Towards Human-Leopard Coexistence in Sri Lanka: Social and Ecological Dimensions

Towards human-leopard coexistence in Sri Lanka: Social and ecological dimensions

Aisha Uduman

B.A., The University of British Columbia, 2015

A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF

MASTER OF SCIENCE

THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES

(Forestry)

THE UNIVERSITY OF BRITISH COLUMBIA

(Vancouver)

December 2019

The following individuals certify that they have read, and recommend to the Faculty of Graduate and Postdoctoral Studies for acceptance, a thesis/dissertation entitled:

Towards human-leopard coexistence in Sri Lanka: Social and ecological dimensions

submitted in partial fulfillment of the by Aisha Uduman requirements for the degree of Master of Science in Forestry

Examining Committee:

Dr. Cole Burton, Forestry Co-supervisor

Dr. Andrew Kittle, Wilderness and Wildlife Conservation Trust Co-supervisor

Dr. Shannon Hagerman, Forestry Supervisory Committee Member

Dr. Robin Naidoo, Institute of Resources, Environment and Sustainability Additional Examiner

Abstract

A leading cause of large carnivore declines is conflict with humans, specifically due to livestock depredation. This conflict threatens both carnivore populations and human communities with livestock-dependent livelihoods. The expanding dairy industry in Sri Lanka, home to the endangered Sri Lankan leopard (Panthera pardus kotiya), provides an opportunity for proactive conflict mitigation. Little is known about incidents of livestock depredation or attitudes of pastoralists whose livelihoods may be threatened by leopard conflict. This thesis aimed to combine social and ecological research methods to address these knowledge gaps.

We surveyed two pastoralist communities that differed in leopard conflict and socioecological factors (Yala and Central Hills) to identify determinants of attitudes towards leopards. We conducted Exploratory Factor Analysis and ran generalised linear models (GLMs) to detect influential variables. In the higher conflict region (Yala), attitudes towards leopards were positively related to socio-demographics (age, number of dependents, years rearing livestock) and an overall desire for wildlife conservation, while attitudes were negatively related to general awareness of leopard ecology and leopard- related tourism. In the lower conflict region (Central Hills), attitudes were positively related to a desire for increased government assistance in cattle rearing. The inability to own land were common concerns for pastoralists in both regions. We recommend assessing programs that may improve attitudes towards leopards, such as involving pastoralists in tourism programs and restricted land ownership. While the Central Hills is currently not experiencing depredation, proactively addressing hardships (e.g. improving roads, subsidizing cattle feed) may facilitate positive attitudes, should incidents of conflict increase.

To investigate potential drivers of cattle depredation in Yala, we used GLMs to test the importance of hypothesised explanatory variables, specifically: native prey availability, cattle husbandry, number of cattle, distance from national park, road density and pastoralist residency time. Model results indicated that depredation frequency increased with the number of cattle and decreased with improved husbandry. Survey responses suggested that

iii stronger cattle enclosures using plastic and light deterrents were husbandry techniques most supported by pastoralists. We recommend testing their efficacy and feasibility.

This thesis illustrates the importance of interdisciplinary research to better inform human-carnivore coexistence grounded in the local context.

iv Lay summary

Large carnivore species are declining globally, with a leading cause of declines being conflict between carnivores, people, and their livestock. This particularly affects rural communities, whose livelihoods rely on livestock and may be threatened by large carnivores. Balancing economic development with carnivore conservation requires knowledge of human communities at the forefront of this conflict, and knowledge of the ecological factors that create conditions for conflict to exist. This research focuses on two communities in Sri Lanka that rear livestock in landscapes shared with endangered Sri Lankan leopards (Panthera pardus kotiya), and aims to i) identify determinants of attitudes towards leopards and understand key issues that may inform support for leopard conservation, and ii) model factors associated with instances of depredation and evaluate the support for different mitigation options. By incorporating both the social and ecological dimensions, we provide a holistic approach to informing human-carnivore coexistence in this tropical biodiversity hotspot.

v Preface

This thesis uses data collected between May and August 2018 and June and August 2019 in Sri Lanka. Andrew Kittle and Anjali Watson aided in project design, acquiring necessary permissions and permits provided extensive local context that helped to ground this research. Drs. Cole Burton, Andrew Kittle, Shannon Hagerman, Edward Kroc advised on study design and statistical analyses and provided helpful feedback and comments on thesis drafts. The WildCo lab (past and present) also provided comments and performed reviews of various thesis sections. This work is collaborative in nature, therefore this thesis is written with the first person “we”. Unless otherwise credited, all photos included in this thesis were taken by Aisha Uduman.

Chapter 2 is based on the results from our surveys and interviews conducted with two pastoralist communities in Sri Lanka. Fieldwork in the Central Hills was facilitated by the Wilderness and Wildlife Conservation Trust, who had existing relationships with many tea estate superintendents, whose permission is required in order to interview residents. UBC Behavioural Research Ethics Board approval (H18-01121) was obtained prior to any surveys being conducted. This chapter will be submitted as a paper co-authored by Cole Burton, Andrew Kittle, Shannon Hagerman, Edward Kroc and Anjali Watson. Analysis was conducted with the guidance of co-authors, and drafts were improved upon following comments from co-authors.

Chapter 3 is based on data obtained from camera traps set up between May and August 2018 and June and August 2019, and data gathered from surveys and farm observations. Permits to set up camera traps were obtained from the Department of Wildlife Conservation (WL/3/2/1/4/1) under the Wilderness and Wildlife Conservation Trust, Sri Lanka. UBC Animal Care approval (A18-0033) was obtained prior to camera set up. This chapter will be subsequently submitted for publication as a paper co-authored with Cole Burton, Andrew Kittle and Anjali Watson. Analysis was conducted with the guidance of co-authors, and drafts were improved upon following comments from co-authors.

vi Table of contents

Abstract ...... iii

Lay summary ...... v

Preface ...... vi

Table of contents ...... vii

List of tables ...... x

List of figures ...... xiii

Acknowledgements ...... xv

Dedication ...... xvii

Chapter 1. General introduction ...... 1 1.1 Harmonising biodiversity conservation with economic development ...... 1 1.2 Human-carnivore conflict ...... 2 1.3 Human-carnivore coexistence requires interdisciplinary research ...... 3 1.4 Rationale and research approach ...... 4

Chapter 2: Attitudes towards leopards in Sri Lanka and considerations for human-leopard coexistence ...... 7 2.1 Introduction ...... 7 2.1.1 Human-carnivore conflict ...... 7 2.1.2 Importance of studying attitudes ...... 7 2.1.3 Sri Lankan context ...... 9 2.1.4 Study objectives ...... 10 2.2 Methods ...... 11 2.2.1 Study area ...... 11 2.2.2 Sampling and survey design ...... 13 2.2.3 Exploratory Factor Analysis ...... 16 2.2.4 Regression models ...... 17 2.2.5 Semi-structured interviews ...... 18 2.3 Results ...... 20 2.3.1 Attitudes towards leopards in Yala ...... 20 2.3.2 Attitudes towards leopards in the Central Hills ...... 23 2.3.3. Constraints to livestock rearing faced by pastoralists in Yala and Central Hills ...... 25 2.3.3.1 Yala ...... 25 2.3.3.2 Central Hills ...... 29

vii 2.4 Discussion ...... 32 2.4.1 Determinants of attitudes towards leopards ...... 32 2.4.2 Potential strategies for human-leopard coexistence ...... 34 2.4.3 Conclusion ...... 38

Chapter 3. Drivers and potential mitigations of livestock depredation by the Sri Lankan leopard (Panthera pardus kotiya) ...... 39 3.1 Introduction ...... 39 3.1.1 Human-carnivore conflict ...... 39 3.1.2 Identifying key drivers of livestock depredation ...... 41 3.1.3 Sri Lankan context ...... 42 3.1.4 Objectives and hypotheses ...... 44 3.2 Methods ...... 45 3.2.1 Study area ...... 45 3.2.2 Surveys and semi-structured interviews ...... 46 3.2.3 Farm visits and observations of husbandry practices ...... 46 3.2.4 Prey availability surveys and predictive biomass models ...... 47 3.2.5 Variation in reported depredation ...... 49 3.3 Results ...... 50 3.3.1 Causes of livestock loss ...... 50 3.3.2 Livestock husbandry practiced by respondents ...... 50 3.3.3 Mitigations supported by respondents ...... 53 3.3.4 Leopard prey species and biomass availability ...... 54 3.3.5 Prey model ...... 55 3.3.6 Depredation model ...... 56 3.4 Discussion ...... 58 3.4.1 Factors involved in depredation incidents ...... 58 3.4.2 Mitigations and husbandry techniques supported by the community ...... 60 3.4.3. Livestock loss due to other causes ...... 61 3.4.4 Reducing livestock densities and preventing overcrowding on the landscape ...... 63 3.4.5 Conclusion ...... 64

Chapter 4: Conclusion ...... 67 4.1 Synthesis and conclusions ...... 67 4.2 Research strengths and limitations ...... 69 4.3 Recommendations and directions for future research ...... 70

References ...... 73

Appendix A: Survey protocol ...... 88

Appendix B: Semi-structured interview protocol ...... 94

Appendix C: Respondent characteristics ...... 95

viii Appendix D: Exploratory Factor Analysis (EFA) ...... 96

Appendix E: Complete model set representing all possible additive combinations to identify determinants of attitudes ...... 98

Appendix F: Camera trap deployment data sheet ...... 102

Appendix G: Additional information on body mass values used for leopard prey species detected on camera traps and Relative Biomass Index (RBI) in kg/day for adult and juvenile prey ...... 104

Appendix H: Camera trap photos ...... 107

Appendix I: Complete model set representing all possible additive combinations to identify factors associated with incidents of livestock depredation in Yala, Sri Lanka .. 111

ix List of tables

Table 2.7. Summary of key issues analysed from individual and group interviews with a subset of respondents in Yala (n=32) and Central Hills (n=19), Sri Lanka, accompanied by an illustrative quote...... 29

x Table 3.4. Independent detections (based on 30 min threshold for independence) of adult and juvenile leopard prey species detected by camera trap sampling (n=28 stations) in Yala, Sri Lanka, between May-August 2018 and June-July 2019...... 55

Table 3.6. Top-ranked generalised linear models of factors relating to reported livestock depredation from 2015-2018 across respondents (n = 61) in Yala, Sri Lanka. Models within 4 ΔAICc units of the model with the lowest AICc, with associated degrees of freedom (df), AICc, ΔAICc and Akaike model weight are shown...... 58

Table E.1. All possible additive combinations to explain determinants of attitudes towards leopards in Yala, Sri Lanka, measured from 61 surveys in May-August 2018 using generalised linear models. Variables include the importance of conservation (CONS), monthly expenditure (COST), awareness of general leopard ecology (ECOL), awareness of leopard-related tourism benefits (TOURISM), the proportion of cattle lost to leopard depredation (LOST_LEO) and socio-demographics (SOCIO). Df is degrees of freedom of the model, AICc is the AIC value adjusted for small sample size, delta AICc is the difference in AICc scores from the top model...... 98

Table E.2. All possible additive combinations to explain determinants of attitudes towards leopards in the Central Hills, Sri Lanka, measured from 52 surveys in May- August 2018 using generalised linear models. Variables include the desire for increased government involvement (GOV), the level of worry that leopard sightings invoke (SIGHTINGS), importance of conservation (CONS), awareness of leopard-related tourism benefits (TOURISM), and the frequency of observing leopards or signs of leopards in the past 3 years (LEO_3). Df is degrees of freedom of the model, AICc is the AIC value adjusted for small sample size, delta AICc is the difference in AICc scores from the top model...... 100

Table G.2. Camera trap stations in Yala, Sri Lanka (n=28), camera trap active days and the Relative Biomass Index (RBI) of adult and juvenile leopard prey species, measured in terms of kg/day. RBI values were obtained by calculating Relative Abundance Index (RAI) for each species at each station and multiplying this by the average bodyweight (see Table G.1). As far as possible, species were identified as adult males, adult females and juveniles, and the respective bodyweights were used...... 105

xi (DIST_NP;km), cattle husbandry (HUS), number of cattle (CATTLE), pastoralist residency time (RESIDENCY; years), road density within a 150m buffer (ROAD; km/km2) Df is degrees of freedom of the model, AICc is the AIC value adjusted for small sample size, delta AICc is the difference in AICc scores from the top model...... 111

xii List of figures

Fig.2.2. Sites where semi-structured surveys were conducted in 8 tea estates across the Central Hills landscape. Peak Wilderness Sanctuary can be seen to the south of the estate locations. The inset map shows Sri Lanka...... 13

