

Discussions of the foraging behaviour of livestock are mostly related to discussions about herbivory. The food of herbivores differs from a carnivorous diet, as do their adaptations to select certain food types, to search for food and to use foraging patches. Foraging by herbivores differs from carnivorous as follows: (1) they do not find all required nutrients in the correct proportions in one type of food; (2) plants have low nutritional quality and high variable composition; (3) herbivores spend a relatively low time searching for food and a long time digesting, especially grass- eater ungulates; (4) food is not provided in well-defined discrete items, and it is therefore difficult to define ‘encounter rates’, ‘handling times’, and ‘prey size’; and (5) nutritional quality is the fundamental variable in herbivore diet (Stephens and Krebs 1986).

M.H. Cassini, Distribution Ecology: From Individual Habitat Use 163 to Species Biogeographical Range, DOI 10.1007/978-1-4614-6415-0_11, © Springer Science+Business Media New York 2013 164 11 Distribution Ecology in Animal Production

Cattle and sheep have frequently been seen as unintelligent animals that have lost their adaptations and cognitive abilities in the course of domestication. This is far from the truth and many studies have demonstrated that they can show complex adaptive behaviour that regulates the way they use and select habitats. In this section, some of these behaviours are briefly described. Penning et al. (1993) show that sheep retain social behaviour that appears to be an adaptation to predation hazard. They investigated the effects of group size on grazing time, recorded for focal sheep within groups that ranged in size from 1 to 15. The animals were continuously stocked on a perennial ryegrass pasture Lolium perenne maintained at a sward surface height of 6 cm. Penning et al. (1993) found that sheep in smaller groups spent less time grazing than sheep in larger groups. Animals in groups of one and two tended to have shorter meals and increased vigilance behaviour than those in larger groups. This study not only showed that sheep adjust their behaviour in the same way as wild mammalian herbivores (Sect. 4.3) but also, most importantly, demonstrated this experimentally, which is more conclu- sive than correlations found in natural contexts. Livestock are often moved from one location to another because of drought or other management considerations. In some cases, animals are moved to areas with foraging and environmental conditions that differ greatly from the conditions to which they are accustomed. To perform satisfactorily in new locations, livestock must be able to adapt to new plants, topographical features, and locations of water, as has been observed in several studies. For example, Bailey et al. (2010) studied the impact of previous experience on grazing patterns and diet selection of Brangus cows in an experiment with three treatments: ‘naïve’ cows that originated from a humid–subtropical environment (Leona, Texas, USA) were brought to the Chihuahuan Desert, USA; ‘native’ cows that spent their life in the Chihuahuan Desert; and ‘tourist’ cows that were born and raised in the Chihuahuan Desert but were moved to Leona during the preceding 3 years. Cows from the three groups were tracked in three extensive pastures (1,000 ha) for three periods of 8–10 days during winter, early summer, and later summer. Faecal near-infrared spectroscopy was used to estimate diet quality. Naïve cows used less area and remained closer to water than cows born and raised in the Chihuahuan Desert (native and tourist cows pooled) when first evaluated in the winter (Fig. 11.1). After pooling all data, native cows were found farther from the water and spent less time at the water than cows that did not spend their life in the desert. During winter and early summer (drought conditions), naïve cows selected diets with lower crude protein than cows born in the desert, but during late summer, after abundant precipitation, naïve cows selected a diet with higher crude protein. Although Brangus cows are highly adaptable, previous experience appears to have a strong influence on the ability of livestock to use habitats. Foster et al. (1996) evaluated another aspect of livestock cognition abilities (a similar experiment was conducted by Matthews and Temple 1979). They estimated the performance of dairy cows in experimental sets designed to test if they followed the behavioural matching law (Sect. 2.8), according to which they must distribute proportionally responses between two sources with different qualities. 11.2 Foraging Behaviour and Distribution of Livestock 165

Fig. 11.1 Effect of experience on landscape use by cattle. ‘Naïve’ cows born in a different region, while ‘tourist’ cows spent 3 years in a different place before the trial. Distances represent maximum distance from water by randomly selected cows. Error bars represent standard error. Different letters within groups of bars represent statistical differences (modified from Bailey et al. 2010)

They used concurrent variable-interval schedules of food delivery, which are standard protocols used in experimental psychology to study choice behaviour. For that purpose, cows were trained in an experimental chamber that contained an ingenious feed delivery apparatus developed by Cate et al. (1978). Cows had to use their nose to press one of two plates that were placed behind two access holes through which food was presented when food hoppers were raised by pneumati- cally operated rams. Foster et al. (1996) found that undermatching was the typical response of cows under these schedule conditions, both in response rate and time- allocation ratios (Fig. 11.2). In other words, although cows learned to use more the best patch, they still tended to use the poor options more intensively and good sites less frequently than the optimum in terms of maximisation of gain rate. An artefact of the experimental design could explain cow undermatching behaviour. In all experiments, a changeover delay of 3 s was used, which specified that food could be earned no sooner than 3 s after switching from one plate to the other. This changeover delay could possibly be of insufficient length and may lead to undermatching (Baum 1974, 1979). Beyond the reasons for this pattern, the most important message from these experiments is that cattle can respond sensitively to subtle variations in the rate of acquisition of resources at different sites. In horses, Dougherty and Lewis (1992) found that these animals trained on several concurrent variable-interval schedules showed a close matching relationship between the relative rates of response and the relative rates of reinforcement. 166 11 Distribution Ecology in Animal Production

Fig. 11.2 An example of the performance of a cow in choice experiments. It showed an undermatching response (see text for explanation of the experimental procedure, modified from Foster et al. 1996)

11.3 Landscape Predictors of Livestock Distribution

Before the 1980s, the main solution to the inefficient use of pastures was intensive grazing systems. Intensive farming involves farmers managing large numbers of animals and pastures. Such techniques involve high costs of both equipment and personnel. Those investigating livestock management in the 1980s began to wonder what would happen if, instead of manipulating the animals and the fields, attention was paid instead to how domestic livestock use space in a continuous grazing system and that knowledge was used to improve ranching. Cassini (1995) reviewed the literature written primarily in the 1980s on habitat selection by ungulates with importance for animal production. These studies are classified by type of animal, activity, and environmental variables (Table 11.1). The most obvious conclusion that emerges from this analysis is that the quality and type of vegetation play a major role in the way that ungulates use space. However, under certain conditions there are other factors that influence the choice of a foraging area. In mountainous areas, slope and altitude may be relevant environmental factors, while in warm areas the distance to the water source becomes important. Certain human activities not directly related to animal handling can also influence habitat selection. The use of herbicides, distance to roads, use of fire for weeding, or fence lines have been described as variables affecting this behaviour. In those studies that discriminated activities in different areas of the environment, there were often found differences in the environmental variables influencing the different behaviours. Overall, grazing is affected by vegetation, while resting appears to be related to abiotic factors. 11.4 Quantitative Analyses of Livestock Distribution 167

Table 11.1 A review of studies on distribution ecology applied to animal production published during the period 1960–1990 (references in Cassini 1995) Study number Animal Activity Analysed variables Anderson, Kothman 1980 Cow Movement Vegetation Barret 1982 Cow General Topography–vegetation Canon et al. 1987 Deer Grazing Fire Cook 1966 Cow General Topography Duncan 1983 Horse 1. Grazing 1. Vegetation 2. Resting 2. Biting insects Edge et al. 1988 Deer Grazing Vegetation Ganskopp, Vaura 1986 Horse General Vegetation–water Gordon 1989 Multiple General Vegetation Grover, Thompson 1986 Deer Grazing Vegetation cover–roads Hall 1988 Cow General Vegetation HewsonWilson 1979 Sheep General Old line fences Low et al. 1981 Cow General Vegetation/sporadic climatic disturbances McDaniel, Tiedeman 1981 Sheep General Slope–bare soil Miller 1983 1. Horse General 1-2. Vegetation–water 2. Cow 1. Topography Pratt et al. 1986 Cow–pony General Vegetation–shadow Rees, Hutson 1983 Cow General Vegetation Roath, Krueger 1982 Cow General Vegetation–water Scifres et al. 1983 Cow Grazing Herbicide Senft et al. 1983 Cow General Seven variables Senft et al. 1985a Cow Resting Topography Senft et al. 1985b Cow Grazing Vegetation–water Shaw, Dodd 1979 Cow Grazing Herbicide Simpson, Gray 1983 Sheep–deer General Topography Skolvin et al. 1983 Deer General Fertiliser–fire Steuter, Wright 1980 Deer General Vegetation cover Tanner et al. 1984 Cow 1. Grazing 1. Vegetation 2. Resting 2. Shadow Tucker, Garner 1983 Antilope Resting Vegetation height Welch 1984 1. Sheep/cow General 1. Forage management 2. Deer 2. Annual migrations Woodward, Ohmart 1976 Donkey General Vegetation–shadow

11.4 Quantitative Analyses of Livestock Distribution

Senft and colleagues have conducted a series of studies on livestock distribution, assuming that it is a key component of management (Senft et al. 1983, 1985a; Senft 1989). In one of these studies, Senft et al. (1985a) derived the habitat-matching law (Sect. 3.5) independently of the other earlier development of this principle. They defined a relative community preference (RCP) as the ratio of the proportion of total 168 11 Distribution Ecology in Animal Production

Table 11.2 Relationship of relative community preference by cattle with plant community variables Regression model Normalised variable Correlation Intercept Coefficient Significance Standing N, preferred species 0.745 0.0023 0.9442 <0.001 Biomass, preferred species 0.712 −0.1604 1.0863 <0.001 Standing N, live plants 0.707 −0.1809 1.0635 0.022 Standing live biomass 0.694 0.0879 0.9393 0.026

grazing time spent during month t in a community i (Git) to the proportion of total pasture area covered by a plant community i (Ai): G = it (11.1) RCPit Ai Then, they defined the relative value of vegetation property (RFC) in plant community i at time t, measured as the absolute measured value for community i at t (AFCit) divided by the mean level of the variable, area-weighted across all communities.

AFC RFC = it (11.2) it AFC å it Ai They selected several potential ‘properties’ that may influence selection of grazing areas. For four of these variables, they found that cattle follow matching law when they are distributed between pasture communities (Table 11.2). Senft et al. (1985a) concluded that matching is the prominent pattern when large herbivores interact with landscape systems by selecting plant communities and other landscape components for feeding. The ratio of the proportion of total feeding time to the proportion of home-range area is generally a linear function of the relative abundance and/or nutritive quality of the preferred plants in the communities. Not only cattle but also sheep, horses, mule deer Odocoileus hemionus, wapiti C. canadensis, North American bison B. bison, and species of kangaroos exhibit landscape-scale matching (Senft et al. 1985a).

11.5 Species Distribution Models Applied to Livestock Management

Senft and collaborators also conducted pioneering work to construct predictive maps of the distribution of livestock. In 1983, prior to the development of most species distribution models, they published predictive models of cattle distribution by applying multiple regression analysis to a body of behavioural observations. Observations of cattle movements were made in two small paddocks, and the total 11.5 Species Distribution Models Applied to Livestock Management 169

Fig. 11.3 Pioneer use of spatial distribution models applied to cattle husbandry (modified from Senft et al. 1983)

observed time was recorded for each cell of a 24 × 24-m grid cell (one per paddock). Data were analysed using stepwise multiple regressions. A larger pasture adjacent to the paddocks was used for validation by mapping faecal pat densities. It was assumed that faecal pat density on a given area would be proportional to time spent in that area. Three behaviour models conducted for grazing, resting, and bedding were weighted by the average amount of time spent per day on each activity. The sum of the weighted behaviour models produced predictions of the distribution of faeces in the validation area. The model was successful in predicting cattle distribution (Fig. 11.3). In a more recent study, Halbritter and Bender (2011) modelled habitat features associated with presence of cattle by using a maximum entropy, species distribution model (Sect. 7.2) to identify habitat attributes associated with areas of increased probability of being used by cattle in the Lincoln National Forest (LNF) of south- central New Mexico. The aim of the study was to evaluate any potential conflict because cattle herbivory should lead to overuse of montane meadow vegetation communities. They attempted to use a quantitative method of estimating cattle distribution instead of the traditional approaches used in that region that are based on 170 11 Distribution Ecology in Animal Production casual observations frequently made from roads, producing a subjective impression of cattle distribution. They recorded 902 observations of cattle and conducted 168 pellet-group transects, between 2004 and 2006. They modelled six variables shown to affect cattle habitat use. Cattle distribution was best predicted by elevation, vegetation cover type, distance to roads, and distance to water and slope, regardless of whether modelled from transect data or observation data. In contrast to observation- based models, randomised transect-based data indicated a much broader distribution of cattle across the landscape, with much lower affinity to open areas and areas immediately adjacent to roads, again probably due to the inclusion of vegetation cover types used for reasons other than feeding and watering. Consequently, while observation-based models indicated that montane meadows were strongly positively associated with cattle distribution in the study area, randomised transect- based models suggested that other vegetation cover types were also positively, and in some cases more strongly, associated with cattle presence. This study showed that rigorous, quantified data modelled with appropriate statistical techniques may provide fundamental information for management decisions. The use of supplemental food blocks in domestic ruminant husbandry implies a new outlook for product evaluation because its success depends not only on farm hands or technicians but also on the choices made by the animals themselves. Detection and consumption of the block in the field are essential to the success of this tool. A frequent recommendation made by manufacturers is to place supplemental food where cattle spend most time per unit area or near to sources of water. Although these recommendations are intuitively appealing, they are not normally justified by scientific study. Cassini and Hermitte (1992) investigated the relationship between supplementation distribution and the pattern of distribution of livestock activities, both spatially and temporally. In the first phase of the study, the environmental factors that affected the use of the environment were studied in two small paddocks. As has been previously observed, cattle spent more time in the shelter and in the area surrounding the watering place than in any other area of the same size within the field and preferred the richest vegetation communities (Fig. 11.4, upper graph). In Phase II they investigated the best field distribution of supplemental food blocks, testing three hypotheses: (1) use is random, (2) use is correlated with cattle presence, and (3) use is directly related to the activity in which cattle engage in each area of the field. Supplemental food blocks were located under shelter, close to water sources, and in meadow areas. Although the meadow area presented the lowest use (per sample unit) at the beginning of the trial, block consumption was significantly greater in this area, indicating that the third hypothesis best predicts block consumption. In Phase III, a large-scale test was conducted in a 3,100-ha field with 1,097 cows. Three types of area were determined in relation to their distance from a watering place. A similar tendency as in the previous phase was observed. Blocks located in the surroundings of the watering place were consumed less than those located in an adjacent foraging area (Fig. 11.4, lower graph). The overall results suggest that placement of supplemental blocks near watering sites will not lead to optimum utilisation and should not therefore be recommended. 11.5 Species Distribution Models Applied to Livestock Management 171

Fig. 11.4 Study on habitat use by cattle oriented to optimise the use of supplemental food blocks. In the small-scale trial cattle used more intensively areas with shelter and water than the meadow. However, during the large-scale trial, they consumed more from blocks located in the meadow, probably due to motivation (modified from Cassini and Hermitte 1992)

Another research area of interest to animal production is related to vegetation management. Distribution modelling has been applied to this type of problem. For example, Karl (2010) studied cover attributes in Southern Idaho, USA, by applying regression kriging and remote sensing. Rangeland management also involves the control of invasive plant species that cause deterioration in the availability of resources for range animals. Distribution models at a coarse scale have been applied to evaluate the impact of invasions. For example, Johnson and Miller (2008) used logistic regression on presence/absence data of pre-settlement juniper Juniperus occidentalis to evaluate the potential for using site characteristics on woodland expansion in Oregon–Idaho, USA. Klinken et al. (2007) mapped Mesquite (Prosopis spp.) distribution and density using visual aerial surveys and geographical information systems in Australia. 172 11 Distribution Ecology in Animal Production

11.6 Species Assemblage Distribution and Livestock Management

While the agricultural frontier continues to expand towards natural landscapes, the interaction between wild ungulates and livestock is becoming an increasing conservation problem around the world. The different approaches described in Chap. 8 that deal with species assemblage distribution may provide theoretical and methodological tools to investigate the consequences of this interaction, from the point of view both of wildlife conservation and livestock management. Competition for resources (grass, forbs, shelter, or water sources) may be an origin of conflict, especially in arid environments where food and water are scarce. Coexistence can occur; however, native species can suffer a displacement to suboptimal habitats. Because native species are normally better adapted to their environment than exotic ungulates, the use of secondary habitats can be wrongly interpreted as habitat preference. Isoleg theory (Sect. 8.2) provides a theoretical framework for interpreting this type of interspecific interaction. It proposes that the possibility of coexistence between two species that share niche components depends on two main factors: degree of specialisation in the use of these components by each competing species and the spatial heterogeneity of the environment. Two conditions increase the probability of coexistence between competing species: (1) when at least one of the species is generalist, i.e. it can use different types of resources, and (2) when the shared resource shows a heterogeneous distribution. Rosenzweig (1981) models are based on the principle that each competitor (normally foragers competing for food) of both populations uses the resource patch that maximises its fitness (indirectly measured as food intake rate). Pimm et al. (1985) developed an isoleg model that is specifically applied to the interaction between a dominant, specialist species and a subordinate, generalist species, both preferring the same type of resources in an environment where this resource is distributed in patches. Coexistence can occur because the subordinate, generalist species uses alternative patches of resources while the density of the dominant, specialist species increases. These models provide a framework to determine if a native species is subordinate to an introduced one, thus requiring conservation consideration. The Puna or Altiplano is an arid environment located at high altitudes of the Andes, with cold and dry weather conditions. Most Puna desert is covered by steppe, characterised by xerophytic vegetation and a high proportion of bare soil. However, in water body margins and swampy soils, there exists a type of habitat known as vegas or bofedales, where pastoral activity is almost exclusively concentrated. These are the ‘oases’ of the Puna desert, with high biomass and vegetation cover throughout the year and with plant species usually of relatively high nutritional content. Domestic ungulates (cattle, sheep, goats, horses, and donkeys) have occupied the Puna region since the Spanish conquest in the sixteenth century. Although local people used to have their own native domestic herds of llamas (Lama glama) and alpacas (Lama pacos), the use of exotic ungulates was gradually adopted as the 11.6 Species Assemblage Distribution and Livestock Management 173

