OF POPULATIONS, HABITAT AND PEOPLE: THE ASIAN … · 2014-04-02 · rendition of our life as children, depicted through the story of Pitki and Dondi. My mother deserves special mention,

1

OF POPULATIONS, HABITAT AND PEOPLE: THE ASIAN ELEPHANT IN A WORLD FAST CHANGING

By

VARUN R. GOSWAMI

A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT

OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

2013

2

© 2013 Varun R. Goswami

3

To the elephant––inherently gentle, socially cognizant and steadfastly familial, a terrestrial titan whose wisdom of the ages is under humanity’s knife across two

continents

4

ACKNOWLEDGMENTS

I have often heard it being said that a PhD can be a long, often tortuous road that

one needs to ultimately tread alone. As I near the end of my journey, and look back at

the road that was, I find myself fortunate to have had an experience largely to the

contrary. Mine has not been a lonesome effort, and I have many to thank.

First and foremost, I express my deepest gratitude to my PhD advisor and guru,

Madan Oli. Through the course of my PhD, Madan has been assiduous in his

mentorship, steadfast in support, and warm in friendship. I admire his perceptiveness as

an advisor, knowing when to push me to strive harder and when to give me space to

grow. I thank Madan for honing my skills as an ecologist, for the ‘what do you want to do

when you grow up’ chats, for the science we practiced together, for this dissertation that

I write, and for being an advisor who moonlights as a gastronome!

I am also immensely thankful to all other members of my PhD committee: Jim

Nichols, for lending direction to my research and instilling much clarity in my analytical

thinking; Rob Fletcher, for his critical, thoughtful insights that vastly strengthened my

science and made my ideas more coherent; Ravi Chellam, for his unstinting support

and encouragement both on and off the field; Jim Austin, for letting me get my hands

dirty in his lab while introducing me to the world of conservation genetics; and Mel

Sunquist, for sharing his wonderful knowledge on the workings of mammals, be it tigers

in the wild, raccoons in Florida, or his elephant in Nepal, Mel Kali.

I am indebted to the School of Natural Resources and Environment for

facilitating my PhD––a journey that could begin and end thanks to the support of

Stephen Humphrey and Tom Frazer, and was sustained by the tireless efforts of Cathy

Ritchie, Meisha Wade and Karen Bray. I also thank the Department of Wildlife Ecology

5

and Conservation, my workspace and home department. In particular, I am grateful to

John Hayes for his wonderful encouragement when I began my PhD, and to Mark

Hostetler, Martin Main, Katie Sieving and Eric Hellgren for continuing the department’s

support. I also thank Elaine Culpepper, Claire Williams, Monica Lindberg, Caprice

McCrae, Tom Barnash and Sam Jones for assistance and support.

I thank the University of Florida for financially supporting me during my PhD. I am

also grateful to various institutions for generously funding my PhD fieldwork: the US

Fish and Wildlife Service – Asian Elephant Conservation Fund (USFWS – AsECF), the

Word Wildlife Fund – Asian Rhino and Elephant Action Strategy (WWF – AREAS), and

the Ashoka Trust for Research in Ecology and the Environment (ATREE) through their

small grants program funded by the John D. and Catherine T. MacArthur Foundation. I

thank Meenakshi Nagendran (USFWS – AsECF), Christy Williams (WWF – AREAS)

and Sarala Khaling (ATREE) for being immensely supportive. I thank Christy also for his

collaboration, and insights into elephant ecology and behavior.

I express my gratitude to the Wildlife Conservation Society (WCS) for facilitating

my fieldwork through logistical support and grant administration; I am particularly

indebted to the WCS – India Program, Centre for Wildlife Studies for serving as the host

institution for my on-ground PhD research. K. Ullas Karanth has always been

encouraging of my work, and I thank him for his support during my PhD. I am hugely

grateful to Samba Kumar and Peter Clyne for their advice, patience and assistance

while managing my research project, and I thank K. V. Phaniraj, A. Haridevan and M.

Srihari for efficiently dealing with fieldwork logistics.

6

I thank the Meghalaya Forest Department for according necessary research

permits and extending their cooperation in the field. I am especially grateful to The

Principal Chief Conservator of Forests (Head Office), V. K. Nautiyal; The Principal Chief

Conservator of Forests (Territorial), T. T. C. Marak; The Additional Principal Chief

Conservator of Forests (R & T) and the Chief Wildlife Warden at the time, S. Kumar;

former Divisional Forest Officer (Balphakram National Park), P. Agrahari; the Divisional

Forest Officer (Garo Hills Wildlife), S. N. Sangma; the Range Forest Officer

(Balphakram National Park), C. G. Momin; and the forest staff on ground.

I am thankful for my collaboration with Samrakshan Trust, who welcomed me to

Garo Hills and offered every form of logistical assistance necessary for my research. I

am grateful to John Fernando Shira, Golebar Sangma, Vikash Sangma and Kendesh

Shira, Janus Marak, Chenang Momin, Babloo Marak, Rollingstone Sangma and other

members of Samrakshan Trust for their dedicated contribution to data collection efforts.

I thank Arpan Sharma for his encouragement, and Kamal Medhi, Yaranajit Deka, Rohan

Mukherjee and Ginseng Sangma for their support and camaraderie during the course of

my work in Garo Hills.

I thank Uma Ramakrishnan for her collaboration, advice and support, and

acknowledge the facilities and expertise provided by her laboratory at National Centre

for Biological Sciences (NCBS), and by Centre for Cellular and Molecular Platforms,

Bangalore. I am particularly grateful to Prasenjeet Yadav and Sheikh Shahnawaz for

their dedication, and thank Malali Gowda and T. N. C. Vidya for their support.

During my PhD, I had the pleasure of interacting with many individuals who

contributed to my overall learning, exposure and social life. I thank: the Olis, Monica,

7

Muna, Maya and Misty; mentors and teachers, Ben Bolker, Arpat Ozgul, Stanley

Latimer, Ignacio Porzecanski, Karen Kainer, Danny Coenen, John Blake and Emilio

Bruna; my lab mates past and present, including in particular, Jeff Hostetler, Kristen

Aaltonen, Eva Kneip, Virginie Rolland, Binab Karmacharya, Sahar Jalal, Oscar Murrilo,

Lizzie Troyer and Madelon van de Kerk; and my other friends, Miguel Acevedo,

Mauricio Nunez-Regueiro, Christina Godoy, Andres Gonzales, Rajeev Pillay, Caroline

Staub, Lara Drizd, Nate Marcy, Grant Sizemore, Jesse Senko, Chris Rota, Raya

Pruner, George Ballantine and Caspian Catahoula. I am especially indebted to many of

them for being an enthusiastic part of my endeavors to play a good game of cricket, so

far away from home!

I gratefully acknowledge my extended wildlife family in India––alumni and faculty

of the WCS-NCBS master’s program in wildlife biology and conservation, Bangalore. In

particular, I thank Sachin Sridhara for his tireless dedication during my PhD fieldwork,

and Ajith Kumar, a father figure, colleague and friend, who I can always turn to for

guidance, spirited discussions and fish fry.

I thank Phoebe Ingty for her wonderful hospitability and mouth-watering food that

kept the energy up for arduous fieldwork. I knew that I could always turn to Eugene

Thomas and the Ingty home in Tura for support during my time in Garo Hills.

I fondly remember the happy times I spent with three generations of family both

on and off the field, and the loving enthusiasm of my doggies each time I returned

home. I express my innermost appreciation to my parents, sister, and family-in-law,

whose unstinting love and support made this PhD possible, and the effort worthwhile. I

am grateful for my interactions with my family-in-law, which have always abounded in

8

fun, high spirits, and overall positivity; such moments can dispel the humdrums of the

dullest of days during the course of a long PhD. My sister has expressed her love and

support through many quiet but thoughtful ways, none more touching than her artistic

rendition of our life as children, depicted through the story of Pitki and Dondi. My mother

deserves special mention, not just for her undying encouragement of my PhD pursuit,

but for nurturing in me a person who has always loved the world of Mowgli, Bagheera

and Baloo, from the day she introduced this wonderful world to a five-year old me. And I

will always remember my father quoting Pink Floyd’s ‘Time’ in his endeavors to inspire

me to run, so I don’t miss the starting gun.

Finally, I thank Divya Vasudev who has been a pillar of strength, source of

inspiration, intellectual confidante and constant companion through the nadirs and

zeniths of my PhD. With her I explored the magical forests of Nokrek and the ‘heart’ of

Balphakram; without her, this PhD would have been an incomplete journey.

9

TABLE OF CONTENTS

page

ACKNOWLEDGMENTS .................................................................................................. 4

LIST OF TABLES .......................................................................................................... 11

LIST OF FIGURES ........................................................................................................ 12

LIST OF ABBREVIATIONS ........................................................................................... 13

ABSTRACT ................................................................................................................... 14

CHAPTER

1 CONSERVATION IN HETEROGENOUS, HUMAN-DOMINATED LANSCAPES: AN OVERVIEW ...................................................................................................... 16

2 COMMUNITY-MANAGED FORESTS AND WILDLIFE-FRIENDLY AGRICULTURE PLAY A SUBSIDIARY BUT NOT SUBSTITUTIVE ROLE TO PROTECTED AREAS............................................................................................. 20

Introduction ............................................................................................................. 20

Methods .................................................................................................................. 23 Study System ................................................................................................... 23

Sampling .......................................................................................................... 24 Occupancy Models and Data Analysis ............................................................. 25

Results .................................................................................................................... 28

Detection of Elephant Presence ....................................................................... 28 Role of Community-Managed Forests and the Importance of Human

Presence ....................................................................................................... 28 Degree of Subsidiarity of Wildlife-Friendly Agriculture ...................................... 29

Discussion .............................................................................................................. 31

3 WHY DO ELEPHANTS RAID CROPS? IMPLICATIONS FOR THE HOLISTIC MANAGEMENT OF HUMAN-WILDLIFE CONFLICT ACROSS AFRICA AND ASIA........................................................................................................................ 41

Introduction ............................................................................................................. 41

Methods .................................................................................................................. 45 Study Area ........................................................................................................ 45 Quantification of Conflicts ................................................................................. 46 Analytical Design and Occupancy Modeling..................................................... 46 Covariates ........................................................................................................ 50

Results .................................................................................................................... 51 Seasonality of Conflicts .................................................................................... 51

10

Probability of Reporting Conflicts ..................................................................... 51

Spatiotemporal Patterns of Crop Depredation .................................................. 52 Discussion .............................................................................................................. 54

4 THE IMPORTANCE OF CONFLICT-INDUCED MORTALITY IN DESIGNING MULTIPLE USE RESERVES FOR WIDE-RANGING SPECIES OF CONSERVATION CONCERN ................................................................................ 67

Introduction ............................................................................................................. 67 Methods .................................................................................................................. 69

Density-Dependent Model of Elephant Demography ....................................... 69 Mortality Due to Human-Elephant Conflict........................................................ 71 Scenarios of Human-Elephant Conflict and Habitat Alteration ......................... 72

Results .................................................................................................................... 74 Effects of HECm on Population Viability ........................................................... 74 Interplay of HECm and Habitat Degradation .................................................... 75

Resource Benefits Vis-à-Vis Mortality Drawbacks of Buffer Habitat ................. 76 Discussion .............................................................................................................. 76

5 CONCLUSIONS AND SUMMARY .......................................................................... 88

APPENDIX

A MODEL SELECTION FOR ELEPHANT SITE USE ................................................ 93

B EFFECTS OF BEHAVIORAL ADAPTATIONS ON ELEPHANT POPULATION VIABILITY ............................................................................................................... 94

LIST OF REFERENCES ............................................................................................... 95

BIOGRAPHICAL SKETCH .......................................................................................... 107

11

LIST OF TABLES

Table page 2-1 Top-ranked models used to assess probabilities of elephant detection and

site use ............................................................................................................... 35

3-1 Top-ranked models used to assess the probability of detection and reporting of elephant crop depredation .............................................................................. 59

3-2 Top-ranked models for the spatial and temporal drivers of elephant crop depredation......................................................................................................... 60

3-3 Top-ranked models used to evaluate the synergistic influence of space and time in driving elephant crop depredation ........................................................... 61

4-1 Annual survival and reproductive parameters used in the females-only elephant population projection model. ................................................................ 83

A-1 Top-ranked multistate occupancy models used to evaluate elephant site use ... 93

12

LIST OF FIGURES

Figure page 2-1 Study area in Garo Hills, India, showing sites sampled to quantify elephant

space-use.. ......................................................................................................... 36

2-2 Probability of detecting elephant presence conditional on low-intensity and high-intensity use of a site. ................................................................................. 37

2-3 Overall probability of elephants using a site, regardless of intensity. .................. 38

2-4 Interactive effects of distance to protected areas and village density on the probability of high-intensity use of a site, conditional on elephant use. .............. 39

2-5 Probability of high-intensity use of a site, conditional on elephant use, as a function of land use and distance to protected areas. ........................................ 40

3-1 Human-elephant conflict locations within the study area in Garo Hills, India ...... 62

3-2 Probability of detection and reporting of crop depredation. ................................ 63

3-3 Probability of crop depredation occurrence estimated for each primary period between 2005 and 2011.. ................................................................................... 64

3-4 Extinction probabilities of elephant crop depredation as a function of mean rainfall and season-specific rainfall variability. .................................................... 65

3-5 Decline in colonization probabilities of crop depredation with distance to forest. ................................................................................................................. 66

4-1 Asian elephant population projections over time. ............................................... 84

4-2 Percentage of core area and HEC-induced mortality interact to induce extinction thresholds.. ......................................................................................... 85

4-3 Interactive effects of HEC-induced mortality rates (HECm) and percentage core area on elephant population dynamics. ...................................................... 86

4-4 Difference in Asian elephant population size between scenarios of buffer width and quality, and a reference population with no buffer. ............................. 87

B-1 Effects of behavioral adaptations on the viability of Asian elephant populations under scenarios of increasing HEC-induced mortality rates. ........... 94

13

LIST OF ABBREVIATIONS

CMF Community-Managed Forest

FD Distance to Forests

GIS Geographic Information Systems

HEC Human-Elephant Conflict

LU Land Use

PA Protected Area

PAD Distance to Protected Areas

RG Ruggedness

VD Village Density

14

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

OF POPULATIONS, HABITAT AND PEOPLE: THE ASIAN ELEPHANT IN A WORLD

FAST CHANGING

By

Varun R. Goswami

December 2013

Chair: Madan K. Oli Major: Interdisciplinary Ecology

Strategies that can reconcile local livelihoods and wildlife habitat needs are

extensively debated in conservation science and policy. Certain land uses outside

protected areas (PAs) have the potential to meet this conservation objective and are

therefore considered to be ‘wildlife-friendly.’ However, the increased human-wildlife

interface in these lands can promote conflicts between people and wildlife, thereby

reducing wildlife survival and devaluing the conservation potential of these areas. In this

dissertation, I used the Asian elephant (Elephas maximus) as a case study to

investigate the extent to which heterogeneous, human-dominated landscapes can

support the conservation needs of endangered and conflict-prone megafauna.

I began by evaluating the conservation value of wildlife-friendly land uses relative

to PAs in a heterogeneous landscape. I used multistate occupancy models to test if

elephant space-use intensity varied among land uses and with human presence outside

PAs. I found that elephants did not differentiate between PAs and wildlife-friendly land

uses in their overall use of a site, but restricted high-intensity use to sites within PAs.

High-intensity use declined with increasing distance to PAs, and this effect was

accentuated at higher village densities.

15

Thereafter, I investigated the spatiotemporal drivers of elephant crop

depredation––the primary form of human-elephant conflict (HEC)––in lands outside

PAs. My novel use of dynamic occupancy models to analyze HEC data supported the

possibility of imperfect detection and reporting of conflicts in human-wildlife conflict

research. I identified nutritional incentives provided by crops and natural forage

limitations brought about by increased rainfall variability, as concomitant drivers of

elephant crop depredation.

Finally, I investigated the ramifications of an upward trend in HEC-induced

mortality outside PAs on elephant population viability, particularly in the face of habitat

loss. I simulated elephant population dynamics under different scenarios of conflict-

induced mortality and within-PA habitat loss. Conflict-induced mortality adversely

affected population persistence, and its detrimental effects were magnified as the

proportion of core habitat within PAs declined.

Taken together, my research provides important insights into the conservation

value of lands outside PAs, and emphasizes the need for effective conflict management

to ensure the persistence of endangered megafauna in heterogeneous, human-

dominated landscapes.

16

CHAPTER 1 CONSERVATION IN HETEROGENOUS, HUMAN-DOMINATED LANSCAPES: AN

OVERVIEW

The influence of humanity on natural ecosystems is pervasive and steadfastly

growing (Vitousek et al. 1997; Sanderson et al. 2002a). A particularly damaging

outcome of increasing human demands for space and resources is the loss of wildlife

habitat. Habitat loss has large, consistently negative effects on biodiversity (Fahrig

2003), and constitutes an important driver of the current global extinction event (Ehrlich

1995; Brooks et al. 2002). In addition to other detrimental effects, habitat loss directly

depresses population size (Lande 1987; Flather & Bevers 2002), reduces survival and

reproduction (Frankham 1995; Reed & Frankham 2003), and contributes to an increase

in the probability of extinction (Saccheri et al. 1998; Westemeier et al. 1998). The

vulnerability of small populations to stochastic processes is well demonstrated in

demographic and population genetic theory (Lande 1988). Strategies that can minimize

the loss of wildlife habitat in the face of a growing human footprint are thus a key

concern in the theory and practice of conservation biology (Green et al. 2005; Fischer et

al. 2008).

Protected areas (PAs), envisioned to separate biodiversity and human resource

consumption to various degrees, constitute a central strategy for the conservation of

remnant wildlife habitat and faunal populations therein (Bruner et al. 2001; Hansen &

DeFries 2007). However, there are growing concerns about the inadequacies of a PA-

centric conservation strategy to meet wildlife habitat needs in fragmented landscapes

(Woodroffe & Ginsberg 1998; DeFries et al. 2005). The currency for conservation is

therefore transitioning to heterogeneous, multiple-use landscapes (e.g., Sanderson et

al. 2002b; Wikramanayake et al. 2004), with particular emphasis on land uses outside

17

PAs that have the potential to meet wildlife habitat needs and are thereby considered

‘wildlife-friendly’ (Daily, Ehrlich & Sanchez-Azofeifa 2001; Ehrlich & Pringle 2008; Scherr

& McNeely 2008).

Lands at the periphery of PAs, however, are becoming increasingly human-

dominated (Wittemyer et al. 2008) and this potentially poses hurdles for wildlife

conservation outside PAs. A particular cause of concern is human-wildlife conflict, a

factor that threatens the coexistence of people and wildlife worldwide (Naughton-Treves

1998; Woodroffe, Thirgood & Rabinowitz 2005; Goswami et al. 2013). Species that

have expansive space and resource requirements (e.g., large bodied mammals) are

particularly conflict-prone (Madhusudan & Mishra 2003). For such species, human-

induced mortality resulting from factors like human-wildlife conflict can seriously

undermine population persistence outside PAs (Woodroffe & Ginsberg 1998; Balme,

Slotow & Hunter 2010; Newby et al. 2013). Consequently, the feasibility of conserving

endangered large mammals in human-dominated landscapes not only hinges on the

conservation value of lands outside PAs, but also on the effective management of

human-wildlife conflicts. The latter can only be achieved through a clearer

understanding of factors that drive conflicts between people and wildlife, and how such

conflicts influence the dynamics and persistence of threatened wildlife populations.