Fig.2.3. Photo of a survey being conducted with a pastoralist in Yala, Sri Lanka, whose calves are kept in the steel enclosure (seen in the background) given to a friend by the nearby hotel...... 15

Fig.2.4. Regression parameter estimates from the top-ranked generalised linear model to explain variation in pastoralist attitudes toward leopards in the Yala study region. Estimates and 95% confidence intervals are shown for the top model (model with the lowest AICc) out of contender models within 2 rAICc...... 22

Fig.2.5. Regression parameter estimates from a candidate GLM to explain variation in pastoralist attitudes toward leopards in the Central Hills study region. Estimates and 95% confidence intervals are shown for the model within 2 rAICc of the top model that had the most parameters (Table 2.5)...... 24

Fig.2.5. Common makeshift pens for cattle that offer little protection against leopards. . 26

Fig.2.6. Steel enclosure with roof given to certain pastoralists in the Yala region...... 26

Fig.2.7. Photo of an underweight cow in the Yala region. Many individuals observed in this study site were similarly underweight...... 27

Fig.2.8. Poor roads that most common modes of transport often refuse to travel on...... 30

Fig.2.9. A typical cattle shed in the Central Hills that lacks adequate drainage systems, causing issues regarding the disposal of sewage water...... 31

Fig.3.1. Female Sri Lankan leopard (Panthera pardus kotiya) seen in Yala National Park, June 2018...... 43

Fig.3.2. Map of camera trap locations in the Yala study site (n=28) used to estimate and extrapolate leopard native prey availability, in the south-east coastal belt of Sri Lanka around Yala National Park. Inset map shows Sri Lanka...... 48

Fig.3.3a. Steel chain-link enclosure with a fixed roof Fig.3.3b. Steel chain-link enclosure without a roof ...... 52

Fig.3.3c. Partial steel chain-link fencing, supplemented with thorns Fig.3.3d. Rudimentary thorn and barbed wire ...... 52

Fig.3.4. Percentage of respondents in Yala (n=61) surveyed in May-August 2018 who supported the future testing of proposed conflict mitigation techniques...... 53

xiii Fig.3.5. Estimated effects of all predictor variables on factors associated with leopard prey availability (measured as Relative Biomass Index; kg/day) for each camera trap location (n = 28) in Yala, Sri Lanka. Predictor variables included distance to roads (km), distance to water (km) and percent forest cover within a 150m buffer. Coefficient estimates of the global generalised linear model are displayed with their 95% confidence intervals. Predictor variables have been standardised to a mean of 0 and standard deviation of 1 to allow for direct comparison...... 56

Fig.3.6. Estimated effects of all predictor variables on factors associated with livestock depredation from 2015-2018 in Yala, Sri Lanka. Predictor variables included leopard prey availability (Relative Biomass Index; kg/day), distance to national park (km), husbandry score, number of cattle, pastoralist residency time (years) and road density within a 150m buffer. Coefficient estimates of the global generalised linear model are displayed with their 95% confidence intervals. Predictor variables have been standardised to a mean of 0 and standard deviation of 1 to allow for direct comparison...... 57

Fig.H.1. Sri Lankan leopard (Panthera pardus kotiya) ...... 107

Fig.H.2. Grey langurs (Semnopithecus priam) ...... 107

Fig.H.3. Ceylon spotted deer (Axis axis ceylonensis) ...... 108

Fig.H.4. Ruddy mongoose (Herpestes smithii) ...... 108

Fig.H.5. Wild boar (Sus scrofa) ...... 109

Fig.H.6. Indian crested porcupine (Hystrix indica) ...... 109

Fig.H.7. Ceylon spotted deer (Axis axis ceylonensis) ...... 110

Fig.H.8. Black-naped hare (Lepus nigricollis) ...... 110

xiv Acknowledgements

There are so many people who have helped make this work possible, and I owe them all my most sincere and genuine appreciation. Firstly, I want to thank my co- supervisor Dr. Cole Burton, for taking me on despite our initial technologically-impaired Skype call while I was sweating with a fever in Sri Lanka. Thank you for encouraging me to grow as researcher, for broadening my perspectives, for always being available despite your chaotic schedule, but most of all for giving me the opportunity to do something that is so close to my heart with the support of your lab. Next, thank you to Dr. Andrew Kittle, my co-supervisor in Sri Lanka, for all the help organizing local logistics, for helping me see this project in a broader perspective, and for helping me understand the challenges of applied conservation in a place like Sri Lanka. A huge thank you to Dr. Shannon Hagerman for helping me as I navigated the social sciences universe, and for offering your help multiple times throughout this process.

There were moments where I felt like this project wasn’t good enough, or I didn’t have the ability to balance both the social and ecological sciences. WildCo lab, I literally could not have done this without you. You are the most supportive, most fun and most genuinely helpful lab I could have asked for – whether it was the discussing model selection over lunch, weekend morning library grant-writing or simply forcing each other to go for a walk in the sun. It all made the world of difference for me. Each and every one of you are doing incredible research and I am so excited to see where we all end up branching out in life. Thank you also to Dr. Edward Kroc for helping me with statistical guidance in order to get the most informative outcome of my data, for teaching me statistics used in social sciences that went completely over my head initially. Collaborations with people like you to provide expertise and guidance is what we all should aspire to do more of.

My friends and family who have supported me, you know who you are. Thank you for believing in me when I was down, for making sure I sleep enough, and for just being the light in my life. Thank you to all the volunteers who have tirelessly helped with various aspects of this work, be it removing false triggers, tagging the hundreds of thousands of

xv camera trap photos or entering data. Morel Liu, Candy Lo, Phoebe Li, August Schwanauer, Ashani Hangawatta, Emily Siemens – you made my life immensely easier!

My field team in Sri Lanka, you are incredible. Al-fayed Mohammed and Parami Pieris: Thank you for cheering me up when my cameras got stolen. Thank you for discovering the best post-heatstroke food places with me. Thank you for going out of your way to make sure we sort through as much data (camera trap and interview) as possible while it was fresh in our heads. Thank you also to Anjali Watson for helping with managing local logistics and acquiring permits and permissions to make this work possible. Thank you to Cinnamon Wild Yala, for providing accommodation and assistance during fieldwork. I hope these results are useful in your efforts to help the beautiful Yala region. I am also grateful to the Department of Wildlife staff who went out of their way to assist with camera set up and checks. Pradeep Kumara and Darrel Bartholomeusz, thank you for your genuine interest and for helping with accommodation and transport, and for your local knowledge of the Yala landscape that helped me see how important this work can be. Thank you Chandana – I don’t know your last name, but you allowed us to get everything done by providing us with local context and local contacts. Your relationship with the pastoralists in Yala helped us gain credibility, and I am forever grateful!

Thank you to my funders: The National Geographic Society, the Rufford Foundation, IDEAWILD, the Greenland Zoo and all those who donated to my GoFundMe page for allowing this surprisingly expensive fieldwork to take place. This work literally wouldn’t have been possible without your help.

Last but not least, thank you to the Sri Lankan communities who welcomed me so graciously, who let me see what their lives are like, and who let me gain insight into their struggles and opinions. I hope I did so with respect and with compassion, and I hope the results can be used towards achieving coexistence for all.

xvi Dedication

To my grandparents, R.K.W. Goonesekere (1928-2014) and Savitri W.E. Goonesekere (1939 - ). You inspire me for many reasons – your pioneering contributions to children’s rights, women’s rights and human rights, for defending those who have been oppressed, and for spending decades of your lives selflessly working towards building a better Sri Lanka for all. But above all, you inspire me because you have both demonstrated that kindness, compassion and integrity is always the best path forward. Thank you for being my strongest supporters throughout my ups and downs, and for showing me that wisdom and knowledge should be accompanied with courage and humility. This is for you.

xvii Chapter 1. General introduction

1.1 Harmonising biodiversity conservation with economic development Balancing ecological resilience and biodiversity conservation with sustainable livelihoods is a complex yet crucial undertaking. The United Nations Sustainable Development Goals 8, 12 and 15 clearly stress the need to strive towards this balance (United Nations 2015). However, reconciling the needs of wildlife communities with the needs of rural – particularly developing – communities is notoriously difficult, and as a result many divided viewpoints exist on what strategies are most effective (Sandker et al. 2009). Many approaches to this issue have been attempted, such as Integrated Conservation and Development Projects (ICDPs; Alpert 1996), which aim to balance conservation with development, “fortress conservation” where investments are focused on protected areas (Brockington 2004), and Community-Based Natural Resource Management (CBNRM; Fabricius et al. 2004), which focuses on ensuring communities are central to the conservation process. However, each approach has been met with criticism, such as the failure of ICDPs to conserve biodiversity (Sandker et al. 2009), the alienation of local stakeholders through “fortress conservation” (Robbins et al. 2006), or the inadequate equity and transparency of CBNRMs (Dickman, Macdonald, and Macdonald 2011).

When addressing the balance between conservation and development, we should take into account not just the ecological aspects, but the social and economic aspects as well. However, a key challenge to most conservation assessments and to the implementation of policies is addressing the complexities of social-ecological systems (Berkes, Colding, and Folke 2002). Challenges also exist when collecting appropriate data that illustrate both opportunities and limitations in a local context, which would help stakeholders make more informed conservation decisions (Cowling et al. 2004). With many rural and impoverished communities striving to improve their livelihoods, there is a clear need to consider local contexts and suggest ways in which development can be consistent with biodiversity conservation. Promising examples include small-scale common property arrangements, redesigning animal husbandry, or working to utilise already-degraded lands (Sanderson and Redford 2003).

1.2 Human-carnivore conflict Globally, we are witnessing a decline in large carnivore populations. A leading cause of this decline is human-carnivore conflict, defined as carnivore-related threats to the livelihood and safety of a person or community that typically result in the persecution of that carnivore species (Inskip and Zimmermann 2009). Often, conflict pertains to livestock depredation, and can result in preventative (e.g. snares) or retaliatory (e.g. shooting) actions being taken against carnivores (Ripple et al. 2014; Inskip and Zimmermann 2009). Human-carnivore conflict arises for many reasons. Carnivores have protein-rich diets and large home ranges, drawing them into competition with humans who are increasingly encroaching into new – and therefore shared – landscapes. Additionally, many carnivores are specialised in ungulate predation, leading to instances of wild carnivores killing domesticated ungulates (Karanth, Sunquist, and Chinnappa 1999; Polisar et al. 2003). This has been observed globally, with wolves and bears killing sheep in North America and Europe (Stone et al. 2017; Kaczensky 1999), leopards and tigers killing livestock in Asia (Karanth et al. 2004; Sangay and Vernes 2008), jaguars and pumas killing cattle in South America (Michalski et al. 2006; Zimmermann, Walpole, and Leader- Williams 2005) and species such as lions, cheetahs, leopards and hyenas killing livestock throughout Sub-Saharan Africa (Kolowski and Holekamp 2006; Dickman et al. 2014; Kissui 2008).

Human-carnivore conflict is particularly damaging for individual households that rear livestock in close proximity to protected areas, presenting a difficult challenge for pastoral development and biodiversity preservation (Treves and Karanth 2003; Loveridge et al. 2010). Incidents of depredation, even if infrequent, can threaten the livelihoods and security of entire households, emphasizing that efforts to resolve human-carnivore conflict cannot consider only humans or carnivores in isolation. Even if livestock losses to carnivores are few in number, this can result in a significant challenge for rural communities whose livelihoods are dependent on their livestock, and who are therefore the most vulnerable to losses (Dickman 2010; Jacobson et al. 2016). Additionally, perceived conflict is often greater than actual conflict experienced, likely due to people feeling unable

2 to control or influence the threat of a large carnivore. This may be compounded by the fact that carnivores are protected by government laws (Madden 2004).