Fig. 11.5 Landscape distribution of vicuñas in relation to the distribution of livestock and feral donkeys in Laguna Blanca, South America. Vicuñas segregated spatially from livestock (modified from Borgnia et al. 2008)

process of colonisation progressed and mixed herds are now preferred. Habitat use by domestic ungulates in the Puna is determined by the shepherds, who led the animals to make an intensive use of vegas. Vicuñas are camelids well adapted to living in high-altitude deserts (see Sect. 3.2). They can use most resources provided by the Puna environment and can be considered generalists in their capacity to use habitats and food resources (Arzamendia et al. 2006). Being generalists, they still prefer to forage in vegas, which are considered the optimal foraging habitat for vicuñas, while steppe areas are seen as suboptimal habitats (Renaudeau d’Arc et al. 2000). Borgnia et al. (2008) investigated the interaction between livestock herds (composed of cows and shoats), feral donkeys, and vicuñas in Laguna Blanca Reserve, Argentinean Puna (Fig. 11.5). They tested two assumptions and one qualitative prediction of the isoleg model of Pimm et al. (1985): (1) both vicuñas and domestics prefer the same type of foraging habitat, (2) vicuñas are generalists while domestics are specialists, and (3) vicuñas and domestics are spatially segregated and vicuñas use suboptimal 174 11 Distribution Ecology in Animal Production habitats more than would be expected from their foraging preferences. To quantify abundance and distribution, census counts were performed along 102 km of trials, recording approximately 1.2 km to each side of the trials (Fig. 11.5). Diet composition was established by taxonomic analysis of ungulate faeces. Habitat preference was measured using foraging effort. This effort was estimated in two ways: (1) the proportion of foraging individuals in each habitat and (2) proportion of plants from each type of habitat in the total diet. Habitat selectivity was also measured in two ways: (1) by evaluating the evenness of distribution in the whole study area and (2) estimating the extent of the habitat niche. Vicuñas and livestock were spatially segregated (Fig. 11.5); however, there was a high overlap of diets between them. Vicuñas were generalists in their use of habitat but invested foraging effort in swamp habitats, while livestock were specialised in swamp habitats, where they were taken by shepherds for grazing. Based on an isoleg model developed by Pimm et al. (1985) for this type of competitive context, Borgnia et al. (2008) concluded that vicuñas and livestock coexist because vicuñas occupy suboptimal habitats while livestock concentrate their foraging in the richest areas. Vicuñas can deal with poor habitats because they have several morphological and physiological adaptations to living in deserts (Hofmann et al. 1983) and can forage on the poor, ligneous, and fibrous vegetation of the Puna region (Vallenas 1991; Sponheimer et al. 2003). Vicuñas will probably move to swamp habitats once shepherds remove livestock from them. Part V Conclusions and Prospects Chapter 12 Final Remarks

12.1 Introduction

Throughout this book, we created a snapshot of most of the approaches and theories regarding ways to study the spatial distribution of organisms. The succession of sections showed a great diversity of viewpoints, which is typical of a growing discipline. There are fields of knowledge that are borne with great strength and are installed in the options of the market theory and other traditional branches of ecology which find ways to adapt to this need to incorporate the geographical dimension. We followed a hierarchical presentation, based on a classification of levels of organisation, which were defined operationally rather than formally. We have defined a total of six main levels—individuals, aggregations, societies, subpopulations, populations, and species—and two supplementary levels: genes and species assemblages. The reason for choosing this organisation of the book is simple and has to do with the history of the emergence of different theoretical models. Each ecological school has approached the problem of spatial distribution based on their own traditions, and many new paradigms have emerged regardless of previous models. In some cases, such as the movement ecology and species distribution models, they originated from new methodological developments (technical or statistical) rather than from the evolution of new ideas. One way of organising the various approaches presented in this book is based on the variables that use models to define the criteria of distribution of organisms. Each model defines a currency that is used to measure the effect of the environment on organisms and decide how the corresponding level unit has to accommodate between different environmental units. When models are quantitative, the currencies usually take the form of dependent variables. Probably the most fundamental challenge for distribution ecology in the future will be to unify all approaches under only one currency. These currencies are: • Individual fitness: For foraging theory and ideal-free, source–sink (original), isoleg, and isodar models

M.H. Cassini, Distribution Ecology: From Individual Habitat Use 177 to Species Biogeographical Range, DOI 10.1007/978-1-4614-6415-0_12, © Springer Science+Business Media New York 2013 178 12 Final Remarks

• Risk of extinction: For metapopulation models • Growth rates: For population models • Local density: For site suitability models • Local occurrence: For most species distribution models This final part is divided into three subparts. In the first section, we will see the most common criticisms that have been made to each of the major approaches presented in this book. Then, there is a comparison between models based on their differences, which occupies several sections. The chapter and the book are closed with a personal view on the way the unification of approaches could be guided.

12.2 Criticisms of Models

The optimisation approach of behavioural ecology has received several criticisms that were analysed by Stephens and Krebs (1985): low testability, difficulty in choosing a strategy set, dependence of design features, ignoring genetic mechanisms, ignoring historical origins, the question of by-products, and extreme atom- ism. Most of these criticisms are applicable to all models that assume that organisms distribute themselves based on a rule of maximising individual payoff, i.e. foraging theory and ideal-free, isoleg, and isodar models. Some of these criticisms are related primarily to the ‘adaptationist’ background implied in these approaches. Others are typical criticisms aimed at any attempt to simplify nature with a modelling technique. The frequent and unavoidable response to most of this criticism is that designing models is the best method to organise ideas in order to produce explanations and predictions. Lomnicki (1999) and Holt and Kimbrell (2007) enumerated some disadvantages of individual-based models. Averages must be computed over large numbers of simulation runs. If the aim is to draw population inferences from individual-based models, this makes it hard to survey the available parameter space thoroughly in complicated models. Their complexity can sometimes make it difficult to deduce which features of the system account for a particular observed outcome. Individual- based models can become so complex that the effort required to understand them distracts the researcher from the original goals. There are many individual-based models because it is normally necessary to develop one for each species and environmental context, with the widespread use of ad hoc assumptions. There is a lack of a theoretical and conceptual approach that unified all of these developments (Grimm and Railsback 2005). Species distribution models have been criticised by an excessive emphasis of methodology. Driven by the rapid advances in the availability and quality of spatially explicit environmental data (Kerr and Ostrovsky 2003), as well as in the plethora of statistical techniques available (Segurado and Araújo 2004; Elith et al. 2006; Hernández et al. 2006), the focus of recent studies has been overwhelm- ingly methodological. This emphasis on the data and statistical component models has come at the expense of a full integration of ecological theory into correlative 12.3 Molecular Versus Molar Models and Processes Versus Associations 179 species distribution modelling and may have undermined the validity of the approach (Guisan and Thuiller 2005; Austin 2007). The discussion concerning what definition of niche should be assigned to species distribution models has thus far failed to provide a clear theoretical framework for them (Guisan and Thuiller 2005; Araújo and Guisan 2006; Jiménez Valverde et al. 2008; Elith and Leathwick 2009). Cassini (2011b) enumerated 13 different types of niches that have been defined in an attempt to relate the output of species distribution models to ecological theory: environmental and trophic niches, fundamental and realised niches, Grinnellian and Eltonian niches, potential and ‘Chase and Leibold’s’ (2003) niches, environmental and climatic niches, and alpha, beta, and gamma niches. The discussion about the fundamental (or potential distribution) and the realised niches appears to actually be a discussion about the degree of predictability of species distribution models. The logic is that if a model is based on data using a current distribution, there is a risk of generating predictions based on a more limited (or wider) environmental range than the species can actually use (Jiménez Valverde et al. 2008). Distribution models originated in traditional population ecology have been readily tested when they were applied to estimate the risk of extinctions to endangered species. Spatial population viability analyses, which aim to predict the persistence of spatially structured populations, are complex models. They require parameter estimates that are hard to obtain, demand a large amount of data, and increase potential for error and the magnitude of uncertainty in models (Conroy et al. 1995; Reed et al. 2002). Lindenmayer et al. (2003) found that predictions of population viability analyses can be poor even for very well-known species, so they suggested that factors other than availability of good life history data can influence model performance. They concluded that the predictive ability of population viability analysis was limited by complex processes typically overlooked in spatial population models, which were interactions between landscape structure and life history attributes.

12.3 Molecular Versus Molar Models and Processes Versus Associations

In the models described in this book, two types of approaches are easily distinguished: molar, associative models that describe patterns, and molecular, mechanistic models that explain processes. Molar models include movement ecology, home-range, site suitability, pattern-based, macroecology, and species distribution models, while molecular models include foraging theory and ideal-free, metapopulation, source–sink, population, isodar, and isoleg models. Molar models make a clear emphasis of the role of the environment. They assume a direct relationship between abundance and habitat suitability and the functional response is assumed to follow a linear pattern. These models also assume that the distribution of organisms balances environmental suitability. These models do not 180 12 Final Remarks explain the processes responsible for the observed patterns and are limited in their explicative ability, and they should only be applied to the range of conditions in which they were taken. Molecular models have the virtue of delving into the mechanisms responsible for the patterns. They are problematic, however, as they generally require more complex information than molar models, which need only measure abundance/presence and habitat quality. Foraging models require some direct or indirect measurement of individual fitness. Population and metapopulation models are demanding because they require measurements of demographic parameters for several generations and several sites simultaneously. Mechanistic models of species distribution are likely to have fewer data requirements, but they are still in a developmental stage. In molecular models, analytical and simulation models can be distinguished between. Most molecular models that are presented in this book are of the analytical type with the aim of disentangling the general rules underlying spatial ecological processes. As with most analytical models, they do not seek predictions on specific conditions. Simulation models, which are represented by spatially explicit individual-based models in this book, have the opposite characteristics (i.e. a low degree of generality where a model is built for each situation), but they can produce predictions that fit specific situations. Another key difference between models is related to their link to fundamental principles. Most models use parameters that are incorporated a posteriori of observing nature. A few approaches have the epistemological advantage of relying more directly on the principle of natural selection to generate explanations and predictions. Foraging theory and ideal-free, isodar, isoleg, and some individual- based models all assume that organisms use space based on a fitness-maximising principle. The types of models are not mutually exclusive and could be used for different functions. Molar models can serve to get results quickly and with some predictive nature that, although limited, can help with the management of biodiversity conservation, and mechanistic models can be used to achieve a more complete explanation of phenomena in distribution ecology. While there could be two complementary approaches, some assumptions are so different that they are incompatible. These differences will be discussed in the following sections.

12.4 Amount of Resources Versus Patch Arrangement

The major conflict between approaches is given in the scale of landscape. At smaller scales, there is a domain of individual-level models, while species level models dominate at larger scales. In contrast, landscape scale is disputed between models at the level of individuals, aggregations, societies, subpopulations, populations, and even species. In addition, there is much debate about what matters most with the landscape scale: the quality of the habitat or the arrangement of habitat patches. Ideal-free, site suitability, and non-spatially structured-population approaches 12.5 Organisation Levels and Ecological Scales Revisited 181 emphasise the role of habitat suitability, while the metapopulation approach is based on the role of the size and distribution of habitat patches. This distinction is relevant for the construction of a sound theory of distribution, as well as for its applied implications. Avoiding habitat fragmentation and increasing connectivity are different conservation strategies to those reducing losses of key resources. While this dispute is not resolved, there is some consensus that there is no single answer to this problem, but that the model that best fits the distribution at the landscape scale depends on the life history characteristics of the species involved. Typically, for relatively large-bodied, long-lived organisms with low reproductive rates, long generation times, and relative low habitat specificity, habitat availability is probably more important than for species characterised by contrasting traits. Short generation times, small body size, high rate of population increase, and high habitat specificity are life history characteristics that probably respond more accurately to a typical metapopulation dynamics. There are groups of species with a greater tendency to form metapopulation structures, such as butterflies and annual grasses. Each metapopulation of this type of species frequently occupies relatively small geographic ranges. The source–sink structures are more common in sessile organisms (Dias 1996; Pulliam 2000). The most widespread metapopulations among vertebrates would occur in a context of structurally similar habitats but with density-independent variables that produce different mortality rates between habitats, for example as a result of hunting (Woodroffe and Ginsberg 1998).

12.5 Organisation Levels and Ecological Scales Revisited

Recognition of the problem of scale in ecology occurred as a sudden shift beginning in the late 1970s and early 1980s (Schneider 2001). Two of the major insights to emerge in ecology since then are that ecological processes are scale-dependent and that many ecological systems are hierarchically structured (Allen and Starr 1982; Wiens 1989). Scale study and the hierarchical theory have provided a framework for thinking about the distribution of organisms at different scales, and some rules emerged. When moving across scales, the dominant process changes. One of the characteristics of the relationship between processes and scales is that mechanisms can operate at different scales to those on which the patterns are observed: in some cases, the patterns must be understood as emerging from the collective behaviours of large ensembles of smaller scale units; in other cases the pattern is imposed by larger scale constraints (Levin 1992). In short, there are bottom-up and top-down links between ecological scales. Hierarchy theory proposed two general characteristics of ecological hierarchies: (1) moving towards larger ecological scales, unstable systems now seem stable, and (2) on large scales, bottom-up controls are less relevant than top-down controls; thus abiotic factors predominate over biotic factors (O’Neill and King 1998). Despite this hierarchical perspective being cited frequently as a strategy for the integration of approaches in distribution ecology, the link between this approach 182 12 Final Remarks and the models presented in this book is not simple. Moreover, the relationship between levels and scales is not always as straightforward as it posits. While there is a trend that low-level processes occur at small ecological scales and that biotic factors may be the most relevant at those scales, it is not clear which is the degree of generalisation of these rules. That is, while there is a relationship between levels and scales, this is not final and can sometimes lead to confusion. Deviations from this relationship are of three types. On the one hand, there are processes at low levels which operate at high scales. This is the case of foraging and reproductive site selection for many marine vertebrates. These animals can travel hundreds of kilometres to choose a site to get food, copulate, or breed. Another plane of confusion occurs when multilevel processes operate on the same scale. For example, the distributions of groups, societies, subpopulations, and populations may occur at the landscape level, depending on the species and environment. Finally, there are approaches that simply do not have a definition of scale, such as models of ideal- free distribution, and there are others that can be applied to different scales, such as use–availability models.

12.6 Equilibrium and the Generalised Matching Law

In ecology there has long been controversy over whether nature can be described with equilibrium states or is dominated by highly complex and chaotic processes, in which equilibrium is an exception. This controversy was especially intense in the 1970s, when new models were shown to still explain ecological phenomena studied until that time, having incorporated chaotic processes. It has been suggested that the notion of equilibrium in ecology rests on a worldview of the universe that stretches back to the origins of Western thought, especially Greek philosophy. Beyond this debate, the fact is that most of the models presented in this book agree to raise the existence of equilibrium distributions to some ecological scale, towards which the distributions converge after a disturbance or change in the environment. This does not imply that the distributions are static. Some models pose dynamic equilibrium conditions (e.g. metapopulations) and others tolerate this condition well, although they do not do it explicitly. In metapopulations, the equilibrium is reached at the molar level but local populations suffer frequent extinctions. A hypothesis that has appeared repeatedly in this book is the matching rule, which states that use increases monotonically with the quality of the site, so the equilibrium state is reached when

= a (12.1) pbiiR where pi is the probability of using site i, Ri is site quality, and b and a are parameters that normally represent biases produced by constant differences between organisms. The matching rule has been mentioned repeatedly thorough this book, and the following list summarises these uses. 12.6 Equilibrium and the Generalised Matching Law 183

• The matching rule was first described by animal psychologists in choice experiments in which animals had to distribute their responses in two sites with different rates

of food delivery. The rate of responses represents pi, the rate of gained reinforce-

ment is Ri, a is a measure of the sensitivity of the individual to differences in reinforcement rates, and b can be a composite of several constant factors of the experimental situation, such as position bias or biases produced by differences in quality and amount of reinforcement. • The matching rule was derived from the marginal value theorem, which was developed by behavioural ecologists to describe the amount of time spent by

individual organisms between foraging patches. The variable pi represents the

optimal duration of a visit to a patch, Ri is the foraging gain, and a and b represent differences in the parameters of the gain curves within the patch. • The matching rule was postulated by ecologists that described the distribution of aggregations of individuals between habitats and landscapes, using ideal-free or

isodar approaches. In this case, pi represents the number of conspecifics in a

habitat patch, Ri is the habitat quality measured as fitness or gain rate, andb and a are related to differences in competitive abilities between types of individuals. • The matching rule was expected to be the equilibrium reached by populations

that occupy isolated locations (from Slobodkin 1953). At this level, pi represents

the carrying capacity of the population, Ri is the amount of habitat, and a and b are proportionality constants that represent biases in the responses and qualities. • The matching rule was deduced by researchers studying the use of pastures by

livestock in animal production. In this applied model, pi is the grazing time spent

during a month and Ri is total pasture area covered by a plant community. Constants represent differences in forage quality and in other traits between plant communities, and differences in competitive abilities between individuals. • The matching rule was implicitly present (and sometimes explicitly assumed) in site suitability and species distribution models. Resource selection functions and multiple regression analyses can be interpreted as a direct decomposition of the matching function in different components that represent the relationship between spatial responses of organisms to different resources provided by the environment. • The matching rule was assumed in macroecological hypotheses on the positive interspecific abundance–range size relationship. The matching rule is probably the most general law of distribution ecology. Equation (12.1) can be called the generalised matching law (after Fagen 1987). Paradoxically, it is probably one of the hypotheses with most opponents in ecology. Critics can be summarised in the following argument: organisms of a species can be abundant at low-quality sites and, more frequently, they may be absent from sites that are suitable. The most frequently cited causes of these deviations are competition, metapopulation and source–sink dynamics, and cyclic environmental changes and disturbances, together with dispersal limits (Van Horne 1983; Thomson et al. 1996; Railsback et al. 2003; Johnson 2008; Pulliam 2000; Anderson et al. 2003; 184 12 Final Remarks