The overarching goal of my dissertation is to investigate whether, and the extent

to which heterogeneous, human-dominated landscapes support the conservation needs

of endangered and conflict-prone megafauna. I use the Asian elephant (Elephas

maximus) as an example species to approach this problem from three important

perspectives. In Chapter 2, I quantitatively assess the conservation value of wildlife-

18

friendly land uses vis-à-vis their potential to fulfill a fundamental function of PAs––the

separation of biodiversity from anthropogenic threats. I specifically ask the question: to

what extent do wildlife-friendly land uses outside PAs provide comparable conservation

benefits to habitats within PAs? I distinguish the role of wildlife-friendly land uses as

being (a) subsidiary, whereby they augment PAs with secondary habitat, or (b)

substitutive, wherein they provide comparable habitat to PAs. I further hypothesize that

this role is modulated by human presence in the case of large mammals that are faced

with threats of poaching and human-wildlife conflict. I tested my hypotheses by

investigating the influence of land use and human presence on space-use intensity of

the Asian elephant in a fragmented landscape comprising a mosaic of PAs and wildlife-

friendly land uses.

In Chapter 3, I focus on the same heterogeneous, human-dominated landscape

as Chapter 2 to investigate what drives existing conflicts between people and elephants

in the area. The primary form of human-wildlife conflict across the world is crop and

livestock depredation. I therefore ask the question: what spatiotemporal factors

contribute to elephant crop depredation patterns in wildlife-friendly land uses outside

PAs? I used data on reports of elephant crop depredation in the landscape between

2005 and 2011 to test whether elephant crop depredation relates to (a) the nutritional

incentive provided by crops, and (b) natural forage limitations with habitats. I analyzed

the conflict data using dynamic occupancy models that provided estimates of crop

depredation occurrence, the extinction of crop depredation in sites raided in the

previous season, and the colonization of crop depredation in sites that had not been

raided in the previous season.

19

In Chapter 4, I assess if human-elephant conflict negatively affects the

conservation potential of wildlife-friendly lands outside PAs. I specifically ask the

question: how might HEC-induced mortality influence long-term persistence of elephant

populations in a scenario where core habitat within PAs is increasingly converted to

multiple-use buffer habitat? Using a single-sex, age-structured density-dependent

matrix population model to simulate elephant population dynamics over a period of 500

years, I (1) assessed the existence of extinction thresholds arising from the interaction

of mortality due to human-elephant conflict (HEC) and habitat degradation, and (2)

evaluated whether and to what extent habitat supplementation by the buffer is devalued

by detrimental effects of conflict-induced mortality.

The three chapters together represent a body of work that makes an important

contribution to our understanding of opportunities and challenges associated with

conserving endangered large mammals in heterogeneous, human-dominated

landscapes. Chapter 2 provides important insights into the conservation value of

wildlife-friendly land uses relative to PAs, particularly in the context of how a strong

human presence in these land uses might modulate their conservation potential.

Chapter 3 investigates the drivers of conflicts between elephants and people––an

inevitable outcome of an increasing human-wildlife interface––to facilitate the design of

effective conflict mitigation strategies. Finally, Chapter 4 emphasizes the need for such

strategies by demonstrating the potential negative ramifications of human-wildlife

conflicts on the long-term persistence of endangered megafauna, such as the Asian

elephant, in multiple-use landscapes.

20

CHAPTER 2 COMMUNITY-MANAGED FORESTS AND WILDLIFE-FRIENDLY AGRICULTURE PLAY A SUBSIDIARY BUT NOT SUBSTITUTIVE ROLE TO PROTECTED AREAS

Introduction

The conservation value of land uses that have the potential to reconcile

biodiversity conservation and human livelihood needs (hereafter, ‘wildlife-friendly’ land

uses), is receiving increasing scientific attention (Daily, Ehrlich & Sanchez-Azofeifa

2001; Ehrlich & Pringle 2008; Norris 2008; Scherr & McNeely 2008). This emergent

trend largely stems from the urgent need for strategies that can minimize costs to

wildlife species in the face of growing human demands on the land (Green et al. 2005;

Fischer et al. 2008), and two key concerns associated with conservation in traditional,

government-managed protected areas (PAs). First, notwithstanding their demonstrated

conservation benefits worldwide (Myers et al. 2000; Bruner et al. 2001), PAs are often

limited in size (Woodroffe & Ginsberg 1998), and are becoming increasingly insular due

to expansion and intensification of human land use around them (Hansen & DeFries

2007). Secondly, the exclusionary policy of strictly inviolate PAs is often in conflict with

livelihoods of local communities (Adams et al. 2004). Wildlife-friendly land uses,

including community-managed forests (CMFs) and certain agricultural systems (Daily,

Ehrlich & Sanchez-Azofeifa 2001), can address livelihood concerns through

mechanisms such as sustainable natural resource extraction, crop harvest, local

enterprise and payments for conservation (Ferraro 2001; Salafsky et al. 2001; Berkes

2007). The integration of wildlife-friendly land uses into conservation plans, however,

hinges on the effectiveness with which they fulfill their purported conservation role.

Protected areas have long been a cornerstone for conservation policy (Margules

& Pressey 2000; Hansen & DeFries 2007). Therefore, a valuable benchmark for the

21

conservation role of wildlife-friendly land uses would be to determine whether or to what

extent these land uses fulfill a fundamental role of PAs––to separate elements of

biodiversity from processes that threaten their persistence (Margules & Pressey 2000).

From such a standpoint, wildlife-friendly land uses can be conceptualized to perform

two distinct conservation roles relative to PAs: (a) Subsidiarity––wildlife-friendly land

uses perform a subsidiary function to PAs by serving as secondary wildlife habitat,

thereby augmenting overall habitat availability and supplementing the conservation

potential of PAs. Under this scenario, wildlife conservation at the landscape scale is

largely facilitated by PAs (Carter et al. 2012); (b) Substitution––wildlife-friendly land

uses provide alternative habitats that are comparable in quality to those offered by PAs

(Western, Russel & Cuthill 2009). Although several studies have documented the

biodiversity in wildlife-friendly land uses (e.g., Daily, Ehrlich & Sanchez-Azofeifa 2001;

Raman 2001; Bali, Kumar & Krishnaswamy 2007), the subsidiary and substitutive roles

of these land uses relative to PAs have not been formally evaluated. Consequently, the

distinction between these roles is not always clear in landscape-scale conservation

plans. If wildlife-friendly land uses play a subsidiary role, the habitat provided by PAs to

species in terms of resources or refuge would be unique and irreplaceable. Therefore, a

conservation model that aims to transfer stewardship of forests to local communities, for

example, would require that CMFs perform a substitutive role to PAs. Although wildlife-

friendly agriculture is not expected to play a substitutive role relative to PAs (Ehrlich &

Pringle 2008), the degree of subsidiarity of such agricultural practices (Raman 2001;

Fischer et al. 2008; Balmford, Green & Phalan 2012) directly speaks to the conservation

debate of land sparing versus land sharing (Green et al. 2005; Fischer et al. 2008),

22

perhaps providing a mechanistic approach to evaluate the viability of these contrasting

strategies in different contexts.

The pervasive influence of a growing human footprint is driving much of the

decline in wildlife populations across the world (Woodroffe & Ginsberg 1998; Sanderson

et al. 2002a; Karanth et al. 2010). Large-bodied mammals are particularly affected

because they are intrinsically extinction-prone (Cardillo et al. 2005), and share a long

history of conflict with people over limited space and resources (Woodroffe, Thirgood &

Rabinowitz 2005). Given the growth of human populations in conservation landscapes

(Wittemyer et al. 2008), and the likely negative influence of this trend on large mammal

populations (Brashares, Arcese & Sam 2001; Cardillo et al. 2004), anthropogenic

factors can potentially modulate the subsidiary or substitutive roles of wildlife-friendly

land uses (e.g., Stokes et al. 2010). The substitution of PAs by wildlife-friendly land

uses would necessitate interventions that minimize threats such as poaching (Blake et

al. 2007) and human-wildlife conflict (Woodroffe, Thirgood & Rabinowitz 2005) in

wildlife-friendly land uses. In the absence of such regulations, hunted or conflict-prone

species might avoid frequent and high-density use (henceforth “high-intensity use”) of

areas outside PAs due to a strong human presence (Ciuti et al. 2012). Therefore, while

species occurrence in wildlife-friendly land uses might indicate that they at least perform

a subsidiary role, the use of the same land uses with high-intensity would affirm their

potential for PA substitution (e.g., Daily, Ehrlich & Sanchez-Azofeifa 2001; Kinnaird &

O’Brien 2012).

I evaluate the subsidiary and substitutive roles of CMFs and wildlife-friendly

agriculture in the context of habitat needs of the endangered Asian elephant (Elephas

23

maximus), a conflict-prone large mammal species. I apply multistate occupancy models

(Nichols et al. 2007) to spatial data on elephant occurrence to differentiate between the

overall probability of elephants using a given site (Ψ1), and the conditional probability of

high-intensity use given that a site is used by the species (Ψ2). Under the subsidiarity

hypothesis, I predicted that elephants would not differentiate between wildlife-friendly

land uses and PAs in their overall space-use patterns but high-intensity use would be

restricted to PAs. Under the substitution hypothesis, however, elephant space-use

intensity would be comparable between wildlife-friendly land uses and PAs. Finally, I

investigate if human presence modulates the conservation potential of wildlife-friendly

land uses by precluding either their subsidiary or their substitutive roles.

Methods

Study System

I used the Asian elephant (Elephas maximus) as an example species based on

the following: (a) its expansive space and resource requirements (Sukumar 2003;

Fernando et al. 2008) necessitates a potential dependency on areas outside PAs for

persistence; and (b) the susceptibility of the species to threats from poaching (Blake &

Hedges 2004) and human-elephant conflict (Williams, Johnsingh & Krausman 2001)

can potentially limit its occurrence in human-dominated areas. These factors, combined

with its significant influence on ecosystem structure and function (Sukumar 2003),

contributes to the recognition of the Asian elephant as a landscape species, whose

conservation can benefit other species and the landscape as a whole (Sanderson et al.

2002c).

I conducted the study in a heterogeneous landscape in Garo Hills, India that

comprises a mosaic of relatively small PAs ( 220 km2), CMFs, areas under slash-and-

24

burn shifting cultivation (locally known as jhum), and monoculture plantations of cashew

(Anacardium occidentale), rubber (Hevea brasiliensis) and areca palm (Areca catechu)

(Fig. 2-1). The study site provided a unique opportunity to: (a) compare forests under

two management regimes (i.e., CMFs and PAs) and test if CMFs can substitute for PAs;

and (b) evaluate if the conservation roles of wildlife-friendly land uses vary by land use

type. Although wildlife-friendly agriculture is not expected to play a substitutive role

relative to PAs (Ehrlich & Pringle 2008), I was able to contrast the degree of subsidiarity

of jhum fallows undergoing regeneration of native vegetation to areas of more intensive

agriculture (i.e., monoculture plantations).

Sampling

I used a grid-based sampling approach to collect data on signs of elephant

presence (e.g., dung piles, tracks and feeding signs). Occupancy studies make a

distinction between ‘occupancy’ and ‘use’ as different parameters of interest: a

landscape unit is ‘occupied’ if a species is physically present somewhere in the unit

during the survey period; use on the other hand may be defined as the species being

present within the unit at random points in time (MacKenzie 2005). Since the objective

of the study was to measure intensity of use by elephants rather than true occupancy, I

defined each study site as a 4 km2 grid such that it is smaller than the minimum

expected home range size of Asian elephants (Fernando et al. 2008). Typically,

occupancy studies account for detection probability through repeated assessments of

sampling units assuming that the occupancy state at each unit is static during these

assessments (i.e., the system is closed to changes in occupancy) (Kendall & White

2009). Repeated surveys within a site may be executed over time (temporal replication),

or across space (spatial replication; i.e., different locations within a site) (Mackenzie et

25

al. 2006). Given the remoteness of a large proportion of the study site, I opted for spatial

replication (Kendall & White 2009) to meet the assumption of closure (Mackenzie et al.

2006). Therefore, each grid encompassed a set of nine uniformly distributed sampling

points (Fig. 2-1), and the approximate Euclidean path between two consecutive points

represented one spatial replicate. Following this design, I sequentially sampled eight

spatial replicates per site, starting at the most accessible sampling point, and walking in

a predetermined direction to all other points within the site.

I sampled a total of 80 sites encompassing 320 km2 between January and May

2011, and an additional 19 sites in January and early February 2012. During this period,

I invested approximately 990 person hours of effort walking a distance of 540.8 km. I

encountered and recorded 2225 elephant signs along sampled spatial replicates within

the 99 sites. I also documented land use at each of the nine sampling points and at the

mid-point of each spatial replicate. Thus, I obtained 17 within-site records of observed

land use. I classified land use as forest, jhum fallow (i.e., areas undergoing varying

stages of successional regeneration following the abandonment of cultivation),

monoculture plantation and human habitation. I used a global positioning system

(Garmin International, Kansas, USA) and available GIS information to map the locations

of PAs, CMFs, jhum fallows, monoculture plantations and villages within the study area.

Occupancy Models and Data Analysis

I used a multistate occupancy model (Nichols et al. 2007) to evaluate the

subsidiary and substitutive roles of lands outside PAs for the Asian elephant.

Occupancy studies generally record the detection or non-detection of species presence

as binary data (Mackenzie et al. 2006). Multistate models are an extension of the

standard occupancy models, and they allow the classification of sites by different

26

categories of occupancy (Nichols et al. 2007). Therefore, these models were ideally

suited for differentiating between low- and high-intensity elephant use of sites (Martin et

al. 2010). I defined the two states of use (low- and high-intensity use), based on median

counts of elephant signs per replicate across all sites (5 signs per replicate). Thus,

elephant sign detection along replicate r in site s was assigned state 1 (low-intensity

use) when signs encountered along r were < 5, and state 2 (high-intensity use) when

encountered signs along r were 5. Using these data, I was able to estimate the

following model parameters: (a) the probability of detecting elephant presence along

replicate r conditional on low-intensity use of site s (psr1); (b) the probability of detecting

elephant presence along replicate r conditional on high-intensity use of site s (psr2); (c)

the overall probability that site s is used, regardless of intensity (Ψs1); (d) the probability

of high-intensity use of site s conditional on elephant use of the site (Ψs2); and (e) the

probability that high-intensity use was observed along replicate r given detection of

elephant presence and that site s was used with high-intensity (δsr).

I estimated the five parameters of interest using Program MARK (White &

Burnham 1999) implemented in R (R Development Core Team 2008) with the help of

the ‘MSOccupancy’ model in the RMark library (Laake & Rexstad 2007). I first identified

the most appropriate model structure for the detection probability parameters (p1, p2 and

δ) based on Akaike’s information criterion corrected for small sample sizes (AICc).

Sample size was the number of surveyed sites. I modeled p1, p2 and δ using the

independent, additive and interactive effects of modal land use, spatial replicate and

mean ruggedness, a measure of variation in elevation within a sampling site. I

considered the following land uses: forest (including both PAs and CMFs), jhum and

27

plantations; I did not have a priori reasons to expect variation in detection probabilities

between PAs and CMFs. During this analysis I allowed the intensity of use parameters

(Ψ1 and Ψ2) to vary as a function of (a) distance to forest (i.e., distance of site s to the

closest forest irrespective of whether it is community-managed or within a PA), (b)

distance of site s to PAs, (c) mean village density (per km2) within site s, and (d) modal

land use within s. My intention was to use the most general model for Ψ1 and Ψ2 while

identifying the best model structure for p1, p2 and δ (e.g., Goswami et al. 2011; Karanth

et al. 2011).

Next, I fixed p1, p2 and δ to the best-supported model structure from the previous

analysis, and evaluated the relative influence of the independent, additive and pair-wise

interactive effects of the aforementioned site-specific covariates (a-d) on Ψ1 and Ψ2. I

also compared all models to an intercept-only model whereby parameters of interest

were constant. Since there were three model structures for p1, p2 and δ that had

comparable support (∆AICc < 2) (Table 2-1) (Burnham & Anderson 2002), I carried out

three sets of analyses whereby p1, p2 and δ were fixed to one structure per analysis set.

Model comparisons were made on the basis of their AICc scores and Akaike weights

(wi). Top models for Ψ1 and Ψ2 were identical in all three sets (Table 2-1 and Table A-1

in Appendix A).

I used ArcGIS v.9.3 to create a distance map for forests (including both PAs and

CMFs), and thereafter extract the minimum Euclidean distance of the centroid of each

sampled site from the nearest forest, irrespective of PA or CMF. I similarly extracted the

Euclidean distance of each site from PAs. I used ArcGIS to create a density map of

village locations within the study area, and obtained the average number of villages per

28

km2 within each site. Finally, I used Quantum GIS v.1.6 to estimate mean ruggedness

for each site from a digital elevation model of the area.

Results

Detection of Elephant Presence

The probability of detecting elephant presence when a given site was used by

the species with low-intensity (p1) depended on the additive effect of land use and

ruggedness within the site (Table 2-1). Land use also influenced detection probability in

sites used with high-intensity (p2). Detection probability for both states of site use

intensity was higher in forests (p1 = 0.57, 95% CI = 0.41–0.72; p2 = 0.83, 95% CI =

0.78–0.86), declining progressively as the land use transitioned to jhum (p1 = 0.14, 95%

CI = 0.05-0.34; p2 = 0.74, 95% CI = 0.64–0.82) and monoculture plantations (p1 = 0.08,

95% CI = 0.03–0.20; p2 = 0.41, 95% CI = 0.27–0.58) (Fig. 2-2A). Site ruggedness had a

negative effect on p1 for all land uses (Table 2-1; Fig. 2-2B). The probability that high-

intensity use was observed given detection of elephant presence and that a site was in

fact used with high-intensity (δ) was estimated to be 0.57 (95% CI = 0.52–0.62).

Role of Community-Managed Forests and the Importance of Human Presence

Asian elephants have been shown to prefer forest fragments and riparian

habitats in heterogeneous landscapes (Kumar, Mudappa & Raman 2010). My results

support this observation whereby distance to forests, including both PAs and CMFs,

was the overwhelming driver of the probability of elephants using a site regardless of

intensity (Ψ1) (Table 2-1). The relationship between Ψ1 and distance to forest was

strongly negative ( = -0.004, 95% CI = -0.008 to -0.0005). Thus, while the probability of

elephants using a site was high within a distance of 1.5 km from forests (Ψ1 > 0.9), it

29

declined sharply to nearly zero (Ψ1 = 0.02) at a distance of just 3 km from the forest

edge (Fig. 2-3).

The probability of high-intensity use conditional on elephant use of a site (Ψ2),

however, was best explained by the interactive effect of distance to PAs and village

density (Table 2-1). For a given distance of a site to PAs, Ψ2 declined with an increment

in village density (Fig. 2-4A). While Ψ2 remained relatively unchanged (mean Ψ2 = 0.84,

SE = 0.12) within a maximum distance of 5.7 km from PAs in the study system when

there were no villages (Fig. 2-4B), it rapidly declined to near zero as village density

increased (Figs. 2-4C, 2-4D and 2-4E). For example, at a mean density of 1.33 villages

per km2, which was the maximum among sampled sites, the probability of high-intensity

use dropped to negligible levels (Ψ2 < 0.01) when distance to PAs exceeded 2.7 km

(Fig. 2-4E). The strong negative influence of village density notwithstanding, Ψ2

remained high at approximately 0.9 within a distance of 1.5 km from the edge of PAs

even with increasing village density (Fig. 2-4A).

Thus, my results suggest that elephants do not differentiate between PAs and

CMFs in their overall space-use patterns but restrict high-intensity use to PAs. This

lends support to the subsidiarity hypothesis with respect to CMFs and contradicts the

hypothesis that CMFs or other wildlife-friendly land uses can substitute for PAs in the

context of elephant habitat requirements. The finding that village density influenced Ψ2

but not Ψ1 (Table 2-1) suggests that although human presence does not detract from

the subsidiary role of wildlife-friendly land uses, it precludes PA substitution.

Degree of Subsidiarity of Wildlife-Friendly Agriculture

I further investigated the degree of subsidiarity of wildlife-friendly agriculture in

the study area using models where Ψ1 varied with land use. Unfortunately, the models

30

failed to converge when other parameters (i.e., p1, p2, Ψ2) were a function of site-

specific covariates. Therefore, I used a simpler model where Ψ1 varied with land use

while all other parameters were fixed to the intercept (i.e., Ψ2, p1, p2 and δ were

constant). Estimates of Ψ1 for different land uses as per this model were: Ψ1FOREST =

1.00, 95% CI = 0.99–1.00; Ψ1JHUM = 0.94, 95% CI = 0.52–0.99; and Ψ1

PLANTATION = 0.65,

95% CI = 0.40–0.84. I did not separate forests into PAs and CMFs because the top

model (Table 2-1) suggested that forests, irrespective of whether they were within PAs

or CMFs, were associated with Ψ1 = 0.99 (95% CI = 0.78–0.99).