1.3 Human-carnivore coexistence requires interdisciplinary research Human-carnivore conflict is inherently difficult to address, due to complexities in the behaviour and ecology of carnivore species, the attitudes and behaviours of human communities, the varying husbandry techniques used, and spatial and temporal variation in the availability of natural carnivore foods (Nyhus 2016). It is therefore difficult to extrapolate results from one context to others with different histories, communities and practices. Though there is increasing acceptance of the need to better integrate social and ecological knowledge, challenges exist in terms of aligning data types, improving cross- discipline communication, and overcoming misunderstandings about the quality and utility of social science data (Fox et al. 2006; Pooley, Mendelsohn, and Milner-Gulland 2014).

Many human-carnivore conflict studies are undertaken from an ecological standpoint, focusing only or primarily on the carnivore species, with little to no social science input (Treves et al. 2006). To ensure wildlife management policies are effective and appropriate to the local contexts, it is important to understand the attitudes of the local people. Assessing attitudes can provide insight on how local communities may behave in the face of increasing economic losses due to wildlife, the degree to which they are willing to coexist, and how interested they are in mitigating against threats (Fulton, Manfredo, and Lipscomb 1996). Successful management of human-carnivore conflicts therefore depends on social science methods of evaluating attitudes, engaging in participatory planning and understanding wider socio-economic practices that are influenced by different spatial and temporal scales than those bounding a particular study region (Dickman 2010; Kansky, Kidd, and Knight 2016).

As rates of species loss increase, there is an urgent need for more applied interdisciplinary research that effectively guides conservation efforts and bridges the gap that too-often exists between scientists and decision-makers (Balme et al. 2014). A greater understanding of the specific factors associated with incidents of conflict (e.g. native prey

3 availability, distance to protected area, livestock husbandry) can give insight into what is driving incidents of conflict to occur. With this knowledge, including the social dimensions of human-wildlife conflict is then crucial to inform the development of context-appropriate solutions to conflict, as attitudes and behaviours of humans are not always static (Marchini and Macdonald 2012). In the context of human-carnivore conflict, such solutions may include education to improve awareness of carnivore species and their ecology (Gore et al. 2016; Marker, Mills, and Macdonald 2003), mitigation efforts to prevent livestock losses (Ogada et al. 2003; McManus et al. 2015), or incentive schemes to compensate pastoralists who have suffered losses (Dickman, Macdonald, and Macdonald 2011; Mishra et al. 2003). Acknowledging and incorporating both the social and ecological dimensions of human- carnivore conflict can allow us to develop appropriate interventions and mitigations at local scales.

Interdisciplinary research has meaningfully informed human-wildlife coexistence globally. For example, in Kenya, it was found that perception of livestock depredation and socio-economic factors were associated with lion killing (Hazzah, Borgerhoff Mulder, and Frank 2009). Following this research, the Lion Guardians conservation organization has worked to improve the attitudes of Maasai warriors towards lions through the use of traditional conflict mitigation techniques. By harnessing cultural values to encourage warriors to engage in lion monitoring, the program has led to a 99% decrease in lion killing (Hazzah et al. 2014; Dolrenry, Hazzah, and Frank 2016). In Hemis National Park, Ladakh, communities were engaged through Appreciative Participatory Planning and Action, which gave them an increased sense of project ownership and empowerment, and ultimately a greater willingness to coexist with snow leopards (Jackson and Wangchuk 2004). These examples illustrate that evidence-based research at local scales can facilitate human- carnivore coexistence.

1.4 Rationale and research approach The focus of this thesis is on the theme of human-carnivore coexistence within the context of an expanding dairy cattle industry in Sri Lanka, a global biodiversity hotspot (Bawa et al. 2007). We approach the term ‘coexistence’ as humans and wildlife co-

4 occurring while maintaining viable wildlife populations and tolerable levels of human- wildlife interaction. As small-scale dairy farms increase in number across the country, there are unquantified reports of livestock depredation by the endangered Sri Lankan leopard (Panthera pardus kotiya). The impact of conflict and associated retaliatory killings on leopard populations is unknown but potentially significant, particularly considering that the dairy industry is set to move forward and grow towards a target of national dairy self- sufficiency (Vernooij, Houwers, and Zijlstra 2015). Engaging pastoral communities at the early stages of expansion is crucial, as coexistence strategies that involve, and have been ‘co-designed’ by, local dairy pastoralists may have greater success than research findings recommended by outside scientists with little local engagement (Treves et al. 2006). The aim of this thesis was to approach human-leopard conflict in Sri Lanka by considering both the social and ecological dimensions in order to assess the key determinants of pastoralist attitudes towards leopards and identify the factors most associated with incidents of livestock depredation. Our goal was to generate results that will help identify key areas in which to focus conservation resources, and specific mitigations supported by the community that can be tested in the future.

In Chapter 2, we examined the determinants of attitudes towards leopards in two regions that contain pastoralist communities in Sri Lanka: Yala and the Central Hills. The communities differ in histories, traditions, local economies, demographics and livestock husbandry, with the key contrast for our study being a difference in the degree of conflict with leopards. In Yala, pastoralists operate with little to no local governance and rear much larger herds of cattle. Husbandry practices are generally weak, with cattle being kept in poorly protected enclosures, resulting in many reports of depredation by leopards. The nearby Yala National Park is the most visited national park in Sri Lanka (Sri Lanka Tourism Development Authority 2017), due to the relatively high chance of leopard sightings. As a result, leopard-related tourism is a significant driver of the local economy, with resulting economic benefits for hotels, safari jeep drivers and others employed in the tourism industry. In contrast, tea estates in the Central Hills are organised through superintendents with a strict chain of command. Estate workers employed as tea pluckers sometimes raise cattle, but in ways that differ from Yala pastoralists. Specifically, Central

5 Hills cattle are a much larger breed, fewer in number, and kept in enclosed sheds without free grazing. This means there are currently few cattle-related issues with leopards in the Central Hills. However, due to people living in close proximity with leopards, the potential for conflict remains, and it is important to assess attitudes towards leopards to inform proactive management. Accordingly, we conducted in-person surveys to obtain data on variables hypothesised to be associated with attitudes towards leopards. We then used Exploratory Factor Analysis and regression modelling to identify and compared key determinants of attitudes in the Yala and Central Hills communities. We further used semi- structured interviews to gather more information on challenges encountered while raising livestock, in order to inform the development of strategies to facilitate sustainable cattle rearing in these communities.

In Chapter 3, we focused on understanding patterns of livestock depredation in the Yala region. We used camera trap surveys and regression models to estimate spatial variation in leopard prey availability based on forest canopy cover, distance to water and distance to roads. We then combined our estimate of prey availability with other factors hypothesized to be important in predicting livestock depredation by leopards, namely husbandry practices, number of cattle, distance from Yala National Park, road density and pastoralist residency time. We used generalised linear models (GLMs) to test the relative influence of these factors on resulting levels of reported depredation. We used an information theoretic approach based on Akaike’s Information Criterion (corrected for small sample size, AICc) to identify the most parsimonious model that best fit the data (Mazerolle 2006; Burnham and Anderson 2002). We assessed model fit by checking the R2 value of the top-ranked model as well as the proportion of deviance explained (Crawley 2007).

In Chapter 4, we summarized results from each data chapter and synthesized conclusions in light of current literature on human-carnivore conflict. We also reflected on the strengths and limitations of our study and provided recommendations for future interdisciplinary research.

6 Chapter 2: Attitudes towards leopards in Sri Lanka and considerations for human-leopard coexistence

2.1 Introduction 2.1.1 Human-carnivore conflict A leading cause of carnivore decline is human-carnivore conflict (hereafter ‘conflict’). This conflict arises when carnivore-related threats occur to the livelihood or safety of a person or community, resulting in persecution of that carnivore species (Inskip and Zimmermann 2009). Conflict is exacerbated when local people feel that wildlife needs are given priority over their own, or when institutions lack the ability to appropriately manage conflict (Madden 2004). Livestock depredation is a major component of conflict globally, particularly within and around protected areas.

Protected areas are vital for the conservation of large carnivores. However, the vast ranges of these species relative to the size of most protected areas means they often use adjacent unprotected landscapes, which are typically shared with humans (Carter and Linnell 2016). These shared landscapes represent a substantial proportion of the remaining geographical ranges of most large carnivore species (Di Minin et al. 2016), underscoring the need to focus research on the interface between human and wildlife communities. Pastoralist communities are central to this interface and may encounter financial difficulty, reduced psychosocial comfort, and opportunity costs as a result of conflict (Rostro-García et al. 2016; Ghoddousi et al. 2016). Communities experiencing significant losses due to depredation may have a lower tolerance toward carnivores, resulting in hostility, reduced support for conservation initiatives and increased retaliatory killings (Chapron et al. 2008).

2.1.2 Importance of studying attitudes It is crucial to consider the social dimensions of conflict when addressing wildlife conservation and coexistence. Numerous interdisciplinary approaches highlight the need to integrate ecological, economic and social perspectives and recognise the complexity of social-ecological systems (SES; Berkes & Folke, 1998; Decker, Riley, & Siemer, 2012;

7 Dickman, 2010; Redpath et al., 2013). Understanding the attitudes of those who live in close proximity to wildlife is important for informing effective wildlife management and conflict mitigation (Kansky and Knight 2014). The attitude construct is a key facet of both social psychology (Fiske and Taylor 2013; Allport 1935) and environmental psychology (Heberlein 2012). Attitudes can be considered a positive or negative evaluation of something specific, differing from beliefs that are more general and associated with broader values (Dietz, Fitzgerald, and Shwom 2005). Importantly, attitudes are not always a predictor of behaviour, but positive attitudes towards an object (e.g. carnivore) or behaviour (e.g. cooperating with wildlife managers to find practical solutions) are necessary conditions for realised behaviour (Kansky and Knight 2014). We define an attitude as a tendency to respond with a degree of favourableness to a psychological object that represents an aspect of an individual’s world (e.g. issue, behaviour, action; (Fishbein and Ajzen 2010). These psychological objects can be framed as either positive or negative. A better understanding of what is influencing community attitudes towards carnivores, particularly factors that negatively influence attitudes, is needed to inform initiatives to improve human-carnivore coexistence.

Several strategies have been proposed to address biodiversity conservation alongside rural development, including community-based natural resource management (CBNRM; Fabricius et al. 2004), and integrated conservation and development projects (ICDP; Alpert 1996), which aim to ensure more equity and distribution of costs and benefits of conservation. Other strategies such as wildlife premium mechanisms (WPMs; Dinerstein et al. 2013), compensation schemes (Dickman, Macdonald, and Macdonald 2011), or community-based livestock insurance programs (Mishra et al. 2003; Gebresenbet et al. 2018) are also potential approaches to consider. However, the effectiveness of these strategies may ultimately depend on local contexts, attitudes and institutional capacities.

No two communities are the same, and it is common to see variation in the key determinants of community attitudes due to political, social, economic, cultural or geographic factors (Madden 2004). Certain common variables underlying conflict do

8 emerge from previous studies, including: economic costs of conflict, livestock demographics and husbandry methods, socio-demographics, socio-economics, knowledge of wildlife behaviour, and prior experience raising livestock and encountering carnivores (Kansky and Knight 2014; Zimmermann, Walpole, and Leader-Williams 2005; Thorn et al. 2015; Störmer et al. 2019). These variables are especially important to consider, to ensure carnivore management policies minimise negative social impact and are well-suited to local levels of support. Managing conflict therefore depends heavily on social science methods such as evaluating attitudes, engaging in participatory planning and understanding long term socio-economic contexts (Dickman 2010; Kansky, Kidd, and Knight 2016). Ultimately, any set of policies or approaches intended to facilitate coexistence will need to consider both social and ecological aspects. This is particularly important in tropical contexts, where trade-offs often occur between human well-being and biodiversity conservation.