Svenning and Skov 2004; Araújo and Pearson 2005; Guisan and Thuiller 2005; Soberón and Peterson 2005; Peterson 2006). Most criticisms are mainly concerned with the wrong assumption made in several studies, especially by some wildlife management studies, that habitat quality is a direct surrogate of population density. Other criticisms are more general, underestimating the role of habitat quality as a direct or indirect determinant of ecological processes. Some of the arguments are described here. In metapopulations, it is common to expect empty patches with a suitable habitat, due to the normal extinction–colonisation processes that characterise them. This dynamic appears to go against the idea of a subpopulation reaching equilibrium. In source– sink scenarios, habitat patches with low quality and a growth rate lower than the unity can contain subpopulations that can be even larger than source populations. This type of distribution, in which poor patches contain more individuals than quality predicts, is a general expectation of ideal despotic models and different versions of ideal-free distribution models that assume competitive differences between phenotypes. Cyclical (daily, seasonal, annual) environmental changes and disturbances produce disequilibrium in the relationship between distributions of organisms and resources. One of the most common causes of deviations from the proportionality is generated by the inevitable delay that occurs between environmental change and the numerical recovery of the species. Small changes in food availability or environmental conditions may generate offsets when species show site tenacity. Van Horne (1983) gives the example of wildlife studies conducted in northern climates, where the identification of habitat quality on the basis of summer densities would be misleading because the availability of the winter range may contribute disproportion- ately to carrying capacity. A lack of equilibrium at landscape or sub-regional scales may be caused by catastrophic events, such as fire or volcanic eruptions, or by anthropogenic impacts, such as different hunting laws between neighbouring states or countries, or different traditions in the use of land. Because the delay in the colonisation of a site depends on the distance from source populations, the effect of the delayed dispersion is expected to be greater at larger scales (Austin 2002). Windstorm, fire, and flooding are sources of variation in landscape structure and produce a mosaic of areas with different severities of disturbance. Another coarse- scale mechanism that has recently begun to be recognised as being responsible for landscape heterogeneity is lateral fluxes of matter and energy between ecosystems (Turner 2005). Processes at regional and continental scales are generally products of climatic changes such as glaciations or global warming. When processes operate at large scales, species recovery can take a long time. In some cases, these processes can set barriers to dispersal that isolate landscapes or subregions. A common example is the effect of glaciations in freshwater environments (Cassini et al. 2009). River basins represent isolated geographic units for aquatic species. The restriction of populations to refugia during glaciations has been followed by the reoccupation of some river basins by certain species that had been displaced by the ice. The basins that had no refugia during glaciations are empty of those species. In both types of basins, the available resources and other environmental variables may be 12.6 Equilibrium and the Generalised Matching Law 185 similar, but the differences in geological histories determine heterogeneity at the scale of basins that must be taken into account. In order to evaluate these criticisms, some clarifications are necessary. First, the generalised matching law, described with (12.1), does not always postulate that the proportion of uses between patches is equal to the proportion of qualities. This is predicted only in the special case in which a and b are equal to 1. So, this equation allows the incorporation of the effect of different factors that can contribute to a change of the value of these constants, both due to differences between organisms and environmental effects. The evolution of the ideal-free models that gradually incorporated different types of competitions is probably the best example. At the same time, a fundamental derivative of the generalised matching law is that, under a lack of differences in competitive abilities and other differences that can bias this relationship, this will tend to proportionality. Therein, the most important conclusion is that the main task is to traduce the characteristics of the system under study in the a and b constants. The other aspect is that any analysis in distribution ecology must be performed in consideration of scales. This is also true when applying the generalised matching law. For example, the theory of metapopulations suggests that steady state conditions can be generated in spatially structured populations. There are two basic types of dynamics: extinction–colonisation and source–sink. Although considerable variation between the abundance of organisms and patch quality is expected within subpopulations, a different scenario appears when the structured population is taken as a whole and compared with other populations. In this case, each whole population, occupying a landscape, can reach a dynamic equilibrium with the environment, and the set of spatially structured populations of a sub-region might follow the matching law. A similar reasoning can be applied when thinking on the temporal changes in the local environment. In other words, the disequilibrium can be encapsulated using average or overall environmental conditions, with sufficiently large cells (in the case of metapopulations) and/or long enough sampling periods (in the case of seasonal cycles). The importance of scale is also important when considering the effect of disturbances on coarse scales. For the effect of historical catastrophes, the procedure involves performing a hierarchical analysis with at least two levels: one at the scale at which these disturbances occurred within the biogeographical range of the species under study. On this scale, it is possible to separate the space in areas with and without the impact of the disturbance. Then, two separate analyses should be conducted, using the environmental variables that operate within disturbed and undis- turbed sectors. Finally, when a species is divided in separated units by barriers generated by catastrophic events, it is expected that in the long term, evolutionary processes will affect genetic pools in a way that units evolve differently depending on local environmental conditions. In other words, this type of spatial problems is more in the orbit of phylogeography and evolution than in that of distribution ecology. The most important conclusion from this section is that the generalised matching law is a potentially useful tool in distribution ecology. It is naive to think that nature can be reduced to such a simple equation, but it must be considered a null hypothesis, 186 12 Final Remarks as other general principles are normally only met in ideal conditions (i.e. Hardy– Weinberg equilibrium principle). This rule can be viewed as the basic conceptual pattern, such that the causes of the actual distributions (not theoretical) may be explained by the deviations found with respect to this basic pattern. The strength of this law is based on two scientific merits. Firstly, it has been derived empirically and independently in several disciplines that study the choice of resources. Secondly, it is based on the most important biological principle, i.e. of evolution by natural selection.

12.7 Holistic or Reductionist Future of Distribution Ecology

The future may be thought of from two different epistemological positions regarding the current status of the distribution ecology. One is that each level of organisation has emergent properties that are specific to that level. This holistic view, which is the method of thinking proposed by hierarchical theory, enables the consideration of different approaches described in this book as legitimate representations of nature, each in charge of explaining and describing patterns and processes that occur in the corresponding level. In this perspective, the future of ecology distribution will consist of advancing the theoretical development of each level and then using inclusive principles and hierarchy theory linking organisational levels. Because this book is organised with this holistic view, no more explanation of this approach appears to be necessary. The other view is that there is one level that is most appropriate to explain the distribution of organisms of a species on any ecological scale considered. If this is the level of individuals, one can speak of a reductionist position in distribution ecology. How can we move forward on this reductionist view in which the individual is the fundamental level? First, we must identify the types of models with an individual approach. There are three types of these models: those using individual fitness maximisation as currency, the individual-based models, and mechanistic models of species distribution. The former are typical analytical models that allow investigating processes and generate general hypotheses. The second category corresponds to simulation models that are appropriate to model-specific cases. The third type of model is based mainly on physiological mechanisms that establish tolerance ranges to environmental conditions. These phylogenetic constraints are a result of a process of natural selection operating at the individual level, and its effect is also expressed individually on the possibilities of expression of adaptations. Thus, even when they were described at the species level because they are associated with the definition of the fundamental niche and apply to the definition of biogeographic ranges, they are based on a principle that operates at the individual level. A possible future scenario of the distribution ecology based on this reductionist perspective could be as follows. The analytical models using individual fitness maximisation should be considered the most correct approach to understanding the general basis of the ecological processes responsible for the distribution on all scales. The underlying assumption is that organisms use adaptive strategies for the use of the environment. The models should also include phylogenetic constraints that limit 12.7 Holistic or Reductionist Future of Distribution Ecology 187

Fig. 12.1 Schematic representation of holistic and reductionist approaches to the future of distribution ecology. Holistic schema is the same as Fig. 1.3, with hierarchical theory connecting all levels of organisation. Reductionist approach is based on the assumption that individual level is the main ecological level and that distribution at any level should be explained with models that use individual fitness maximisation as currency, subject to phylogenetic constraints. These analytical models are linked to simulation and associative models through the principle established by the generalised habitat-matching law the expression of these strategies under certain conditions. At the edges of the geographical distribution, in the aftermath of disturbances, and at the ends of the environmental cycles, the constraints are dominant in determining the distribution process, while in the other conditions there are adaptations that best explain the distribution patterns observed. The landscape-scale phenomena should be interpreted with these models, from a conceptual integration process that has already been initiated between the study of adaptive behaviour and population ecology. There are several fundamental reviews that have demonstrated that it is possible to produce such integration (Sibly and Smith 1985; Sutherland 1996; Fryxell and Lundberg 1998; Stephens et al. 2007). This scenario is a reductionist method of analysis of those ecological processes that are responsible for the distribution at any scale. However, the question remains: how is this type of analytical approach related to the products of molar models, which represent a different methodology for studying distribution? As a first step, molar models, such as site suitability and species distribution models, serve as quick methods for predicting distribution based on simple associations of the distribution of organisms with the distribution of environmental qualities, which is a useful tool for management. In addition, it is possible to explore a way of integrating molecular models based on individuals with these molar models. This link is the generalised matching law. Figure 12.1 shows schematic representations of the holistic and the reductionist scenarios. Only future theoretical and empirical developments will determine which 188 12 Final Remarks will be the most appropriate direction of distribution ecology. I want to end the book with a final thinking. Western culture and its fascination with technological development have been responsible for that nature reaches its homeostatic capacity limits. Those who love to live in a biodiverse world and contemplate the interactions of living organisms with the environment, we are astonished at the perpetuation of national and international policies that promote the destruction and disregard for nature. This book has something of a paradox. In its introduction, it was stated that the main objective of distribution ecology is untangle ecological interactions that determine that the organisms of a particular species distributed heterogeneously in space. However, natural biotic and abiotic interactions have been increasingly per- turbated by human activities. One may simply hope that in the future not only anthropogenic factors be responsible for the distribution of living beings. References

Abba AM, Cassini MH (2008) Ecology and conservation of three species of armadillos in the Pampean Region, Argentina. In: Vizcaino S, Loughry J (eds) Biology of the Xenarthra. University of Florida Press, Florida Abba AM, Cassini MH (2010) Ecological segregation between two sympatric species of armadillos (Xenarthra, Mammalia) with different masticatory morphology. Acta Theriol 55:35–44 Allen TFH, Starr TB (1982) Hierarchy: perspectives for ecological complexity. University of Chicago Press, Chicago Amezcua AB, Holyoak M (2000) Empirical evidence for predator–prey source-sink dynamics. Ecology 81:3087–3098 Anderson DJ (1982) The home range: a new nonparametric estimation technique. Ecology 63:103–112 Anderson DM, Kothmann MM (1980) Relationship of distance traveled with diet and weather for hereford heifers. J Range Manage 33:217–220 Anderson MT, Seifert HS (2008) Neisseria gonorrhoeae and humans perform an evolutionary LINE dance. Mob Genet Elements 1:85–87 Anderson RP, Lew D, Peterson AT (2003) Evaluating predictive models of species’ distributions: criteria for selecting optimal models. Ecol Model 162:211–232 Araújo MB, Guisan A (2006) Five (or so) challenges for species distribution modelling. J Biogeogr 33:1677–1688 Araújo MB, Pearson RG (2005) Equilibrium of species’ distributions with climate. Ecography 28:693–695 Arduino SM, Quiroga MP, Ramírez MS, Merkier AK, Errecalde L, Di Martino A, Smayevsky J, Kaufman S, Centrón D (2012) Transposons and integrons in colistin resistant clones with epi- demic or sporadic behavior. J Med Microbiol 61:1417–1420 Arzamendia Y, Cassini MH, Vilá BL (2006) Habitat use by vicuñas, Vicugna vicugna, in Laguna Pozuelos Reserve (Jujuy, Argentina). Oryx 40:1–6 Austin MP (2002) Spatial prediction of species distribution: an interface between ecological theory and statistical modelling. Ecol Model 157:101–118 Austin MP (2007) Species distribution models and ecological theory: a critical assessment and some possible new approaches. Ecol Model 200:1–19 Avise JC, Helfman GS, Saunders NC, Hales LS (1986) Mitochondrial DNA differentiation in North Atlantic eels: population genetic consequences of an unusual life history pattern. Proc Natl Acad Sci U S A 83:4350–4354 Avise JC, Arnold J, Ball RM, Bermingham E, Lamb T, Neigel JE, Reeb CA, Saunders NC (1987) Intraspeciﬁc phylogeography: the mitochondrial DNA bridge between population genetics and systematics. Annu Rev Ecol Syst 18:184–216

M.H. Cassini, Distribution Ecology: From Individual Habitat Use 189 to Species Biogeographical Range, DOI 10.1007/978-1-4614-6415-0, © Springer Science+Business Media New York 2013 190 References

Baguette M, Van Dyck HV (2007) Landscape connectivity and animal behavior: functional grain as a key determinant for dispersal. Landsc Ecol 22:1117–1129 Bailey DW, Thomas MG, Walker JW, Whitmore BK, Tolleson D (2010) Effect of previous experience on grazing patterns and diet selection of Brangus cows in the Chihuahuan Desert. Range Ecol Manag 63:438–446 Bailleul F, Cotté C, Guninet C (2007) Southern elephant seals from Kerguelen Islands confronted by Antarctic Sea ice: changes in movements and in diving behaviour. Deep Sea Res II 54:343–355 Baker RR (1978) The evolutionary ecology of animal migration. Hodder & Sotughton, London Barrett RH (1982) Habitat preferences of feral hogs, deer, and cattle on a Sierra Foothill range 1982. J Range Manage 35:342–346 Barry SC, Elith J (2006) Error and uncertainty in habitat models. J Appl Ecol 43:413–423 Bartholomew GA (1970) A model for the evolution of pinniped polygyny. Evolution 24:546–559 Barton NH, Whitlock MC (1997) The evolution of metapopulations. In: Hanski I, Gilpin ME (eds) Metapopulation biology: ecology, genetics, and evolution. Academic, San Diego, pp 183–210 Bartumeus F, Levin SA (2008) Fractal intermittence in locomotion: linking animal behavior to statistical patterns of search. Proc Natl Acad Sci U S A 105:19072–19077 Bartumeus F, Da Luz MGE, Viswanathan GM, Catalan J (2005) Animal search strategies: a quantitative random-walk analysis. Ecology 86:3078–3087 Bascompte J, Vilà C (1997) Fractals and search paths in mammals. Landsc Ecol 12:13–221 Baselga A, Araujo MB (2009) Individualistic vs community modelling of species distributions under climate change. Ecography 32:55–65 Baum WM (1974) On two types of deviation from the matching law: bias and undermatching. J Exp Anal Behav 22:231–242 Baum WM (1979) Matching, undermatching, and overmatching in studies of choice. J Exp Anal Behav 32:269–281 Baum WM (1981) Changing over and choice. In: Bradshaw CM, Szabadi E, Lowe CF (eds) Quantification of steady-state operant behaviour. Elsevier, Amsterdam, pp 118–135 Bazzaz FA (1991) Habitat selection in plants. Am Nat 137:116–130 Beaumont LJ, Hughes L, Poulsen M (2005) Predicting species distributions: use of climatic parameters in BIOCLIM and its impact on predictions of species’ current and future distributions. Ecol Model 186:250–269 Begon M, Townsend CR, Harper JL (2006) Ecology: from individuals to ecosystems. Blackwell, Oxford Bélisle M (2005) Measuring landscape connectivity: the challenge of behavioral landscape ecology. Ecology 86:1988–1995 Bell AD (1984) Dynamic morphology: a contribution to plant population ecology. In: Dirzo R, Sarukhan J (eds) Perspectives on plant population ecology. Sinauer, Sunderland, pp 48–65 Bell WJ (1991) Searching behavior: the behavioral ecology of finding resources. Chapman and Hall, New York Bennett ATD (1996) Do animals have cognitive maps? J Exp Biol 199:219–224 Berec L (2002) Techniques of spatially explicit individual-based models: construction, simulation, and mean-field analysis. Ecol Model 150:55–81 Bernstein C, Kacelnik A, Krebs JR (1991a) Distribution of birds among habitats: theory and practice. In: Perrins CM (ed) Bird population studies: relevance to conservation and management. Oxford University Press, Oxford, pp 317–345 Bernstein C, Kacelnik A, Krebs JR (1991b) Individual decisions and the distribution of predators in a patchy environment. II. The influence of travel costs and structure of the environment. J Anim Ecol 60:205–225 Berry PM, Dawson TP, Harrison PA, Pearson G (2002) Modelling potential impacts of climate change on the bioclimatic envelope of species in Britain and Ireland. Glob Ecol Biogeogr 11:453–462 Berryman AA (2002) Population: a central concept for ecology? Oikos 97:439–442 References 191