The best-supported model that included land use as a covariate for Ψ2 (Table 2-

1: model 3) suggested an additive effect of distance to PAs and within-site land use on

high-intensity use. This model further supports the finding that wildlife-friendly land uses

do not substitute for PAs. As per this model, there was a strong likelihood of high-

intensity use inside PAs (Ψ2PA = 0.93, 95% CI = 0.82–0.98) (Fig. 2-5A). At an average

distance of 1.5 km from the outer edge of PAs, CMFs and jhum fallows were also fairly

likely to support high-intensity use by elephants (Ψ2CMF = 0.83, 95% CI = 0.66–0.92;

Ψ2JHUM = 0.90, 95% CI = 0.66–0.98) (Fig. 2-5A). The estimate of Ψ2 in plantations at the

same distance from PAs was lower and less precise (Ψ2 = 0.60, 95% CI = 0.29–0.85).

As distance to PAs increased, however, Ψ2 decreased rapidly in all land use types

(Figs. 2-5B, C and D), with the sharpest decline in plantations (Fig. 2-5D), and the

shallowest in jhum fallows (Fig. 2-5C). Mean village densities per km2 in CMFs, jhum

and plantations were 0.20 (SE = 0.04), 0.17 (SE = 0.05) and 0.64 (SE = 0.06),

respectively. Thus, it is conceivable that a ~200% higher village density in plantations

31

compared to CMFs and jhum contributed to a considerably lower probability of high-

intensity use of this land use.

Discussion

Community-based conservation stems directly from the debate on

preservationism versus sustainable use as contrasting paradigms for wildlife

conservation and natural resource management (Robinson 1993; Madhusudan &

Raman 2003). The purported failure of state-run exclusionary conservation is a key

factor that contributed to the development of different community-based conservation

approaches (Berkes 2007). Thus, sustainable use in CMFs presents an alternative

model to preservation in nationally mandated PAs. This view of community-based

conservation, however, implicitly assumes that CMFs can substitute for PAs in terms of

the benefits they offer to wildlife species of conservation concern. Results from this

study suggest that this assumption may not be valid for species that compete with

humans for resources: high-intensity space use by elephants was largely confined to

PAs with CMFs playing a subsidiary conservation role. Therefore, the inclusion of CMFs

in conservation plans should be conceived as a strategy that can augment space and

resource availability for wildlife, while meeting sustainable livelihood needs of local

communities. Depending on the spatial location of CMFs, they could effectively fulfill the

function of a ‘go-between’ buffer zone, supporting sustainable use of natural resources

while facilitating greater preservation within PAs (Noss 1983). They could also serve as

corridors or movement conduits between PAs (Ricketts 2001).

As current conservation policy grapples with strategies that can meet wildlife

habitat needs in the face of increasing anthropogenic land transformation (Margules &

Pressey 2000; Sanderson et al. 2002a), much debate has centered on balancing

32

agriculture with wildlife conservation (Green et al. 2005; Fischer et al. 2008). Land

sharing through wildlife-friendly farming can be a suitable conservation strategy for

adaptable species that can persist in a ‘soft-matrix’ landscape (Green et al. 2005;

Fischer et al. 2008). For generalist species such as the Asian elephant, (Sukumar

2003), wildlife-friendly farming therefore has the potential to play a subsidiary role to

PAs, providing secondary habitat to the species. I tested this hypothesis in the context

of jhum, a wildlife-friendly farming technique, and evaluated its conservation value

relative to intensive agriculture in monoculture plantations.

Both Ψ1 and Ψ2 were ~30% greater in jhum fallows than in plantations,

suggesting that the former is the more elephant-friendly land use. Compared to CMFs,

jhum fallows retained a marginally higher chance of supporting high-intensity use and of

sustaining it farther away from PAs (Fig. 2-5). Jhum fallows are often characterized by

grasses, that at a later stage of succession, are replaced by dense bamboo culms

(Raman, Rawat & Johnsingh 1998). Both grasses and bamboo are valuable resources

for elephants (Sukumar 2003), and their prevalence likely attracts elephants to these

fallows. It is important to note, however, that high-intensity use of these agricultural

areas was contingent on elephant presence, which was strongly dependent on proximity

to forests within PAs or CMFs (Table 2-1 and Fig. 2-3). The importance of neighboring

forests has been highlighted by other studies investigating the conservation value of

wildlife-friendly farming (Raman 2001; Bali, Kumar & Krishnaswamy 2007), and their

maintenance will likely hold the key to successful elephant conservation within

heterogeneous landscapes. If intact forests are difficult to maintain in conjunction with

wildlife-friendly farming due to lower agricultural yields in the latter (Green et al. 2005), a

33

combination of land sparing and wildlife-friendly farming might be an optimal landscape-

scale conservation strategy (Fischer et al. 2008). Where intensive agriculture is

unavoidable, conservation plans must ensure that large patches of forests are

conserved and connectivity among these patches is maintained (Fischer et al. 2008).

Human-wildlife coexistence, particularly in the context of conflict-prone

megafauna, has been the subject of much research (Woodroffe, Thirgood & Rabinowitz

2005; Carter et al. 2012) and recent debate (Goswami et al. 2013; Harihar et al. 2013;

Karanth et al. 2013b). Coexistence is generally difficult to achieve because of

competing, and often conflicting, resource needs of large mammals and people

(Woodroffe, Thirgood & Rabinowitz 2005), and the resultant decline in large mammal

populations in the face of a burgeoning human footprint (Brashares, Arcese & Sam

2001; Cardillo et al. 2004). My results clearly demonstrate this contention, whereby the

substitutive role of wildlife-friendly land uses was mediated by human presence. Thus,

even though CMFs and wildlife-friendly farming might ‘soften’ the matrix between PAs in

heterogeneous landscapes, threats imposed by prevalent human populations can

substantially limit species like the Asian elephant from using the matrix. This apparent

avoidance response to a human-dominated “landscape of fear” is analogous to

observed species behavioral responses to human disturbance (Ciuti et al. 2012).

Therefore, the mitigation of anthropogenic threats outside PAs is essential to realize the

conservation potential of wildlife-friendly land uses.

The currency for conservation is increasingly transitioning towards multiple-use,

heterogeneous landscapes to meet the habitat requirements of wide-ranging species

(Sanderson et al. 2002c). Conservation investment and planning in lands outside PAs

34

clearly need to be based on empirical evidence vis-à-vis their conservation value

(Sutherland et al. 2004; Ferraro & Pattanayak 2006). Here, I highlight the potential for

CMFs and wildlife-friendly agriculture to strengthen the conservation benefits offered by

PAs to wide-ranging species like the Asian elephant. However, my results emphasize

that these wildlife-friendly land uses do not substitute for PAs in their ability to support

the habitat needs of elephant populations. I show that the strong presence of humans

outside PAs has an overriding negative influence on the conservation potential of CMFs

and wildlife-friendly farming. Therefore, global conservation policy not only needs to

recognize the multiple roles that wildlife-friendly land uses can fulfill, but also their

conservation limitations. This holds the key to effecting successful conservation

initiatives in human-dominated landscapes.

35

Table 2-1. Top-ranked models used to assess probabilities of elephant detection and site use

Model K AICc ∆AICc wi

Detection Probability Ψ1(FD) Ψ2(PAD) p1(LU + RG) p2(LU) δ(.) 12 1184.76 0.00 0.30 Ψ1(FD) Ψ2(PAD) p1(LU + RG) p2(LU + RG) δ(.) 13 1186.30 1.54 0.14 Ψ1(FD) Ψ2(PAD) p1(LU + RG) p2(LU) δ(RG) 13 1186.59 1.83 0.12 Ψ1(FD) Ψ2(PAD) p1(LU + RG) p2(LU) δ(LU) 14 1186.96 2.20 0.10 Ψ1(FD) Ψ2(PAD) p1(LU + RG) p2(LU) δ(LU × RG) 17 1187.55 2.80 0.07 Ψ1(FD) Ψ2(PAD) p1(LU + RG) p2(LU + RG) δ(RG)

14 1188.20 3.44 0.05

Site Use Probability Ψ1(FD) Ψ2(PAD × VD) p1(LU + RG) p2(LU) δ(.) 14 1181.61 0.00 0.53 Ψ1(FD) Ψ2(PAD) p1(LU + RG) p2(LU) δ(.) 12 1184.76 3.15 0.11 Ψ1(FD) Ψ2(LU + PAD) p1(LU + RG) p2(LU) δ(.) 13 1184.97 3.36 0.10 Ψ1(FD) Ψ2(PAD + VD) p1(LU + RG) p2(LU) δ(.) 13 1185.85 4.24 0.06 Ψ1(FD + VD) Ψ2(PAD) p1(LU + RG) p2(LU) δ(.) 13 1186.21 4.60 0.05 Ψ1(FD × VD) Ψ2(PAD) p1(LU + RG) p2(LU) δ(.) 14 1186.97 5.36 0.04

Probabilities of detecting low and high intensity use (p1 and p2, respectively), and the probability of observing high intensity use given elephant detection and actual high intensity use of a site (δ), were modeled to vary with land use (LU), ruggedness (RG) and spatial replicate. Covariates for the probability of site use (Ψ1) and the probability of high intensity site use (Ψ2) included distance to forests (FD), distance to PAs (PAD), village density (VD) and land use. Covariate structure for p1, p2 and δ were fixed to the top detection probability model to estimate site use probability; Ψ1 and Ψ2 were modeled as a function of the same covariates as for detection probability, as well as their additive and interactive effects. The table includes the top six models from the two respective analyses. AICc represents Akaike’s information criterion corrected for small sample size; differences in AICc between each model and the most parsimonious model are denoted by ΔAICc. K is the number of parameters and wi is the AICc model weight. Model notation follows that of linear models: a × b includes additive and interactive effects of a and b, whereas a + b includes additive effects only. A model where parameter a was held constant is represented by a(.). AICc of intercept-only models in (A) i.e., p1(.) p2(.) δ(.) and (B) i.e., Ψ1(.) Ψ2(.) were 1233.61 and 1217.1, respectively.

36

Figure 2-1. Study area in Garo Hills, India (black polygon in the inset) comprising

protected areas (PAs) and community-managed forests (CMFs) interspersed within an agricultural landscape. Sampled sites included 99 grids of size 4 km2 distributed across the landscape. Within each grid, there were nine uniformly distributed sampling points, and a walk between to adjacent points represented a spatial replicate. Land use in grids outside PAs and CMFs largely included slash-and-burn shifting cultivation and plantations of cashew, rubber and areca palm.

37

A B Figure 2-2. Probability of detecting elephant presence conditional on low-intensity and

high-intensity use of a site (p1 and p2, respectively) as a function of A) land use within the site and B) ruggedness. A) Shaded bars represent detection probability estimates and the error bars are 95% confidence intervals. Ruggedness for p1 was the average value across sites. B) The influence of ruggedness within a site on p1 when land use is forest (solid line), shifting cultivation or jhum (dashed line) and monoculture plantations (dotted line). Shaded circles represent observed elephant detections or non-detections in sampled sites.

Forest Jhum Plantation

p1

p2

Land use

De

tection

pro

bab

ility

0.0

0.2

0.4

0.6

0.8

1.0

(a)

5 10 15 20 25

0.0

0.2

0.4

0.6

0.8

1.0

Ruggedness indexD

ete

ction

pro

bab

ility

(b)

38

Figure 2-3. Overall probability of elephants using a site, regardless of intensity (Ψ1), as

distance to forests (within protected areas or community-managed forests) increases. The solid line represents Ψ1 estimates, and shaded polygon, the 95% confidence interval around these estimates. Shaded circles represent observed elephant detections or non-detections in sampled sites.

0 1000 2000 3000 4000

0.0

0.2

0.4

0.6

0.8

1.0

Distance to forest (m)

Pro

ba

bili

ty o

f use

39

Figure 2-4. Interactive effects of distance to protected areas (PAs) and village density

(VD) on the probability of high-intensity use of a site conditional on elephant use of the site (Ψ2). A) Contour lines represent Ψ2 estimates across the range of distance to PAs and mean village densities associated with sampled sites within the study area. Variation in estimates of Ψ2 (solid lines) as a function of distance to PAs, are plotted for increasing village density ranging from B) VD = 0 (first quartile), through C) VD = 0.25 (mean), and D) VD = 0.43 (third quartile) to E) VD = 1.33 (maximum). VD was quantified as the number of villages per km2 averaged across each 4 km2 sampling site. Shaded polygons represent 95% confidence intervals.

40

Figure 2-5. Probability of high-intensity use of a site conditional on elephant use of the

site (Ψ2) as a function of land use and distance to protected areas (PAs). Estimates of Ψ2 in community-managed forests (CMF), jhum (JHM) and plantations (PLN) are presented for A) sites at an average distance of 1.5 km from the edge of PAs (shaded bars), and with increasing distance to PAs (solid lines in B, C and D, respectively). Error bars and shaded polygons represent 95% confidence intervals.

41

CHAPTER 3 WHY DO ELEPHANTS RAID CROPS? IMPLICATIONS FOR THE HOLISTIC

MANAGEMENT OF HUMAN-WILDLIFE CONFLICT ACROSS AFRICA AND ASIA

Introduction

Conflict between people and wildlife—typically involving species that compete

with humans for space and resources (Woodroffe, Thirgood & Rabinowitz 2005)—is one

of the most pervasive and intractable issues of current conservation concern (Dickman

2010; Karanth et al. 2013a). Crop or livestock depredation by wildlife imposes

substantial costs on local people and their livelihoods (e.g., Naughton-Treves 1998;

Madhusudan 2003). Concurrently, human incursions into wildlife habitat (Sanderson et

al. 2002a) and retributive killing of conflict-prone species (e.g., Woodroffe & Frank 2005)

threaten the persistence of endangered fauna living in close proximity to people

(Dickman 2010). Recurrent conflicts not only undermine the well-being of both people

and wildlife (Madhusudan 2003), they also encumber local support for conservation

(Naughton-Treves, Grossberg & Treves 2003). Therefore, the effective management of

human-wildlife conflict is an essential precondition for the coexistence of wildlife and

people across space and over time (Madden 2004; Dickman, Macdonald & Macdonald

2011).

Human-elephant conflict (HEC) exemplifies the challenges associated with

conserving conflict-prone species in human-dominated systems (Guerbois, Chapanda &

Fritz 2012). Depredation of cultivated crops by elephants is the primary source of HEC

in both Africa and Asia (Williams, Johnsingh & Krausman 2001; Sukumar 2003; Sitati,

Walpole & Leader-Williams 2005), and understanding the drivers of crop depredation is

central to designing effective HEC mitigation strategies. Crop depredation, and HEC in

42

general, is inevitable in the fragmented, human-impacted landscapes that comprise a

large proportion of existing elephant habitat (Leimgruber et al. 2003; Chartier,

Zimmermann & Ladle 2011; Guerbois, Chapanda & Fritz 2012). However, not all

elephants that have access to a crop field will raid it (Williams, Johnsingh & Krausman

2001; Sukumar 2003). The key question then is: what drives elephants to raid crops?

Cultivated crops generally have higher nutritional value than forest plants

(Sukumar 1989), and crop fields as such represent resource rich patches. However,

there is also an element of risk associated with crop raiding due to human retaliation

(Sukumar & Gadgil 1988). Given the presence of crops, two non-mutually exclusive

hypotheses can potentially determine when elephants take the ‘risk’ of raiding them. (1)

Elephants adopt a “high-risk-high-gain” strategy (sensu Sukumar & Gadgil 1988)

whereby they venture out of forest refuges to maximize nutritional benefits from crops

even when forage in the natural habitat is not in short supply (henceforth, “nutritional

incentive hypothesis”). Under this hypothesis, crop depredation should peak during

maximum crop availability. (2) Elephants raid crops when natural forage is limited or

depleted (Sukumar 2003; Osborn 2004) (henceforth, “forage limitation hypothesis”).

Crop depredation under this hypothesis could coincide, for example, with (a) seasons or

climatic conditions when resources are limiting within natural habitats, and (b) dispersal

events or movement forays between habitat patches, where food resources might be

largely restricted to crops. Limited natural forage availability could add to the nutritional

incentive hypothesis, intensifying the frequency with which elephants raid crops. Crop

depredation could either be an outcome of opportunistic encounters of crop fields, or

may be influenced by prior knowledge of crop field locations.

43

Rainfall is a key determinant of primary productivity in terrestrial ecosystems,

influencing plant phenology and periods of peak forage availability (van Shaik, Terborgh

& Wright 1993; Ostfeld & Keesing 2000). Crop depredation patterns predictably

demonstrate rainfall-related seasonal signatures, coinciding with crop growing and

harvest periods (Sukumar 2003; Osborn 2004; Gubbi 2012). Therefore, it is conceivable

that average rainfall during seasons of plant growth would be positively correlated with

resource availability within both habitat patches and crop fields. On the other hand,

irregularity of rainfall during the same period could potentially limit the abundance of

plant resources and introduce variability, and thus unpredictability, in forage availability

(e.g., Knapp et al. 2002). The relationship between rainfall—manifested through its

influence on resource availability—and crop depredation is important to disentangle to

manage HEC over time, particularly because climate-change models predict

increasingly variable rainfall regimes (Easterling et al. 2000; Weltzin et al. 2003).

Human-wildlife conflict research largely relies on data obtained from local

informants or respondents to questionnaires (e.g., Sitati et al. 2003; Karanth et al.

2013a). Within such a framework, if a crop depredation event is reported, we know with

certainty that it has occurred (assuming that the report is verified to avoid false

positives). However, if a crop depredation event is not recorded or reported, it could

imply either that a crop raid has not occurred, or that it did occur but was not detected or

reported. Therefore, the reporting of conflicts is analogous to the detection of animals

when attempting to estimate population abundance or species occurrence (Williams,

Nichols & Conroy 2002; Mackenzie et al. 2006). Biases in conflict reporting probability

may occur due to factors such as variable search effort whereby conflicts from remote

44

or inaccessible locations may be under-reported because of lower sampling effort in

such areas. Such concerns have recently been raised about citizen science surveys

whereby imperfect detection or reporting of data generated opportunistically under a

participatory research framework can result in biased trends (Kéry et al. 2010; Szabo et

al. 2010; van Strien, van Swaay & Termaat 2013). Occupancy models are designed to

appropriately account for imperfect detection and variable observation efforts

(MacKenzie et al. 2002; Mackenzie et al. 2006), and have been shown to provide

reliable estimates of species occurrence in participatory research surveys (Karanth et

al. 2009; Kéry et al. 2010; Pillay et al. 2011). These models could be just as effective

and robust in quantifying human-wildlife conflict when conflicts are likely to be

imperfectly detected and/or reported.

In this paper, I present a novel application of occupancy modeling to human-

wildlife conflict research, using data on conflicts between people and the Asian elephant

in Garo Hills, India. I combined five years of elephant crop depredation data with

appropriate multi-season occupancy models (MacKenzie et al. 2003) to provide

important insights into potential drivers of depredation in a fragmented landscape, while

accounting for imperfect detection and reporting of conflicts. I began by testing if

accessibility of sampling units influences the probability of detection and reporting of

conflicts ( ). I then estimated and modeled probabilities of HEC extinction ( ; probability

that sites raided by elephants in season t are not raided in season t + 1) and

colonization ( ; probability that sites not raided by elephants in season t are raided in

season t + 1) as a function of various spatiotemporal covariates to test the nutritional

incentive and forage limitation hypotheses as drivers of elephant crop depredation.