2.1.3 Sri Lankan context Sri Lanka is one such context, facing pressing needs in both conservation and development. Being a global biodiversity hotspot, Sri Lanka has high rates of species endemism and a reliance on its biodiversity for tourism and socio-economic activities, yet faces threats by its dense human population (Bawa et al. 2007). The Sri Lankan leopard (Panthera pardus kotiya) is an example of an endangered subspecies endemic to Sri Lanka (Stein et al. 2016; Miththapala, Seidensticker, and O’Brien 1996), increasingly threatened by habitat loss and degradation by accelerating human development (Vernooij, Houwers, and Zijlstra 2015; Ranaweera 2015). One form of development that is encroaching on remaining leopard habitats is dairy farming (Kittle, Watson, Kumara, & Sanjeewani, 2012). Targets for growth in the dairy sector have been promoted as a means for improving livelihoods and broadening economic development following Sri Lanka’s emergence from a protracted civil war (1983-2009) and the 2004 South Asian tsunami (Development 2013; Barlow 2011). Historically, Sri Lanka has been a low dairy consuming country, with buffalos (Bubalus bubalis) favoured for production of curd. However, recent economic growth has led to a shift in local demand for cow milk and related products that has not been matched by an increase in domestic production (Perera

9 and Jayasuriya 2008; Vernooij, Houwers, and Zijlstra 2015). In 2013, a government effort began to reduce dairy imports and increase from 35% self-sufficiency to 100% by 2020. As a result, there has been an expansion of dairy farms, processing centres, and importing of cattle (Vernooij, Houwers, and Zijlstra 2015). A majority of dairy producers are smallholder farms that are distributed across the country. In certain areas, these farms are in close proximity to national parks and key wildlife species such as the Sri Lankan leopard.

In addition to being a species of high economic value, the Sri Lankan leopard is the island’s apex predator and therefore maintaining viable leopard populations is important for their ecological, economic and intrinsic value (Kittle et al. 2017). However, leopards do prey on livestock, primarily dairy cattle. Even when cattle losses due to leopard predation are numerically few, they can represent a significant challenge for the vulnerable rural communities who are dependent on these cattle for their livelihood (Dickman 2010; Jacobson et al. 2016).

2.1.4 Study objectives In this study, our objectives were to inform human-leopard coexistence strategies by a) identifying determinants of pastoralist community attitudes towards leopards and b) suggest coexistence strategies to be considered that are grounded in local context. We used a survey to examine factors associated with observed attitudes towards leopards in two pastoralist communities in Sri Lanka. We used multivariate regression models and an information theoretic framework (based on Akaike’s Information Criterion, AIC (Burnham and Anderson 2002)) to evaluate the degree to which these variables explained variation in local attitudes in two sites with differing socio-ecological contexts. We also conducted semi-structured interviews to gain greater understanding of respondents and the history of the communities. This approach can give us important insights into the specific components that should be addressed in order to facilitate coexistence.

10 2.2 Methods 2.2.1 Study area We selected two study sites where cattle are being raised in landscapes where leopards occur but differing in socio-ecological contexts. Pastoralists in the Yala region (Yala; Fig 2.1) have experienced incidents of cattle depredation, while there have been no such incidents reported in the Central Hills (Fig.2.2). The regions also differ in terms of dominant religions, ethnicities and cattle farming histories, providing the opportunity to assess whether there are common determinants of attitudes between the two regions.

Yala is in the south-east coastal belt region of Hambantota District, outside the boundaries but in close proximity to Yala National Park (YNP; Fig.2.1). YNP is the second largest national park in Sri Lanka and reputed to have one of the highest leopard densities in the world (Kittle, Watson, and Fernando 2017). Block 1 of YNP (closest to the western park boundary) is characterised by thorny scrub forest, with taller climax forest and thin understory comprising the northern riparian zone (Mueller-Domboi 1972). Average temperatures at sea level are 25 - 28 ºC and the mean annual rainfall is approximately 900mm/year, with an extensive dry season (<50mm/month) that occurs from mid-May to mid-October (IUCN 1990; Mueller-Domboi 1972). YNP is the most visited national park in Sri Lanka, generating around 670 million rupees (around 3.68 million USD) in revenue each year (Sri Lanka Tourism Development Authority 2017). Tourism is a key contributor to Sri Lanka’s GDP, and foreign tourists are responsible for around 93% of YNP’s annual revenue (Sri Lanka Tourism Development Authority 2017). Due to the high density of leopards in YNP, some resident leopards have ranges that extend beyond park boundaries and some transient or dispersing individuals may be pushed out into peripheral areas around the park where cattle are grazed, increasing the chances of encounters between leopards and cattle.

The Central Hills is a UNESCO World Heritage Site in the wet zone of Sri Lanka, and one of the areas being targeted for dairy farm expansion, as tea estate companies are increasingly diversifying to include a dairy development sector in their operations (Ranaweera 2009; Prasannath 2015; Development 2013). This region has a high human

11 population density, as well as a leopard population shown to utilise and inhabit a wide variety of landscapes (e.g. heavily impacted secondary forest; forest patches; plantations of eucalyptus, pinus, and tea; and areas in close proximity to human settlements) (Kittle et al. 2012), a phenomenon also seen in India where female leopards were found to raise cubs and establish territory inside tea estates, due to the cover they provide (Bhattacharjee and Parthasarathy 2013). The mean annual rainfall is between 2540mm and the average temperature is 15.4ºC (IUCN 1990). Our study area included 8 tea estates located north of the 22, 779 ha Peak Wilderness Sanctuary, the region’s largest protected area (Fig.2.2).

Fig.2.1. Cattle farms where in-person surveys were conducted in the Yala region from May-August 2018, in the south-east coastal dry zone of Sri Lanka. Yala National Park is to the north east of survey locations. The inset map shows Sri Lanka.

Fig.2.2. Cattle farms where in-person surveys were conducted in 8 tea estates across the Central Hills landscape. Peak Wilderness Sanctuary can be seen to the south of the estate locations. The inset map shows Sri Lanka.

2.2.2 Sampling and survey design Our target sampling population for the surveys was pastoralists that penned and/or grazed their cattle in lands adjacent to protected areas. In Yala, we initially selected pastoralists who had already participated in a Corporate Social Responsibility (CSR) program operated by a nearby hotel, which gives out steel cattle enclosures as protection for cattle. To ensure we sampled a broader range of husbandry practices, cattle locations and cattle rearing experiences, we added other cattle owners in the study area through purposeful sampling (i.e. intentionally seeking out respondents who provided a wider range of the variables considered in this study; Cresswell and Plano Clark 2011). In the Central Hills, we contacted 8 tea estates located near Peak Wilderness Sanctuary and gained permission from each estate superintendent to survey cattle owners. UBC Behavioural Research Ethics Board approval was obtained prior to any surveys being conducted.

13 Between May-August 2018, we conducted 113 in-person surveys, the design of which included both closed-ended and open-ended questions (61 surveys in the Yala region and 52 surveys in the Central Hills; Appendix A). Each survey involved a single pastoralist and generated both quantitative and qualitative data, reflecting formats used in similar conflict studies in Kashmir (Mir, Noor, Habib & Veeraswami 2015), South Africa (Thorn et al. 2015) and the Pantanal (Zimmermann et al. 2005). The survey covered 7 themes related to human-leopard conflict: socio-demographics, cattle demographics, husbandry techniques and mitigations used, leopard sightings and cattle loss, importance of conservation, general awareness about leopard ecology and economic benefits related to tourism, and attitudes towards leopards (Table 2.1). Surveys were conducted without leading or directional questions about potential conflict with leopards, so as to avoid bias in the responses (Treves et al. 2006).

Table 2.1. Survey components and details about questions covered in each component. Surveys were conducted in the Yala region (n=61) and the Central Hills (n=52) of Sri Lanka from May to August 2018. Survey component Example questions included Socio-demographics Age, gender, number of dependents, forms of livelihood, time spent rearing cattle Cattle demographics Age and number of cattle owned, how cattle number and type has changed over time Husbandry techniques What husbandry techniques are practiced, what and mitigations used mitigations have been tested and to what success, limitations experienced Leopard sightings and Leopard sightings and cattle depredation, loss of guard cattle loss dogs, other causes of cattle loss and actions taken following depredation Importance of The desire to conserve multiple wildlife species and conservation whether their population levels are acceptable General awareness about Awareness of leopard endangerment level, habitat, prey leopard ecology and species, and general awareness of the leopard’s economic economic benefits related value through tourism (e.g. YNP revenue, hotel programs to tourism focused on leopards nearby) Attitudes towards How respondents felt about leopards in relation to the leopards country and their livelihood, actions taken if cattle depredation occurred and willingness to collaborate with scientists and government to find a common solution; see Tables 2.3 and 2.5

Fig.2.3. Photo of a survey being conducted with a pastoralist in Yala, Sri Lanka, whose calves are kept in the steel enclosure (seen in the background) given to a friend by the nearby hotel.

We characterized pastoralist attitudes towards leopards using a composite score based on 8 attitude statements adapted from Zimmerman et al. (2005). Statements were ranked on a 5-point Likert scale (Carr and Halvorsen 2001; Thorn et al. 2015), and balanced in terms of positive and negative statements (Dunlap et al. 2000). See Appendix A for the survey protocol and Appendix B for interview topics.

To reduce the chances of exaggerated reports of depredation, motivated by the hope of receiving greater benefits (Rasmussen 1999), we told each respondent prior to the survey that their responses would be anonymous. We pre-tested the survey at each study site to ensure appropriate language was used and respondents had a clear understanding of questions (Khumalo and Yung 2015). After pilot testing, we

15 administered surveys in Sinhala languages in Yala (using a bilingual research assistant) and in Tamil in the Central Hills (using bilingual estate workers). Pictures of wildlife species were shown to each respondent, which acted as a validation test to ensure no misattributions took place (e.g. mistaking leopards for fishing cats, a medium size cat also present in these habitats). In efforts to reduce reporting biases and establish greater trust, we were accompanied during surveys by long-term community members who had familiarity with pastoralists, which was key to improving our credibility with the community.

2.2.3 Exploratory Factor Analysis We conducted exploratory factor analysis (EFA) to group survey responses and attitude statement responses into variables for subsequent analysis. EFA is a method of multivariate data reduction that looks for factors that account for at least 50% of the common variation in the observed data (Fricker, Kulzy, and Appleget 2012). Survey questions were pre-grouped into categories within which questions would elicit responses relating to the same latent phenomenon. EFA then reduces survey questions into a smaller number of factors, identifying which variables are collinear to each other and have the strongest association to a given factor (DiStefano, Zhu, and Mîndrila 2009). This form of data reduction allows for as much data to be incorporated as possible, while preventing overparameterization in subsequent explanatory models (Harrison et al. 2018). For example, the variables age, number of dependents, and length of time rearing cattle were grouped into the single factor of ‘socio-demographics’ to use in subsequent modelling.

We then created a weighted composite attitude score for each respondent by multiplying each response to individual attitude statements (Tables 2.3 and 2.5) by the factor loading of the statement (i.e. the level to which responses to that question related to the overall composite score). This treats each question and its response individually, instead of applying equal weights across all questions. Therefore, questions that had a lower association with the overall attitude score were given a lower weight. The resulting composite attitude score for each individual was a continuous variable on a scale relative to other respondents within the sample. We ran Cronbach’s coefficient alpha test on the composite attitude scores. This test determines whether the differences between

16 individual respondents was due to differences in true scores or due to measurement error (Deng and Chan 2017), with standard requirements for internal consistency of alpha > 0.70 for non-clinical studies (Bland and Altman 1997). The resulting alpha value in the Yala region was 0.76 and in the Central Hills 0.65, with the latter value marginally below the suggested value, though still considered ‘acceptable’, particularly given to the small sample size (van Griethuijsen et al. 2015). See Appendix D for a more detailed explanation on EFA and the various decision points in the process.