Bertonatti C, Corcuera J (2000) Situación ambiental Argentina 2000. Fundación Vida Silvestre Argentina, Buenos Aires Birch CPD, Hutchings MJ (1994) Exploitation of patchily distributed soil resources by the clonal herb Glechoma hederacea . J Ecol 82:653–664 Boness DJ (1991) Determinants of mating systems in the Otariidae (Pinnipedia). In: Renouf D (ed) Behavior of Pinnipeds. Chapman and Hall, London, pp 1–44 Boorman SA, Levitt PR (1973) Group selection on the boundary of a stable population. Theor Popul Biol 4:85–128 Borgnia M, Galante ML, Cassini MH (2000) Diet of the coypu (Myocastor coypus ) in agro-systems of the Argentinean Pampas. J Wildl Manage 64:354–361 Borgnia M, Vilá BL, Cassini MH (2008) Interaction between vicuñas, feral donkeys and livestock in Argentinean Puna. J Arid Environ 72:2150–2158 Bornemann H, Kreyscher M, Ramdohr S (2000) Southern elephant seal movements and Antarctic sea ice. Antarct Sci 12:3–15 Botkin DB, Janak JF, Wallis JR (1972) Some ecological consequences of a computer model of forest growth. J Ecol 60:849–872 Boulinier T, Danchin E (1997) The use of conspeciﬁc reproductive success for breeding patch selection in terrestrial migratory species. Evol Ecol 11:505–517 Bowler DE, Benton TG (2005) Causes and consequences of animal dispersal strategies: relating individual behaviour to spatial dynamics. Biol Rev Camb Philos Soc 80:205–225 Boyce MS, McDonald LL (1999) Relating populations to habitats using resource selection functions. Trends Ecol Evol 14:268–272 Boyce MS, Vernier PR, Nielsen SE, Schmiegelow FKA (2002) Evaluating resource selection functions. Ecol Model 157:279–298 Boyer D, Ramos Fernández G, Miramontes O, Mateos JL, Cocho G, Larralde H, Ramos H, Rojas F (2006) Scale-free foraging by primates emerges from their interaction with a complex environment. Proc R Soc Lond B 273:1743–1750 Bradshaw AD, McNeilly T (1981) Evolution and pollution. Arnold, London Breiman L, Friedman JH, Olshen RA, Stone CJ (1984) Classiﬁcation and regression trees. Wadsworth, Belmont Brereton R, Bennett S, Mansergh I (1995) Enhanced greenhouse climate change and its potential effect on selected fauna of South-Eastern Australia: a trend analysis. Biol Conserv 72:339–354 Bristow KD, Ockenfels RA (2004) Pairing season habitat selection by Montezuma quail in southeastern Arizona. J Range Manage 57:532–538 Broennimann O, Fitzpatrick MC, Pearman PB, Petitpierre B, Pellissier L, Yoccoz NG (2012) Measuring ecological niche overlap from occurrence and spatial environmental data. Glob Ecol Biogeogr 21:481–497 Brown JL (1969) Territorial behaviour and population regulation: a review and re-evaluation. Wilson Bull 81:293–329 Brown JH (1984) On the relationship between abundance and distribution of species. Am Nat 124:255–279 Brown JS (1988) Patch use as an indicator of habitat preference, predation risk, and competition. Behav Ecol Sociobiol 22:37–47 Brown JH (1995) Macroecology. University of Chicago Press, Chicago Brown JS (1998) Game theory and habitat selection. In: Dugatkin LA, Reeves HK (eds) Game theory and animal behavior. Oxford University Press, Oxford, pp 188–220 Brown JH, Maurer BA (1989) Macroecology: the division of food and space among species on continents. Science 243:1145–1150 Brussard PF (1984) Geographic patterns and environmental gradients: the central-marginal model in Drosophila revisited. Annu Rev Ecol Syst 15:25–64 Bugmann H (2001) A review of forest gap models. Clim Chang 51:259–305 Burnham KP, Anderson DR (2002) Model selection and multimodel inference: a practical information-theoretic approach. Springer, New York 192 References

Burt WH (1943) Territoriality and home range concepts ad applied to mammals. J Mammal 24:346–352 Cade BS, Noon BR, Flather CH (2005) Quantile regression reveals hidden bias and uncertainty in habitat models. Ecology 86:786–800 Cain ML (1994) Consequences of foraging in clonal plant species. Ecology 75:933–944 Calabrese JM, Fagan WF (2004) A comparison-shopper’s guide to connectivity metrics. Front Ecol Environ 2:529–536 Campagna C, Quintana F, Le Boeuf B, Blackwell S, Crocker D (1998) Diving behaviour and foraging ecology of female southern elephant seals from Patagonia. Aquat Mamm 4:1–11 Canon SK, Urness PJ, DeByle NV (1987) Habitat selection, foraging behavior and dietary nutrition of elk in burned aspen forest. J Range Manage 40:433–438 Cappozzo HL, Tunez JI, Cassini MH (2008) Role of sexual harassment in the evolution of breeding colonies in the Southern sea lion Otaria flavescens . Naturwissenschaften 95:625–630 Carter J, Leonard BP (2002) A review of the literature on the worldwide distribution, spread of, and efforts to eradicate the coypu ( Myocastor coypus ). Wildl Soc Bull 30:162–175 Cassini MH (1991) Foraging under predation risk in the Wild Guinea Pig Cavia aperea . Oikos 62:20–24 Cassini MH (1993) Searching strategies within food patches in the armadillo Chaetophractus vellerosus . Anim Behav 46:400–402 Cassini MH (1995) Etología aplicada a resolver problemas de alimentación del ganado. Etología 4:33–46 Cassini MH (1999) The evolution of reproductive systems of pinnipeds. Behav Ecol 10:612–616 Cassini MH (2000) A model on female breeding dispersion and the reproductive systems of pinnipeds. Behav Process 51:93–99 Cassini MH (2001) Behavioral responses of South American fur seals to approach by tourists—a brief report. Appl Anim Behav Sci 71:341–346 Cassini MH (2011a) Consequences of local Allée effects in spatially structured populations. Oecologia 165:547–552 Cassini MH (2011b) Ecological principles of species distribution models: the habitat matching law. J Biogeogr 38:2057–2065 Cassini MH, Fernández-Juricic E (2003) Costs and benefits of joining South American sea lion breeding groups: testing the assumptions of a model on female breeding dispersion. Can J Zool 81:1154–1160 Cassini MH, Galante ML (1992) Foraging under predation risk in the wild guinea pig: the effect of vegetation height on habitat utilization. Ann Zool Fenn 4:289–294 Cassini MH, Hermitte G (1992) Patterns of environmental use by cattle in relation to the consumption of supplemental food blocks. Appl Anim Behav Sci 32:297–312 Cassini MH, Krebs JR (1994) Behavioural responses to food addition by hedgehogs. Ecography 17:289–296 Cassini MH, Kacelnik A, Segura ET (1990) The tale of the Screaming Hairy Armadillo, the Guinea Pig and the marginal value theorem. Anim Behav 39:1030–1050 Cassini MH, Szteren D, Fernández Juricic E (2004) Fence effects on the behavioural responses of South American fur seals to tourist approaches. J Ethol 22:127–133 Cassini MH, Fasola L, Chéhebar C, MacDonald DW (2009) Scale-dependent analysis of an otter- crustacean system in Argentinean Patagonia. Naturwissenschaften 96:593–599 Cassini MH, Fasola L, Chehébar C, Macdonald DW (2010) Defining conservation status using limited information: the case of Patagonian otters Lontra provocax in Argentina. Hydrobiologia 652:389–394 Cate LR, Mullord M, Temple W, Matthews LR (1978) A feed system for the study of choice behavior in cows. Behav Res Methods Instrum 10:27–28 Caudill CC (2003) Empirical evidence for nonselective recruitment and a source–sink dynamic in a mayfly metapopulation. Ecology 84:2119–2132 Caughley G (1977) Analysis of vertebrate populations. Wiley, Washington References 193

Charnov EL (1976) Optimal foraging, the marginal value theorem. Theor Popul Biol 9:129–136 Chase JM, Leibold MA (2003) Ecological niches: linking classical and contemporary approaches. The University of Chicago Press, Chicago Chaverri G, Kunz TH (2011) Response of a specialist bat to the loss of a critical resource. PLoS One 6:e28821. doi: 10.1371/journal.pone.0028821 Chilton NB (1989) Life cycle adaptations and their implications in the distribution of two parapat- ric species of tick. Flinders University of South Australia, Adelaide Chuine I, Beaubien E (2001) Phenology is a major determinant of temperate tree range. Ecol Lett 4:500–510 Ciszak M, Comparini D, Mazzolai B, Baluska F, Arecchi FT, Vicsek T, Mancuso S (2012) Swarming behavior in plant roots. PLoS One 7:e29759. doi: 10.1371/journal.pone.0029759 Clavero M, Garcia-Berthou E (2005) Invasive species are a leading cause of animal extinctions. Trends Ecol Evol 20:110 Clayton NS, Krebs JR (1994) Hippocampal growth and attrition in birds affected by experience. Proc Natl Acad Sci U S A 91:7410–7414 Climent J, Aránzazu Prada M, Calama R, Chambel MR, Sánchez de Ron D, Alía R (2008) To grow or to seed: ecotypic variation in reproductive allocation and cone production by young female Aleppo pine ( Pinus halepensis , Pinaceae). Am J Bot 95:833–842 Clobert J, GalliardJF L, Cote J, Meylan S, Massot M (2009) Informed dispersal, heterogeneity in animal dispersal syndromes and the dynamics of spatially structured populations. Ecol Lett 12:197–209 Clutton-Brock TH, Price O, MacColl A (1992) Mate retention, harassment and the evolution of ungulate leks. Behav Ecol 3:234–242 Conroy MJ, Cohen Y, James FC, Matsinos YG, Maurer BA (1995) Parameter estimation, reliability and model improvement for spatially explicit models of animal populations. Ecol Appl 5:17–19 Cook RE (1966) Factors affecting utilization of mountain slopes by cattle. J Range Manage 19:200–204 Cornuet JM, Oldroyd BP, Crozier RH (1995) Unequal thermostability of allelic forms of malate dehydrogenase in honey bees. J Apic Res 34:45–47 Courchamp F, Clutton Brock T, Grenfell B (1999) Inverse density dependence and the Allée effect. Trends Ecol Syst 14:405–410 Courchamp F, Berek L, Gascoigne J (2008) Allée effects in ecology and conservation. Oxford University Press, Oxford Cox BC, Moore PD (2005) Biogeography: an ecological and evolutionary approach. Blackwell, Oxford Crowder LB, Hazen EL, Avissar N, Bjorkland R, Latanich C, Ogburn MB (2008) The impacts of ﬁsheries on marine ecosystems and the transition to ecosystem-based management. Annu Rev Ecol Evol Syst 39:259–278 Curtis H, Barnes S, Schnek A, Massarini A (2008) Biología. Editorial Médica Panamericana, Madrid D’adamo P, Guichón ML, Bó RF, Cassini MH (2000) Habitat use of coypus ( Myocastor coypus ) in agro-systems of the Argentinean Pampas. Acta Theriol 45:25–33 Dale MRT, Dixon PM, Fortin MJ, Legendre P, Myers DE, Rosenberg MS (2002) Conceptual and mathematical relationships among methods for spatial analysis. Ecography 25:558–577 Dalziel BD, Morales JM, Fryxell JM (2008) Fitting probability distributions to animal movement trajectories: using artiﬁcial neural networks to link distance, resources, and memory. Am Nat 172:248–258 Damschen EI, Brudvig LA, Haddad NM, Levey DJ, Orrock JL, Tewksbury JJ (2008) The movement ecology and dynamics of plant communities in fragmented landscapes. Proc Natl Acad Sci U S A 105:19078–19083 Davies NB (1991) Mating systems. In: Krebs JR, Davies NB (eds) Behavioural ecology: an evolutionary approach. Blackwell, Oxford, pp 263–294 194 References

Daunt F, Afanasyev V, Adam A, Croxall JP, Wanless S (2007) From cradle to early grave: juvenile mortality in European shags Phalacrocorax aristotelis results from inadequate development of foraging proﬁciency. Biol Lett 3:371–374 de Kroon H, Visser EJW, Huber H, Mommer L, Hutchings MJ (2009) A modular concept of plant foraging behaviour: the interplay between local responses and systemic control. Plant Cell Environ 32:704–712 Del Lama MA, Oliveira Souza R, Araneda Durán XA, Egea Soares AE (2004) Clinal variation and selection on MDH allozymes in honeybees in Chile. Hereditas 140:149–153 Dias PC (1996) Source and sinks in population biology. Trends Ecol Evol 11:326–330 Diffendorfer JE (1998) Testing models of source-sink dynamics and balanced dispersal. Oikos 81:417–433 Dixon KR, Chapman JA (1980) Harmonic mean measure of animal activity areas. Ecology 61:1040–1044 Doncaster CP, Macdonald DW (1991) Drifting territoriality in the red fox Vulpes vulpes. J Anim Ecol 60:423–439 Dougherty DM, Lewis P (1992) Matching by horses on several concurrent variable-interval schedules. Behav Processes 26:69–76 Dukas R, Ratcliffe J (2009) Cognitive ecology II. The University of Chicago Press, Chicago Dunbar RIM (1992) Time: a hidden constraint on the behavioral ecology of baboons. Behav Ecol Sociobiol 31:35–49 Dunbar RIM (1996) Determinants of group size in primates: a general model. Proc Br Acad 88:33–57 Duncan P (1983) Determinants of the use of habitat by horses in a mediterranean wetland. J Anim Ecol 52:93–109 Dussault C, Ouellet JP, Courtois R, Huot J, Breton L, Jolicoeur H (2005) Linking moose habitat selection to limiting factors. Ecography 28:619–628 Dyer FC (1998) Cognitive exology of navigation. In: Dukas R (ed) Cognitive ecology. The evolutionary ecology of information processing and decision making. The University of Chicago Press, Chicago, pp 201–260 Eckert CG, Samis KE, Lougheed SC (2008) Genetic variation across species’ geographical ranges: the central–marginal hypothesis and beyond. Mol Ecol 17:1170–1188 Edge WD, Les Marcum C, Olson-Edge SL (1988) Summer habitat selection by elk in Western Montana: a multivariate approach. J Wildl Manage 51:844–851 Eeley HA, Lawes MJ, Piper SE (1999) The inﬂuence of climate change on the distribution of indigenous forest in Kwa-Zulu- Natal, South Africa. J Biogeogr 26:595–617 Eigenbrod F, Hecnar SJ, Fahrig L (2008) Accessible habitat: an improved measure of the effects of habitat loss and roads on wildlife populations. Landsc Ecol 23:159–168 Elith J, Leathwick JR (2009) Species distribution models: ecological explanation and prediction across space and time. Annu Rev Ecol Evol Syst 40:677–697 Elith J, Graham CH, Anderson RP, Dudik M, Ferrier S, Guisan A, Hijmans RJ, Huettmann F, Leathwick JR, Lehmann A, Li J, Lohmann LG, Loiselle BA, Manion G, Moritz C, Nakamura M, Nakazawa Y, Overton JM, Townsend Peterson A, Phillips SJ, Richardson K, Scachetti- Pereira R, Schapire RE, Soberón J, Williams S, Wisz MS, Zimmermann NE (2006) Novel methods improve prediction of species’ distributions from occurrence data. Ecography 29:129–151 Emery NJ, Clayton NS (2001) Effects of experience and social context on prospective caching strategies by scrub jays. Nature 414:443–446 Emlen ST, Oring LW (1977) Ecology, sexual selection, and the evolution of mating systems. Science 197:215–223 Epperson BK (2003) Geographical genetics. Princeton University Press, Princeton Erasmus BFN, Van Jaarsveld AS, Chown SL, Kshatriya M, Wessels KJ (2002) Vulnerability of South African animal taxa to climate change. Glob Chang Biol 8:679–693 Ernest HB, Boyce WM, Bleich VC, May B, Stiver SJ, Torres SG (2003) Genetic structure of mountain lion ( Puma concolor ) populations in California. Conserv Genet 4:353–366 References 195

Evans DL, Zhu Z (1993) AVHRR for forest mapping: national applications and global implications. In: Lewis A (ed) Looking to the future with an eye on the past. Proceedings of the 1993 ACSM/ASPRS convention. ASPRS, Maryland, pp 76–79 Evans PR (1976) Energy balance and optimal foraging strategies in Shorebirds: some implications for their distribution and movements in the non-breeding season. Ardea 64:117–139 Fagen R (1987) A generalized habitat matching rule. Evol Ecol 1:5–10 Ferrier S, Guisan A (2006) Spatial modelling of biodiversity at the community level. J Appl Ecol 43:393–404 Fielding A, Bell J (1997) A review of methods for the assessment of prediction errors in conservation presence/absence models. Environ Conserv 24:38–49 Foster TM, Temple W, Robertson B, Nair V, Poling A (1996) Concurrent-schedule performance in dairy cows: persistent undermatching. J Exp Anal Behav 65:57–80 Frankham R, Ballou JD, Briscoe DA (2002) Introduction to conservation genetics. Cambridge University Press, Cambridge Freckleton RP, Watkinson AR (2002) Large-scale spatial dynamics of plants: metapopulations, regional ensembles and patchy populations. J Ecol 90:419–434 Fretwell SD (1972) Populations in a seasonal environment. Princeton University Press, Princeton Fretwell SD, Lucas HL Jr (1970) On territorial behaviour and other factors inﬂuencing habitat distribution in birds. I. Theoretical development. Acta Biotheor 19:16–36 Fryxell JM (2001) Habitat suitability and source-sink dynamics of beavers. J Anim Ecol 70:310–316 Fryxell JM, Lundberg P (1998) Individual behavior and community dynamics. Chapman & Hall, London Gamarra JGP (2005) Metapopulations in multifractal landscapes: on the role of spatial aggregation. Proc R Soc Lond B 272:1815–1822 Ganskopp D, Vavra M (1986) Habitat use by feral horses in the northern sagebrush steppe. J Range Manage 39:207–212 Garant D, Kruuk LEB, McCleery RH, Sheldon BC (2007) The effects of environmental heterogeneity on multivariate selection on reproductive traits in female great tits. Evolution 61:1546–1559 Garshelis DL (2000) Delusions in habitat evaluation: measuring use, selection, and importance. In: Boitani L, Fuller TK (eds) Research techniques in animal ecology: controversies and consequences. Columbia University Press, New York, pp 111–164 Gaston KJ (2003) The structure and dynamics of geographic ranges. Oxford University Press, Oxford Gaston KJ, Blackburn TM (2000) Pattern and process in marcroecology. Blackwell, Oxford Gaston KJ, Blackburn TM, Gregory RD (1997) Abundance–range size relationships of breeding and wintering birds in Britain: a comparative analysis. Ecography 20:569–579 Getz WM, Saltz D (2008) A framework for generating and analyzing movement paths on ecological landscapes. Proc Natl Acad Sci U S A 105:19066–19071 Gibson LA, Wilson BA, Cahill DM, Hill J (2004) Spatial prediction of rufous bristle bird habitat in a coastal heath land: a GIS-based approach. J Appl Ecol 41:213–223 Gilpin ME (1991) The genetic effective size of a metapopulation. Biol J Linn Soc 42:165–175 Giuggioli L, Bartumeus F (2010) Animal movement, search strategies and behavioural ecology: a cross-disciplinary way forward. J Appl Ecol 79:906–909 Godley BJ, Blumenthal JM, Broderick AC, Coyne MS, Godfrey MH, Hawkes LA (2008) Satellite tracking of sea turtles: where have we been and where do we go next? Endanger Species Res 4:3–22 González Suárez M, Gerber LR (2008) A behaviorally-explicit demographic model integrates habitat selection and population dynamics in California sea lions. Conserv Biol 22:1608–1618 Goodall C (1990) A survey of smoothing techniques. In: Fox J, Long JS (eds) Modern methods of data analysis. Sage, Beverly Hills, pp 126–176 196 References