45

Under the nutritional incentive hypothesis, I expected crop depredation within a year to

peak during seasons of maximal crop availability. In addition, I predicted that crop

depredation would increase with average seasonal rainfall for the same season across

years. Under the forage limitation hypothesis, on the other hand, I predicted that crop

depredation would increase with rainfall variability, and occur maximally along elephant

movement corridors. For either hypothesis, I expected crop depredation to increase with

village density, as the latter is likely to be positively related to the probability of

elephants encountering crops in the study landscape where local livelihoods are largely

agriculture-dependent. However, I note that areas with high village densities might also

be ‘riskier’ to raid because of increased human retaliation. I conclude with a discussion

on the implications of my study findings for HEC management.

Methods

Study Area

The study area in Garo Hills was a fragmented landscape comprising a mosaic of

community-managed forests and four government-managed protected areas

(Baghmara Reserve Forest, Balphakram National Park, Siju Wildlife Sanctuary and

Rewak Reserve Forest) interspersed in a matrix of agriculture and human habitation

(Fig. 3-1). Dominant agricultural land uses in the matrix include slash-and-burn shifting

cultivation (locally known as jhum), paddy cultivation and monoculture cash-crop

plantations. Road accessibility within the landscape is largely limited to two major roads

running in a north-south and east-west direction.

Garo Hills receives an average annual rainfall of ca. 1900 mm, with most of the

rainfall occurring between April and September (mean rainfall – ca. 3360 mm) (Indian

Institute of Tropical Meteorology, unpublished data). Agricultural seasons in the region

46

are determined by rainfall patterns, and can be broadly classified as (a) jhum season

involving the sowing, growth and harvest of crops such as rice, maize and millet on hill

slopes, (b) paddy season coinciding with wet rice cultivation in flooded valleys, and (c)

fallow season during which agriculture is limited to vegetable gardens (Ramakrishnan

1992; Datta-Roy, Ved & Williams 2009).

Quantification of Conflicts

In collaboration with the Samrakshan Trust, India, I obtained data on HEC from

the study area between June 2005 and October 2011, using an adaptation of

methodologies that have been successfully used to quantify HEC in Africa (Hoare 1999;

Sitati et al. 2003). The work began by training a team of 17 informants to record and

report conflicts from communal lands owned and managed by residents of 49 villages

across the study area (Fig. 3-1). Each informant verified HEC reports from 2-3 villages,

recorded the locations of these conflicts using a global positioning system (GPS), and

collated the information on a standardized data collection form. Each informant was

visited once a month to monitor the recording of HEC and to retrieve the data collection

forms. Crop depredation was the primary form of HEC; reports of property damage and

human injury were negligible in the study area.

Analytical Design and Occupancy Modeling

An occupancy modeling framework typically utilizes binomial/multinomial data on

species detection or non-detection to estimate the probability of occupancy or use of a

given sampling unit (Mackenzie et al. 2006). Repeated assessments of each sampling

unit additionally allow the estimation of species detection probability (MacKenzie et al.

2002). I used this statistical approach to estimate the probability of elephant crop

depredation using binary conflict data (i.e., HEC reported/not reported). My sampling

47

units were grid cells of size 4 km2 which ensured that the scale at which the HEC data

were being analyzed was coarse enough to minimize spatial autocorrelation (Sitati et al.

2003), while still allowing for inference on the underlying drivers of elephant crop

depredation (Guerbois, Chapanda & Fritz 2012). I used ArcGIS v.9.3 to overlay a set of

43 such grid cells across the study site, spanning an area of 172 km2 within the

agricultural matrix (Fig. 3-1). I then extracted reported conflicts from within each grid cell

during the sampling period, and assessed them for the ‘detection or non-detection’ of

HEC on a monthly basis. Therefore, one or more reports of HEC from within a grid cell

during a month was assigned a ‘1’ while the lack of HEC reports was assigned a ‘0’. In

this manner, I developed detection/reporting histories of HEC, in the form of crop

depredation, over 77 months (June 2005–October 2011) and across 43 sampled grid

cells. The total number of months during which HEC data were collected, however, was

41 months. Therefore, I incorporated the months with no HEC data into the detection

history as missing values (see Mackenzie et al. 2006). Similarly, grid cells that were not

surveyed during a given secondary occasion were treated as missing data.

I analyzed HEC data described above using multi-season occupancy models

(MacKenzie et al. 2003; Mackenzie et al. 2006). These models are analogous to

Pollock’s robust design (Pollock 1982; Pollock et al. 1990) in mark-recapture studies

whereby the occupancy state of sites remains unchanged (or changes randomly) across

secondary sampling occasions within a primary sampling period, but may change

non-randomly among T primary sampling periods due to colonization or local extinction

(MacKenzie et al. 2003; Mackenzie et al. 2006). I estimated the following parameters:

(1) the probability of detection and subsequent reporting of crop depredation in a grid

48

cell during survey j within primary period t ( ); (2) the probability of crop depredation

occurrence in a grid cell during the first primary period ( ); (3) the probability that a grid

cell with no crop depredation in primary period t experienced crop depredation in

primary period t + 1 ( ); and (4) the probability that a grid cell with crop depredation in

primary period t did not experience crop depredation in primary period t + 1 ( ). The

parameters and are akin to colonization and extinction probabilities in traditional

dynamic occupancy studies (MacKenzie et al. 2003). For the analyses, I used Program

MARK (White & Burnham 1999) implemented in R (R Development Core Team 2008)

with the help of the ‘RDOccupEG’ model in the RMark library (Laake & Rexstad 2007).

This parameterization of the multi-season occupancy model provided estimates of and

for each primary period while accounting for across all secondary occasions.

As a starting point, I used seasons of crop growth and harvest to define three

primary periods in a year, each comprising four secondary monthly sampling occasions:

(a) fallow season (January-April); (b) jhum crop season (May-August); and (c) paddy

crop season (September-December). I expected changes in the occurrence of crop

depredation events between primary periods because of the seasonality of crop growth

and harvest. I partitioned the 77-month HEC detection history into 20 primary periods

including 9 primary periods during which the grid cells were not sampled. The

occupancy modeling approach is flexible insofar as to allow parameter estimation for

periods with missing values using data from sampled periods (Mackenzie et al. 2006).

My interpretation of within this approach relates to the concept of ‘use’, defined as the

occurrence of a target species—in this case, the occurrence of HEC—within a sampling

unit at random points in time (MacKenzie 2005).

49

I began the occupancy modeling by first identifying the most appropriate model

structure for based on Akaike’s information criterion corrected for small sample sizes

(AICc). I used distance to major roads (defined as paved or unpaved roads accessible

by vehicle), and elevation gradient (or ruggedness) as covariates to test the effects of

grid cell accessibility on the probability of detection and subsequent reporting of crop

depredation events. I also modeled as a function of primary period. I compared these

models to an intercept-only model where was constant. During this analysis I allowed

and to vary as a function of (a) crop season (fallow, jhum or paddy), (b) distance to

forest (i.e., distance of grid s to the closest forest irrespective of whether it is

community-managed or within a PA), and (c) distance to movement corridors. My

intention was to use general models for and while identifying the best model

structure for . I did not model crop depredation occurrence in the first primary period

( ) as a function of spatial covariates; I therefore fixed it to be constant in all models.

Next, I fixed to the best-supported model structures from the previous analysis,

and investigated the spatial drivers of and . I used the independent, additive and

pair-wise interactive effects of distance to (a) forest, and (b) corridors, as well as village

density. I similarly identified the top temporal covariates for and fixing and as

earlier. For this analysis, I used the following temporal covariates: (a) crop season; (b)

; (c) with a 1-month time lag ( ), i.e., mean rainfall during a four-month duration

starting a month prior to each primary period; (d) with a 2-month time lag ( ); (e)

coefficient of variation of rainfall during each primary period ( ); (f) lagged by 1-

month ( ); and (g) with a 2-month time lag (

). Models included the influence

of the different rainfall covariates, as well as their additive and pair-wise interactive

50

effects with crop season. Finally, I combined the best-supported spatial and temporal

covariates from the above analyses to test if spatiotemporal factors act in synchrony to

drive and . Model comparisons in all sets of analyses were made on the basis of their

AICc scores and Akaike weights ( ).

Covariates

I used a combination of approaches to quantify the various spatiotemporal

covariates. I first used a GPS and available GIS information to map the locations of

villages, protected areas (PAs), community-managed forests and major roads within the

study landscape. I then created a map of village density in ArcGIS v.9.3 and obtained

the average number of villages per km2 within each sampled grid. I similarly created a

distance map for the closest forest (including both community-managed forests and

PAs), and major roads, and extracted the minimum Euclidean distance of the centroid of

each sampling grid from each of the three features. To quantify distance to movement

corridors, I first obtained satellite imagery of the study landscape at a resolution of 23.5

m (IRS P6 satellite using a LISS-III sensor) from the National Remote Sensing Centre,

Indian Space Research Organization (http://www.nrsc.gov.in/). This image was

classified into forest, monoculture plantations, crop fields and human habitation using

on-ground land use information (Chapter 2). I then used the classified image to create

cost-weighted distances between major elephant habitat patches (i.e., PAs), assigning

minimum movement cost to forests and highest to habitation. Finally, I used the cost-

weighted distance raster to identify potential least cost paths between PAs (Fig. 3-1),

and extracted Euclidean distances of sampling grids from these corridors as described

earlier. I note that the least cost paths coincide with elephant movement corridors

previously identified for the landscape (Tiwari et al. 2005). In addition to the above

51

analyses carried out in ArcGIS 9.3, I used Quantum GIS v.1.6 to estimate mean

ruggedness for each grid from a digital elevation model of the study area. I also

obtained average monthly rainfall records for the region from a database maintained by

the Indian Institute of Tropical Meteorology (http://www.tropmet.res.in/).

Results

Seasonality of Conflicts

Between 2005 and 2011, a total of 636 crop depredation events were reported

from agricultural lands belonging to residents of the 49 surveyed villages. Annual mean

number of conflict reports for the fallow, jhum and paddy seasons were ca. 38, 76 and

52, respectively. Mean monthly rainfall recorded during each season (averaged across

all years i.e., 2005–2011) was (a) mean ± SE = 906.25 ± 103.77 mm (fallow), (b)

3097.39 ± 116.77 mm (jhum) and (c) 904.89 ± 62.40 mm (paddy). Across years, the

coefficient of variation of rainfall for each season was on average 0.84 ± 0.06 (fallow),

0.20 ± 0.01 (jhum) and 0.56 ± 0.05 (paddy).

Probability of Reporting Conflicts

The overall probability of detection and subsequent reporting of crop depredation

( ) based on a model where was constant (i.e., (.) model) was 0.37 ± 0.03 (SE). The

best-supported models for , however, varied among primary periods, and included the

additive effects of distance to major roads and ruggedness within a grid cell (Table 3-1);

∆AICc between the top model in Table 3-1 and the constant model was 15.09. As per

the top model in Table 3-1, estimates of during each primary period for grid cells that

were at a mean distance of 2.9 km from major roads, ranged between 0.12 ± 0.04 and

0.63 ± 0.11 (Fig. 3-2A). However, was negatively influenced by distance to major

roads ( = –0.17 , SE = 0.46 ) (Fig. 3-2B). Ruggedness too had a

52

negative effect on ( = –0.03, SE = 0.03). I fixed to the two model structures that

had comparable support (∆AICc < 2; Table 3-1) (Burnham & Anderson 2002) while

further modeling the crop depredation parameters (i.e., and ) as a function of

different spatiotemporal covariates.

Spatiotemporal Patterns of Crop Depredation

Overall estimates of and based on the intercept-only model (i.e., a (.), (.)

model), were 0.30 ± 0.06 and 0.36 ± 0.05, respectively. However, there was substantial

evidence that spatiotemporal covariates affected both and (Table 3-2). Distance to

forest and its additive and pair-wise interactive effects with village density were the best-

supported spatial covariates for (Table 3-2). Distance to forest was also the most

important spatial covariate for . Akaike weights ( ) for models that included distance

to forest as a covariate summed 0.52 and 0.63 for and , respectively. There was

some evidence, albeit not as well supported, that distance to corridors might also

influence crop depredation: AICc for the top model that included distance to corridors

as a covariate for and was < 3 (Table 3-2) and Akaike weights for distance to

corridor models were 0.33 and 0.28 for and , respectively.

Mean rainfall lagged by two months ( ), crop season, and its additive effects

with , best explained the temporal variation in (Table 3-2). Average rainfall and

rainfall variability had comparable influence on whereby total for models that

included mean rainfall and rainfall coefficient of variation as covariates, were 0.48 and

0.38, respectively. There was less evidence of temporal variation in colonization

probabilities: was constant in the best-supported model in Table 3-2. However, top

models did include mean rainfall as well as rainfall variation covariates for .

53

I combined the best-supported spatial and temporal covariates in Table 3-2 to

further test if these factors synergistically drive elephant crop depredation. Top models

from this analysis (Table 3-3) had more support than either spatial or temporal factors

acting alone (∆AICc > 2 between the top model in Table 3-3 and best-supported models

in Table 3-2). Mean rainfall lagged by two months ( ), and its additive influence with

distance to forest and village density, were the best-supported model structures for

(Table 3-3). However, there was also support for the additive effects of crop season and

on . As per the top models in Table 3-3, was negatively influenced by (a) (

= –0.84 , SE = 0.28 ), (b) village density ( = –2.32, SE = 1.44), (c) the

jhum ( = –4.66, SE = 1.85) and paddy seasons ( = –7.11, SE = 2.58) and (d) (

= –6.20, SE = 2.91). The fallow season ( = 7.54, SE = 3.42) and distance to forest ( =

0.001, SE = 0.60 ) had a positive effect on .

Variation in was best explained by distance to the closest forest, although

mean rainfall and rainfall variation covariates also found support (Table 3-3). Distance

to forest had a negative influence on ( = –0.62 , SE = 0.27 ) as did

mean rainfall with no time lags ( ) ( = –0.33 , SE = 0.27 ). Rainfall

variation without time lags ( ) positively influenced but this effect was associated

with a high level of uncertainty ( = 0.25, SE = 0.58). Estimates of based on the top

model in Table 3-3 ranged between 0.39 (± 0.12) and 0.72 (± 0.08) across the 20

primary periods (Fig. 3-3).

The effects of the different spatiotemporal covariates on and , and the

resultant temporal variation in , suggested the following overall trends in elephant crop

depredation patterns: (1) probabilities of crop depredation occurrence ( ) increased as

54

the crop season transitioned from fallow through jhum to paddy (Fig. 3-3); (2) extinction

probabilities declined with mean rainfall lagged by two months ( ) (Fig. 3-4A); (3)

extinction probabilities also declined with rainfall variability during a four-month period

starting two months prior to each season ( ) (Fig. 3-4B). The effects of rainfall

variation were season-specific. As the season transitioned from paddy to fallow,

remained high (> 0.9) at 0.85 but decreased steadily thereafter. The decline in

with rainfall variability occurred at progressively lower levels of as the season

transitioned from fallow to jhum ( > 0.9 at 0.10), and from jhum to paddy ( =

0.61 ± 27 at minimum ). (4) Crop depredation declined with distance to forest––

extinction probabilities increased with distance to the closes forest, while colonization

probabilities declined away from forest refuges (Fig. 3-5).

Discussion

I found strong season-specific signatures in elephant crop depredation patterns.

Within a year, crop availability clearly varies among seasons, and this variation was

mirrored by (a) an increase in probabilities of crop depredation occurrence ( ) from the

fallow season to seasons of crop growth (i.e., jhum and paddy) (Fig. 3-3); and (b) the

finding that elephants were more likely to keep raiding crops as the season transitioned

to the jhum and paddy seasons (Figs. 3-4B). Extinction probabilities of crop depredation

events were the lowest during the paddy season, possibly indicating a preference for

paddy as a food resource. Similar peaks in elephant crop depredation during growing

periods of ‘preferred’ crops have been reported earlier (e.g., Sukumar 1989; Osborn

2004; Gubbi 2012). Nevertheless, I note that such a pattern could also be a function of

55

the variation in spatial location of jhum crops (cultivated on hill slopes) and paddy

(cultivated in flooded valleys).

My results provide evidence for the nutritional incentive hypothesis. For one, a

progressive decline in extinction probabilities with transitions in season from fallow

through jhum to paddy indicate peaks in elephant crop depredation during periods of

maximal crop availability within a year. More importantly, extinction probabilities of crop

depredation events declined with mean rainfall lagged by two months (Fig. 3-4A). This

suggests that elephants continued to raid crops during periods of peak crop availability

even when natural forage may not have been limited. However, mean rainfall lagged by

two months also had a negative effect on colonization probabilities, thereby suggesting

that elephants were less likely to raid new crop fields during these periods.

I found support for the forage limitation hypothesis as an important and

concomitant driver of elephant crop depredation. Extinction probabilities demonstrated a

consistent decline across all three seasons with rising rainfall variability two months

prior to each of these seasons (Fig. 3-4B). Natural forage availability at a given point in

time is likely limited by rainfall variability at an earlier time, and when this coincides with

high crop availability during seasons of crop growth and harvest, it can lead to high-

levels of crop depredation. Under conditions of high rainfall variability, crop depredation

continued even in the fallow season when crops are largely limited to vegetable

gardens–– ranged between 0.14 and 0.79 when the season transitioned to fallow and

> 1. Therefore, foraging opportunity (provided by crop fields) and natural forage

limitations within elephant habitats likely act in synchrony to drive elephant crop

depredation patterns.

56

Perceived risk from human presence is hypothesized to increase with distance

from refuges (Naughton-Treves 1998; Graham et al. 2009). Consistent with this

hypothesis, results from this study indicate that elephants were more reluctant to raid

crops farther away from refuges, or habitat patches. This conclusion is supported by

data on elephant space use patterns in the study area, which suggests that frequent

and high-density occurrence of elephants declines with distance to protected refuges as

village density (potentially from perceived risk due to human presence) increases

(Chapter 2). Conversely, regions of higher village density could provide for greater crop

raiding opportunity in landscapes where local livelihoods are largely agriculture-

dependent. I found support for this in the study landscape, whereby extinction

probabilities decreased with village density. I note that all sampled grid cells were within

2.5 km from the closest forest. Observations from both Africa and Asia suggest that

elephant crop depredation, although negatively affected by distance to forest refuges,

can occur up to a distance of 4–6 km from such areas (Gubbi 2012; Guerbois,

Chapanda & Fritz 2012). It may therefore be that in landscapes where villages are

located at larger distances from refuges, the positive relationship of conflict with village

density observed in this study may not hold.

Support for the nutritional incentive hypothesis emphasizes the need to adopt

strategies that can minimize elephant forays into cultivated lands. In the short-term,

crop-raiding deterrents (e.g., chili fences and spotlights) and physical barriers (e.g.,

electric fences) can prove to be effective, as previously demonstrated (Davies et al.

2011). Where possible, zonation of land use in a manner that ensures the cultivation of

57

depredation-prone crops (e.g., paddy and millets) away from forested refugia, might

serve as a holistic, longer-term solution.

The negative influence of rainfall variability on natural forage availability (Knapp

et al. 2002) appears to be an important driver of crop depredation. Climatic extremes,

including increasingly variable rainfall regimes, are a consistent prediction of climate

change models (Easterling et al. 2000; Weltzin et al. 2003). Therefore, it is foreseeable

that a changing climate could aggravate conflicts between people and wildlife by

increasing unpredictability in forage availability due to highly variable rainfall patterns.

Therefore, I also recommend the adoption of measures such as conservation education

and participatory conflict compensation schemes (e.g., Mishra et al. 2003) that can

potentially increase tolerance for conflict-prone species and alleviate conflict-induced

property and economic loss.