2.2.4 Regression models We used generalised linear models (GLMs) to test the relationships between the composite attitude score (response variable) and determinants of attitudes (predictor variables). We first explored the data for collinearity using the ‘polycor’ package (J. Fox 2016) to address both continuous and ordinal variables simultaneously. We tested predictor variables using Pearson correlation coefficients (r), removing one variable in each pair that had r > 0.5 (our threshold for collinearity). This resulted in six predictor variables for Yala, and five predictor variables for the Central Hills (Table 2.2). We further explored the data by checking for outliers, relationships between the response and predictor variables and potential interactions using Cleveland dotplots, scatterplots and coplots, respectively (Zuur et al. 2010). Continuous variables were standardised to a mean of 0 and a standard deviation of 1 prior to running regression models to allow for direct comparison of effect sizes. We ran GLMs with a gamma distribution and log link due to the continuous response variable and right-skew of the data (Ng and Cribbie 2017). In the case of surveys, outliers in the response variable may represent true process variation as opposed to measurement error, thus we chose the more flexible gamma probability distribution to account for larger mean values in the response variable, without requiring the removal or transformation of data (Zuur, Ieno, and Elphick 2010). Gamma distributions have a dispersion parameter, so overdispersion or under-dispersion of the response variable was not further considered.

We used an information theoretic approach based on Akaike’s Information Criterion, corrected for small sample size (AICc), to rank the performance of candidate

17 models (Mazerolle 2006; Burnham and Anderson 2002). We ran separate global models for each of the two study regions, including additive effects of the full set of predictor variables. We compared these global models with a candidate set of models representing all possible additive combinations of the variables (64 models for Yala, 32 for Central Hills; Appendix E). We ranked candidate models by their AICc score and considered all models within 2 ∆AICc of the top-ranked (lowest AICc) model as plausible models (Tables 2.4 and 2.6). We assessed the adequacy of the top-ranked models using standard residual diagnostics in the DHARMa package (Hartig 2019) and effect size stability (Heinze, Wallisch, and Dunkler 2018). Estimates of regression coefficients and their 95% confidence intervals are reported. We analysed all data using R statistical software, version 3.5.1 (R Core Team 2018).

2.2.5 Semi-structured interviews In order to provide more context and nuance to the quantitative analysis of the surveys, we conducted semi-structured interviews (hereafter referred to as ‘interviews’) with a subset of willing respondents. Out of 113 respondents who completed the survey, 51 respondents (32 in the Yala region and 19 in the Central Hills) participated in these interviews. Based on a literature review and existing knowledge on the system, we developed an interview protocol designed to explore themes of leopards, specific husbandry techniques and places were institutional capacity was lacking. It became clear after the initial few surveys that some changes needed to be made, specifically by including questions pertaining to issues faced rearing livestock in these landscapes (see results in Table 2.7), which can help contextualise the results from the quantitative survey portions. This iterative process of conducting interviews to inform subsequent interviews helped us to better understand respondents and note emergent questions (Corbin and Strauss 2008; Khumalo and Yung 2015; Bradley, Curry, and Devers 2007). When analyzing each transcript, the initial round of coding focused on husbandry techniques and the testing of potential mitigations. Additionally, subsequent rounds of iterative coding identified some unanticipated themes (Saldana 2012). This approach is described as inductive grounded theory, where instead of using predefined codes, we create codes by defining what we see in the data (Charmaz 2006).

18 Though the focus of these interviews varied between respondents, Appendix B outlines the stable list of questions that were explored. Respondents were free to introduce topics that they wanted to discuss. We did not ask questions regarding illegal activities, such as killing of leopards, which could elicit distrust with the pastoralists who may fear penalty; however, if pastoralists volunteered such information it was noted. Every interview was subsequently transcribed and translated. In the Yala region, this translation was done from Sinhala to English through a bilingual research assistant. In the Central Hills, a Tamil to Sinhala translation first occurred through an estate manager, and the subsequent Sinhala to English translation was done by the same bilingual research assistant as in the Yala region. The Central Hills interviews and respective transcripts were shorter in length due to the double-translation required. Translated transcripts were examined collaboratively with local field assistants shortly after each interview, to ensure subtext and nuances were understood.

Table 2.2. Predictor variables included in regression models with their explanations and examples of survey questions used to infer variable data. The range of response values are given across both study sites. Variables with a (*) were only included in Yala models. Variables with a (**) were only included in Central Hills models. Survey section refers to complete survey found in Appendix A. Variable Explanation and survey questions Range Variable Survey name used to infer variable data type section Reported Proportion of cattle reported to have 0 - 0.5 Continuous 4 depredation* been killed by leopards Socio- EFA variable comprised of respondent 1.13 – 10.0 Continuous 1 demographics* age, number of dependents and time spent rearing cattle Cost* Monthly expenditure (Rs) on cattle 10,000 – Continuous 3 100,000 General EFA variable comprised of 1.45 – 10.0 Continuous 6 awareness of respondents’ awareness of leopard leopard endangerment, general movement and ecology prey species. Questions included: Do you know that the Sri Lankan leopard is endangered? Do you know you graze your cattle very close to a national park where leopards and other animals live? Do you know that leopards exist outside of these protected national parks and can live and reproduce in smaller forest patches?

19 Variable Explanation and survey questions Range Variable Survey name used to infer variable data type section General EFA variable comprised of 1.01 – 10.0 Continuous 6 awareness of respondents’ awareness of high leopard leopard-related YNP tourism and tourism* economic value. Questions included: Do you know that leopards produce economic benefits both in this area and across Sri Lanka? Do you know that YNP is the most visited national park in Sri Lanka? Do you know that YNP generates over 600 million rupees each year? Importance of EFA variable comprised of 1.02 – 10.0 Continuous 5 conservation respondents’ opinions on the conservation of Sri Lanka’s mammals, birds, reptiles and amphibians Leopard How leopard sightings or signs have 1 - 7 Continuous 4 sightings changed over the last 3 years and/or signs** Worry** If respondents felt worried for the 0 - 1 Categorical 4 safety of themselves or their cattle after seeing a leopard Government Desire for increased government 0 - 1 Categorical 4 involvement** involvement to facilitate an improvement in cattle rearing (e.g. vaccines or land leases)

2.3 Results

We found that determinants of attitudes towards leopards differed significantly in the two regions, so we present the results separately for each region. Attitudes towards leopards were in general less favourable in Yala compared to the Central Hills.

2.3.1 Attitudes towards leopards in Yala Respondents in Yala displayed less favourable attitudes towards leopards (Table 2.3), as seen by their level of agreement to attitude statements 3, 5, 6 and 7 that pertain to cattle depredation and leopard populations. Variables in the top model included a respondents’ level of awareness about general leopard ecology, level of awareness about leopard tourism, opinion on the importance of conservation, and socio-demographics (Table 2.4). There were negative relationships between attitudes toward leopards and respondents’ levels of awareness about leopard ecology (-3.41, 95% CI [-5.77, -0.95]) and leopard tourism (-0.38, 95% CI [-0.65, -0.11]), and positive relationships between

20 attitudes towards leopards and the importance of conservation (0.15, 95% CI [0.10, 0.21]) and socio-demographics (0.04, 95% CI [-0.01, 0.10]. All covariates except for socio- demographics were statistically significant as their 95% confidence intervals did not overlap zero (Fig.2.4). The adjusted proportion of deviance accounted for by the top model was 0.46. Models within 2 rAICc of the top model are described in Table 2.4. The residuals for the top model displayed no visible patterns, and there was non- significant deviation resulting from the scaled residual QQ plot.

Table 2.3. Percentage of responses to 8 statements relating to attitudes towards leopards in the Yala region (n = 61), scored on a Likert scale. Responses are from the raw data, prior to creating weighted composite scores for each individual through EFA. No. Statement Strongly Agree Neutral Disagree Strongly agree disagree 1 The nature and wildlife of Sri Lanka 68.85 3.28 26.2 1.64 0.00 is a national treasure and should be conserved 2 I respect leopards for the economic 50.82 14.75 27.87 3.28 3.28 value they bring to the country through wildlife tourism 3 My livelihood is more important 91.80 6.56 0.00 1.64 0.00 than current leopard populations 4 It does not matter if a leopard kills a 26.23 13.11 21.31 8.20 31.15 few of my cattle, they are wild animals trying to survive 5 At this farm we cannot tolerate 68.85 19.67 6.56 4.92 0.00 leopards killing any cattle at any time 6 I would be happier if there were 65.57 4.92 14.75 3.28 11.48 fewer leopards around where I live and raise my cattle 7 I do not want to kill leopards, but if 72.13 3.28 14.75 1.64 8.20 they kill my cattle I might have to 8 I would like to communicate and 52.46 9.84 32.79 1.64 3.28 work together with scientists, government staff and organizations to find a solution that works for everyone

Table 2.4. Predictor variables included in generalized linear models to identify determinants of pastoralists attitudes (n=61) toward leopards in the Yala region. Models shown are those that were within 2 AICc units of the model with the lowest AICc, with associated degrees of freedom (df), AICc, rAICc and AIC weight. Model df AICc rAICc AIC weight Importance of conservation + awareness of general 6 236.6 0.00 0.152 leopard ecology + awareness of leopard-related tourism + socio-demographics Monthly expenditure + awareness of general leopard 6 236.7 0.14 0.141 ecology + awareness of leopard-related tourism + socio- demographics Importance of conservation + awareness of general 5 237.0 0.39 0.125 leopard ecology + awareness of leopard-related tourism Monthly expenditure + awareness of general leopard 5 237.8 1.26 0.081 ecology + socio-demographics Importance of conservation + awareness of general 5 238.0 1.43 0.074 leopard ecology + awareness of leopard-related tourism

22 2.3.2 Attitudes towards leopards in the Central Hills In contrast to Yala, respondents in the Central Hills generally displayed more neutral attitudes towards leopards, as seen by their responses to statements 1, 2, 6 and 7, noting that there were high levels of support for two negatively-framed questions (statements 4 and 5; Table 2.5). However, the statement that scored the highest was positive in nature: I would like to communicate and work together with scientists, government staff and organizations to find a solution that works for everyone. This spirit of collaboration was evident in the fact that 86.5% of respondents desired increased government involvement in facilitating cattle rearing. Similarly, the top-ranked regression model (Table 2.6) suggested that a desire for increased government involvement in improving cattle rearing was positively associated with attitudes towards leopards (0.31, 95% CI [0.16, 0.46]). However, the top model displayed patterns in the residuals, indicating a lack of model fit, possibly due to model misspecification (i.e. the variables included were insufficient to adequately explain respondents’ attitudes towards leopards). The adjusted proportion of deviance accounted for by the top model was 0.21.

Table 2.5. Percentage of responses to 8 statements relating to attitudes towards leopards in the Central Hills (n = 52), scored on a Likert scale. Responses are from the raw data, prior to creating weighted composite scores for each individual through EFA. No. Statement Strongly Agree Neutral Disagree Strongly agree disagree 1 The nature and wildlife of Sri Lanka is a 13.46 0.00 78.85 7.69 0.00 national treasure and should be conserved 2 I respect leopards for the economic value they 9.62 0.00 82.69 7.69 0.00 bring to the country through wildlife tourism 3 My livelihood is more important than current 57.69 38.46 3.85 0.00 0.00 leopard populations 4 It does not matter if a leopard kills a few of 0.00 0.00 15.38 7.69 76.92 my cattle, they are wild animals trying to survive 5 At this farm we cannot tolerate leopards 75.00 3.85 19.23 0.00 1.92 killing any cattle at any time 6 I would be happier if there were fewer 11.54 19.23 61.54 7.69 9.62 leopards around where I live and raise my cattle 7 I do not want to kill leopards, but if they kill 15.38 3.84 76.95 0.00 3.84 my cattle I might have to 8 I would like to communicate and work 86.54 3.84 9.62 0.00 0.00 together with scientists, government staff and organizations to find a solution that works for everyone

Table 2.6. Predictor variables included in generalized linear models to identify determinants of pastoralist (n = 52) attitudes toward leopards in the Central Hills. Models shown are those that were within 2 AICc units of the model with the lowest AICc, with associated degrees of freedom (df), AICc, rAICc and AIC weight. Model df AICc rAICc AIC weight Government involvement 3 214.7 0.00 0.204 Government involvement + awareness of general leopard 4 215.1 0.39 0.168 ecology Government involvement + awareness of general leopard 5 216.4 1.76 0.085 ecology + worry Government involvement + sightings 4 216.5 1.85 0.081 Government involvement + importance of conservation + 5 216.6 1.90 0.079 awareness of general leopard ecology

24 2.3.3. Constraints to livestock rearing faced by pastoralists in Yala and Central Hills Interviews with a subset of respondents highlighted differences in key constraints and issues faced between pastoralists in our two study regions.