Gordon IJ (1989) Vegetation community selection by ungulates on the isle of Rhum. II. Vegetation community selection. J Appl Ecol 26:53–64 Gould JL (1982) Ethology. The mechanisms and evolution of behavior. WW Norton, New York Greegor DH Jr (1985) Ecology of the little hairy armadillo Chaetophractus vellerosus . In: Montgomery GG (ed) The evolution and ecology of armadillos, sloths, and vermilinguas. Smithsonian Institution Press, Washington, DC, pp 397–405 Greene CM, Stamps JA (2001) Habitat selection at low population densities. Ecology 82:2091–2100 Grime JP (1979) Plant strategies and vegetation processes. Wiley, Chischester Grimm V, Railsback SF (2005) Individual-based modeling and ecology. Princeton University Press, New York Grover KE, Thompson MJ (1986) Factors influencing spring feeding site selection by elk in the Elkhorn Mountains, Montana. J Wildl Manage 50:466–470 Gruntman M, Novoplansky A (2004) Physiologically mediated self/non-self discrimination. Proc Natl Acad Sci U S A 101:3863–3867 Guichón ML, Cassini MH (1999) Local determinants of the distribution of the coypu (Myocastor coypus ) along Luján River, East-Central Argentina. J Wildl Manage 63:895–900 Guichón ML, Cassini MH (2005) Effect of hunting pressure on population structure of coypus in the Pampas region (Argentina). Acta Theriol 50:125–132 Guichón ML, Benítez VB, Abba A, Borgnia M, Cassini MH (2003a) Foraging behaviour of coypus Myocastor coypus in the Argentinean Pampas: why do coypus consume aquatic plants? Acta Oecol 24:241–246 Guichón ML, Borgnia M, Fernández Righi C, Cassini GH, Cassini MH (2003b) Social behaviour of coypus ( Myocastor coypus ) in the Argentinean pampas. J Mammal 84:169–177 Guichón ML, Doncaster CP, Cassini MH (2003c) Population structure of coypus ( Myocastor coypus ) in their region of origin and comparison with introduced populations. J Zool 261:265–272 Guillot G, Estoup A, Mortier F, Cosson JF (2005) A spatial statistical model for landscape genetics. Genetics 170:1261–1280 Guisan A, Thuiller W (2005) Predicting species distribution: offering more than simple habitat models. Ecol Lett 8:993–1009 Guisan A, Zimmermann NE (2000) Predictive habitat distribution models in ecology. Ecol Model 135:147–186 Gurarie E, Andrews RD, Laidre KL (2009) A novel method for identifying behavioural changes in animal movement data. Ecol Lett 12:395–408 Gutierrez D (2005) Effectiveness of existing reserves in the long-term protection of a regionally rare butterfly. Conserv Biol 19:1586–1597 Judson OP (1994) The rise of the individual-based model in ecology. Trends Ecol Evol 9:9–14 Halbritter H, Bender LC (2011) Contrasting observation and transect-based models of cattle distribution on Lincoln National Forest, New Mexico. Rangel Ecol Manage 64:514–520 Hall ER (1981) The mammals of North America. Wiley, New York Hall SJG (1988) Chillingham Park and its herd of white cattle: relationships between vegetation classes and patterns of range use. J Appl Ecol 25:777–789 Hall RM, Stokes HW (2004) Integrons or super integrons? Microbiology 150:3–4 Hamilton WD (1971) Geometry of the selfish herd. J Theor Biol 31:295–311 Hanski I (1994) A practical model of metapopulation dynamics. J Anim Ecol 63:151–162 Harding KC, McNamara JM (2002) A unifying framework for metapopulation dynamics. Am Nat 160:173–185 Hargis CD, Bissonette JA, David JL (1997) Understanding measures of landscape patterns. In: Bissonette JA (ed) Wildlife and landscape ecology. Springer, New York, pp 231–261 Harper KA, Macdonald SE (2001) Structure and composition of Riparian Boreal forest: new methods for analyzing edge. Ecology 82:649–659 Harrison S, Hastings A (2005) Genetic and evolutionary consequences of metapopulation structure. Trends Ecol Evol 11:180–183 References 197

Hartl D, Clark A (1997) Principles of population genetics, 3rd edn. Sinauer Associates, Sunderland Harvey PH, Pagel MD (1991) The comparative method in evolutionary biology. Oxford University Press, Oxford Hastie T, Tibshirani RJ, Friedman JH (2001) The elements of statistical learning: data mining, inference and prediction. Springer, New York Hastings A, Cuddington K, Davies KF, Dugaw CJ, Elmendorf S, Freestone A, Harrison S, Holland M, Lambrinos J, Malvadkar U, Melbourne BA, Moore K, Taylor C, Thomson D (2005) The spatial spread of invasions: new developments in theory and evidence. Ecol Lett 8:91–101 Hayne DW (1949) Calculation of size of home range. J Mammal 30:1–18 Heikkinen R, Luoto M, Aráujo MB, Virkkala R, Thuiller W, Sykes MT (2006) Methods and uncer- tainties in bioclimatic envelope modelling under climate change. Prog Phys Geogr 30:751–777 Heil M, Karban R (2010) Explaining evolution of plant communication by airborne signals. Trends Ecol Evol 25:137–144 Hellberg ME (1995) Stepping stone gene ﬂow in the solitary coral Balanophyllia elegans : equilibrium and nonequilibrium at different spatial scales. Mar Biol 123:573–581 Hernández PA, Graham CH, Master LL, Albert DL (2006) The effect of sample size and species characteristics on performance of different species distribution modeling methods. Ecography 29:773–785 Herrnstein RJ (1961) Relative and absolute strength of responses as a function of frequency of reinforcement. J Exp Anal Behav 4:267–272 Hewson R, Wilson CJ (1979) Homo range and movements of Scottish Blackface sheep in Lochaber, north-west Scotland. J Appl Ecol 16:743–751 Hirzel A, Hausser J, Perrin N (2001) Biomapper 1.0. Lausanne. Laboratory for Conservation Biology. Available at: http://www.unil.ch/biomapper Hirzel AH, Hausser J, Chessel D, Perrin N (2002) Ecological-niche factor analysis: how to com- pute habitat suitability maps without absence data? Ecology 83:2027–2036 Hofgaard A, Ball JP, Danell K, Callaghan TV (1999) Animal responses to global change in the North. Ecol Bull 47:1–187 Hofmann RK, Otte K, Ponce CF, Ríos MA (1983) El Manejo de la Vicuña Silvestre. Editorial Cooperación Técnica Alemana GTZ, Lima Holderegger R, Wagner HH (2006) A brief guide to landscape genetics. Landsc Ecol 21:793–796 Holt RD (1985) Population dynamics in two patch environments: some anomalous consequences of an optimal habitat distribution. Theor Popul Biol 28:181–208 Holt RD (1993) Ecology at the mesoscale: the inﬂuence of regional processes on local communities. In: Ricklefs RE, Schluter D (eds) Species diversity in ecological communities. University Chicago Press, Chicago, pp 77–88 Holt RD, Keitt TH (2000) Alternative causes for range limits: a metapopulation perspective. Ecol Lett 3:41–47 Holt D, Kimbrell T (2007) Foraging and population dynamics. In: Stephens DW, Brown JS, Ydenberg RC (eds) Foraging behavior and ecology. The University Chicago Press, Chicago, pp 365–396 Holyoak M, Casagrandi R, Nathan R, Revilla E, Splegel O (2008) Trends and missing parts in the study of movement ecology. Proc Natl Acad Sci U S A 105:19060–19065 Hoogland JL (1995) The black-tailed prairie dog: social life of a burrowing mammal. The University of Chicago Press, Chicago Huntley B, Bartlein PJ, Prentice IC (1989) Climatic control of the distribution and abundance of beech (Fugus L) in Europe and North America. J Biogeogr 16:551–560 Husband BC, Barrett SCH (1998) Spatial and temporal variation in population size of Eichhornia paniculata in ephemeral habitats: implications for metapopulation dynamics. J Ecol 86:1021–1031 Huston M, DeAngelis D, Post B (1988) New computer models unify ecological theory. Bioscience 38:682–691 198 References

Hutchings MJ, de Kroon H (1994) Foraging in plants: the role of morphological plasticity in resource acquisition. Adv Ecol Res 25:159–238 Hutchinson GE (1947) A note on the theory of competition between two social species. Ecology 28:319–321 International Whaling Commission (2001) Report of the workshop on the comprehensive assessment of right whales: a worldwide comparison. J Cetacean Res Manage 2:1–60 IUCN (2001) IUCN red list categories and criteria: version 3.1. IUCN Species Survival Commission, Gland IUCN (2003) Guidelines for application of IUCN red list criteria at regional levels: version 3.0. IUCN Species Survival Commission, Gland Ives AR, Klopfer ED (1997) Spatial variation in abundance created by stochastic temporal variation. Ecology 78:1907–1913 Jacob J, Brown JS (2000) Microhabitat use, giving-up densities and temporal activity as short- and long-term anti-predator behaviors in common voles. Oikos 91:131–138 Jacobs LF, Schenk F (2003) Unpacking the cognitive map: the parallel map theory of hippocampal function. Psychol Rev 110:285–315 Jander R (1975) Ecological aspects of spatial orientation. Annu Rev Syst Ecol 6:171–188 Jarrel KF, McBride MJ (2008) The surprisingly diverse ways that prokaryotes move. Nat Rev Microbiol 6:466–476 Jennrich RI, Turner FB (1969) Measurement of non-circular home range. J Theor Biol 22:227–237 Jiménez Valverde A, Lobo JM (2006) The ghost of unbalanced species distribution data in geographic model predictions. Divers Distrib 12:521–524 Jiménez Valverde A, Lobo JM, Horta J (2008) Not as good as they seem: the importance of concepts in species distribution modelling. Divers Distrib 14:885–890 Johnson C (2008) Sensitivity of species-distribution models to error, bias, and model design: an application to resource selection functions for woodland caribou. Ecol Model 213:143–155 Johnson DD, Miller RF (2008) Intermountain presettlement juniper: distribution, abundance, and inﬂuence on postsettlement expansion. Rangel Ecol Manage 61:82–92 Johnson JB, Omland KS (2004) Model selection in ecology and evolution. Trends Ecol Evol 19:101–108 Jonzén N, Wilcox C, Possingham HP (2004) Habitat selection and population regulation in temporally ﬂuctuating environments. Am Nat 164:103–114 Karl JW (2010) Spatial predictions of cover attributes of rangeland ecosystems using regression kriging and remote sensing. Rangel Ecol Manage 63:335–349 Kaschner K, Watson R, Trites AW, Pauly D (2006) Mapping worldwide distributions of marine mammals using a relative environmental suitability (RES) model. Mar Ecol Prog Ser 316:285–310 Kattan GH (1992) Rarity and vulnerability: the birds of the Cordillera Central of Colombia. Conserv Biol 6:64–70 Kauhala K (1995) Changes in distribution of the European badger Meles meles in Finland during the rapid colonization of the raccoon dog. Ann Zool fenn 32:183–191 Kearney M, Porter WP (2009) Mechanistic niche modeling: combining physiological and spatial data to predict species’ ranges. Ecol Lett 12:334–350 Kearny M (2006) Habitat, environment and niche: what are we modelling? Oikos 115:186–191 Keating KA, Cherry S (2004) Use and interpretation of logistic regression in habitat selection studies. J Wildl Manage 68:774–789 Keeley ER, Parkinson EA, Taylor EB (2007) The origins of ecotypic variation of rainbow trout: a test of environmental vs. genetically based differences in morphology. J Evol Biol 20:415–829 Keitt TH, Urban DL, Milne BT (1997) Detecting critical scales in fragmented landscapes. Conserv Ecol 1. Available at: www.consecol.org.vol1/iss1/art4 References 199

Keller I, Largiader CR (2003) Recent habitat fragmentation caused by major roads leads to reduction of gene flow and loss of genetic variability in ground beetles. Proc R Soc Lond B 270:417–423 Kerkhoff AJ, Milne BT, Maehr DS (2000) Toward a panther-centered view of the forests of south Florida. Conserv Ecol 4:1 Kerr JT, Deguise I (2004) Habitat loss and the limits to endangered species recovery. Ecol Lett 7:1163–1169 Kiflawi M, Enquist BJ, Jordan MA (2000) Position within the geographic range, relative local abundance and developmental instability. Ecography 23:539–546 Kimura M (1953) Stepping stone model of population. Annu Rep Natl Inst Genet 3:62–63 Kimura M, Weiss GH (1964) The stepping stone model of population structure and the decrease of genetic correlation with distance. Genetics 49:561–576 King JE (1983) Seals of the world. University of Queensland Press, St. Lucia King AW (1997) Hierarchy theory: a guide to system structure for wildlife biologists. In: Bissonette JA (ed) Wildlife and landscape ecology. Effects of pattern and scale. Springer, New York, pp 185–212 Kissling WD, Dormann CF, Groeneveld J, Hickler T, Kühn I, McInerny GJ, Montoya JM, Ro¨mermann C, Schiffers K, Schurr FM, Singer A, Svenning JC, Zimmermann NE, O’Hara RB (2011) Towards novel approaches to modelling biotic interactions in multispecies assemblages at large spatial scales. J Biogeogr. doi: 10.1111/j.1365-2699.2011.02663.x Klinken R, Shepherd D, Parr R, Robinson T, Anderson L (2007) Mapping mesquite (prosopis) distribution and density using visual aerial surveys. Rangel Ecol Manage 60:408–416 Klvana I, Berteaux D, Cazelles B (2004) Porcupine feeding scars and climatic data show ecosystem effects of the solar cycle. Am Nat 164:283–297 Knowles LL (2009) Statistical phylogeography. Annu Rev Ecol Evol Syst 40:593–612 Kotliar NB, Wiens JA (1990) Multiple scales of patchiness and patch structure: a hierarchical framework for the study of heterogeneity. Oikos 59:253–260 Krakauer DC (1995) Groups confuse predators by exploiting perceptual bottlenecks: a connection- ist model of the confusion effect. Behav Ecol Sociobiol 36:421–429 Krasnov B, Khokhlova I, Shenbrot G (2003) Density-dependent host selection in ecotoparasites: an application of isodar theory to fleas parasitizing rodents. Oecologia 134:365–373 Krebs CJ (1985) Ecology: the experimental analysis of distribution and abundance. Harper & Row, New York Kreft JU (2004) Conflicts of interest in biofilms. Biofilms 1:265–276 Krivan V, Cressman R, Schneider C (2008) The ideal free distribution: a review and synthesis of the game-theoretic perspective. Theor Popul Biol 73:403–425 Lacher TE, Cassini MH (2001) Cavies. In: MacDonald DW (ed) New encyclopaedia of mammals. Oxford University Press, Oxford, pp 672–675 Lamont MM, Vida JT, Harvey JT, Jeffries S, Brown R, Huber HH, DeLong R, Thomas WK (1996) Genetic substructure of the pacific harbor seal (Phoca vitulina richardsi) off Washington, Oregon, and California. Mar Mamm Sci 12:402–413 Larsen RT, Flinders JT, Mitchell DL, Perkins ER, Whiting DG (2007) Chukar watering patterns and water site selection. Range Ecol Manag 60:559–565 Laundre JW, Hernandez L, Altendrof KB (2001) Wolves, elk, and bison: reestablishing the landscape of fear in Yellowstone National Park, USA. Can J Zool 79:1401–1409 Laundre JW, Hernandez L, Ripple WJ (2010) The landscape of fear: ecological implications of being afraid. Open Ecol J 3:1–7 Laurance WF (1991) Ecological correlates of extinction proneness in Australian tropical rain forest mammals. Conserv Biol 5:79–89 Laurie H, Perrier E (2009) A multifractal model for spatial variation in species richness. Ecol Complex 7:32–35 Le Boeuf BJ (1991) Pinniped mating systems on land, ice and in the water: emphasis on the Phocidae. In: Renouf D (ed) Behavior of Pinnipeds. Chapman and Hall, London, pp 45–65 200 References