The idea of citizen science has emerged from the recognition that certain types

of information can only be gathered through a participatory research framework (Kéry et

al. 2010). The quantification of human-wildlife conflicts largely rely on citizen science

data, collected either with the help of an informant network or through questionnaire

surveys (e.g., Sitati et al. 2003; Karanth et al. 2013a). Notwithstanding the value of

citizen science research, some authors have recently raised concerns about biases

arising due to imperfect detection or reporting of the resultant data (Kéry et al. 2010;

Szabo et al. 2010; van Strien, van Swaay & Termaat 2013). I found that the probability

of detection and subsequent reporting of elephant crop depredation events ( ) was less

than 1.0, showed spatial and temporal variation, and was particularly low for remote

sites. I therefore suggest that an assumption of perfect detectability may not always be

58

valid in human-wildlife conflict studies, and recommend the adoption of sampling

designs and analytical frameworks that can account for imperfect detection and

reporting of conflicts. My study is the first to apply an occupancy modeling framework to

the study of human-wildlife conflict. I demonstrate that this approach effectively

addresses the issue of imperfect detection, forms a reliable and robust monitoring

protocol, and allows for inference on mechanisms underlying spatiotemporal patterns of

human-wildlife conflict.

59

Table 3-1. Top-ranked models used to assess the probability of detection and reporting of elephant crop depredation

Model K ∆AICc wi

(S) (FD) p(PP + RD) 18 0.00 0.20 (S) (FD) p(PP + RD + RG) 19 0.28 0.18 (S) (CD) p(PP + RD) 18 0.60 0.15 (S) (CD) p(PP + RD + RG) 19 1.59 0.09 (S) (FD) p(PP + RG) 18 3.39 0.04 (S) (CD) p(PP + RG) 18 3.71 0.03 (CD) (CD) p(PP + RD + RG) 18 3.76 0.03

Probability of detection and reporting of crop depredation was modeled as a function of primary sampling period (PP), distance to major roads (RD), and ruggedness or elevation gradient (RG) within a grid cell. Probability of crop depredation occurrence for

the first primary period ( 1) was fixed to the intercept. Covariates for probabilities of

extinction ( ) and colonization ( ) included distance to forests (FD), distance to movement corridors (CD), and crop season. Only the top 7 models are reported here. AICc is Akaike’s information criterion corrected for small sample size; ΔAICc the differences in AICc between each model and the most parsimonious model, K the

number of parameters and the AICc model weight.

60

Table 3-2. Top-ranked models for the spatial and temporal drivers of elephant crop depredation

Model K ∆AICc wi

Spatial Drivers (FD VD) (FD) p(PP + RD + RG) 20 0.00 0.22 (FD + VD) (FD) p(PP + RD + RG) 19 1.42 0.11 (CD) (CD) p(PP + RD + RG) 18 2.99 0.05 (FD) (FD) p(PP + RD + RG) 18 3.72 0.03 (CD) (FD) p(PP + RG) 17 3.99 0.03 (CD + VD) (CD) p(PP + RG + RG) 19 4.24 0.03 (CD VD) (CD) p(PP + RD + RG) 20 4.29 0.03 Temporal Drivers

( ) (.) p(PP + RD) 16 0.00 0.03

( ) ( ) p(PP + RD) 17 0.82 0.02

(S + ) (.) p(PP + RD) 18 1.21 0.02

( ) ( ) p(PP + RD) 17 1.31 0.01

( ) ( ) p(PP + RD) 17 1.48 0.01

( ) (.) p(PP + RD + RG) 17 1.58 0.01

( ) ( ) p(PP + RD) 17 1.61 0.01 (S) (.) p(PP + RD) 17 1.80 0.01

( ) (S) p(PP + RD) 18 2.04 0.01

Probabilities of extinction ( ) and colonization ( ) of crop depredation were modeled to vary with (a) spatial covariates including distance to forests (FD), distance to movement corridors (CD) and village density (VD); and (b) temporal covariates comprising crop season (S), mean rainfall during each 4-month primary sampling period with 1-month

( ) and 2-month time lags ( ) and without time lags ( ), and rainfall coefficient of

variation during each primary period without time lags ( ) as well as with 1-month and 2-month time lags (

and , respectively). Only the top 7 models are reported for

both (a) and (b). AICc represents Akaike’s information criterion corrected for small sample size; ΔAICc the differences in AICc between each model and the most parsimonious model, K the number of parameters and the AICc model weight.

61

Table 3-3. Top-ranked models used to evaluate the synergistic influence of space and time in driving elephant crop depredation

Model K ∆AICc wi

( ) (FD) p(PP + RD) 17 0.00 0.15

( ) (FD + ) p(PP + RD) 18 0.64 0.11

(FD + VD + ) ( ) p(PP + RD) 19 1.11 0.08

(S + ) (FD) p(PP + RD) 19 1.16 0.08

(FD + ) (.) p(PP + RD) 17 1.36 0.07 (S +

) (FD) p(PP + RG + RG) 20 1.44 0.07

(FD + ) (.) p(PP + RD + RG) 18 2.60 0.04

Probabilities of extinction ( ) and colonization ( ) of crop depredation were modeled to vary with the combined effects of the best-supported spatial and temporal covariates in Table 3-2, including distance to forest (FD), village density (VD), crop season (S),

rainfall coefficient of variation during each primary period without time lags ( ) as well

as with a 2-month time lag ( ), and mean rainfall with a two-month time lag ( )

and without time lags ( ). Only the top 7 models are reported here. AICc is Akaike’s information criterion corrected for small sample size; ΔAICc the differences in AICc between each model and the most parsimonious model, K the number of parameters

and the AICc model weight.

62

Figure 3-1. Study area in Garo Hills, India (black polygon in the inset) comprising

protected areas—Baghmara Reserve Forest (BRF), Balphakram National Park (BNP), Siju Wildlife Sanctuary (SWS) and Rewak Reserve Forest (RRF)—and community-managed forests interspersed within an agricultural matrix of slash-and-burn shifting cultivation, paddy cultivation and monoculture cash-crop plantations. Sampling sites included 36 grids of size 4 km2 distributed across the agricultural lands of 40 villages from which HEC data were collected. Least cost paths or movement corridors between protected areas were identified on the basis of the cost-weighted distance raster that shows areas of low and high cost to elephant movement.

63

A

B

Figure 3-2. Probability of detection and reporting of crop depredation ( ) as a function of primary period and distance to major roads. A) Detection probability during each four-month primary period at mean distance to major roads. Crop seasons coinciding with sampled primary periods during 2005–2011 (abbreviated as 05–11) are also indicated. B) Decline in detection probability with distance to major roads. Error bars and shaded polygons represent 95% confidence intervals.

05 05 06 07 07 07 08 09 09 11 11

Primary period

Dete

ctio

n p

rob

ab

ility

0.0

0.2

0.4

0.6

0.8

1.0 Jhum

Paddy

Fallow

0.00

0.25

0.50

0.75

1.00

0.0 2.5 5.0 7.5

Distance to major roads (km)

Dete

ction p

robab

ility

64

Figure 3-3. Probability of crop depredation occurrence ( ) estimated for each primary period between 2005 and 2011. Year names are abbreviated, and crop seasons coinciding with each primary period within a year are indicated. Error bars represent 95% confidence intervals.

05 05 06 06 06 07 07 07 08 08 08 09 09 09 10 10 10 11 11 11

Primary period

Pro

bab

ility

of o

ccu

rre

nce

0.0

0.2

0.4

0.6

0.8

1.0 Jhum

Paddy

Fallow

65

A

B

Figure 3-4. Extinction probabilities of elephant crop depredation ( ) as a function of mean rainfall and season-specific rainfall variability. A) Decline in overall extinction probabilities with an increment in mean rainfall lagged by two

months ( ). B) Effects of rainfall variability with a 2-month time lag ( )

on extinction probabilities during crop season transitions from paddy to fallow, from fallow to shifting cultivation or jhum, and from jhum to the paddy cultivation season.

0.00

0.25

0.50

0.75

1.00

0 1000 2000 3000

Mean rainfall (mm)

Extin

ction p

robab

ility

Paddy to Fallow Fallow to Jhum Jhum to Paddy

0.00

0.25

0.50

0.75

1.00

0.0 0.5 1.0 1.5 0.0 0.5 1.0 1.5 0.0 0.5 1.0 1.5

Rainfall variability

Extin

ction p

robabili

ty

66

Figure 3-5. Decline in colonization probabilities of crop depredation ( ) with distance to forest. Shaded polygons represent 95% confidence intervals.

0.00

0.25

0.50

0.75

1.00

0.0 0.5 1.0 1.5 2.0 2.5

Distance to forest (km)

Colo

niz

atio

n p

robabili

ty

67

CHAPTER 4

THE IMPORTANCE OF CONFLICT-INDUCED MORTALITY IN DESIGNING MULTIPLE USE RESERVES FOR WIDE-RANGING SPECIES OF CONSERVATION

CONCERN

Introduction

The establishment of protected areas (PAs) is widely advocated as the leading

strategy to safeguard species and their habitats from anthropogenic threats (Bruner et

al. 2001; DeFries et al. 2005). However, PAs can often be limited in size (Woodroffe &

Ginsberg 1998), and they increasingly exist as habitat islands embedded in a human-

dominated matrix (DeFries et al. 2005). Species persistence in fragmented landscapes

therefore hinges both on the size of habitat patches and the ability of faunal

assemblages to use the intervening matrix (Saunders, Hobbs & Margules 1991;

Ricketts 2001). Conversely, local people residing along park boundaries are often

dependent on resources within reserves for their livelihood (Saberwal, Rangarajan &

Kothari 2001; Adams et al. 2004). In recognition of this reality, the Fourth World

Congress on National Parks and Protected Areas in 1992 came to the consensus that

PA management should extend beyond park boundaries (McNeely 1993).

Buffer zones, comprising secondary habitat around PAs, are often demarcated

as ‘go-between’ areas where sustainable use can be practiced whilst facilitating pure

preservation in a protected core (Hjortsø, Stræde & Helles 2006). These buffer zones

can augment the conservation potential of PAs by (1) reducing extinction risk through

an increase in available habitat and population size (Pimm, Jones & Diamond 1988),

and (2) diluting edge effects and anthropogenic pressures such as hunting and resource

extraction (Brashares, Arcese & Sam 2001; Mills 2012). Furthermore, buffer zones can

68

potentially supplement the livelihood needs of local people by providing timber, fuel

wood, and other non-timber forest produce (Mackinnon et al. 1986).

Studies from across the globe suggest that the greatest challenge to the mutual

well being of humans and wildlife outside PAs arises from the conflict that occurs

between them (Naughton-Treves 1998; Woodroffe, Thirgood & Rabinowitz 2005;

Karanth et al. 2013a). Large-bodied mammals are particularly prone to conflicts with

humans as their expansive home range needs (Karanth & Sunquist 2000; Fernando et

al. 2008) force them to directly compete with people for limited space and resources.

Although a certain level of conflict is inevitable at any interface between humans and

large mammals (Madhusudan 2003), loss and degradation of core habitat are likely to

cause a greater spillover of animals into the surrounding buffer. Given that large

mammal persistence outside PAs is increasingly threatened by human-induced

mortality arising due to factors such as human-wildlife conflict, buffer zones could act as

population sinks (Woodroffe & Ginsberg 1998; Balme, Slotow & Hunter 2010; Newby et

al. 2013). As such, within-buffer anthropogenic mortality of wide-ranging megafauna can

greatly devalue its role as proxy habitat around PAs.

The Asian elephant Elephas maximus is a prime example of a large-bodied,

wide-ranging, conflict-prone species that is increasingly threatened by anthropogenic

activities across its geographical range. Almost half of the 873,000 km2 of habitat that

harbors elephants in Asia today is both fragmented and heavily impacted by humans

(Leimgruber et al. 2003). Growing rural populations, deforestation and agro-

developmental land conversion are resulting in ever-shrinking habitat islands

interspersed within a human-dominated landscape (Leimgruber et al. 2003). This, in

69

combination with an upswing in incidents of human-elephant conflict (HEC) (Sukumar

2003), is severely exacerbating the endangered status of the species. Long-term

conservation of the Asian elephant therefore requires a greater understanding of the

implications of increasing HEC, particularly in the context of habitat loss and

degradation (Goswami, Madhusudan & Karanth 2007).

The impact of conflict-induced mortality on the viability of elephant populations in

multiple-use landscapes is largely unknown. I address this issue by extending a density-

dependent age-structured matrix population model (Armbruster & Lande 1993;

Armbruster, Fernando & Lande 1999) to project Asian elephant population viability in

the light of HEC-induced mortality within the buffer. My goal was to understand how the

interplay of anthropogenic mortality and modifications to existing habitat could shape

the future of elephant populations. I defined my hypothetical area of interest as

comprising a protected core surrounded by sub-optimal buffer habitat, and explored

elephant population persistence under different scenarios of core-buffer configurations

and HEC-induced mortality. I use my results to make recommendations about elephant

conservation in the face of growing anthropogenic pressures.

Methods

Density-Dependent Model of Elephant Demography

I constructed a females-only age-structured matrix model (Caswell 2001) with 60

1-year age classes using previously reported age-specific survival and fecundity rates

for female Asian elephants (Sukumar, Ramakrishnan & Santosh 1998). I estimated

fertility rates for age class i (Fi) using a post-breeding census formulation, as the

product of the age-specific survival (Pi) and fecundity mi rates, i.e. Fi = Pi ×mi. I then

parameterized a population projection matrix using Pi and Fi. Population growth rate

70

(dominant eigenvalue, λ), stable age distribution (right eigenvector, w) and reproductive

value (left eigenvector, v) were estimated as per Caswell (2001).

I adopted the approach used by Ambruster and Lande (1993), and Ambruster et

al. (1999) to model density-dependence such that elephant density (D; females per km2)

affected reproductive parameters (age of first reproduction and calving intervals). I

modeled age at first reproduction (α) and calving interval in years (CI) as functions of

density (D), whereby α = αmin + β1D and CI = CImin + β2D. Here, αmin and CImin

correspond to α and CI prior to the effects of density-dependence. I set αmin as 16 years

following Sukumar et al. (1998) and calculated CImin as the reciprocal of annual

fecundity rates (Armbruster & Lande 1993). I identified an appropriate density for female

Asian elephants in protected core habitats as 2.35 female elephants km-2 based on a

population survey within a PA in southern India (Goswami, Madhusudan & Karanth

2007). The study reported a density of 3.13 elephants km-2 and a female-biased sex

ratio (0.75 females, 0.25 males). I then estimated β1 = β2 = 3.77 such that equilibrium

population size in core areas yielded a density of 2.35 female elephants km-2.

I estimated density-dependent fecundity rates (mi[DD]) for each age class as 1/CI

following Ambruster and Lande (1993), and used these to derive density-dependent

fertility rates (Fi[DD]). For each set of simulations, I used a starting population size of

2350 females. This initial population size approximated the equilibrium population size

for a 1000 km2 core area (Kcore), estimated as the abundance of female elephants given

an equilibrium density of 2.35 female elephants km-2. The starting population size was

distributed according to the estimated stable age distribution, which was used as the

71

initial population vector. All simulations were run for 500 years. Parameters values are

provided in Table 4-1.

Mortality Due to Human-Elephant Conflict

My model imposes two forms of control on elephant populations: (a) density-

dependence through its influence on reproduction, and (b) HEC-induced mortality in the

buffer affecting elephant survival rates. The objective of this paper is to test for the

effects of the latter on elephant population viability. To achieve this I conceptualized a

hypothetical modeling space Areatot comprised of a circular core region of size Areacore

and radius Rcore, surrounded by a multiple-use buffer of width BW. I defined the core

area as an inviolate space free of human presence to represent an undisturbed PA. The

multiple-use buffer region, on the other hand, accommodates human activities to

varying extents, thus leading to differing buffer quality and HEC-induced mortality

scenarios. As such, HEC-induced mortality was restricted to the buffer.

I allowed elephants in the model to move between the core and buffer regions

within the limitations of their biological movement capabilities and quantified by a

movement parameter (M). I assumed that elephants could move from the core to the

buffer provided they are within a distance M from the core-buffer edge. I estimated M =

7.9 km as the radius of a circle of size 196 km2, equivalent in area to the average of

reported minimum convex polygon home range sizes for Asian elephants in telemetry

studies (n = 29 individuals/herds) (Olivier 1978; Fernando & Lande 2000; Williams 2002;

Venkataraman et al. 2005; Fernando et al. 2008; Alfred et al. 2012). Effectively, the core

area of my modeling space could be partitioned into (1) an interior core of size Areaint

and radius Rint where Rint = Rcore - M and (2) a movement zone of width M (hereafter,

movement zone). I assumed that elephants inhabiting the interior core did not move to

72

the buffer, and thus were not exposed to HEC-induced mortality. However, elephants

were allowed to move between the movement zone and the buffer, and this section of

the population encountered HEC-induced mortality.

The realized HEC-induced mortality rate (HECmreal) was modeled as a function

of the probability of human-elephant encounters p(encounter) in the buffer and naïve

annual HEC-induced mortality rate (HECm) as:

HECmreal = p(encounter) × HECm

(4-1)

where,

(encounter) = Areatot – Areacore

Areatot – Areaint

(4-2)

I assumed that HECm was additive to natural mortality due to factors such as

predation, disease, malnutrition, and accidents (Sukumar 2003). Therefore, I estimated

realized survival (Pi[real]) for the fraction of the population exposed to HEC-induced

mortality as the product of baseline survival rates (Pi; i.e., survival rates in the absence

of HECm), and the probability of surviving HEC as:

Pi[real] = Pi × (1 - HECmreal)

(4-3)

Survival rates of elephants inhabiting the core interior were assumed to be

unaffected by HECm.

Scenarios of Human-Elephant Conflict and Habitat Alteration

I started with a set of scenarios focused on evaluating the ramifications of HEC-

induced mortality for populations faced with increasing habitat degradation i.e., core

habitat converted to buffer of varying quality. I first simulated the dynamics of an initial

population of 2350 female elephants, distributed among age classes as per the

73

estimated stable age distribution, in a core area of 1000 km2. I then simulated scenarios

of increasing conversion of core area to buffer habitat. I assumed four scenarios of

buffer quality: (a) the carrying capacity of the buffer, Kbuffer = 0.25 × carrying capacity of

the core, Kcore, (b) Kbuffer = 0.5Kcore, (c) Kbuffer = 0.75Kcore, and (d) Kbuffer = Kcore. To isolate

the effects of HEC-induced mortality I controlled for equilibrium population size in the

modeled space Areatot by supplementing core area loss with a buffer area of equivalent

carrying capacity. For example, when Kbuffer = 0.5Kcore, 1 km2 of core habitat lost was

compensated by 2 km2 of buffer habitat to maintain equilibrium population size of

Areatot. I considered four scenarios of habitat degradation: 0, 25, 50 and 100% loss of

core habitat. For this simulation I used HECm = 0.05. I considered this to be a moderate

rate of HEC-induced mortality as this value approximately corresponds to natural

mortality rates faced by adult Asian elephants (Armbruster, Fernando & Lande 1999).

My second set of scenarios was designed to identify extinction thresholds for the

Asian elephant brought about by the interaction of HEC-induced mortality and habitat

degradation. Like above, I allowed Areacore to range from 0 to 1000 km2, and

compensated the loss of core habitat with buffer. Since I did not have data-based

estimates of HEC-induced mortality, I used a range of HECm from 0 to 0.2, highlighting

the following rates: low (0.025), moderate (0.05), and high (0.10 and 0.20). My definition

of high HECm was based on Sukumar et al. (1998), who considered annual female

elephant mortality rates of ≥8% as high. For each simulation I quantified the time taken

by the hypothetical population to reach ‘quasi-extinction’, defined as a decline in

elephant abundance to 10% of its original size as per earlier studies (Engen, Lande &

Saether 2002; Inchausti & Halley 2003). Thereafter, I estimated equilibrium population

74

size with respect to Areacore as the average population size for the last 100 years of

simulation. I confirmed that the population had in fact attained equilibrium during this

time period by ensuring that λ ~ 1. For each HECm scenario, I evaluated minimum

Areacore required to avoid quasi-extinction in the hypothetical population.