2.3.3.1 Yala The most recurring issue raised by survey respondents in Yala was their lack of ability to lease (and therefore develop) the land on which they raise their cattle (Table 2.7). Respondents were in consensus that installing stronger and taller enclosures for their cattle would better protect them from leopard attacks, and as respondent Y26 claimed, calves “provide an easy meal” for leopards. However, they are unable to build any semi- permanent structures due to the lack of land permits and leases. This problem has persisted for decades, and some respondents seemed to accept incidents of depredation as a consequence of living in this landscape. As respondent Y44 said: “My father worked here, I work here, my son will work here. We have to get used to this.” All respondents knew about the CSR project involving the distribution of steel enclosures to certain pastoralists (Fig.2.6). Many felt that this was the best option to reduce depredation risk, compared to the standard penning method using a combination of thorn bushes and barbed wire (Fig.2.5), as the steel enclosures were stronger and are able to be dismantled and moved with relatively little effort. As respondent Y18 said:

“Separating the cattle isn’t enough, we have to put them in pens with a strong roof above – a leopard can climb a tree nearby and stalk. There are too many trees around that leopards can always go up”.

The lack of leases also means that competition exists for grazing lands, and many pastoralists now operate within certain areas to limit potential clashes with those having larger herds. This same issue of competition was brought up as a contributing factor to why many are not willing to relocate to potentially better areas further away from YNP.

Fig.2.5. Common makeshift pens for cattle that offer little protection against leopards.

Fig.2.6. Steel enclosure with roof given to certain pastoralists in the Yala region.

The interviews revealed that rearing cattle in this landscape has become increasingly harder due to the development of the land for hotels. Before the end of the civil war (1983-2009), there was less development, fewer vehicles and pastoralists were allowed to camp by their cattle pens overnight. They have since been banned from such

26 camping. The combined effects of not being able to perform patrols of their cattle at night, and the lack of ability to lease the land and develop semi-permanent structures, were common themes of frustration expressed by many respondents. As respondent Y14 stated:

“Before [when the civil war was taking place] there weren’t as many people or hotels. It was easier to have more cattle. Now there are more people, more roads and less foraging lands”.

As a result of the development, sections of the land are visibly overgrazed and many cattle are malnourished (Fig. 2.7) with less access to fresh water, causing them to produce less milk.

Fig.2.7. Photo of an underweight cow in the Yala region. Many individuals observed in this study site were similarly underweight.

Our interviews with pastoralists illuminated a common distrust many respondents felt with the Sri Lankan government’s Department of Wildlife Conservation (DWC), which has jurisdiction over the buffer zone habitats adjacent to Yala National Park.

27 Historical unrest and distrust have existed between these two groups, as cattle rearing is illegal within a certain buffer of the boundary, but has persisted for multiple generations. As respondent Y32 said:

“many of us have gone to the department about these problems, but they are not interested. We used to be allowed to stay in the night near our pens but the department banned us from doing that. They do not want us to change our [husbandry] methods, since guard dogs will require more food, more permits will be needed from them and as the number of herders increases more land would be needed. Still, on the 31st of each year we give milk to the DWC to give thanks”.

Pastoralists told us that even the presence of one or two open-minded DWC staff have made a difference in the communication they have had with the department. However, lack of continuity and turnaround times have resulted in little progress in increasing communication. The surveys revealed that when an incident of depredation occurs, only 9.84% of respondents reported this to the DWC for verification. Due to the lack of good faith, far more either ignored the depredation (37.7%), tied up the carcass (18.0%), buried the carcass (14.8%), or poisoned it (4.92%).

28 Table 2.7. Summary of key issues analysed from individual and group interviews with a subset of respondents in Yala (n=32) and Central Hills (n=19), Sri Lanka, accompanied by an illustrative quote. Study site Key issue Illustrative quote and respondent number Yala Lack of ability “Even though we have the ability, we cannot build a to own or better pen. We need a roof that is strong, the leopard I develop land saw [on top of one of the steel enclosures] came in the night and was lying on the roof until morning”; Y23 Yala Post-war “Before [when the civil war was taking place] there development of weren’t as many people or hotels. It was easier to have surrounding area more cattle. Now there are more people, more roads and less foraging lands”; Y14 Yala Distrust of the “More than 100 people have approached the DWC Department of about this issue [leopards, lack of land] but we are not Wildlife taken seriously anymore”; Y15 Conservation (DWC) Central Infrastructure “Even when we pay to hire a vehicle and for the Hills doctor fee, the doctor will not come because of the bad roads. It is wasted cost and many cows and calves die from infection”; CH14 Central Loss of dogs “We know many other people who have lost dogs. We Hills heard of leopards coming near to people in other areas, but so far not here”; CH51 Central Lack of ability “We are on a private estate, so we cannot lease this Hills to own or land. These sheds for our cattle are extremely develop land expensive to build”; CH42

2.3.3.2 Central Hills The most recurring issue brought up by respondents in the Central Hills was the need to improve infrastructure in order to better rear their cattle. Specifically, improvements are desired for the roads leading to the tea estates, which are unpaved and extremely bumpy (Fig.2.8), and the sheds in which cattle were kept (Fig.2.9). Inaccessible locations due to poor roads mean that pastoralists do not have access to proper veterinary care, a serious issue given the high rate of miscarriages, infections or hypothermia in their cattle. As respondent CH48 said:

29 “The vet doesn’t come up here even if we get together and have the money to pay for his transport, across the lake and up to here. We have tried so many times. I have lost 2 cows to miscarriages”.

In the Central Hills, the breeds of cattle are much larger and more productive, thus pastoralists had on average 3.5 individuals (compared to 84 in the Yala region). Poor roads mean milk collection services are reduced as vehicles are unable to traverse the roads, forcing many to travel long distances to drop off milk, often multiple times a day.

Fig.2.8. Poor roads that most common modes of transport often refuse to travel on.

Fig.2.9. A typical cattle shed in the Central Hills that lacks adequate drainage systems, causing issues regarding the disposal of sewage water.

Interviews with respondents also revealed that many had lost dogs to leopards and fishing cats, though many could not give exact numbers of dogs lost to each species. These dogs were not used to guard the cattle from predators, but to warn the respondents if animals such as deer or wild boar were approaching the cattle shed, where they keep the large quantity of food (50-150kg) needed to feed the cattle for a day. No respondent claimed to have any specific issues with leopards apart from experiencing dog loss; however, multiple respondents claimed to have seen leopards around the estate, particularly in higher elevation ridges further away from their houses. As respondent CH11 said:

“We have heard leopards are coming closer to people in other areas, but not here. They have taken many dogs in this estate, two of mine”.

The inability to own or lease land on the tea estates was another recurring issue in the Central Hills. Rearing cattle on the sampled tea estates was a sort of unwritten agreement, but as a result no substantial development of cattle sheds could take place. Many respondents struggle to acquire funds to build these sheds, which often cost

31 between Rs. 200,000-300,000 to construct. The lack of space results in the inability to install a filtering system to manage the high volumes of sewage. A main issue was this inadequate drainage of dung water, which inevitably flows downstream into rivers and reservoirs. The police in these areas are said to be strict, and fine pastoralists up to Rs. 25,000 if this is detected. Eight respondents stated they had to sell one of their cows in order to pay for these fines. Several respondents were keen that the government should start a loan program (as opposed to bank loans, which have higher interest and more restrictions) to allow them to adequately rear their cattle.

2.4 Discussion 2.4.1 Determinants of attitudes towards leopards Yala and the Central Hills differ in terms of history, socio-demographics, habitat, issues faced, husbandry techniques practiced and cattle characteristics. Unsurprisingly, determinants of attitudes towards leopards differed as well. Importantly, there were no reported instances of cattle depredation by leopards in the Central Hills, whereas 93% of respondents in the Yala region reported some level of leopard depredation. The fact that attitudes towards leopards in Yala were more negative with increased awareness of leopard ecology and leopard-related tourism benefits is contrary to findings that have shown increased knowledge of conservation or about a species being a predictor of pro- environmental attitudes (Klöckner 2013; Kansky and Knight 2014). Other studies on attitudes regarding conflict have revealed similarly counter-intuitive insights, such as how a subset of pastoralists who have experienced a higher amount of damage displayed more positive attitudes towards carnivores, contrary to assumptions that damage drives negative attitudes (Dickman 2010). We suggest that the more respondents knew about how leopards use this shared landscape, the more they could anticipate conflicts with their own cattle and livelihoods. During conversations with respondents, many said that leopards prefer calves over native prey. Respondents claimed that domesticated cattle do not react to leopards the same way deer, boar, langurs and hares do. This perception of leopard prey selection made those respondents feel more threatened in terms of their calves and the stability of their livelihoods, even if native leopard prey was abundant in the area.

32 Similarly, being more aware of the economic value leopards bring to the country influenced respondents to be less favourable towards leopards. Particularly in YNP, where leopard sightings are a main tourist attraction, those who face the realities of living alongside leopards expressed their frustration that they receive no compensation and few benefits from the tourism industry. Given that the benefits go to the nearby hotels and government, the increased awareness of benefits combined with lack of involvement in the tourism industry could lead to hostilities and resentment felt by pastoralists. This finding indicates that developing tourism programs that include pastoralists could be an important step towards improving attitudes towards leopards. Examples may include eco- tourism programs or finding employment in safari camps. This is supported by other studies that have shown attitudes in favour of human-carnivore coexistence were correlated with benefits from ecological tourism (Hemson et al. 2009; Lindsey, Du Toit, and Mills 2005).

When respondents desired the conservation of all wildlife species around them, they had more favourable attitudes towards leopards. The wildlife listed included mammals, birds, reptiles and mammals. Similarly, as socio-demographic metrics increased (age, number of dependents and number of years spent rearing cattle), attitudes towards leopards were more favourable. Similar findings have been shown in other studies that found older respondents were more likely to be tolerant of depredation incidents (Mkonyi et al. 2017). We suggest that increased age, and thus a longer presence in the local landscape, imparts older respondents with more perspective of the realities of living in this shared landscape, and have come to accept it as a natural occurrence when rearing cattle in this habitat. Likewise, multiple respondents with young children often told us stories of how happy it made them and their children to see wildlife around them, especially “cute” species such as the rusty spotted cat (Prionailurus rubiginosus).

In contrast to the distrust and frustration felt toward government in Yala, respondents in the Central Hills expressed more hope that the government would lend assistance in resolving conflict with leopards. This difference is likely driven by poorer government relations over longer periods of time among Yala pastoralists than those in

33 the Central Hills, where a desire for government involvement was the only predictor of attitudes towards leopards in the top-ranked model. Pastoralists in the Central Hills rear cattle on tea estates with the agreement of the estate superintendent, though they also do not have the ability to own this land. This is the one common element that came out of interviews in both study sites: the lack of opportunities to own and develop the land they rear livestock in. Nevertheless, our model results were weaker for the Central Hills, which may be due to the fact that the survey was designed based on similar conflict surveys done in communities where carnivores and communities have come into conflict. We were expecting more issues with leopards to be present in the Central Hills communities but found that respondents did not have issues with leopards as much as they did with other aspects of raising cattle (e.g. infrastructure, access to medication and veterinary care).

2.4.2 Potential strategies for human-leopard coexistence Our results from Yala indicate that being unable to own or develop land, along with distrust of the DWC, were barriers to human-leopard coexistence. Increasing communication and familiarity between pastoralists and the DWC may help address this barrier. Developing a training program on the identification and reporting of depredation incidents could facilitate a more positive relationship between pastoralists and the DWC, as depredation incidents rarely get reported currently. Individuals and institutions involved with any level of conflict should undergo training and education regarding the development and implementation of government procedures, local procedures and programs, and what relevant governing laws exist (Madden 2004). This is, however, contingent on the long-term capacity of the DWC, which may currently lack funding and staff for such programs. We also suggest clarifying the currently loosely-defined “buffer zone” term from the Flora and Fauna Protection Ordinance, as well as the respective boundaries between DWC and Forest Department (FD) jurisdiction, and for areas on the landscape to be delineated appropriately. Only after this point can any land use agreements be made with the DWC, FD and pastoralists, so it remains an important first step. Given that respondents felt frustrated with a perceived lack of communication and assistance from the government, we also suggest greater efforts be made to create a more

34 open dialogue between pastoralists, DWC, the dairy co-operative and relevant NGOs that can provide support for facilitating coexistence and building capacity.