Leaper R, Cooke J, Trathan P, Reid K, Rowntree VJ, Payne R (2006) Global climate drives southern right whale (Eubalaena australis ) population dynamics. Biol Lett 2:289–292 Leathwick JD, Rowe J, Richardson J, Hastie T (2005) Using multivariate adaptive regression splines to predict the distributions of New Zealand’s freshwater diadromous fish. Freshw Biol 50:2034–2052 Leggieri L, Guichón ML, Cassini MH (2011) Distribution of coypu (Myocastor coypus ) in Argentinean Pampas: landscape correlates and integration with smaller ecological scales. Ital J Zool 78:124–129 Lehman N, Wayne RK, Stewart BS (1993) Comparative levels of genetic variability in harbour seals and northern elephant as determined by genetic fingerprints. Symp Zool Soc Lond 66:49–60 Levin SA (1992) The problem of pattern and scale in ecology: The Robert H. MacArthur Award. Ecology 73:1943–1967 Levins R (1969) Some demographic and genetic consequences of environmental heterogeneity for biological control. Bull Entomol Soc Am 15:237–240 Levins R (1970) Extinction. Lect Math Life Sci 2:75–107 Lewontin RC (1974) The genetic basis of evolutionary change. Columbia University Press, New York Li H, Reynolds JF (1995) On definition and quantification of heterogeneity. Oikos 73:280–284 Li HB, Wu JG (2004) Use and misuse of landscape indices. Landsc Ecol 19:389–399 Lidicker WZ Jr (1975) The role of dispersal in the demography of small mammals. In: Golley FB, Petrusewicz K, Ryszkowski L (eds) Small mammals: their productivity and population dynamics. Cambridge University Press, New York, pp 103–128 Lima S, Dill LM (1990) Behavioural decisions made under the risk of predation: a review and prospectus. Can J Zool 68:619–640 Lima SL, Zollner PA (1996) Towards a behavioral ecology of ecological landscapes. Trends Ecol Evol 11:131–134 Lindenmayer DB, Possingham HP, Lacy RC, McCarthy MA, Pope ML (2003) How accurate are population models? Lessons from landscape-scale tests in a fragmented system. Ecol Lett 6:41–47 Lohmann KJ, Putman NF, Lohmann CMF (2008) Geomagnetic imprinting: a unifiying hypothesis of long-distance natal homing in salmon and sea turtles. Proc Natl Acad Sci U S A 105:19096–19101 Lomnicki A (1999) Individual-based models and the individual-based approach to individual- based model. Ecology 83:1817–1830 Low WA, Dudzinski ML, Muller WJ (1981) The influence of forage and climatic conditions on range community preference of shorthorn cattle in Central Australia. J Appl Ecol 18:11–26 Lucifora LO, García VB, Worm B (2011) Global diversity hotspots and conservation priorities for sharks. PLoS One 6:219356 Lucifora LO, García VB, Menni RC, Worm B (2012) Spatial patterns in the diversity of sharks, rays, and chimaeras (Chondrichthyes) in the Southwest Atlantic. Biodivers Conserv 21:407–419 Lundquist JE, Sommerfeld RA (2002) Use of Fourier transforms to define landscape scales of analysis for disturbances: a case study of thinned and unthinned forest stands. Landsc Ecol 17:445–454 Lundy MG, Buckley DJ, Boston ESM, Scott DD, Prodöhl PA, Marnell F, Teeling EC, Montgomery WI (2012) Behavioural context of multi-scale species distribution models assessed by radio- tracking. Basic Appl Ecol 13:188–195 MacArthur RH, Pianka ER (1966) On optimal use of patchy environment. Am Nat 100:603–609 MacCall AD (1990) Dynamic geography of marine fish populations. Washington Sea Grant Program. University of Washington Press, Washington Machpherson E (1989) Influence of geographical distribution, body size and diet on population density of benthic fishes of Namibia (South West Africa). Mar Ecol Prog Ser 50:295–299 References 201

MacNally R (2000) Regression and model-building in conservation biology, biogeography and ecology: the distinction between—and reconciliation of—‘predictive’ and ‘explanatory’ models. Biodivers Conserv 9:655–671 Malécot G (1968) The mathematics of heredity. Freeman, San Francisco Mandel JT, Bildsteinb KL, Bohrerc G, Winklera DW (2008) Movement ecology of migration in turkey vultures. Proc Natl Acad Sci U S A 105:19102–19107 Mandelbrot BB (1982) The fractal geometry of nature. WH Freeman, New York Manel S, Schwartz MK, Luikart G, Taberlet P (2003) Landscape genetics: combining landscape ecology and population genetics. Trends Ecol Evol 18:189–197 Manier MK, Arnold SJ (2005) Population genetic analysis identifies source–sink dynamics for two sympatric garter snake species ( Thamnophis elegans and Thamnophis sirtalis). Mol Ecol 14:3965–3976 Manly BB, McDonald L, Thomas DL, Erickson WP (1992) Resource selection by animals: statistical design and analysis for field studies, 2nd edn. Kluwer Academic, Dordrecht Martiny JB, Bohannan BJ, Brown JH, Colwell RK, Fuhrman JA, Green JL, Horner-Devine MC, Kane M, Krumins JA, Kuske CR, Brown JH, Colwell RK, Fuhrman JA, Gree JL, Horner- Devine MC, Kane M, Krumins JA, Kuske CR, Morin PJ, Naeem S, Ovreas L, Reysenbach A-L, Smith VH, Staley JT (2006) Microbial biogeography: putting microorganisms on the map. Nat Rev Microbiol 4:102–112 Mason C, Macdonald S (1986) Otters: ecology and conservation. Cambridge University Press, London, UK Matthews LR, Temple W (1979) Concurrent schedule assessment of food preference in cows. J Exp Anal Behav 32:245–254 Maynard Smith J (1982) Evolution and the theory of games. Cambridge University Press, Cambridge McCann TS (1981) Population structure and social organization of southern elephant seals, Mirounga leonina (L). Biol J Linn Soc 14:133–150 McDaniel KC, Tiedeman JA (1981) Sheep use on mountain winter range in New Mexico. J Range Manage 34:102–104 McGarigal K, Marks BJ (1995) FRAGSTATS: spatial pattern analysis program for quantifying landscape structure. U.S. Forest Service Pacific Northwest Research Station General Technical Report PNW-GTR-351 McIntyre NE, Wiens JA (1999) Interactions between landscape structure and animal behavior: the roles of heterogeneously distributed resources and food deprivation on movement patterns. Landsc Ecol 14:437–447 McIntyre NE, Wiens JA (2000) A novel use of the lacunarity index to discern landscape function. Landsc Ecol 15:313–321 McNab BK (1974) The energetics of endotherms. Ohio J Sci 74:370–380 McNab BK (2002) The physiological ecology of vertebrates: a view from energetics. Cornell University Press, New York McNaughton SJ, Wolf LL (1979) General ecology. Holt, Rinehart & Winston, New York Medri I, Mourão G (2005) Home range of giant anteaters ( Myrmecophaga tridactyla ) in the Pantanal wetland, Brazil. J Zool (Lond) 266:365–375 Meynecke JO (2004) Effects of global climate change on geographic distributions of vertebrates in North Queensland. Ecol Model 174:347–357 Midgley GF, Hannah L, Millar D, Rutherford MC, Powrie LW (2002) Assessing the vulnerability of species richness to anthropogenic climate change in a biodiversity hotspot. Glob Ecol Biogeogr 11:445–451 Milinski M, Parker GA (1991) Competition for resources. In: Davies NB, Krebs JR (eds) Behavioural ecology. Blackwell Scientific, Oxford, pp 137–168 Milne BT (1997) Applications of fractal geometry in wildlife biology. In: Bissonette JA (ed) Wildlife and landscape ecology. Effects of pattern and scale. Springer, New York, pp 32–69 202 References

Miller R (1983) Habitat use of feral horses and cattle in Wyoming’s Red Desert. J Range Manage 36:195 Mock DW, Lamey TC, Thompson SBA (1988) Falsiﬁability and the information centre hypothesis. Ornis Scand 19:231–248 Moline MA, Claustre H, Frazer TK, Schoﬁelds O, Vernet M (2004) Alterations of the food web along the Antarctic Peninsula in response to a regional warming trend. Glob Chang Biol 10:1973–1980 Moore LR, Ostrowski M, Scanlan DJ, Feren K, Sweetsir T (2005) Ecotypic variation in phosphorus-acquisition mechanisms within marine picocyanobacteria. Aquat Microb Ecol 39:257–269 Moreno S, Villafuerte R (1995) Traditional management of scrubland for the conservation of rabbits Oryctolagus cuniculus and their predators in Doñana National Park, Spain. Biol Conserv 73:81–85 Morin X, Lechowicz MJ (2008) Contemporary perspectives on the niche that can improve models of species range shifts under climate change. Biol Lett 4:573–576 Morris DW (1987) Ecological scale and habitat use. Ecology 68:362–369 Morris DW (1989) Habitat-dependent estimates of competitive interaction. Oikos 55:111–120 Morris DW (1990) Temporal variation, habitat selection and community structure. Oikos 59:303–312 Morris PA (1991a) Hedgehogs Erinaceus europaeus. In: Corbet GB, Hams S (eds) The handbook of Bntish mammals. Blackwell, Oxford, pp 37–43 Morris DW (1991b) On the evolutionary stability of dispersal to sink habitats. Am Nat 137:907–911 Morris DW (1992) Scales and costs of habitat selection in heterogeneous landscapes. Evol Ecol 6:412–432 Morris DW (1996) Temporal and spatial population dynamics among patches connected by habitat selection. Oikos 75:207–219 Morris DW (2002) Measuring the Allée effect: positive density dependence in small mammals. Ecology 83:14–20 Morris DW (2004) Some crucial consequences of adaptive habitat selection by predators and prey: apparent mutualisms, competitive ghosts, habitat abandonment, and spatial structure. Isr J Zool 50:207–232 Mourant AE, Kopéc AC, Domaniewsha-Sobczak K (1976) The distribution of the human blood groups and other polymorphisms. Oxford University Press, London Mysterud A (2000) The relationship between ecological segregation and sexual body-size dimorphism in large herbivores. Oecologia 124:40–54 Nardelli M, Scalzo PM, Quiroga MP, Ramírez MS, Cassini MH, Centrón D (2012) Class 1 integrons in environments with different degrees of urbanization. PLos One 7:39223 Nathan R (2008) An emerging movement ecology paradigm. Proc Natl Acad Sci U S A 105:19050–19051 Nathan R, Getz WM, Revilla E, Holyoak M, Kadmon R, Saltz D, Smouse PE (2008) A movement ecology paradigm for unifying organismal movement research. Proc Natl Acad Sci U S A 105:19052–19059 Nielsen D, Page RE, Crosland MWJ (1994) Clinal variation and selection of allozymes in honey bee populations. Experientia 50:867–871 Nolet BA, Klaassen M (2009) Retrodicting patch use by foraging swans in a heterogeneous environment using a set of functional responses. Oikos 118:431–439 Nonacs P (2001) State dependent behavior and the Marginal Value Theorem. Behav Ecol 12:71–83 Novoplansky A (2009) Picking battles wisely: plant behavior under competition. Plant Cell Environ 32:726–741 O’Neill RV, King AW (1998) Homage to St. Michael; or, why are there so many books on scale? In: Peterson DL, Parker VT (eds) Ecological scale. Theory and applications. Columbia University Press, New York, pp 3–15 References 203

O’Neill RV, Johnson AR, King AW (1989) A hierarchical framework for the analysis of scale. Landsc Ecol 3:193–205 Olden JD, Schooley RL, Monroe JB, Poff NL (2004) Context-dependent perceptual ranges and their relevance to animal movements in landscapes. J Anim Ecol 73:1190–1194 Omernik JM (1987) Ecoregions of the conterminous United States Map (scale 1:7,500,000). Ann Assoc Am Geogr 77:118–125 Orians GH (1969) On the evolution of mating systems in birds and mammals. Am Nat 103:589–603 Osborne PE, Suarez-Seoane S, Alonso JC (2007) Behavioural mechanisms that undermine species envelope models: the causes of patchiness in the distribution of great bustards Otis tarda in Spain. Ecography 30:819–828 Oswald SA, Huntley B, Collingham Deborah YC, Russell JF, Anderson BJ, Arnold JM (2011) Physiological effects of climate on distributions of endothermic species. J Biogeogr 38:430–438 Ovaskainen O, Smith AD, Osborne JL, Reynolds RD, Carreck NL, Martin AP, Niitepõld K, Hanski I (2008) Tracking butterfly movements with harmonic radar reveals an effect of population age on movement distance. Proc Natl Acad Sci U S A 105:19090–19095 Oyugi JO, Brown JS (2003) Giving-up densities and habitat preference of European Starlings and American Robins. Condor 105:130–135 Parker GA (1970) The reproductive behaviour and the nature of sexual selection in Scatophaga stercoraria L. (Diptera: Scatophagidae) IV. Epigamic recognition and competition between males for the possession of females. Behaviour 37:113–139 Partridge SR, Tsafnat G, Coiera E, Iredell JR (2009) Gene cassettes and cassette arrays in mobile resistance integrons. FEMS Microbiol Rev 33:757–784 Patalas K (1971) Crustacean plankton in forty-five lakes in the Experimental Lakes Area, north- western Ontario. J Fish Res Board Can 28:231–244 Patriquin KJ, Barclay RMR (2003) Foraging by bats in cleared, thinned, and unharvested boreal forest. J Appl Ecol 40:646–657 Patterson TA, Basson M, Bravington MV, Gunn JS (2009) Classifying movement behaviour in relation to environmental conditions using hidden Markov models. J Anim Ecol 78:1113–1123 Peakall R, Ruibal M, Lindenmayer DB (2003) Spatial autocorrelation analysis offers new insights into gene flow in the Australian bush rat, Rattus fuscipes . Evolution 57:1182–1195 Pearce JL, Boyce MS (2006) Modelling distribution and abundance with presence-only data. J Appl Ecol 43:405–412 Pearson DL, Cassola F (1992) World-wide species richness patterns of tiger beetles (Coleoptera: Cincindelidae): indicator taxon for biodiversity and conservation studies. Conserv Biol 6:376–391 Pearson RG, Dawson TE (2003) Modelling species distributions in Britain: a hierarchical integration of climate and land-cover data. Ecography 27:285–298 Penning PD, Parson AK, Newman JA, Orr RJ, Harvey A (1993) The effects of group size on grazing time in sheep. Appl Anim Behav Sci 27:101–109 Percival DB, Wang M, Overland JE (2004) An introduction to wavelet analysis with applications to vegetation time series. Community Ecol 5:19–30 Peterson AT (2006) Uses and requirements of ecological niche models and related distributional models. Biodivers Inform 3:59–72 Peterson AT, Ortega Huerta MA, Bartley J, Sanchez-Cordero V, Soberon J, Buddemeier RH, Stockwell DRB (2002) Future projections for Mexican faunas under global climate change scenarios. Nature 416:626–629 Petty JT, Grossman GD (2010) Giving-up densities and ideal pre-emptive patch use in a predatory benthic stream fish. Freshw Biol 55:780–793 Phillips SJ, Anderson RP, Schapire RE (2006) Maximum entropy modelling of species geographic distributions. Ecol Model 190:231–259 Pianka ER (1974) Evolutionary ecology. Harper & Row, New York 204 References

Piglucci M, Barbujani G (1991) Geographical patterns of gene frequencies in Italian populations of Ornithogalum montanum (Liliaceae). Genet Res 58:95–104 Pimm SL, Rosenzweig ML, Mitchell W (1985) Competition and food selection: field test of a theory. Ecology 66:798–807 Platt T, Denman KL (1975) Spectral analysis in ecology. Annu Rev Ecol Syst 6:189–210 Plotnick RE, Gardner RH, O’NeilI RV (1993) Lacunarity indices as measures of landscape texture. Landsc Ecol 8:201–211 Pogson GH, Taggart CT, Mesa KA, Boutilier RG (2001) Isolation by distance in the Atlantic cod, Gadus morhua , at large and small geographic scales. Evolution 55:131–146 Porter WP, Kearney M (2009) Size, shape, and the thermal niche of endotherms. PNAS 106: 19666–19672 Powell RA (2000) Animal home ranges and territories and home range estimators. In: Boitani L, Fuller TK (eds) Research techniques in animal ecology: controversies and consequences. Columbia University Press, New York, pp 65–110 Pratt RM, Putman RJ, Ekins JR, Edwards PJ (1986) Use of habitat by free ranging cattle and ponies in the new forest, Southern England. J Appl Ecol 23:539–557 Pulliam HR (1988) Sources, sinks, and population regulation. Am Nat 132:652–661 Pulliam HR (2000) On the relationship between niche and distribution. Ecol Lett 3:349–361 Purves DW, Dushoff J (2005) Directed seed dispersal and metapopulation response to habitat loss and disturbance: application to Eichhornia paniculata . J Ecol 98:658–669 Raes N, Ter Steege H (2008) A null-model for significance: testing of presence-only species distribution models. Ecography 30:727–736 Railsback SF, Stauffer HB, Harvey BC (2003) What can habitat preference models tell us? Tests using a virtual trout population. Ecology 13:1580–1594 Ramos-Fernández G, Mateos JL, Miramontes O, Larralde H, Cocho G, Ayala Orozco B (2004) Lévy walk patterns in the foraging movements of spider monkeys ( Ateles geoffroyi ). Behav Ecol Sociobiol 55:223–230 Real R, Barbosa AM, Vargas JM (2006) Obtaining environmental favourability functions from logistic regression. Environ Ecol Stat 13:237–245 Reed JM, Mills LS, Dunning JB Jr, Menges ES, McKelvey KS, Frye R, Beissinger SR, Anstett MC, Miller P (2002) Emerging issues in population viability analysis. Conserv Biol 16:7–19 Rees VH, Hutson GD (1983) The behaviour of free-ranging cattle on an alpine range in Australia. J Range Manage 36:740–743 Regan HM, Auld TD, Keith DA, Burgman MA (2003) The effects of fire and predators on the long-term persistence of an endangered shrub, Grevillea caleyi . Biol Conserv 109:73–83 Renaudeau d’Arc N, Cassini MH, Vilá BL (2000) Habitat use of vicuñas in Laguna Blanca Reserve (Catamarca, Argentina). J Arid Environ 46:107–115 Revilla E, Wiegand T (2008) Individual movement behavior, matrix heterogeneity and the dynamics of spatially structured populations. Proc Natl Acad Sci U S A 105:19120–19125 Revilla E, Wiegand T, Palomares F, Ferreras P, Delibes M (2004) Animal dispersal in a heterogeneous matrix: from individual behavior to population-level parameters. Am Nat 164:130–153 Richner H, Heeb P (1995) Is the information center theory a flop? Adv Study Behav 24:1–45 Ricketts TH (2001) The matrix matters: effective isolation in fragmented landscapes. Am Nat 158:87–99 Ricklefs RE (1979) Ecology. Chiron Press, Oregon Roach JL, Stapp P, Van Horne B, Antolin MF (2001) Genetic structure of a metapopulation of black-tailed prairie dogs. J Mammal 82:946–959 Roath LR, Krueger WC (1982) Cattle grazing and behavior on a forested range. J Range Manage 35:322–338 Rodríguez D, Bastida R (1998) Four hundred years in the history of pinniped colonies around Mar del Plata, Argentina. Aquat Conserv 8:721–735 Roobitaille JF, Laurance S (2002) Otter, Lutra lutra, occurrence in Europe and in France in relation to landscape characteristics. Anim Conserv 5:337–344 References 205