For my final set of simulations, I investigated whether the positive effects of

increased resource availability through the addition of buffer habitat is negated by the

detrimental effects of HEC-induced mortality. I considered a reference population of

Asian elephants in a core habitat of area 1000 km2 supplemented by a buffer of varying

size and habitat quality. Thus, I included scenarios where (1) buffer width (BW) was

equal to and half of M and (2) Kbuffer was equal to and half of Kcore. For each set of

simulations, I used HECm values ranging from no HECm (0), low (0.025), moderate

(0.05) and one representative value for high (0.10). For time t, I evaluated each

scenario based on the difference in projected population size (Nt) for that particular

scenario and the projected population size for the reference population, where the

habitat was composed entirely of core (Base Nt). Therefore, for a scenario

corresponding to x buffer width, y buffer carrying capacity, and z HECm, the parameter

of interest, difference in population size for time t (∆Nt), was given by:

∆Nt[x, y, z] = Nt[x, y, z] – Base Nt

(4-4)

All models were implemented in MATLAB (MATLAB 2009).

Results

Effects of HECm on Population Viability

HEC-induced mortality had a substantial negative impact on the hypothetical

Asian elephant population (Fig. 4-1). My results indicate a disproportionate HEC-

induced decline in elephant numbers with increasing conversion of core habitat to

75

buffer. For example, when the carrying capacity of the buffer and the core were

equivalent (Kbuffer = Kcore), the initial population of 2350 female elephants declined by

36% with a 25% reduction in Areacore, and by 65% when Areacore was reduced to 50% of

the original 1000 km2 of core habitat (Fig. 4-1A). This population decline was more

severe as buffer quality (i.e., its carrying capacity Kbuffer), decreased relative to the

carrying capacity of the core, Kcore. For HECm = 0.05, scenarios of 0, 25, 50 and 100%

loss of Areacore led to equilibrium population sizes of ca. 2654, 952, 432 and 56,

respectively when Kbuffer = 0.25Kcore as compared to ca. 2654, 1693, 926 and 56,

respectively when Kbuffer = Kcore.

Interplay of HECm and Habitat Degradation

My study population declined to a functionally quasi-extinct state (10% of the

original population) when HECm > 0.05 and available core habitat was less than critical

minimum levels (Fig. 4-2A). For example, Areacore of 25% and 50% were required to

avoid quasi-extinction for HECm of 0.05 and 0.10, respectively (Figs. 4-2A and 4-2B).

Below these levels of Areacore, time to quasi-extinction diminished with core habitat loss

for a given HECm (Fig. 4-2B). Time to quasi-extinction reduced more rapidly for HECm

of 0.05 than HECm > 0.10. For example, with a decline in Areacore from 25% to 5%, time

to quasi-extinction reduced by 64% for HECm = 0.05 and by 18% for HECm = 0.10.

These results indicate (1) the existence of extinction thresholds arising from the

interaction of mortality due to HEC and habitat degradation, and (2) that this threshold is

most sensitive to HECm in the range of moderate (0.04) to high (0.08). Small

increments of HECm in this range necessitated disproportionately large increases in

Areacore to prevent quasi-extinction (Figs. 4-3A and 4-3B). With the exception of high

Areacore (>80%), this threshold effect was also reflected in substantial declines in

76

equilibrium population size with increasing HECm (Fig. 4-3B). Therefore, a small

increase in HECm needed to be accompanied by large additions to Areacore extents to

maintain a given equilibrium population size (Fig. 4-3B). On the other hand, elephant

populations faced with low HECm (e.g., 0.02) did not become quasi-extinct (Fig. 4-3A),

but were maintained at approximately 30-40% of the original population even when core

habitat was ≤ 10% of the total habitat area (Fig. 4-3B). Results presented in Figs. 4-2

and 2-3 were for Kbuffer = 0.5Kcore; these results were comparable to a scenario of Kbuffer

= Kcore.

Resource Benefits Vis-à-Vis Mortality Drawbacks of Buffer Habitat

Although the addition of a buffer zone to core habitat was associated with

benefits of increased elephant population size, these benefits were negated by

moderate (0.05) to high (0.10) HECm (Fig. 4-4). In the absence of HEC-induced

mortality, the size and quality of the buffer zone around a core was positively associated

with an increase in elephant population size. The benefits from a smaller buffer with the

same quality as the core (BW = 0.5M, Kbuffer = Kcore) (Fig. 4-4B) were comparable to a

larger buffer of poorer quality (BW = M, Kbuffer = 0.5Kcore) (Fig. 4-4C). Interestingly

however, buffer quality was more important than buffer size for elephant population

viability in the presence of low conflict-induced mortality (HECm = 0.025).

Discussion

Conservation focus in the last two decades has undergone a decided shift

towards embracing lands outside strictly inviolate PAs (e.g., Sanderson et al. 2002b;

Wikramanayake et al. 2004). This move addresses ethical concerns of human

displacement (Adams & Hutton 2007) while increasing the space and resources

available to wildlife (Bali, Kumar & Krishnaswamy 2007). A consequence of this

77

paradigm shift, however, is increased interactions between people and wildlife, which

for some species can imply heightened incidence of human-wildlife conflict.

Conservation of the Asian elephant typifies this chain of events, whereby the wide-

ranging nature of the species necessitates conservation measures in areas larger than

those encompassed by PAs. Yet, the increased interface between such species and

people in the larger landscape aggravates the incidence of human-wildlife conflict (e.g.,

Naughton-Treves 1998; Madhusudan 2003; Karanth et al. 2013a). It is imperative

therefore to carefully consider the implications of human-wildlife conflict for the long-

term viability of species before undertaking landscape-scale conservation measures. My

results emphasize the need for caution and indicate that human-wildlife conflict can

have a strong detrimental influence on wildlife populations.

Theoretical studies posit that there exists a critical habitat level below which a

given population cannot persist (Lande 1987; Fahrig 2001). I assessed the synergistic

influence of habitat degradation and HEC on elephant population viability, and found

evidence for a threshold determined by interactions between core habitat availability

and HECm (Figs. 4-2 and 4-3). Although there is little information on HECm, data on

other conflict-prone species indicate that human-caused mortality rates outside PAs can

be very high: 20.3% for African wild dogs Lycaon pictus (Woodroffe et al. 2007), 39.0%

for African lions Panthera leo (Woodroffe & Frank 2005), and 35.8% for leopards

Panthera pardus (Balme, Slotow & Hunter 2010). My results suggest that as conflict

induced mortality increases, a greater percentage of core area may be required to avoid

quasi-extinction (Fig. 4-3A), or substantial population decline (Fig. 4-3B). As per my

model, Asian elephants are most sensitive to a specific range of HECm considered to

78

be moderate to moderate-high. Within this range, a small increment in HECm

necessitated a disproportionately large increase in Areacore to avoid quasi-extinction.

Thus conservation efforts will need to maintain conflict-induced mortality rates at much

lower levels, and more significantly, be sensitive to even small increments in mortality to

ensure long-term viability of populations. Furthermore, we note that species with long

generation times might show a delayed response to factors that reduce their

populations (e.g., conflict-induced mortality), as well as those that ameliorate threats to

survival (i.e., conservation interventions) (Tilman et al. 1994). This is evident in my

results on time to quasi-extinction (Fig. 4-2B).

The interactive effects of habitat availability and HEC on population persistence

assume significance in the context of PA sizes in India, a country that holds nearly 60%

of the extant Asian elephant population (Sukumar 2003) despite accounting for just 17%

of its geographical range (Leimgruber et al. 2003). While the average area covered by

Elephant Reserves in India is 2383 km2, only 37% of this area on average is occupied

by PAs (Government of India 2005). National Parks, equivalent to the inviolate core in

my model, comprise 23% of the PA network in India (Government of India 2012)

thereby encompassing an even smaller proportion of these Elephant Reserves.

Therefore as little as 8.5% of elephant habitat in India can be regarded as inviolate and

potentially conflict-free. This situation may apply to many other species such that they

face human-induced mortality in significant proportions of their habitat (Woodroffe &

Ginsberg 1998; Robinson & Bennett 2000; Mills 2012). Based on my results, I strongly

suggest that any increase in conflict-induced mortality needs to be simultaneously

compensated by augmentation of core habitat and vice-versa. In contrast, habitat loss

79

and conflict-induced mortality are likely both increasing in many contexts (e.g., Chartier,

Zimmermann & Ladle 2011), suggesting that these threats are synergistically pushing

conflict-prone species like the Asian elephant towards extinction.

The addition of a buffer to core areas is an effective conservation measure to

augment population viability through increased space- and resource-availability.

However, there exist few guidelines to aid in conservation decisions regarding trade-offs

between the quality of buffer areas—defined for example in terms of resource

availability and existence of anthropogenic threats—and their size. Results from this

study indicate that buffer areas can in fact augment elephant abundance provided HEC-

induced mortality is low (HECm < 0.05) (Fig. 4-4). Importantly, at a low HECm of 0.025,

a relatively small but good quality buffer was associated with greater benefits (Fig. 4-4B)

than a buffer twice its size but inferior in quality (Fig. 4-4C). Thus, my results suggest

that habitat restoration and conflict resolution in the buffer may be as important as, or

more important than, securing large buffer areas. This is a key finding in the context of

densely populated countries where acquiring land for conservation is a challenge.

Given the generalist feeding habits of the Asian elephant and their potential

preference for resources available in a moderately disturbed buffer (Fernando 2006;

Fernando & Leimgruber 2011), I acknowledge the possibility of scenarios where Kbuffer >

Kcore. Under such scenarios, equilibrium population size would be higher than those

reported here when HECm is low. At higher HECm, however, detrimental effects of

anthropogenic mortality will likely exceed benefits of additional habitat offered by the

buffer. Asian elephants generally occur in lower densities in primary rainforests than dry

and more open forests (Sukumar 2003). Therefore, scenarios where Kbuffer > Kcore can

80

potentially occur in regions where PAs are largely comprised of rainforests, and

surrounding buffer regions are relatively less disturbed. However, protected areas in

countries that support a majority of the extant Asian elephant population (India and Sri

Lanka) include drier forests and grasslands (Sukumar 2003; Fernando & Leimgruber

2011). In addition, there is little evidence from Asia or Africa to suggest a higher

carrying capacity for elephants in the buffer as compared to the core; in fact, studies

from Africa suggest that elephant population density in buffer regions is at best

equivalent to that in the core (Stokes et al. 2010), and can be as low as 17% of core

densities (Blake et al. 2007).

I recognize that behavioral adaptations of species, arising from their perception

of the buffer, might lead to results different than those reported here. For example,

animals might be drawn towards the buffer because of the availability of nutrient-rich

resources in crop fields or prey in the form of livestock (Sukumar 2003; Treves &

Karanth 2003). This possible preference in combination with heightened mortality risk

might render the buffer an ecological trap (Robertson & Hutto 2006). Conversely,

species might show behavioral adaptations to avoid the risk-prone buffer. I undertook

preliminary analyses to test the implications of these behavioral adaptations: when

HECm ≥ 0.10, buffer avoidance facilitated population persistence while an ecological

trap scenario reduced equilibrium population size in relation to a ‘no behavioral

adaptation’ scenario (Appendix B). In the context of elephants, adult males may be

more likely to be affected by the ecological trap scenario as they have a higher

propensity to raid crop fields (Madhusudan & Mishra 2003), ostensibly because they are

more willing to risk encounters with humans in return for nutritional benefits from crops

81

(Sukumar & Gadgil 1988). However, elephant fecundity given a polygynous mating

system is unlikely to be affected until adult sex ratio is highly skewed to about 1 male to

20-25 females (Sukumar 2003). Therefore, results from this study are unlikely to change

except when HECm is extremely male-biased.

Finally, I note that reliable estimates of conflict-induced mortality are difficult to

obtain due to lack of reportage. A large proportion of studies on human-wildlife conflict

in countries like India have focused on the loss of human life and property (e.g.,

Saberwal et al. 1994; Madhusudan 2003). It is imperative that data on conflict-related

mortality of endangered species like the Asian elephant also be collected regularly and

systematically. Threats such as poaching need to be quantified in a similar manner as

they can have an additional deleterious effect on survival (Blake & Hedges 2004) and

thereby lead to population declines at much lower conflict-induced mortality rates. Site-

based law enforcement monitoring programs can yield well documented data on

human-induced mortality (Stokes 2012). Such data, in combination with reliable

population monitoring techniques (e.g., Goswami et al. 2012) will help assess the health

of existing populations, and also permit a direct assessment of threats arising from

factors such as habitat loss, poaching and human-wildlife conflict.

Effective conservation of wide-ranging large mammals must necessarily

encompass areas of human presence in addition to inviolate PAs. However, I suggest a

note of caution while promoting conservation measures in regions delegated as buffers

to core habitat. With continuing human-wildlife conflict, retaliatory killings by people in

the buffer will likely have a strong impact on population viability, as evidenced by this

study. I therefore emphasize the need to take counter-measures to effectively reduce

82

human-wildlife conflict and associated mortality (reviewed in Treves, Wallace & White

2009). For example, measures that deter crop-raiding elephants (Davies et al. 2011)

and participatory compensation schemes for livestock depredation (Mishra et al. 2003)

have shown conservation success. The close juxtaposition of people and wildlife clearly

poses a challenge for the conservation of endangered, conflict-prone fauna; ironically, it

is these human-dominated landscapes that perhaps offer species like the Asian

elephant the best chance of survival over time.

83

Table 4-1. Annual survival and reproductive parameters used in the females-only elephant population projection model a.

Age class (i)

Baseline survival rates (Pi)

Fecundity rates (mi)

Initial fertility ratesb

(Fi)

Minimum calving intervalb

(CImin) 1 0.9000 0 0 0 2-5 0.9600 0 0 0 6-15 0.9850 0 0 0 16-20 0.9970 0.2250 0.2243 4.4444 21-50 0.9850 0.2250 0.2216 4.4444 51-60 0.9000 0.2000 0.1800 5.0000

a Age at first reproduction (αmin = 16), and annual survival and fecundity rates were obtained from Sukumar et al. (1998). b Fi was estimated as Pi × mi, and CImin as 1/mi

84

A B

C D Figure 4-1. Asian elephant population projections over time for scenarios where the

carrying capacity of the buffer is A) equivalent to the core, B) three-quarters of the core, C) half of the core and D) one-fourth that of the core. For all models, initial population size was 2350 female elephants in a core area of 1000 km2, and the rate of HECm was 0.05.

0 100 200 300 400 5000

500

1000

1500

2000

2500

3000

3500

Time (years)

Pro

jecte

d p

op

ula

tio

n s

ize

Population decline = 36%



Core habitat loss = 0%




0 100 200 300 400 5000

500

1000

1500

2000

2500

3000

3500

Time (years)

Pro

jecte

d p

op

ula

tio

n s

ize








0 100 200 300 400 5000

500

1000

1500

2000

2500

3000

3500

Time (years)

Pro

jecte

d p

op

ula

tio

n s

ize








0 100 200 300 400 5000

500

1000

1500

2000

2500

3000

3500

Time (years)

Pro

jecte

d p

op

ula

tio

n s

ize








85

A B Figure 4-2.Percentage of core area and HEC-induced mortality interact to induce

extinction thresholds. A) Equilibrium population size and B) time to quasi-extinction as a function of declining percentage of core area and increasing HEC-induced mortality. Dashed vertical lines show the critical percentage core area below which elephant populations declined to 10% of their original size given HECm of 0.5, 0.1 and 0.2, respectively.

86

A B Figure 4-3. Interactive effects of HEC-induced mortality rates (HECm) and percentage

core area on elephant population dynamics shown by A) critical percentage core area requirements to avoid quasi-extinction with varying HECm and B) a contour plot of equilibrium population size with varying HECm and percentage core area. Contour lines represent percentages of a starting population size of 2350 individuals. Threshold effects are seen around moderate levels of conflict-induced mortality (HECm = 0.04–0.08).

87

A

B

C

D

Figure 4-4. Difference in Asian elephant population size (∆Nt) between scenarios of

buffer width and quality, and a reference population in a core area of 1000 km2 with no buffer. I used a core area of 1000 km2 supplemented by a buffer where (1) size (buffer width, BW) was relative to the average movement distance for elephants (M) and (2) quality (carrying capacity of buffer, Kbuffer), was compared to the carrying capacity of the core (Kcore). The four graphs represent sets of scenarios where A) BW equals 0.5M, Kbuffer equals 0.5Kcore; B) BW equals 0.5M, Kbuffer equals Kcore; C) BW equals M, Kbuffer equals 0.5Kcore; D) BW equals M, Kbuffer equals Kcore. In each set of scenarios, HEC-induced mortality ranges from absent (HECm = 0) through low (HECm = 0.025) and moderate (HECm = 0.05) to high (HECm = 0.10). For each scenario, advantages of buffer supplementation in terms of additional habitat availability are evaluated against the potential detrimental effects of HEC-induced mortality (HECm) on population sizes.

88

CHAPTER 5 CONCLUSIONS AND SUMMARY

As human populations and associated developmental pressures expand across

the globe, there is an increasing need for strategies that can simultaneously promote

biodiversity conservation and meet human resource requirements (Green et al. 2005;

Fischer et al. 2008). Protected areas (PAs) are envisioned to achieve this goal by

separating biodiversity and human resource consumption to various degrees (Bruner et

al. 2001; Hansen & DeFries 2007), while strategies under a sustainable-use paradigm

suggest the sharing of land to meet the needs of both humans and wildlife (Adams et al.

2004; Berkes 2007). Although shared lands can support various land uses that have the

potential to be wildlife-friendly (e.g., Daily, Ehrlich & Sanchez-Azofeifa 2001; Raman

2001; Bali, Kumar & Krishnaswamy 2007), the conservation value of these shared lands

is still a subject of scientific debate (Green et al. 2005; Fischer et al. 2008). An important

concern is the increased human-wildlife interface in lands outside PAs, which can be

detrimental to the mutual well being of people and species whose space and resource

requirements have historically conflicted with those of people (Woodroffe, Thirgood &

Rabinowitz 2005). The overarching goal of my dissertation was to investigate whether,

and the extent to which heterogeneous, human-dominated landscapes support the

conservation needs of endangered and conflict-prone megafauna.

In Chapter 2, I focused on the potential of shared lands to provide habitat for the

conflict-prone Asian elephant to ask the question: to what extent do wildlife-friendly land

uses outside PAs provide comparable conservation benefits to habitats within PAs? I

tested whether wildlife-friendly land uses play a subsidiary role to PAs by providing

secondary habitat, or whether they can potentially substitute for PAs in meeting the

89

habitat requirements of the Asian elephant. I further tested if the conservation role of

shared land is modulated by human presence. I used multistate occupancy models to

quantify elephant space-use intensity in a heterogeneous, human-dominated landscape

comprising a mosaic of PAs and wildlife-friendly land uses. My results suggest that

elephants do not differentiate between PAs and wildlife-friendly land uses in their overall

use of a site. However, high-intensity use conditional on elephants using a given site

declined with increasing distance to PAs, and this effect was accentuated by an

accretion of villages. Therefore, while wildlife-friendly land uses did play a subsidiary

conservation role, their potential to substitute for PAs was offset by a strong human

presence. These findings demonstrate heterogeneity among land uses in the benefits

they offer to wildlife species, and the feasibility of quantifying this heterogeneity through

the use of robust yet easy to implement methods under an occupancy modeling

framework. I emphasize the need to evaluate the role of wildlife-friendly land uses in

landscape-scale conservation; for species that have conflicting resource requirements

with people, refuge from growing anthropogenic threats is likely to be provided only by

PAs. The effectiveness with which wildlife-friendly land uses can perform the same role

will hinge on the mitigation of factors that undermine their conservation potential.

Effective management of anthropogenic threats such as human-wildlife conflict

hinges on a clear understanding of its drivers. Depredation of cultivated crops by

elephants is the primary source of human-elephant conflict (HEC) in both Africa and

Asia (Williams, Johnsingh & Krausman 2001; Sukumar 2003; Sitati, Walpole & Leader-

Williams 2005). Therefore, in Chapter 3 I asked the question: what spatiotemporal

factors contribute to elephant crop depredation patterns in wildlife-friendly land uses

90

outside PAs? I addressed this question by applying dynamic occupancy models to

elephant crop depredation data collected between 2005 and 2011 from my study area.

This approach allowed me to estimate the occurrence of elephant crop depredation, and

also model the colonization and extinction of crop depredation events among seasons.