A potential coexistence measure that can be considered is to grant restricted land ownership to pastoralists. This was the most common constraint identified by respondents in the Yala region, with 93% of respondents desiring some form of land rights and ability to build structures to better house their cattle. Group land ownership results in community members sharing the risk of cattle loss, and has been seen to facilitate coexistence between humans and wildlife in Peru (Naughton-treves et al. 2003). However, the impact of granting restricted land ownership in ecologically sensitive regions (such as buffer zones) should be considered. For example, in Yala this could involve a requirement that owned land be a certain minimum distance away from YNP, and a limit on cattle herd size in order to reduce potential overgrazing pressure.

Pastoralists in Yala face high conflict but receive no benefits from leopard-related tourism, despite being aware of the regional benefits. This is a key challenge outside protected areas: how to get local communities to view large carnivores as net economic assets despite experiencing livestock depredation and loss of livelihood (Dickman, Macdonald, and Macdonald 2011). A more equitable distribution of benefits is key, and has been a central theme in initiatives that aim to balance conservation and development. In theory, CBNRM will allow communities to benefit economically from resource management in ways that can contribute to rural development, provide stable tenure over land and resources, and help to conserve biodiversity. The Communal Area Management Programme for Indigenous Resources (CAMPFIRE) program in Zimbabwe is an example of CBNRM recognised for having initial success due to transparency, investment in capacity-building of communities, and high levels of accountability between community leaders (Taylor 2009). CAMPFIRE inspired other programs of benefit sharing and devolution of wildlife management from the state to the community. Success in such programs has been demonstrated in Namibia, where the CBNRM program is credited with playing a major role in the recovery of wildlife across communal conservancies, while also generating revenue to the communities through tourism activities and

35 sustainable wildlife use, particularly trophy hunting (Störmer et al. 2019; Naidoo et al. 2016). However, compared to many countries in Africa, using wildlife tourism to generate revenue to share with communities is a relatively newer concept in South Asia, and potential CBNRM programs should be evaluated appropriately (Sekhar 2003).

Similarly, ICDPs attempt to strengthen relationships between state-managed protected areas and adjacent communities, but typically without granting land or natural resource ownership to those communities. In the Annapurna Conservation Area in Nepal, ICDPs have built capacity and increased interest in conservation amongst local communities, over a period of 10 years (Baral, Stern, and Heinen 2007). Both CBNRM and ICDP vary across contexts, but criticisms have highlighted the instability of funding, insufficient alternative economic benefits provided to communities, and failure to achieve conservation outcomes once institutionalised and put into practice (Pailler et al. 2015; Cernea and Schmidt-Soltau 2006). Therefore, such approaches exist as potential options for the Yala pastoralist community, but should be considered with caution and in collaboration with local NGOs, government departments and communities who have the necessary knowledge and context to determine whether adequate capacity for such programs currently exists. Wildlife Premium Mechanisms are another option to provide economic incentives to encourage pastoralist communities to conserve leopards outside the national parks (Dinerstein et al. 2013). For example, communities that received economic benefits through tourism in Nepal were more tolerant of tiger presence (Dinerstein et al. 1999). We therefore recommend consideration of a similar program for leopard conservation in Sri Lanka.

Compensation schemes are another method used to alleviate loss of income due to cattle depredation. It has been implemented in places such as Brazil, India and Kenya (Maclennan et al. 2009; Marchini and Macdonald 2012; Mishra 1997). In our study, 34% of respondents expressed an interest in receiving such compensation, but no such scheme currently exists for cattle depredation in Sri Lanka. Given the complex, multifaceted nature of conflict, compensation schemes should be part of a larger solution including other mitigation efforts that cover a wider range of social, biological and economic

36 aspects (e.g. awareness and education programs, community-funded maintenance and veterinary programs). Pastoralists in the Yala region estimated a payment of around Rs. 100,000 (USD 550) would be appropriate for each cattle confirmed lost to depredation, based on the profit earned per litre of milk (Rs. 70/L) over 7 months of the year. However, compensation schemes may be prone to “problems of perverse incentives” (Dickman, Macdonald, and Macdonald 2011). For example, pastoralists may demand an increasing price paid per animal, or new pastoralists may be incentivised to rear cattle in this landscape. Compensation schemes may also be ineffective if there are barriers to on-site verification or delays in payments, leading to an unsustainable system (Jackson and Wangchuk 2004).

An alternative to compensation that may be more suitable for Sri Lanka is a community-based livestock insurance program, similar to the program set up in the Spiti Valley, northeastern Himalayas, in an effort to improve coexistence between pastoralists and snow leopards (Mishra et al. 2003). This has also been practiced in the Kafa highlands, Ethiopia, where communities were shown to report livestock losses to their neighbours more than to the local administration (Gebresenbet et al. 2018). However, initially these programs require incentives, start-up money and monitoring efforts, which at this current time do not exist. This may be a future step, and provides an opportunity for the tourism sector to get involved with more comprehensive CSR programs.

Though there were no incidents of cattle depredation in the Central Hills, pastoralist attitudes towards leopards were related to a desire for increased government involvement, which presents an opportunity to stay ahead of conflict problems if cattle farming increases in the future. Greater involvement from the government – which is promoting domestic dairy production – to help alleviate some key challenges faced by the communities may help sustain these positive views. Improving estate roads and milk collection services, subsidizing prices for nutritional cattle feed and offering improved loans to help provide cattle sheds with adequate drainage are all options with local support that could be considered. Unlike Yala, pastoralism in the surveyed tea estates is contained in small pockets where cattle do not graze, but remain inside small sheds. Therefore, there

37 is less risk of ecological damage as it is unlikely pastoralists will increase the number of cattle, but rather focus on improving the productivity and health of each individual.

2.4.3 Conclusion It is common for trade-offs to occur between biodiversity conservation and human well-being, particularly in tropical areas (Salerno et al. 2016). Engaging pastoralist communities in conservation is crucial; human-carnivore coexistence strategies that involve, and have even been ‘co-designed’ by, local pastoralists may have a greater impact than those based on research findings from outside scientists with little local engagement (Treves et al. 2006). By attempting to understand attitudes towards leopards and the greater context of cattle-rearing, the coexistence strategies we recommend acknowledge and incorporate the experiences of the pastoralist communities. By undertaking this important first step, it is our hope that these strategies will have a greater chance of long term success (Hill 2004).

We recognize that pastoralist communities and wildlife will likely continue to exist in this landscape, and suggest coexistence strategies that take into account the local context, attitudes and issues communicated to us by respondents. Issues of cattle depredation in Yala and issues of infrastructure and cattle health in the Central Hills should be addressed. Our results suggest that policies aimed at these communities towards reducing incidents of cattle depredation will be more successful if they consider the attitudes and issues experienced by the respective communities affected. The coexistence strategies suggested here may facilitate positive attitudes towards leopards, but it is imperative to consider other factors that were not included in this study, such as personal and social motivations, behaviours and actions towards leopards, barriers to retaliatory killings, and whether a culture of tolerance or beliefs in long-term positive wealth impacts exist (Marchini and Macdonald 2012; Gebresenbet et al. 2018). Our study highlights that the relationships between attitudes and conflict are context-dependent and cannot be easily generalized. However, we recognize a need to move beyond snapshots of current attitudes towards longer term studies that evaluate how attitudes change with changing social and environmental factors, in order for comprehensive coexistence strategies to be developed and implemented successfully.

38 Chapter 3. Drivers and potential mitigations of livestock depredation by the Sri Lankan leopard (Panthera pardus kotiya)

3.1 Introduction 3.1.1 Human-carnivore conflict Large carnivores can greatly influence biological communities through predation and interspecific competition, regulating abundance of prey and smaller carnivores, and thereby having the potential to impact the structure of entire ecosystems (Berger et al. 2001; Estes et al. 1998). Generally, carnivores are particularly sensitive to anthropogenic threats, as their positions at higher trophic levels and requirements for extensive spatial ranges often restrict them to populations of lower density (Cardillo et al. 2004). Furthermore, large carnivores are prone to conflict with humans and are potentially sensitive to fragmentation; the increasing lack of large, connected habitats makes them inherently vulnerable to stochastic events. Wide-ranging species in particular are more exposed to threats on reserve borders, and are likely to disappear from smaller reserves faster, regardless of population size (Woodroffe and Ginsberg 1998). As a result, carnivore conservation and management is a growing concern (Ripple et al. 2014). A leading cause of carnivore populations declines is human-carnivore conflict (hereafter ‘conflict’), defined as carnivore-related threats to the livelihood and safety of a person or community, resulting in persecution of that carnivore species (Inskip and Zimmermann 2009).

Globally, conflict arises for many reasons. The large home ranges and protein-rich diets of carnivores often causes them to move outside protected areas and into more human- dominated landscapes (Treves and Karanth 2003). Additionally, many carnivores are specialised in ungulate predation, leading to instances of wild carnivores killing domesticated ungulates (e.g. cattle, sheep), when conditions are favourable to do so (Karanth, Sunquist, and Chinnappa 1999; Polisar et al. 2003). There are also cases where humans are killed by large carnivore species, such as lion attacks in Tanzania (Packer et al. 2005) and tiger and leopard attacks in India (Dhanwatey et al. 2013), illustrating the complexities in balancing conservation of carnivore populations with ensuring the safety of human livelihoods. A leading cause of carnivore decline is conflict with people and their

39 livestock (i.e. livestock depredation). Carnivores are often the victims of preventative or retaliatory killings (Ripple et al. 2014; Inskip and Zimmermann 2009; Ogada et al. 2003), which result from the product of complex sociological norms and attitudes, socio- economic standing, and the relative risk to livelihoods (Treves and Karanth 2003; Inskip and Zimmermann 2009; Gálvez et al. 2018).

Factors affecting conflict vary between contexts but include: previous exposure to conflict, livestock ownership, husbandry practices, historical grievances, native prey abundance and predator behavioural ecology (Dickman 2010; Hazzah, Borgerhoff Mulder, and Frank 2009; Inskip and Zimmermann 2009; Bagchi and Mishra 2006). Increased understanding of the ecological variables associated with livestock depredation can guide management policies and identify further areas of study. However, due to differing social and ecological contexts, there is unlikely to be a universally applicable conflict mitigation strategy, and potential strategies need to be tailored to site-specific circumstances (Loveridge et al. 2017). Conflict mitigation studies are particularly lacking outside protected areas, creating a barrier to the design of effective conservation and management plans (Samarkoon 1999). Such plans are particularly needed in tropical biodiversity “hotspots”, where threatened and endemic large carnivores compete with pressing needs for economic development (Brooks et al. 2006; Myers et al. 2000).

The leopard (Panthera pardus) is one mammalian carnivore that has been associated with livestock depredation and increasing levels of conflict (Kolowski and Holekamp 2006; Khorozyan et al. 2017). Leopards are the most adaptable and widespread large felid, remaining extant in areas where other large felids have been extirpated (Jacobson et al. 2016; Stein et al. 2016). The relative persistence ultimately increases the potential for interactions with expanding human populations. Of the nine leopard subspecies, five are classified as endangered or critically endangered by the IUCN (Jacobson et al. 2016) and increased conflict puts these leopard subspecies at even greater risk (Inskip and Zimmermann 2009). Even if livestock losses due to leopard predation are numerically low, this can be a significant challenge for vulnerable communities who are dependent on livestock for their livelihood (Dickman 2010) as perceived harm to livelihood or lives will

40 often result in retaliatory attacks on leopards (Woodroffe, Thirgood, and Rabinowitz 2005).

3.1.2 Identifying key drivers of livestock depredation Incidents of leopard-livestock conflict are driven by the decision of the leopard to take livestock as prey. Studies have found the frequency of depredation by carnivores at individual farms to increase with the number of livestock owned by a farmer (Hemson et al. 2009; Bradley and Pletscher 2005), and the length of time a farmer has spent rearing livestock in the landscape (residency time; Mkonyi et al. 2017; Arjunan et al. 2006). Predators have also been shown to prey upon livestock in areas where livestock densities are higher than native prey densities (Bagchi and Mishra 2006; Woodroffe et al. 2005).