Rood JP (1972) Ecological and behavioral comparisons of three genera of Argentina cavies. Anim Behav Monogr 5:1–83 Rosenberg MS (2001) PASSAGE: pattern analysis, spatial statistics, and geographic exegesis, 1.1 edition. ASU Department of Biology, Tempe, AZ Rosenzweig ML (1974) On the optimal above ground activity of bannertail kangaroo rats. J Mammal 55:193–199 Rosenzweig ML (1981) A theory of habitat selection. Ecology 62:327–335 Rosewarne CP, Pettigrove V, Stokes HW, Parsons YM (2010) Class 1 integrons in benthic bacterial communities: abundance, association with Tn402-like transposition modules and evidence for coselection with heavy-metal resistance. FEMS Microbiol Ecol 72:35–46 Rotenberry JT (1986) Habitat relationships of shrubsteppe birds: even “good” models cannot predict the future. In: Verner J, Morrison ML, Ralph CJ (eds) Wildlife 2000: modeling habitat relationships of terrestrial vertebrates. University of Wisconsin Press, Madison, pp 217–222 Royama T (1971) Evolutionary significance of predators’ response to local differences in prey density: a theoretical study. In: den Boer PJ, Gradwell GR (eds) Dynamics of populations. Centre for Agricultural Publishing and Documentation, Wageningen, pp 344–357 Ruckstuhl KE, Neuhaus P (2000) Sexual segregation in ungulates: a new approach. Behaviour 137:361–377 Runge JP, Runge MC, Nichols JD (2006) The role of local populations within a landscape context: defining and classifying sources and sinks. Am Nat 167:925–938 Saltz D, Rowen M, Rubenstein DI (2000) The impact of space use patterns of reintroduced Asiatic wild ass on effective population size. Conserv Biol 14:1852–1862 Schaefer JA (2006) Towards maturation of the population concept. Oikos 112:236–240 Schneider DC (2001) The rise of the concept of scale in Ecology. Bioscience 51:545–553 Schooley RL, Wiens JA (2003) Finding habitat patches and directional connectivity. Oikos 102:559–570 Schröder B, Seppelt R (2006) Analysis of pattern-process interactions based on landscape models overview, general concepts, methodological issues. Ecol Model 199:505–516 Scifres CJ, Scifres JR, Kothmann MM (1983) Differential grazing use of herbicide-treated areas by cattle. J Range Manage 36:65–67 Segurado P, Araújo MB (2004) An evaluation of methods for modelling species distributions. J Biogeogr 31:1555–1568 Senft RL (1989) Hierarchical foraging models: effects of stocking and landscape composition on simulated resource use by cattle. Ecol Model 46:283–303 Senft RL, Rittenhouse LR, Woodmansee RG (1983) The use of regression models to predic spatial pattterns of cattle behavior. J Range Manage 36:553–557 Senft RL, Rittenhouse LR, Woodmansee RG (1985a) Factors influencing patterns of grazing behavior on shortgrass steppe. J Range Manage 38:81–87 Senft RL, Rittenhouse LR, Woodmansee RG (1985b) Factors infl uencing selection of resting sites by cattle on shortgrass steppe. J Range Manage 38:295–299 Sergio F, Marchesi L, Pedrini P (2004) Integrating individual habitat choices and regional distribution of a biodiversity indicator and top predator. J Biogeogr 31:619–628 Shaw RB, Dodd JD (1979) Cattle activities and preferentes following strip application of herbicide. J Range Manage 32:449–452 Shaw JH, Machado Neto J, Carter TS (1987) Behavior of free-living giant anteaters (Myrmecophaga tridactyla ). Biotropica 19:255–259 Sibly RM, Smith RH (1985) Behavioural ecology: ecological consequences of adaptive behaviour. Blackwell Scientific, Oxford Simpson CD, Gray GG (1983) Topographic and habitat use by sympatric Barbary sheep and mule deer in Palo Duro Canyon, Texas. J Range Manage 36:190–194 Singer RS, Ward MP, Maldonado G (2006) Can landscape ecology untangle the complexity of antibiotic resistance? Nat Rev Microbiol 4:943–952 Skolvin JM, Edgerton PJ, McConnell BR (1983) Elk use of winter range as affected by cattle grazing, fertilizing and burning in Southeastern Washington. J Range Manage 36:184–189 206 References

Slade AJ, Hutchings MJ (1987) The effects of nutrient availability on foraging in the clonal herb Glechoma hederacea . J Ecol 75:95–112 Slobodkin BL (1953) On social single species populations. Ecology 34:430–434 Smith JW, Benkman, Coffey K (1999) The use and misuse of public information by foraging red crossbills. Behav Ecol 10:54–62 Smouse PE, Focardi S, Moorcroft PR, Kie JG, Forester JD, Morales JM (2010) Stochastic modelling of animal movement. Philos Trans R Soc B 365:2201–2211 Soberón JM (1986) The relationship between use and suitability of resources and its consequences to insect population size. Am Nat 127:338–357 Soberón J (2007) Grinnellian and Eltonian niches and geographic distributions of species. Ecol Lett 10:1115–1123 Soberón J, Peterson AT (2005) Interpretation of models of fundamental ecological niches and species’ distributional areas. Biodivers Inform 2:1–10 Sponheimer M, Robinson T, Roeder B, Hammer J, Ayliffe L, Passey B, Cerling T, Dearing D, Ehleringer J (2003) Digestion and passage rates of grass hays by llamas, alpacas, goats, rabbits, and horses. Small Rumin Res 48:149–154 Staddon JER (1983) Adaptive behaviour and learning. University of Cambridge Press, Cambridge Stephens DW, Krebs JR (1986) Foraging theory. Princeton University Press, Princeton, NJ Stephens PA, Sutherland WJ, Freckleton RP (1999) What is the Allée effect? Oikos 85:185–190 Stephens PA, Buskirk SW, Hayward GD, Martinez del Rio C (2005) Information theory and hypothesis testing: a call for pluralism. J Appl Ecol 42:4–12 Stephens DW, Brown JS, Ydenberg (2007) Foraging: behavior and ecology. The University of Chicago Press, Chicago Steuter AA, Wright HA (1980) White-tailed deer densities and brush cover on the Rio Grande plain 1980. J Range Manage 33:328–331 Stokes HW, Gillings MR (2011) Gene ﬂow, mobile genetic elements and the recruitment of antibiotic resistance genes into Gram-negative pathogens. FEMS Microbiol Rev 35:790–819 Stokes HW, Nesbø CL, Holley M, Bahl MI, Gillings MR, Boucher Y (2006) Class 1 integrons potentially predating the association with tn402-like transposition genes are present in a sediment microbial community. J Bacteriol 188:5722–5730 Storfer A, Murphy MA, Evans JS, Goldberg CS, Robinson S, Spear SF, Dezzani R, Delmelle E, Vierling L, Waits LP (2007) Putting the ‘landscape’ in landscape genetics. Heredity 98:128–142 Storfer A, Murphy MA, Spear SF, Holderegger R, Waits LP (2010) Landscape genetics: where are we now? Mol Ecol 19:3496–3514 Storz JF (1999) Genetic consequences of mammalian social structure. J Mammal 80:553–569 Sutherland WJ (1983) Aggregation and the ‘ideal free’ distribution. J Anim Ecol 52:821–828 Sutherland WJ (1996) From individual behaviour to population ecology. Oxford University Press, Oxford Sutherland WJ, Anderson CW (1993) Predicting the distribution of individuals and the consequences of habitat loss: the role of prey depletion. J Theor Biol 160:223–230 Sutherland WJ, Parker GA (1985) Distributions of unequal competitors. In: Sibly RM, Smith RH (eds) Behavioural ecology: ecological consequences of adaptive behaviour. Blackwell, Oxford, pp 225–274 Svenning JC, Skov F (2004) Limited ﬁlling of the potential range in European tree species. Ecol Lett 7:565–573 Svensson BW (1992) Changes in occupancy, niche breadth and abundance of three Gyrinus species as their respective range limits are approached. Oikos 63:147–156 Tanner GW, Sandoval LD, Martin FG (1984) Cattle behavior on a South Florida Range. J Range Manage 37:248–251 Taylor PD, Fahrig L, Henein K, Merriam G (1993) Connectivity is a vital element of landscape structure. Oikos 68:571–573 References 207

Telleria JL, Santos T (1993) Distributional patterns of insectivorous passerines in the Iberian forests: does abundance decrease near the border? J Biogeogr 20:235–240 Templeton JJ, Giraldeau LA (1996) Vicarious sampling: the use of personal and public information by starlings foraging in a simply patchy environment. Behav Ecol Sociobiol 38:105–114 Terrell JW, Cade BS, Carpenter J, Thompson JM (1996) Modelling stream ﬁsh habitat limitations from wedged-shaped patterns of variation in standing stock. Trans Am Fish Soc 125:104–117 Thomas CD (1994) Extinction, colonization, and metapopulations: environmental tracking by rare species. Conserv Biol 8:373–378 Thomas CD, Kunin WE (1999) The spatial structure of populations. J Anim Ecol 68:647–657 Thomas CD, Cameron A, Green RE, Bakkenes M, Beaumont LJ, Collingham YC, Erasmus BFN, de Siqueira MF, Grainger A, Hannah L, Hughes L, Huntley B, Van Jaarsveld AS, Midgley GF, Miles L, Ortega Huerta MA, Peterson AT, Phillips OL, Williams SE (2004) Extinction risk from climate change. Nature 427:145–148 Thomson JD, Weiblen G, Thomson BA, Alfaro S, Legendre P (1996) Untangling multiple factors in spatial distributions: lilies, gophers, and rocks. Ecology 77:1698–1715 Thuiller W (2003) BIOMOD optimizing predictions of species distributions and projecting potential future shifts under global change. Glob Chang Biol 9:1353–1362 Tinbergen N (1963) On aims and methods in ethology. Z Tierpsychol 20:410–433 Tolman EC (1948) Cognitive maps in rats and men. Psychol Rev 55:189–208 Tosh CA, McIntyre T, Bester MN, Frickenhaus S, Plötz J (2008) Adult male southern elephant seals from King George Island utilize the Weddell Sea. Antarct Sci 21:113–121 Touchon M, Hoede C, Tenaillon O, Barbe V, Baeriswyl S, Bidet P, Bingen E, Bonacorsi S, Bouchier C, Bouvet O, Calteau A, Chiapello H, Clermont O, Cruveiller S, Danchin A, Diard M, Dossat C, Karoui ME, Frapy E, Garry L, Ghigo JM, Gilles AM, Johnson J, Le Bouguénec C, Lescat M, Mangenot S, Martinez Jéhanne V, Matic I, Nassif X, Oztas S, Petit MA, Pichon C, Rouy Z, Ruf CS, Schneider D, Tourret J, Vacherie B, Vallenet D, Médigue C, Rocha EP, Denamur E (2009) Organised genome dynamics in the Escherichia coli species results in highly diverse adaptive paths. PLoS Genet 5:1000344 Trebilco R, Halpern BS, Mills Flemming J, Field C, Blanchard W, Worm B (2011) Mapping species richness and human impact drivers to inform global pelagic conservation prioritization. Biol Conserv 144:1758–1766 Tregenza T (1995) Building on the ideal free distribution. Adv Ecol Res 26:253–302 Tucker RD, Garner GW (1983) Habitat selection and vegetational characteristics of antelope fawn bed sites in west Texas. J Range Manage 36:110–113 Túnez JI, Centrón D, Cappozzo HL, Cassini MH (2007) Geographic distribution and diversity of mitochondrial DNA haplotypes in South American sea lions ( Otaria ﬂavescens) and fur seals ( Arctocephalus australis ). Mamm Biol 72:193–203 Túnez JI, Cappozzo HL, Cassini MH (2008) Natural and anthropogenic factors associated with the distribution of South American sea lion along the Atlantic coast. Hydrobiologia 598:191–202 Túnez JI, Guichón ML, Centrón D, Henderson A, Callahan C, Cassini MH (2009) Kinship and social organisation in coypus of Argentinean pampas. Mol Ecol 18:147–155 Turchin P (1998) Quantitative analysis of movement. Sinauer, Sunderland Turner M (2005) Landscape ecology: what is the state of the science. Annu Rev Ecol Evol Syst 36:319–344 Tyre AJ, Tenhumberg B, Bull CM (2006) Identifying landscape scale patterns from individual scale processes. Ecol Model 199:442–450 Uchmánski J, Grimm V (1996) Individual-based modelling in ecology: what makes the difference? Trends Ecol Syst 11:437–441 Urbano F, Cagnacci F, Calenge C, Dettki H, Cameron A, Neteler M (2010) Wildlife tracking data management: a new vision. Philos Trans R Soc B 365:2177–2185 US Fish and Wildlife Service (1981) Standards for the development of Suitability Index Models. Ecology Service Manual 103, Division of Ecological Services. Washington, DC 208 References

Valenzuela L, Sironi M, Rowntree V, Seger J (2009) Isotopic and genetic evidence for culturally inherited site fidelity to feeding grounds in southern right whales ( Eubalaena australis ). Mol Ecol 18:782–791 Vallenas A (1991) Características anatomofisiológicas. In: Fernández Baca S (ed) Avances y per- spectivas del conocimiento de camélidos sudamericanos. Oficina Regional de la FAO para América Latina y el Caribe, Santiago de Chile Valone TJ (1989) Group foraging, public information, and patch estimation. Oikos 56:357–363 Van Ballenberghe V, Ballard WB (1994) Limitations and regulation of moose populations: the role of predation. Can J Zool 72:2071–2077 Van Horne B (1983) Density as a misleading indicator of habitat quality. J Wildl Manage 47:893–901 Van Rossum F, Vekemans X, Meerts P, Gratia E, Lefebvre C (1997) Allozyme variation in relation to ecotypic differentiation and population size in marginal populations of Silene nutans . Heredity 78:552–560 Vaz S, Martin CS, Eastwood PD, Ernande B, Carpentier A, Meaden GJ, Coppin F (2008) Modelling species distributions using regression quantiles. J Appl Ecol 45:204–217 Velando A, Munilla I (2011) Disturbance to a foraging seabird by sea-based tourism: implications for reserve management in marine protected areas. Biol Conserv 144:1167–1174 Verbeylen G, De Bruyn L, Adriaensen F, Matthysen E (2003) Does matrix resistance influence red squirrel ( Sciurus vulgaris L. 1758) distribution in an urban landscape? Landsc Ecol 18:791–805 Vilá BL, Cassini MH (1990) Agresividad entre hembras y separación madre-cría en el Lobo Marino del Sur. Rev Chil Hist Nat 63:169–176 Viswanathan GM, Buldyrev SV, Havlin S, da Luz MGE, Raposo EP, Stanley HE (1999) Optimizing the success of random searches. Nature 401:911–914 Wang IJ (2010) Recognizing the temporal distinctions between landscape genetics and phylogeography. Mol Ecol 19:2605–2608 Ward P, Zahavi A (1973) The importance of certain assemblages of birds as ‘information centres’ for food finding. Ibis 115:517–534 Weiss GH, Kimura M (1965) A mathematical analysis of the stepping stone model of genetic correlation. J Appl Probab 2:129–149 Welch D (1984) Studies in the grazing of heather moorland in north-east Scotland. J Appl Ecol 21:179–195 West SA, Griffin AS, Gardner A, Diggle SP (2006) Social evolution theory for microorganisms. Nat Rev 4:597–609 Wiens JA (1976) Population responses to patchy environments. Annu Rev Ecol Syst 7:81–120 Wiens JA (1989) Spatial scaling in ecology. Funct Ecol 3:385–397 Wiens JJ (2009) Paleontology, genomics, and combined-data phylogenetics: can molecular data improve phylogeny estimation for fossil taxa? Syst Biol 58:87–99 Willems EP, Hill RA (2009) Predator-specific landscapes of fear and resource distribution: effects on spatial range use. Ecology 90:546–555 Williams GC, Koehn RK, Thorsteinsson V (1984) Icelandic eels: evidence for a single species of Anguilla in the North Atlantic. Copeia 1984:221–223 Williams SE, Bolitho EE, Fox S (2003) Climate change in Australian tropical rainforests: an impending environmental catastrophe. Proc R Soc Lond B 270:1887–1892 Woodroffe R, Ginsberg JR (1998) Edge effects and the extinction of populations inside protected areas. Science 280:2126–2128 Woodward SL, Ohmart RD (1976) Habitat use and fecal analysis of feral burros (Equus asinus ), Chemehuevi Mountains, California, 1974. J Range Manage 29:482–485 Worm B, Duffy JE (2003) Biodiversity, productivity and stability in real food webs. Trends Ecol Evol 18:628–632 Wright S (1931) Evolution in Mendelian populations. Genetics 16:97–159 Wright S (1938) Size of population and breeding structure in relation to evolution. Science 87:430–431 References 209