Moreover, I was able to account for imperfect detection and reporting of conflicts, a

factor that may bias inference on the spatiotemporal drivers of conflict, but has thus far

not been accounted for in human-wildlife conflict research. My results support the

possibility that the detection and reporting of human-wildlife conflicts may be imperfect,

and that it may vary with the accessibility of sampled sites. My findings also highlight

variation in crop depredation patterns, explained in part by key spatial and temporal

covariates. Distance to forest refuges negatively affected the probability of crop

depredation in areas that were not raided in the previous season (colonization), and

positively influenced the probability of crop depredation not occurring in areas raided by

elephants in the previous season (extinction). Both mean rainfall (lagged by two

months) and rainfall variation (also lagged by two months) had a negative effect on

extinction probabilities. These results point to the importance of both nutritional

incentives provided by crops, and natural forage limitations brought about by increased

rainfall variability, as potential determinants of elephant crop depredation patterns. In

addition, elephant crop depredation is likely modulated by crop availability and

accessibility. This novel application of occupancy models to human-wildlife conflict

research provides a flexible and robust approach to quantify and understand factors that

influence spatiotemporal patterns of conflict, while accounting for imperfect

detectability––a critical step for effective conflict management.

91

In Chapter 4, I focused on investigating whether human-induced mortality arising

due to human-wildlife conflict devalues the conservation potential of wildlife-friendly

lands outside PAs. I specifically asked the question: how might HEC-induced mortality

influence long-term persistence of elephant populations in a scenario where core habitat

within PAs is increasingly converted to multiple-use buffer habitat? I simulated elephant

population dynamics under different scenarios of conflict-induced mortality and

conversion of within-PA habitat to multiple-use buffer areas. Mortality rates induced by

HEC (HECm) adversely affected population persistence, and its detrimental effects

were magnified as the proportion of core habitat in PAs declined. Under moderate

HECm, small increments in mortality rates necessitated disproportionately large

increases in core area availability to avoid quasi-extinction. Furthermore, benefits of

buffer supplementation were driven more by buffer quality than size, and these benefits

declined as HECm increased. I emphasize the need for effective management of

human-wildlife conflict through mitigation measures, such as the implementation of

crop-raiding deterrents and participatory compensation schemes, while designing

multiple-use reserves for conflict-prone species.

Taken together, my PhD dissertation provides important empirical insights into

the challenges associated with the conservation conflict-prone but charismatic

megafauna in increasingly fragmented and human-dominated landscapes. To begin

with, my research underscores the irreplaceable role of PAs as refugia for such species.

Securing remnant PAs is therefore a necessary precondition to ensure the long-term

persistence of endangered, conflict-prone large mammals in human-dominated

landscapes. However, it is equally important for conservation prioritization of lands

92

outside PAs to meet the extensive habitat and resource requirements of these wide-

ranging species, while ensuring the livelihoods of local communities. I provide a

generalized approach to evaluate the conservation value of different land uses in

heterogeneous landscapes, and show that wildlife-friendly land uses can play an

important conservation role, albeit subsidiary to that of PAs. However, as demonstrated

by my research, the conservation potential of these land uses can only be realized if

anthropogenic threats prevalent in shared land outside PAs––such as human-wildlife

conflict––are effectively managed. I provide insights into potential mechanisms that

underlie human-wildlife conflicts in the context of elephant crop depredation. I also show

how small increments in conflict-induced mortality rates outside PAs can have large

negative impacts on long-term population viability of species like the Asian elephant.

Therefore, my research demonstrates that human-wildlife conflict not only devalues the

conservation potential of wildlife-friendly lands outside PAs, it also threatens the

persistence of endangered megafauna in human-dominated landscapes. As such, there

is need for future research on human-wildlife conflict, with an emphasis on further

investigating its context-specific drivers and experimentally testing the efficacy of

conflict-mitigation strategies over time. There is also need to evaluate the conservation

role of shared lands outside PAs in other landscapes, specifically in the context of other

‘wildlife-friendly’ land uses and conflict-prone species. Scientific endeavors of the kind,

consolidated across space, can shape conservation policy in a manner that ensures the

long-term persistence of endangered megafauna, such as the Asian elephant, in a

world fast changing.

93

APPENDIX A MODEL SELECTION FOR ELEPHANT SITE USE

Table A-1. Top-ranked multistate occupancy models used to evaluate elephant site use

Model K AICc ∆AICc wi

Analysis Set 2 Ψ1(FD) Ψ2(PAD × VD) p1(LU + RG) p2(LU + RG) δ(.) 15 1183.25 0.00 0.53 Ψ1(FD) Ψ2(PAD) p1(LU + RG) p2(LU + RG) δ(.) 13 1186.30 3.05 0.11 Ψ1(FD) Ψ2(LU + PAD) p1(LU + RG) p2(LU + RG) δ(.) 14 1186.64 3.39 0.10 Ψ1(FD) Ψ2(PAD + VD) p1(LU + RG) p2(LU + RG) δ(.) 14 1187.46 4.21 0.06 Ψ1(FD + VD) Ψ2(PAD) p1(LU + RG) p2(LU + RG) δ(.) 14 1187.81 4.56 0.05 Ψ1(FD × VD) Ψ2(PAD) p1(LU + RG) p2(LU + RG) δ(.)

15 1188.63 5.38 0.04

Analysis Set 3 Ψ1(FD) Ψ2(PAD × VD) p1(LU + RG) p2(LU) δ(RG) 15 1183.61 0.00 0.52 Ψ1(FD) Ψ2(PAD) p1(LU + RG) p2(LU) δ(RG) 13 1186.59 2.98 0.12 Ψ1(FD) Ψ2(LU + PAD) p1(LU + RG) p2(LU) δ(RG) 14 1186.90 3.29 0.10 Ψ1(FD) Ψ2(PAD + VD) p1(LU + RG) p2(LU) δ(RG) 14 1187.75 4.14 0.07 Ψ1(FD + VD) Ψ2(PAD) p1(LU + RG) p2(LU) δ(RG) 14 1188.11 4.50 0.05 Ψ1(FD × VD) Ψ2(PAD) p1(LU + RG) p2(LU) δ(RG) 15 1188.92 5.31 0.04

Probabilities of detecting low and high intensity use (p1 and p2, respectively), and the probability of observing high intensity use given elephant detection and actual high intensity use of a site (δ), were fixed to the second model in Table 2-1 (Analysis Set 2) and the third model in Table 2-1 (Analysis Set 3). Covariates for the probability of site use (Ψ1) and the probability of high intensity site use (Ψ2) included the independent, as well as additive and interactive effects of distance to forests (FD), distance to PAs (PAD), village density (VD) and land use (LU). Both (a) and (b) include the top six models from the two respective analyses. AICc represents Akaike’s information criterion corrected for small sample size; differences in AICc between each model and the most parsimonious model are denoted by ΔAICc. K is the number of parameters and wi is the AICc model weight. Model notation follows that of linear models: a × b includes additive and interactive effects of a and b, whereas a + b includes additive effects only. A model where parameter a was held constant is represented by a(.).

94

APPENDIX B

EFFECTS OF BEHAVIORAL ADAPTATIONS ON ELEPHANT POPULATION VIABILITY

A

B

C

D

Figure B-1. Effects of behavioral adaptations on the viability of Asian elephant

populations under scenarios of increasing HEC-induced mortality rates (HECm = 0.025, 0.05, 0.01 and 0.02; A through D). Adaptations of buffer avoidance and preferential selection of the buffer (potential ecological trap) are contrasted with a scenario of no positive or negative selection of the buffer.

100 90 80 70 60 50 40 30 20 10 00

500

1000

1500

2000

2500

3000

Equ

ilibrium

po

pula

tion s

ize

Percentage core area

HECm = 0.025

Avoidance of buffer

No selection

Ecological Trap

100 90 80 70 60 50 40 30 20 10 00

500

1000

1500

2000

2500

3000

Equili

briu

m p

opu

lation s

ize


HECm = 0.05

Avoidance of buffer

No selection

Ecological Trap

100 90 80 70 60 50 40 30 20 10 00

500

1000

1500

2000

2500

3000

Equili

brium

pop

ula

tion s

ize


HECm = 0.01

Avoidance of buffer

No selection

Ecological Trap

100 90 80 70 60 50 40 30 20 10 00

500

1000

1500

2000

2500

3000

Equ

ilibrium

po

pula

tion s

ize


HECm = 0.02

Avoidance of buffer

No selection

Ecological Trap

95

LIST OF REFERENCES

Adams, W.M., Aveling, R., Brockington, D., Dickson, B., Elliott, J., Hutton, J., Roe, D., Vira, B. & Wolmer, W. (2004) Biodiversity conservation and the eradication of poverty. Science, 306, 1146-1149.

Adams, W.M. & Hutton, J. (2007) People, parks and poverty: political ecology and biodiversity conservation. Conservation and Society, 5, 147-183.

Alfred, R., Ahmad, A.H., Payne, J., Williams, C., Ambu, L.N., How, P.M. & Goossens, B. (2012) Home range and ranging behaviour of Bornean elephant (Elephas maximus borneensis) females. PLoS ONE, 7, e31400.

Armbruster, P., Fernando, P. & Lande, R. (1999) Time frames for population viability analysis of species with long generations: an example with Asian elephants. Animal Conservation, 2, 69-73.

Armbruster, P. & Lande, R. (1993) A population viability analysis for African Elephant (Loxodonta africana): how big should reserves be? Conservation Biology, 7, 602-610.

Bali, A., Kumar, A. & Krishnaswamy, J. (2007) The mammalian communities in coffee plantations around a protected area in the Western Ghats, India. Biological Conservation, 139, 93-102.

Balme, G.A., Slotow, R. & Hunter, L.T.B. (2010) Edge effects and the impact of non-protected areas in carnivore conservation: leopards in the Phinda–Mkhuze Complex, South Africa. Animal Conservation, 13, 315-323.

Balmford, A., Green, R. & Phalan, B. (2012) What conservationists need to know about farming. Proceedings of the Royal Society Biological Sciences Series B, 279, 2714-2724.

Berkes, F. (2007) Community-based conservation in a globalized world. Proceedings of the National Academy of Sciences of the United States of America, 104, 15188-15193.

Blake, S. & Hedges, S. (2004) Sinking the flagship: the case of forest elephants in Asia and Africa. Conservation Biology, 18, 1191-1202.

Blake, S., Strindberg, S., Boudjan, P., Makombo, C., Bila-Isia, I., Ilambu, O., Grossmann, F., Bene-Bene, L., de Semboli, B., Mbenzo, V., S’hwa, D., Bayogo, R., Williamson, L., Fay, M., Hart, J. & Maisels, F. (2007) Forest elephant crisis in the Congo Basin. PLoS Biology, 5, e111.

Brashares, J.S., Arcese, P. & Sam, M.K. (2001) Human demography and reserve size predict wildlife extinction in West Africa. Proceedings of the Royal Society of London B Biological Sciences, 268, 2473-2478.

96

Brooks, T.M., Mittermeier, R.A., Mittermeier, C.G., da Fonseca, G.A.B., Rylands, A.B., Konstant, W.R., Flick, P., Pilgrim, J., Oldfield, S., Magin, G. & Hiton-Taylor, C. (2002) Habitat loss and extinction in the hotspots of biodiversity. Conservation Biology, 16, 910-923.

Bruner, A.G., Gullison, R.E., Rice, R.E. & da Fonseca, G.A.B. (2001) Effectiveness of parks in protecting tropical biodiversity. Science, 291, 125-128.

Burnham, K.P. & Anderson, D.R. (2002) Model Selection and Inference: A Practical Information-Theoretic Approach, 2nd edn. Springer-Verlag, New York, NY, USA.

Cardillo, M., Mace, G.M., Jones, K.E., Bielby, J., Bininda-Emonds, O.R.P., Sechrest, W., Orme, C.D.L. & Purvis, A. (2005) Multiple causes of high extinction risk in large mammal species. Science, 309, 1239–1241.

Cardillo, M., Purvis, A., Sechrest, W., Gittleman, J.L., Bielby, J. & Mace, G.M. (2004) Human population density and extinction risk in the world’s carnivores. PLoS Biology, 2, e197.

Carter, N.H., Shrestha, B.K., Karki, J.B., Pradhan, N.M.B. & Liu, J. (2012) Coexistence between wildlife and humans at fine spatial scales. Proceedings of the National Academy of Sciences of the United States of America, 109, 15360–15365.

Caswell, H. (2001) Matrix population models: construction, analysis and interpretation, Second edition edn. Sinauer Associates, Inc., Sunderland, Massachusetts.

Chartier, L., Zimmermann, A. & Ladle, R.J. (2011) Habitat loss and human-elephant conflict in Assam, India: does a critical threshold exist? Oryx, 45, 528-533.

Ciuti, S., Northrup, J.M., Muhly, T.B., Simi, S., Musiani, M., Pitt, J.A. & Boyce, M.S. (2012) Effects of humans on behaviour of wildlife exceed those of natural predators in a landscape of fear. PLoS ONE, 7, e50611.

Daily, G.C., Ehrlich, P.R. & Sanchez-Azofeifa, G.A. (2001) Countryside biogeography: use of human-dominated habitats by the avifauna of southern Costa Rica. Ecological Applications, 11, 1-13.

Datta-Roy, A., Ved, N. & Williams, A.C. (2009) Participatory elephant monitoring in South Garo Hills: efficacy and utility in a human-animal conflict scenario. Tropical Ecology, 50, 163-171.

Davies, T.E., Wilson, S., Hazarika, N., Chakrabarty, J., Das, D., Hodgson, D.J. & Zimmermann, A. (2011) Effectiveness of intervention methods against crop-raiding elephants. Conservation Letters, 4, 346-354.

DeFries, R., Hansen, A., Newton, A.C. & Hansen, M.C. (2005) Increasing isolation of protected areas in tropical forests over the past twenty years. Ecological Applications, 15, 19-26.

97

Dickman, A.J. (2010) Complexities of conflict: the importance of considering social factors for effectively resolving human–wildlife conflict. Animal Conservation, 13, 458-466.

Dickman, A.J., Macdonald, E.A. & Macdonald, D.W. (2011) A review of financial instruments to pay for predator conservation and encourage human-carnivore coexistence. Proceedings of the National Academy of Sciences of the United States of America, 108, 13937–13944.

Easterling, D.R., Meehl, G.A., Parmesan, C., Changnon, S.A., Karl, T.R. & Mearns, L.O. (2000) Climate extremes: observations, modeling, and impacts. Science, 289, 2068-2074.

Ehrlich, P.R. (1995) The scale of the human enterprise and biodiversity loss. Extinction rates (eds J.H. Lawton & R.M. May), pp. 214-226. Oxford University Press, Oxford.

Ehrlich, P.R. & Pringle, R.M. (2008) Where does biodiversity go from here? A grim business-as-usual forecast and a hopeful portfolio of partial solutions. Proceedings of the National Academy of Sciences of the United States of America, 105, 11579 –11586.

Engen, S., Lande, R. & Saether, B.E. (2002) The spatial scale of population fluctuations and quasi-extinction risk. American Naturalist, 160, 439-451.

Fahrig, L. (2001) How much habitat is enough? Biological Conservation, 100, 65-74.

Fahrig, L. (2003) Effects of habitat fragmentation on biodiversity. Annual Review of Ecology Evolution and Systematics, 34, 487-515.

Fernando, P. (2006) Elephant Conservation in Sri Lanka: Integrating Scientific Information to Guide Policy. Principles of Conservation Biology (eds M.J. Groom, G.K. Meffe & C.R. Carroll). Sinauer Associates, Sunderland, MA, USA.

Fernando, P. & Lande, R. (2000) Molecular genetic and behavioral analysis of social organization in the Asian elephant (Elephas maximus). Behavioral Ecology and Sociobiology, 48, 84-91.

Fernando, P. & Leimgruber, P. (2011) Asian Elephants and Seasonally Dry Forests. The Ecology and Conservation of Seasonally Dry Forests in Asia (eds W.J. McShea, S.J. Davies & N. Bhumpakphan). Smithsonian Institution Scholarly Press, Washington D.C., USA.

Fernando, P., Wikrarnanayake, E.D., Janaka, H.K., Jayasinghe, L.K.A., Gunawardena, M., Kotagama, S.W., Weerakoon, D. & Pastorini, J. (2008) Ranging behavior of the Asian elephant in Sri Lanka. Mammalian Biology, 73, 2-13.

98

Ferraro, P.J. (2001) Global habitat protection: limitations of development interventions and a role for conservation performance payments. Conservation Biology, 15, 990–1000.

Ferraro, P.J. & Pattanayak, S.K. (2006) Money for nothing? A call for empirical evaluation of biodiversity conservation investments. PLoS Biology, 4, 482-488.

Fischer, J., Brosi, B., Daily, G.C., Ehrlich, P.R., Goldman, R., Goldstein, J., Lindenmayer, D.B., Manning, A.D., Mooney, H.A., Pejchar, L., Ranganathan, J. & Tallis, H. (2008) Should agricultural policies encourage land sparing or wildlife-friendly farming? Frontiers in Ecology and the Environment, 6, 380–385.

Flather, C.H. & Bevers, M. (2002) Patchy reaction-diffusion and population abundance: The relative importance of habitat amount and arrangement. American Naturalist, 159, 40-56.

Frankham, R. (1995) Inbreeding and extinction - a threshold effect. Conservation Biology, 9, 792-799.

Goswami, V.R., Getz, L.L., Hostetler, J.A., Ozgul, A. & Oli, M.K. (2011) Synergistic influences of phase, density, and climatic variation on the dynamics of fluctuating populations. Ecology, 92, 1680-1690.

Goswami, V.R., Lauretta, M.V., Madhusudan, M.D. & Karanth, K.U. (2012) Optimizing individual identification and survey effort for photographic capture–recapture sampling of species with temporally variable morphological traits. Animal Conservation, 15, 174-183.

Goswami, V.R., Madhusudan, M.D. & Karanth, K.U. (2007) Application of photographic capture-recapture modelling to estimate demographic parameters for male Asian elephants. Animal Conservation, 10, 391-399.

Goswami, V.R., Vasudev, D., Karnad, D., Krishna, Y.C., Krishnadas, M., Pariwakam, M., Nair, T., Andheria, A., Sridhara, S. & Siddiqui, I. (2013) Conflict of human–wildlife coexistence. Proceedings of the National Academy of Sciences of the United States of America, 110, E108.

Government of India (2005) Elephant Reserves with estimated population in India. Project Elephant Directorate, Ministry of Environment and Forest, New Delhi. Available from http://moef.nic.in/pe/ELEPHANTRESERVESPOPULATION .pdf (accessed May 2012).

Government of India (2012) Protected area network in India. Ministry of Environment and Forest, New Delhi. Available from http://moef.nic.in/downloads/public-information/protected-area-network.pdf (accessed May 2012).

99

Graham, M.D., Douglas-Hamilton, I., Adams, W.M. & Lee, P.C. (2009) The movement of African elephants in a human-dominated land-use mosaic. Animal Conservation, 12, 445-455.

Green, R.E., Cornell, S.J., Scharlemann, J.P.W. & Balmford, A. (2005) Farming and the fate of wild nature. Science, 307, 550-555.

Gubbi, S. (2012) Patterns and correlates of human–elephant conflict around a south Indian reserve. Biological Conservation, 148, 88-95.

Guerbois, C., Chapanda, E. & Fritz, H. (2012) Combining multi-scale socio-ecological approaches to understand the susceptibility of subsistence farmers to elephant crop raiding on the edge of a protected area. Journal of Applied Ecology, 49, 1149-1158.

Hansen, A. & DeFries, R. (2007) Ecological mechanisms linking protected areas to surrounding lands. Ecological Applications, 17, 974-988.

Harihar, A., Chanchani, P., Sharma, R.K., Vattakaven, J., Gubbi, S., Pandav, B. & Noon, B. (2013) Conflating “co-occurrence” with “coexistence”. Proceedings of the National Academy of Sciences of the United States of America, 110, E109.

Hjortsø, C.N., Stræde, S. & Helles, F. (2006) Applying multi-criteria decision-making to protected areas and buffer zone management: A case study in the Royal Chitwan National Park, Nepal. Journal of Forest Economics, 12, 91-108.

Hoare, R.E. (1999) Determinants of human-elephant conflict in a land-use mosaic. Journal of Applied Ecology, 36, 689-700.

Inchausti, P. & Halley, J. (2003) On the relation between temporal variability and persistence time in animal populations. Journal Of Animal Ecology, 72, 899-908.

Karanth, K.K., Gopalaswamy, A.M., Prasad, P.K. & Dasgupta, S. (2013a) Patterns of human–wildlife conflicts and compensation: insights from Western Ghats protected areas. Biological Conservation, 166, 175-185.

Karanth, K.K., Nichols, J.D., Hines, J.E., Karanth, K.U. & Christensen, N.L. (2009) Patterns and determinants of mammal species occurrence in India. Journal of Applied Ecology, 46, 1189–1200.

Karanth, K.K., Nichols, J.D., Karanth, K.U., Hines, J.E. & Christensen Jr., N.L. (2010) The shrinking ark: patterns of large mammal extinctions in India. Proceedings of the Royal Society Biological Sciences Series B, 277, 1971–1979.

Karanth, K.U., Gopalaswamy, A.M., Karanth, K.K., Goodrich, J., Seidensticker, J. & Robinson, J.G. (2013b) Sinks as saviors: Why flawed inference cannot assist tiger recovery. Proceedings of the National Academy of Sciences of the United States of America, 110, E110.

100

Karanth, K.U., Gopalaswamy, A.M., Kumar, N.S., Vaidyanathan, S., Nichol, J.D. & MacKenzie, D.I. (2011) Monitoring carnivore populations at the landscape scale: occupancy modelling of tigers from sign surveys. Journal Of Applied Ecology, 48, 1048-1056.

Karanth, K.U. & Sunquist, M.E. (2000) Behavioural correlates of predation by tiger (Panthera tigris), leopard (Panthera pardus) and dhole (Cuon alpinus) in Nagarahole, India. Journal Of Zoology, 250, 255-265.

Kendall, W.L. & White, G.C. (2009) A cautionary note on substituting spatial subunits for repeated temporal sampling in studies of site occupancy. Journal of Applied Ecology, 46, 1182-1188.

Kéry, M., Royle, J.A., Schmid, H., Schaub, M., Volet, B., Häfliger, G. & Zbinden, N. (2010) Site-occupancy distribution modeling to correct population-trend estimates derived from opportunistic observations. Conservation Biology, 24, 1388-1397.

Kinnaird, M.F. & O’Brien, T.G. (2012) Effects of private-land use, livestock management, and human tolerance on diversity, distribution, and abundance of large African mammals. Conservation Biology, 26, 1026-1039.

Knapp, A.K., Fay, P.A., Blair, J.M., Collins, S.L., Smith, M.D., Carlisle, J.D., Harper, C.W., Danner, B.T., Lett, M.S. & McCarron, J.K. (2002) Rainfall variability, carbon cycling, and plant species diversity in a mesic grassland. Science, 298, 2202-2205.

Kumar, M.A., Mudappa, D. & Raman, T.R.S. (2010) Asian elephant Elephas maximus habitat use and ranging in fragmented rainforest and plantations in the Anamalai Hills, India. Tropical Conservation Science, 3, 143-158.

Laake, J.L. & Rexstad, E.A. (2007) RMark. Program MARK: a gentle introduction (eds E.G. Cooch & G.C. White). http://www.phidot.org/software/mark/docs/book

Lande, R. (1987) Extinction thresholds in demographic models of territorial populations. American Naturalist, 130, 624-635.

Lande, R. (1988) Genetics and demography in biological conservation. Science, 241, 1455-1460.

Leimgruber, P., Gagnon, J.B., Wemmer, C., Kelly, D.S., Songer, M.A. & Selig, E.R. (2003) Fragmentation of Asia's remaining wildlands: implications for Asian elephant conservation. Animal Conservation, 6, 347-359.

MacKenzie, D.I. (2005) What are the issues with presence-absence data for wildlife managers? Journal of Wildlife Management, 69, 849-860.

101

MacKenzie, D.I., Nichols, J.D., Hines, J.E., Knutson, M.G. & Franklin, A.B. (2003) Estimating site occupancy, colonization, and local extinction when a species is detected imperfectly. Ecology, 84, 2200-2207.

MacKenzie, D.I., Nichols, J.D., Lachman, G.B., Droege, S., Royle, J.A. & Langtimm, C.A. (2002) Estimating site occupancy rates when detection probabilities are less than one. Ecology, 83, 2248-2255.

MacKenzie, D.I., Nichols, J.D., Royle, J.A., Pollock, K.H., Bailey, L.L. & Hines, J.E. (2006) Occupancy Estimation and Modeling: Inferring Patterns and Dynamics of Species Occurrence. Elsevier San Diego.

Mackinnon, J.R., Mackinnon, K., Child, G. & Thorsell, J. (1986) Managing protected areas in the tropics IUCN, Gland, Switzerland.

Madden, F. (2004) Creating coexistence between humans and wildlife: global perspectives on local efforts to address human–wildlife conflict. Human Dimensions of Wildlife, 9, 247-257.

Madhusudan, M.D. (2003) Living amidst large wildlife: livestock and crop depredation by large mammals in the interior villages of Bhadra Tiger Reserve, south India. Environmental Management, 31, 466-475.

Madhusudan, M.D. & Mishra, C. (2003) Why big, fierce animals are threatened: conserving large mammals in densely populated landscapes. Battles over nature: the science and politics of conservation in India (eds M. Rangarajan & V. Saberwal), pp. 31-55. Permanent Black, New Delhi.

Madhusudan, M.D. & Raman, T.R.S. (2003) Conservation as if biological diversity matters: Preservation versus sustainable use in India. Conservation and Society, 1, 49-59.

Margules, C. & Pressey, R. (2000) Systematic conservation planning. Nature, 405, 243-253.

Martin, J., Chamaillé-Jammes, S., Nichols, J.D., Fritz, H., Hines, J.E., Fonnesbeck, C.J., MacKenzie, D.I. & Bailey, L.L. (2010) Simultaneous modeling of habitat suitability, occupancy, and relative abundance: African elephants in Zimbabwe. Ecological Applications, 20, 1173-1182.

McNeely, J.A. (1993) Parks for life: report of the IVth World Congress on National Parks and Protected Areas. IUCN, Gland, Switzerland.

Mills, L.S. (2012) Conservation of Wildlife Populations: Demography, Genetics, and Management (2nd Ed.). Wiley / Blackwell Publishers, Sussesx, UK.

102

Mishra, C., Allen, P., McCarthy, T., Madhusudan, M.D., Bayarjargal, A. & Prins, H.H.T. (2003) The role of incentive programs in conserving the snow leopard. Conservation Biology, 17, 1512-1520.

Myers, N., Mittermeier, R.A., Mittermeier, C.G., da Fonseca, G.A.B. & Kent, J. (2000) Biodiversity hotspots for conservation priorities. Nature, 403, 853-858.

Naughton-Treves, L. (1998) Predicting patterns of crop damage by wildlife around Kibale National Park, Uganda. Conservation Biology, 12, 156-168.

Naughton-Treves, L., Grossberg, R. & Treves, A. (2003) Paying for tolerance: rural citizens' attitudes toward wolf depredation and compensation. Conservation Biology, 17, 1500-1511.

Newby, J.R., Mills, L.S., Ruth, T.K., Pletscher, D.H., Mitchell, M.S., Quigley, H.B., Murphy, K.M. & DeSimone, R. (2013) Human-caused mortality influences spatial population dynamics: pumas in landscapes with varying mortality risks. Biological Conservation, 159, 230-239.

Nichols, J.D., Hines, J.E., MacKenzie, D.I., Seamans, M.E. & Gutierrez, R.J. (2007) Occupancy estimation and modeling with multiple states and state uncertainty. Ecology, 88, 1395-1400.

Norris, K. (2008) Agriculture and biodiversity conservation: opportunity knocks. Conservation Letters, 1, 2-11.

Noss, R.F. (1983) A regional landscape approach to maintain diversity. Bioscience, 33, 700-706.

Olivier, R.C.D. (1978) On the ecology of the Asian elephant. PhD, University of Cambridge.

Osborn, F.V. (2004) Seasonal variation of feeding patterns and food selection by crop-raiding elephants in Zimbabwe. African Journal of Ecology, 42, 322-327.

Ostfeld, R.S. & Keesing, F. (2000) Pulsed resources and community dynamics of consumers in terrestrial ecosystems. Trends in Ecology & Evolution, 15, 232-237.

Pillay, R., Johnsingh, A.J.T., Raghunath, R. & Madhusudan, M.D. (2011) Patterns of spatiotemporal change in large mammal distribution and abundance in the southern Western Ghats, India. Biological Conservation, 144, 1567-1576.

Pimm, S.L., Jones, H.L. & Diamond, J.M. (1988) On the risk of extinction. American Naturalist, 132, 757-785.

Pollock, K.H. (1982) A capture–recapture design robust to unequal probability of capture. Journal of Wildlife Management, 46, 752-757.

103

Pollock, K.H., Nichols, J.D., Brownie, C. & Hines, J.E. (1990) Statistical inference for capture-recapture experiments. Wildlife Monographs, 107, 3-97.

R Development Core Team (2008) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria.

Ramakrishnan, P.S. (1992) Shifting agriculture and sustainable development: an interdisciplinary study from north-eastern India. United Nations Educational, Scientific, and Cultural Organization, Paris.

Raman, T.R.S. (2001) Effect of slash-and-burn shifting cultivation on rainforest birds in Mizoram, northeast India. Conservation Biology, 15, 685-698.

Raman, T.R.S., Rawat, G.S. & Johnsingh, A.J.T. (1998) Recovery of tropical rainforest avifauna in relation to vegetation succession following shifting cultivation in Mizoram, north-east India. Journal of Applied Ecology, 35, 214-231.

Reed, D.H. & Frankham, R. (2003) Correlation between fitness and genetic diversity. Conservation Biology, 17, 230-237.

Ricketts, T.H. (2001) The matrix matters: effective isolation in fragmented landscapes. American Naturalist, 158, 87-99.

Robertson, B.A. & Hutto, R.L. (2006) A framework for understanding ecological traps and an evaluation of existing evidence Ecology, 87, 1075-1085.

Robinson, J.G. (1993) The limits to caring - sustainable living and the loss of biodiversity. Conservation Biology, 7, 20-28.

Robinson, J.G. & Bennett, E.L. (2000) Hunting for sustainability in tropical forests. Columbia University Press, New York.

Saberwal, V., Rangarajan, M. & Kothari, A. (2001) People, parks and coexistence. Orient Longman Hyderabad, Andhra Pradesh, India.

Saberwal, V.K., Gibbs, J.P., Chellam, R. & Johnsingh, A.J.T. (1994) Lion-human conflict in the Gir forest, India. Conservation Biology, 8, 501-507.

Saccheri, I., Kuussaari, M., Kankare, M., Vikman, P., Fortelius, W. & Hanski, I. (1998) Inbreeding and extinction in a butterfly metapopulation. Nature, 392, 491-494.

Salafsky, N., Cauley, H., Balachander, G., Cordes, B., Parks, J., Margoluis, C., Bhatt, S., Encarnacion, C., Russel, D. & Margoluis, R. (2001) A systematic test of an enterprise strategy for community-based biodiversity conservation. Conservation Biology, 15, 1585-1595.

Sanderson, E.W., Jaiteh, M., Levy, M.A., Redford, K.H., Wannebo, A.V. & Woolmer, G. (2002a) The human footprint and the last of the wild. Bioscience, 52, 891-904.

104

Sanderson, E.W., Redford, K.H., Chetkiewicz, C.L.B., Medellin, R.A., Rabinowitz, A.R., Robinson, J.G. & Taber, A.B. (2002b) Planning to save a species: the jaguar as a model. Conservation Biology, 16, 58-72.

Sanderson, E.W., Redford, K.H., Vedder, A., Coppolillo, P.B. & Ward, S.E. (2002c) A conceptual model for conservation planning based on landscape species requirements. Landscape and Urban Planning, 58, 41-56.

Saunders, D.A., Hobbs, R.J. & Margules, C.R. (1991) Biological consequences of ecosystem fragmentation: a review. Conservation Biology, 5, 18-32.

Scherr, S.J. & McNeely, J.A. (2008) Biodiversity conservation and agricultural sustainability: towards a new paradigm of ‘ecoagriculture’ landscapes. Philosophical Transactions of the Royal Society of London B Biological Sciences, 363, 477-494.

Sitati, N.W., Walpole, M.J. & Leader-Williams, N. (2005) Factors affecting susceptibility of farms to crop raiding by African elephants: using a predictive model to mitigate conflict. Journal of Applied Ecology, 42, 1175-1182.

Sitati, N.W., Walpole, M.J., Smith, R.J. & Leader-Williams, N. (2003) Predicting spatial aspects of human-elephant conflict. Journal of Applied Ecology, 40, 667-677.

Stokes, E. (2012) Assessing threats and monitoring law enforcement. Monitoring Elephant Populations and Assessing Threats: a manual for researchers, managers and conservationists (ed. S. Hedges). Universities Press (India), Hyderabad, India.

Stokes, E.J., Strindberg, S., Bakabana, P.C., Elkan, P.W., Iyenguet, F.C., Madzoké, B., Malanda, G.A.F., Mowawa, B.S., Moukoumbou, C., Ouakabadio, F.K. & Rainey, H.J. (2010) Monitoring great ape and elephant abundance at large spatial scales: Measuring effectiveness of a conservation landscape. PLoS ONE, 5, e10294.

Sukumar, R. (1989) The Asian elephant: ecology and management. Cambridge University Press, Cambridge, UK.

Sukumar, R. (2003) The Living Elephants: Evolutionary Ecology, Behavior and Conservation. Oxford University Press, New York.

Sukumar, R. & Gadgil, M. (1988) Male-female differences in foraging on crops by Asian elephants. Animal Behaviour, 36, 1233-1235.

Sukumar, R., Ramakrishnan, U. & Santosh, J.A. (1998) Impact of poaching on an Asian elephant population in Periyar, southern India: a model of demography and tusk harvest. Animal Conservation, 1, 281-291.

Sutherland, W.J., Pullin, A.S., Dolman, P.M. & Knight, T.M. (2004) The need for evidence-based conservation. Trends in Ecology & Evolution, 19, 305-308.

105

Szabo, J.K., Vesk, P.A., Baxter, P.W.J. & Possingham, H.P. (2010) Regional avian species declines estimated from volunteer-collected long-term data using List Length analysis. Ecological Applications, 20, 2157-2169.

Tilman, D., May, R.M., Lehman, C.L. & Nowak, M.A. (1994) Habitat destruction and the extinction debt. Nature, 371, 65-66.

Tiwari, S.K., Karyong, S.S., Sarkar, P., Choudhary, A. & Williams, A.C. (2005) Elephant corridors in north-eastern India. Right of passage: elephant corridors of India (eds V. Menon, S.K. Tiwari, P.S. Easa & R. Sukumar), pp. 157. Wildlife Trust of India, New Delhi.

Treves, A. & Karanth, K.U. (2003) Human-carnivore conflict and perspectives on carnivore management worldwide. Conservation Biology, 17, 1491-1499.

Treves, A., Wallace, R.B. & White, S. (2009) Participatory planning of interventions to mitigate human–wildlife conflicts. Conservation Biology, 23, 1577-1587.

van Shaik, C.P., Terborgh, J.W. & Wright, S.J. (1993) The phenology of tropical forests: adaptive significance and consequences for primary consumers. Annual Review of Ecology and Systematics, 24, 353-377.

van Strien, A.J., van Swaay, C.A.M. & Termaat, T. (2013) Opportunistic citizen science data of animal species produce reliable estimates of distribution trends if analysed with occupancy models. Journal of Applied Ecology, doi: 10.1111/1365-2664.12158.

Venkataraman, A.B., Saandeep, R., Baskaran, N., Roy, M., Madhivanan, A. & Sukumar, R. (2005) Using satellite telemetry to mitigate elephant-human conflict: An experiment in northern West Bengal, India. Current Science, 88, 1827-1831.

Vitousek, P.M., Mooney, H.A., Lubchenco, J. & Melillo, J.M. (1997) Human domination of Earth’s ecosystems. Science, 277, 494-499.

Weltzin, J.F., Loik, M.E., Schwinning, S., Williams, D.G., Fay, P.A., Haddad, J.H., Huxman, T.E., Knapp, A.K., Lin, G., Pockman, W.T., Shaw, M.R., Small, E.E., Smith, M.D., Smith, S.D., Tissue, D.T. & Zak, J.C. (2003) Assessing the response of terrestrial ecosystems to potential changes in precipitation. Bioscience, 53, 941-952.

Westemeier, R.L., Brawn, J.D., Simpson, S.A., Esker, T.L., Jansen, R.W., Walk, J.W., Kershner, E.L., Bouzat, J.L. & Paige, K.N. (1998) Tracking the long-term decline and recovery of an isolated population. Science, 282, 1695-1698.

Western, D., Russel, S. & Cuthill, I. (2009) The status of wildlife in protected areas compared to non-protected areas of Kenya. PLoS ONE, 4, e6140.

106

White, G.C. & Burnham, K.P. (1999) Program MARK: survival estimation from populations of marked animals. Bird Study, 46 Supplement, 120-138.

Wikramanayake, E., McKnight, M., Dinerstein, E., Joshi, A., Gurung, B. & Smith, D. (2004) Designing a conservation landscape for tigers in human-dominated environments. Conservation Biology, 18, 839-844.

Williams, A.C. (2002) Elephants (Elephas maximus), their habitats in Rajaji-Corbett National Parks, Northwest India. PhD, Saurashtra University.

Williams, A.C., Johnsingh, A.J.T. & Krausman, P.R. (2001) Elephant-human conflicts in Rajaji National Park, northwestern India. Wildlife Society Bulletin, 29, 1097-1104.

Williams, B.K., Nichols, J.D. & Conroy, M.J. (2002) Analysis and management of animal populations. Academic Press, San Diego, California, USA.

Wittemyer, G., Elsen, P., Bean, W.T., Burton, A.C.O. & Brashares, J.S. (2008) Accelerated human population growth at protected area edges. Science, 321, 123-126.

Woodroffe, R., Davies-Mostert, H., Ginsberg, J., Graf, J., Leigh, K., McCreery, K., Mills, G., Pole, A., Rasmussen, G., Robbins, R., Somers, M. & Szykman, M. (2007) Rates and causes of mortality in endangered African wild dogs Lycaon pictus: lessons for management and monitoring. Oryx, 41, 215-223.

Woodroffe, R. & Frank, L.G. (2005) Lethal control of African lions (Panthera leo): local and regional population impacts. Animal Conservation, 8, 91-98.

Woodroffe, R. & Ginsberg, J.R. (1998) Edge effects and the extinction of populations inside protected areas. Science, 280, 2126-2128.

Woodroffe, R., Thirgood, S. & Rabinowitz, A. (2005) People and Wildlife: Conflict or Coexistence? . Cambridge University Press, Cambridge, UK.

107

BIOGRAPHICAL SKETCH

Varun R. Goswami obtained a Bachelor of Science (B.Sc.) degree in zoology

from Delhi University, India. Thereafter, he pursued a Master of Science (M.Sc.) degree

in wildlife biology and conservation at the National Centre for Biological Sciences

(NCBS), Bangalore, India. The M.Sc. program was initiated by the Wildlife Conservation

Society – India Program, Centre for Wildlife Studies in collaboration with NCBS and

National Institute of Advanced Studies, and the degree was awarded by Manipal

Academy of Higher Education, India. Varun began his PhD in Interdisciplinary Ecology

at the University of Florida in August 2008 and graduated in December 2013.

Documents

OF POPULATIONS, HABITAT AND PEOPLE: THE ASIAN … · 2014-04-02 · rendition of our life as children, depicted through the story of Pitki and Dondi. My mother deserves special mention,