Wright S (1943a) An analysis of local variability of ﬂower color in Linanthus parryae . Genetics 28:139–156 Wright S (1943b) Isolation by distance. Genetics 28:114–138 Wroot AJ (1984) Feeding ecology of the European hedgehog, Erinaceus europaeus L. Ph D thesis, University of London, London Zanette L (2000) Fragment size and the demography of an area-sensitive songbird. J Anim Ecol 69:458–470 Zhang HP, Be’er A, Florin EL, Swinney HL (2010) Collective motion and density ﬂuctuations in bacterial colonies. Proc Natl Acad Sci U S A 107:13626–13630 Zollner PA (2000) Comparing the landscape level perceptual abilities of forest sciurids in fragmented agricultural landscapes. Landsc Ecol 15:523–533 Zollner PA, Lima SL (1997) Landscape-level perceptual abilities in white-footed mice perceptual range and the detection of forested habitat. Oikos 80:51–60 Index

A B Actinosphaerium nucleofi lum , 82–83 Balanophyllia elegans , 134–135 Adult white butter fl ies , 19 Bat , 113, 149 Aggregation distribution , 44–45 Black-chinned hummingbirds , 121 conservation biology Black-tailed prairie dog , 136 ideal-free distribution approach , Blue-throated hummingbirds , 121 153–154 Bofedales , 172 individual-based models , 151–153 Brangus cows , 164 site suitability models , 149–152 Butter fl ies , 89, 161, 181 habitat-matching rule , 54–56 ideal free distribution models Fretwell and Lucas’ model , 49–52 C habitat selection , 48–49 California sea lions , 99 interference constant , 52 Carabus violaceus , 135 intraspeci fi c competition , 48–49 Cavia aperea , 60 resource depletion , 54 Cercopithecus aethiops , 114 unequal competitors , 52–53 Cervus canadensis , 63 individual-based models Chaetophractus vellerosus , 33, 34 bottom-up models , 46 Chihuahuan Desert , 164 characteristics , 46 Class 1 integron genes host–parasite system , 47 antibiotic resistance spatially explicit model , 45 functional components , 139 vicuñas (Vicugna vicugna) , 46 gram-negative bacteria , 138 Alces alces , 28–29 Salmonella genomic island 3 , 139 Alpacas , 172 transposon Tn402 , 139 Anguilla rostrata , 130 global scale , 140–141 Animal production. See Livestock distribution habitat scale , 142 Apis mellifera , 131 landscape scale , 140–142 Aponomma hydrosauri , 47 Conservation biology Archilochus alexandri , 121 aggregation distributions Armadillos , 32–34, 36 ideal-free distribution approach , Asiatic wild ass , 148 153–154 Ateles geoffroyi , 24 individual-based models , 151–153 Atlantic cod , 133 site suitability models , 149–152

M.H. Cassini, Distribution Ecology: From Individual Habitat Use 211 to Species Biogeographical Range, DOI 10.1007/978-1-4614-6415-0, © Springer Science+Business Media New York 2013 212 Index

Conservation biology (cont.) intra-and interspeci ﬁ c competition , 3 ensemble distributions levels , 8–9 chondirichthyan , 161–162 livestock distribution (see Livestock insects , 161 distribution) macroecological approach , 160 meta-population theory , 4 tiger beetles , 161 migration , 11–13 vertebraes , 160 movement and dispersal , environmental challenges 11–13 biodiversity and protected areas , 146 phylogenetic constraints , 186 climate change , 147 population distribution ( see Population ecosystem preservation , 146–147 distribution) endangered species , 145–146 population dynamics models , 3 individual and gene distribution scales Asiatic wild ass (Equus hemionus) , 148 cartographic scale , 4 bat foraging behaviour , 149 ecoregion , 7 captive breeding and grain and extent , 5 reintroductions, 148 habitat , 6 disc-winged bat (Thyroptera landscapes , 6–7 tricolor) , 149 patch , 6 habitat requirements , 147 resource , 6 roosting habitat , 149 society ( see Society) southern right whales (Eubalaena spatial revolution , 4 australis) , 147–148 species assemblages (see Species specialist species , 148 assemblages) T. tricolor , 149 species distribution models (see Species landscape ecology and pattern-based distribution models) models , 156–158 subpopulations distribution metapopulations and source-sinks , 156 ( see Subpopulations distribution) species distribution models bioclimatic envelope , 160 climate change , 160 E habitat requirements , 158 Ecological scales Patagonian otters (Lontra cartographic scale , 4 provocax) , 159 ecoregion , 7 pests/invasive species , 159 grain and extent , 5 spatial prediction , 158 habitat , 6 Coypu , 150–152 landscapes , 6–7 Cynomys ludovicianus , 136 organisation levels , 181–182 patch , 6 resource , 6 D Eichhornia paniculata , 80–81 Diadromous ﬁ sh , 125, 126 Elephant seals , 106 Disc-winged bat , 149 Elk , 28–29, 63 Distribution ecology , 187 Ensemble distributions aggregation distribution (see Aggregation chondirichthyan , 161–162 distribution) insects , 161 conservation biology (see Conservation macroecological approach , 160 biology) tiger beetles , 161 gene distribution (see Gene distribution) vertebraes , 160 habitat selection , 9–11 Equus hemionus , 148 holistic and reductionist approaches , 187 Erinaceus europaeus , 22, 23 individual-based models , 186 Esporal , 45 individuals distribution ( see Individuals Eubalaena australis , 147 distribution) Eugenes fulgens , 121 Index 213

Eukaryotes genetic discontinuity , 135–136 central–marginal model , 137 isolation by distance , 133 Clinal variation , 130–131 metapopulations , 136–137 ecotypes , 131–133 random patterns , 130 genetic discontinuity , 135–136 source–sink model , 137 isolation by distance , 133 stepping-stone model , 133–135 metapopulations , 136–137 Generalised linear model , 28, 105–106, random patterns , 130 124, 162 source–sink model , 137 Generalised matching law stepping-stone model , 133–135 aggregations distribution , 183 European shag , 153 animal production , 183 ecological scales , 185 habitat quality and river basin , 184 F ideal-free models , 185 Flightless ground beetle , 135 site suitability and species distribution Florida panther , 157 models , 183 Foraging behaviour , 30–31 source–sink model , 184 bat , 149 Geomagnetic imprinting , 20 livestock distribution Giant anteaters , 27 Brangus cows , 164 Glechoma hederacea , 39 cattle and sheep , 164 Global positional system (GPS) , 13 cognition abilities , 164 Great bustards , 113–114 dairy cows , 164 Grevillea caleyi , 151, 153 environmental conditions , 164 Grey squirrels , 91 herbivores , 163 Groups horses , 165 Allée effects subpopulations distribution , 84–85 habitat quality , 69 Foraging theory immigration , 70–71 foraging behaviour , 30–31 local densities , 70 heterogeneous habitats, plants , 38–39 local population density , 69 marginal value theorem , 31–34 population costs and bene ﬁ ts , 68 patches population growth rate and density , 68 prey selection , 37–38 sink populations , 70 selection of , 35–37 avoidance, effect of , 60 types , 31 confusion effect , 62 Forest gap models , 111 defence effect , 62 Fruit eating spider monkeys , 24 dilution effect , 61 encounter effect , 60 foraging , 62–63 G local environment , 63 Gadus morhua , 133 sel ﬁ sh herd model , 61 Garter snake , 137 vigilance effect , 60–61 Gene distribution bacterial communities class 1 integron genes (see Class 1 H integron genes) Habitat-matching rule , 54–56, 94 lateral genetic transfer , 138 Habitat selection , 9–11 biological and environmental features , aggregation distribution , 48–49 128–129 individuals distribution , 28–29 eukaryotes landscape scale , 29 central–marginal model , 137 species assemblages , 119 clines , 130–131 Habitat selection hypothesis , 123 ecotypes , 131–133 Harems , 65 214 Index

Hedgehogs , 22, 23 moose , 28, 30 Honey bee , 131 random sampling design , 28 Hunting , 28, 150–151 selection ratios , 29 simple sample comparisons , 28 Isodar theory , 84, 85 I Isoleg theory , 118–119, 172 Iberian lynxes , 86 Ideal free distribution model Fretwell and Lucas’ model , 49–52 J group formation Juniperus occidentalis , 171 Allée effects , 67–71 avoidance, effect of , 60 confusion effect , 62 L defence effect , 62 Lama glama , 172 dilution effect , 61 Lama pacos , 172 encounter effect , 60 Landscape genetics. See Gene distribution foraging , 62–63 Landscape scale , 180–181 local environment , 63 class 1 integron genes , 140–142 sel ﬁ sh herd model , 61 habitat selection , 29 vigilance effect , 60–61 moose home ranges , 29–30 habitat selection , 48–49 types , 6–7 interference constant , 52 Lapornis clemenciae , 121 intraspeci ﬁ c competition , 48–49 Lasionycteris noctivagans , 149 mating systems Lek , 65 polygyny-threshold model , 72–73 Lincoln National Forest (LNF) , 169 sexual harassment , 73–75 Livestock distribution resource depletion , 54 foraging behaviour territories , 71–72 Brangus cows , 164 unequal competitors , 52–53 cattle and sheep , 164 Individual-based models , 178 cognition abilities , 164 bottom-up models , 46 dairy cows , 164 characteristics , 46 environmental conditions , 164 host–parasite system , 47 herbivores , 163 spatially explicit model , 45 horses , 165 vicuñas (Vicugna vicugna) , 46 landscape predictors , 166–167 Individuals distribution quantitative analyses , 167–168 biased matching , 42 species assemblage distribution foraging theory (see Foraging theory) habitat preference , 174 home ranges , 25–27 Isoleg theory , 172 movement ecology native species , 172 adaptive approach , 22 Puna/Altiplano environment , 172–173 animal behaviour research , 18 vicuñas , 173–174 cognitive aspects , 20–22 species distribution models external factors , 17 cattle distribution , 169 neuro-ethology , 18–20 Juniperus occidentalis , 171 sessile species , 17 paddocks , 168–169 statistical tools , 24–25 supplemental food blocks , 170, 171 refuges , 39–41 vegetation communities , 170 site suitability models llamas , 172 case–control sampling , 28 LNF. See Lincoln National Forest (LNF) generalised linear model , 28 Lolium perenne , 164 habitat selection , 28–29 Lontra provocax , 159 hunting , 28 Lynx pardinus , 86 Index 215

M subpopulations distribution Marginal value theorem , 31–34 ( see Subpopulations distribution) Matching rule , 182–183 Puma concolor coryi , 157 Mirounga leonina , 106 Molar model , 179–180 Molecular model , 179–180 R Moose/elk , 28–29 Random forager , 35 Movement ecology RCP. See Relative community preference adaptive approach , 22–23 (RCP) animal behaviour research , 18 Red-cockaded woodpeckers , 90 cognitive aspects , 20–22 Relative community preference (RCP) , external factors , 17 167–168 neuro-ethology Rivoli’s hummingbirds , 121 environmental cues , 19 Robins , 40, 41 geomagnetic imprinting , 20 insect cognition , 19 male moths , 18 S sessile species , 17 Salmonella genomic island 1 (SGI1) , 139 statistical tools , 24–25 Scales. See Ecological scales Myocastor coypus , 150–152 Sciurus carolinensis , 91 Myotis nattereri , 113 Screaming hairy armadillo , 33 Myrmecophaga tridactyla , 26–27 Sea lions , 67 Sel fi sh herd model , 61 Sessile species , 17 N Silene nutans , 132–133 Naïve cows , 164 Simulation models , 180 Native cows , 164 Site suitability models North Atlantic eel , 130 aggregation distribution , 44–45 conservation biology , 149–152 individuals distribution O case–control sampling , 28 Optimum response surface model , 109 generalised linear model , 28 Ornithogalum montanum , 130 habitat selection , 28–29 Otaria fl avescens , 67, 134, 135 hunting , 28 Otariidae , 66 moose , 28, 30 Otis tarda , 113–114 random sampling design , 28 selection ratios , 29 simple sample comparisons , 28 P Society Pajonal , 45 de fi nition , 57 Patagonian otters , 159 dispersion patterns , 58–59 Phalacrocorax aristotelis , 153 group formation Phocidae , 66 Allée effects , 67–71 Pieris rapae , 19 avoidance, effect of , 60 Pinnipedia , 66 confusion effect , 62 Polygyny-threshold model , 72–73 defence effect , 62 Population distribution dilution effect , 61 behavioural mechanisms , 98–99 encounter effect , 60 close populations , 95–97 foraging , 62–63 demographic space , 94 local environment , 63 mobility axis , 94 sel fi sh herd model , 61 spatially structured populations , 94, 97–98 vigilance effect , 60–61 216 Index

Society (cont.) density-independent habitat habitat use and reproductive systems selection, 118 in birds , 65 density threshold , 119 in mammals , 65 dominant species , 121 mating harassment , 66–67 habitat selection , 119 phylogenetic history , 65 isoleg theory , 118–119 Pinnipedia , 66 Rosenzweig’s terminology , 118 polygyny-threshold model , 72–73 macroecology resource defence polygyny , 65 coarse-scale approach , 121 sexual harassment , 73–75 habitat selection hypothesis , 123 types , 64 locally abundant and rare species , 122 territory metapopulation hypothesis , 123 de fi nition , 64 niche breadth hypothesis , 123 ideal despotic distribution , 71–72 niche position hypothesis , 123 territorial fi ghts and home range , 64 phylogenetic non-independence , 122 Source−sink model principal patterns , 121 eukaryotes , 137 sampling area artefact , 122–123 habitat patches , 184 sampling effort artefact , 122 sessile organisms , 181 spatial and temporal scales , 117 subpopulations distribution Species distribution models , 178–179 Actinosphaerium nucleofi lum , abiotic interactions , 117 82–83 Akaike’s information criterion , 104 conservation biology , 81 associative models , 102 evolutionary stable strategy , 82 behavioural adaptation habitat quality , 81 bat (Myotis nattereri) , 113 habitat suitability and social behaviour , behavioural traits , 112 81–82 correlative approach , 112–113 Holt’s model , 82 great bustards (Otis tarda) , 113–114 methodological approaches , 82 physiology-based models , 112 Tetrahymena pyriformis , 82–83 radio-tracking , 113 South American sea lion , 134, 135 time-budget model , 114 Southern right whales , 147 vervet monkey (Cercopithecus Spatial population viability aethiops) , 114 analyses, 179 biogeographical distribution , 102 Species assemblages biogeographical ranges distribution model abundance surface , 108 assemble and predict together gain and loss surface model , 109–110 strategy , 124 local demographic (basin) model , 109 biotic interaction , 125 methodologies , 109 classi fi cation-then-modelling optimum response surface model , 109 strategy , 124 stochastic temporal variation community-level distribution model , 110 models , 124 vital rates model , 109 diadromous fi sh , 126 characteristics , 101 fi sh species, predictors , 126 classi fi cation tree analysis , 104–105 individual distributions , 126 climatic variables , 104 multivariate adaptive regression splines conservation biology analysis, 125, 126 bioclimatic envelope , 160 segment scale predictor , 126 climate change , 160 species interactions, incorporation of , habitat requirements , 158 124, 125 land management , 102 isoleg model Patagonian otters (Lontra analytical approach , 117 provocax) , 159 competitive abilities , 119–121 pests/invasive species , 159 density dependence , 118 spatial prediction , 158 Index 217

correlation statistics , 102 metapopulation approach , 88 ecological niche factor analysis , 107–108 presence/abundance species , 88–89 ecological tool , 103 spatial heterogeneity , 87 elephant seals, sexual segregation, Wavelet analysis , 89 106–107 landscape utilisation , 90–91 environmental envelope models , 103 metapopulations environmental variable , 104–105 Eichhornia paniculata , 80–81 generalized additive models , 104 environmental heterogeneity , 78 generalized linear models , 105–106 extinction-colonisation process , 78 global analyses , 102 geometry and stochasticity , 78 habitat variables , 103–104 habitat patches , 78 individuals and physiological mechanism Levins model , 79–80 forest gap models , 111 predictive tests , 80 geometric model , 111 source-sink model limitations , 110 Actinosphaerium nucleofi lum , 82–83 PHENOFIT , 111 conservation biology , 81 information-theoretical methods , 102, 104 evolutionary stable strategy , 82 marine mammal , 106, 108 habitat quality , 81 maximum entropy , 107 habitat suitability and social behaviour , Mediterranean species , 104 81–82 natural distributions , 103 Holt’s model , 82 niche and geographical models , 104 methodological approaches , 82 principal component analysis , 104 Tetrahymena pyriformis , 82–83 quantile regression analysis , 102 spatially structured populations , 86–87 regression technique , 103–104 Systematic forager , 35 resource selection , 102 restore ecosystem , 103 risk rejecting variables , 104 T slope and elevation , 105 Tetrahymena pyriformis , 82–83 soil type , 105 Thamnophis elegans , 137 spatial regression model , 102 Thamnophis sirtalis , 137 tree-based model , 103 Thyroptera tricolor , 149 uses , 102 Tiger beetles , 161 validity and reliability , 103 Tourist cows , 164, 165 Starlings , 40 Turdus migratorius , 40, 41 Stepping-stone model , 133–135 Stochastic temporal variation model , 110 Sturnus vulgaris , 40 V Subpopulations distribution Vegas , 173 fi tness maximisers , 84–85 Vervet monkey , 114 foraging behaviour , 84–85 Vicugna vicugna , 45, 46 individual-based models , 86–87 Vicuñas , 45, 46, 173–174 landscape distribution , 84–85 Vital rates model , 109 landscape ecology butter fl ies , 89 environmental compositions , 88 W fractals , 89–90 Wild guinea pigs , 60 FRAGSTATS , 88 historical factors and individual species, 88 Z matrix resistance effects , 89 Zalophus californianus , 9 9