Depredation has been shown to decrease with increasing distance from protected area boundaries (Nyahongo, and Røskaft 2007; Patterson et al. 2004). For example, one study reported that depredation by large felids occurred primarily within 3km of Serengeti National Park, Tanzania (Holmern, Nyahongo, and Røskaft 2007). It has also been suggested that predator willingness to approach livestock enclosures may be influenced by distance from forest cover (Ogada et al. 2003). Despite encountering higher risks of mortality, leopards have been shown to move outside protected areas and into habitats where prey were easier to catch compared to areas where prey were more abundant (Balme, Slotow, and Hunter 2010). Depredation has also been shown to decrease with improved husbandry (e.g. characteristics of livestock enclosures) (Woodroffe et al. 2007; Ogada et al. 2003). Livestock husbandry, particularly poor forms of husbandry, contributes to incidents of depredation (Wang and Macdonald 2006; Mkonyi et al. 2017; Treves and Karanth 2003). Carnivores may avoid areas of human development, and as such, roads have also been suggested to be negatively related to depredation by large felids, in terms of distance from the nearest road (Soh et al. 2014) and density of roads within a buffer (Treves et al. 2004).

Native prey abundance has also been shown to be negatively related to livestock depredation, as leopards may be more inclined to target domesticated prey when native

41 prey availability is lower (Khorozyan, Ghoddousi, et al. 2015; Bagchi and Mishra 2006). This has been seen in the Ahmednagar district in Maharashtra, India where scat analysis determined that 87% of leopard prey biomass was comprised of domestic animals (livestock as well as domestic dogs and cats) (Athreya et al. 2016). Diets of large cats can shift in response to changes in the abundance of available prey species (Kitchener 1991), with leopards tending to preferentially prey upon species that are 10-40kg in weight (Hayward et al. 2006). Predators may also be enticed to prey upon livestock due to the latter’s lack of defence mechanisms and escape abilities comparable to wild prey (Jackson and Nowell 1996).

Globally, pastoralists and leopards frequently share the same landscapes, often adjacent to national parks (Kabir et al. 2014; Shehzad et al. 2015; Kissui 2008; Chase Grey, Bell, and Hill 2017). In general, depredation incidents largely occur during the night, where livestock are in poorly constructed pens and in the absence of artificial lights. In many cases, this leads researchers to promote strategies to improving livestock corrals/enclosures, restricting grazing times, improved government responses to reported depredation and increased vaccination programs to improve livestock health.

3.1.3 Sri Lankan context Sri Lanka is another context where landscapes are being shared between leopards and pastoralists nationwide. The Sri Lankan leopard (Panthera pardus kotiya) (Fig.3.1) is an endangered, endemic subspecies (Stein et al. 2016; Miththapala, Seidensticker, and O’Brien 1996). It is the only leopard subspecies that is the apex predator throughout its range (Turner 1997), having evolved in the absence of intra-guild competition for 5000- 10000 years (Miththapala, Seidensticker, and O’Brien 1996). Apex predators, even at low densities, can greatly affect composition and structure of the ecosystems they are in (Ripple et al. 2014). Thus, the Sri Lankan leopard is of potentially profound ecological significance and therefore high conservation value. The lack of competition has resulted in subtle ecological and behavioural differences compared to other leopard subspecies, such as an apparent preference for larger prey species (Kittle, Watson, and Fernando 2017). These wide-ranging leopards, which occupy all habitat zones in Sri Lanka, may serve as useful

42 “umbrella species”, meaning that protecting viable leopard populations directly helps to protect other spatially restricted species (Kittle et al. 2017).

Fig.3.1. Female Sri Lankan leopard (Panthera pardus kotiya) seen in Yala National Park, June 2018.

An expanding dairy cattle industry may be an emerging threat for leopards in Sri Lanka. Following the 2004 South Asian tsunami and the Sri Lankan civil war (1983-2009), dairy cattle farming in Sri Lanka has been promoted as a means of improving livelihood and broadening development (Barlow 2011; Ranaweera 2009; Development 2013). The focus of these efforts are targeting the large rural population of the country (Sri Lanka Department of Census and Statistics 2012). Recent economic growth has led to a push in domestic production to meet increased demands for cow milk and associated milk products (Perera and Jayasuriya 2008; Vernooij, Houwers, and Zijlstra 2015). Most of the dairy production growth will come from smallholder farms (as opposed to large, factory farming operations), with many now increasing from small to medium size (Vernooij, Houwers, and Zijlstra 2015). These farms exist throughout the country, but are of particular concern when they occur in lands adjacent to protected areas where leopards also occur.

43 Economic development and poverty alleviation are essential, but for long-term sustainability they must allow for coexistence with valued wildlife species, such as leopards, that compete with humans for habitat and potential prey (e.g. cattle). Physically separating humans and carnivores through wildlife-proof fencing may be an effective strategy to preserve viable populations (Packer et al. 2013), but may be unrealistic in many contexts. Effective conservation of healthy carnivore populations will likely depend on landscape-level initiatives that reduce conflict outside protected areas (Creel et al. 2013). Understanding site-specific challenges and developing appropriate coexistence strategies is therefore imperative to recover large carnivore populations while ensuring human well- being and livelihoods in shared landscapes (Carter and Linnell 2016). Livestock depredation by leopards (specifically of calves) occurs in Sri Lanka, despite makeshift mitigations and the presence of natural prey species. Levels of livestock depredation have yet to be quantified, and as a result it is currently unknown what factors are related to depredation incidents. As a result of economic losses, pastoralists sometimes take revenge through preventative or retaliatory killing of leopards using snares, shooting or poison (Fernando 2016).

3.1.4 Objectives and hypotheses Given the concerted effort to expand the dairy industry and increase production in Sri Lanka, human-leopard conflicts will likely rise unless practical mitigation efforts to reduce potential conflict are tested and implemented. To help facilitate trust and build a relationship with pastoralists, we wanted to also provide tangible options for conflict mitigation or reducing leopard-related losses (Treves et al. 2006). Our objectives were to 1) quantify levels of livestock depredation and model this against hypothesised causal factors, 2) assess current livestock husbandry practices, and 3) gauge community support for proposed mitigations that could be tested and potentially implemented in the future to reduce livestock depredation.

We used surveys, interviews, farm observations, and camera trapping to collect data on leopard depredation of livestock and the factors hypothesized to influence incidents of depredation. We hypothesized that livestock depredation would be related to the following

44 variables: distance of farm to protected forest, native prey availability, livestock husbandry, number of cattle, road density and pastoralist residency time (Table 3.1).

Table 3.1. Predictor variables measured and used in regression models to explain observed variation in reported incidents of leopard depredation on livestock in Yala, Sri Lanka, with specific hypotheses and explanation for expected patterns. Variable Hypothesized Explanation relationship to depredation Prey biomass Negative Leopards may go after domestic prey (livestock) if availability native prey is limited Livestock Negative Stronger, taller fences and protections (enclosures) will husbandry prevent leopards from depredating livestock Distance to Negative Yala National Park is a stronghold for leopards and protected forest likely a source of individuals depredating nearby farms Road density Negative Vehicle traffic and noise on roads deters leopards from approaching farms Number of Positive Greater numbers of livestock may attract leopards to cattle areas where cattle are kept overnight, and increase depredation risk Pastoralist Positive Pastoralists raising livestock in this landscape for longer residency time would have experienced more incidents of depredation and are able to better identify leopard attacks

3.2 Methods 3.2.1 Study area Our study area was the Yala region (hereafter Yala; Fig.3.2.), as described in Chapter 2, where pastoralists rear and graze their livestock in close proximity to Yala National Park (YNP) which has one of the highest recorded leopard densities in the world (Kittle, Watson, and Fernando 2017). While it is known that these pastoralists (Fig.2.1, Chapter 2) have experienced livestock depredation, the frequency of depredation has not previously been quantified. Pastoralists owned between 20-350 cattle, with the average number of cattle per pastoralist being 84 with a standard deviation of 70, in the 71km2 study area.

45 3.2.2 Surveys and semi-structured interviews We used in-person surveys and semi-structured interviews (hereafter referred to as ‘interviews’) with pastoralists in Yala to quantify and obtain information on livestock and pastoralist characteristics, leopard depredation levels, and mitigations used. Details of these methods are given in Chapter 2 (see also Appendix A and B). Data on reported livestock depredation in the past 3 years were obtained from survey respondents. We used the reference period of 3 years as this coincided with a change of government, and during pre-testing of surveys, more respondents were familiar with this compared to a 1-year reference point. Potential mitigation measures were at times adapted following feedback from pastoralists, making them more feasible and contextually appropriate (low cost, low maintenance and mobile). Mitigation options evaluated in our survey included: using trained guard dogs, increasing human presence while livestock are grazing, using higher thorn fencing, increasing human patrols during the night (when reported depredation incidents take place most often), relocating cattle to lower-risk areas (e.g. further from the national park boundary), using PVC-lined enclosures, using light and sound distractions, and adopting better husbandry techniques in the form of reinforced chain-link enclosures. The different causes of livestock loss (e.g. disease, depredation, theft) were also quantified in order to assess the relative impact of leopard depredation.

3.2.3 Farm visits and observations of husbandry practices During visits to cattle farms, we collected information on the type of husbandry practices and enclosures used by pastoralists. Photos were taken to document general sizes, structures, and materials used in cattle enclosures, in order to verify what respondents reported to us. The field team spent a total of three months between the two sampling periods living in the community and regularly engaging with pastoralists. This helped to build trust between community members and ourselves, giving us greater confidence in the data collected (Hazzah, Borgerhoff Mulder, and Frank 2009), especially during the unplanned conversations we had with many pastoralists who recognized us from before. We created a husbandry score for each respondent. Each observed type of husbandry (specifically livestock enclosure) was assigned a value, where the lowest score represented thorn pens lined with barbed wire, and the highest score represented a steel chain-link

46 enclosure with a firm roof overhead (indicating the structurally strongest form of livestock penning; see Figs. 3.3a-d).

3.2.4 Prey availability surveys and predictive biomass models To estimate the availability of wild prey for leopards in the study area, we set remote camera traps (Bushnell Trophy Cam, Essential E2 and E3) in the dry season from May-August 2018 and June-August 2019, (see Appendix F for camera trap deployment data sheet). Relevant permits (WL/3/2/1/4/1) were obtained from the Sri Lankan Department of Wildlife Conservation (DWC) prior to camera setup. We had a total of 28 camera stations for 694 active days, ranging from 8 to 55 days (mean of 26 days; shorter durations due to camera theft). At each location, a single camera was placed 40-70cm above the ground, and angled depending on the slope of each location, with the aim of detecting species ranging from buffalo to small rodents (rats, squirrels, hares). Cameras were intended to be distributed evenly across the landscape to encompass varying habitat types, dominant vegetation species and distances to water. However, cameras were ultimately placed opportunistically within the focal area, due to constraints imposed by the required DWC approval, with certain areas being avoided due to high risk of theft (Fig. 3.2).

Camera trap images were processed using Timelapse2 software (Greenberg 2018) to identify species. We identified all individuals by their age and sex classes when possible, in order to more accurately estimate the relative biomass index, and as certain species (e.g. buffalo) are only preyed upon by leopards as juveniles (Kittle, Watson, and Fernando 2017). We calculated a relative abundance index (RAI) for each prey species at each camera trap station, using a time interval of 30 minutes to define independent detections (O’Brien, Kinnaird, and Wibisono 2003). In order to provide a better estimate of food availability for leopards across prey of widely varying sizes, we also calculated relative biomass indices (RBI) for each camera station by multiplying the mean weights of prey species (obtained from the literature and from local field guides; Appendix G) in both adult and juvenile stages by the RAI (Balme, Slotow, and Hunter 2010).

47 To allow us to extrapolate from our camera locations to nearby farms (Fig.3.2), we constructed predictive models of prey RBI using variables from the literature which have been shown to influence prey presence and abundance. Specifically, we considered distance to water (Ogutu et al. 2014), distance to paved road (Fahrig 2010), and forest density (i.e. canopy cover; (Sodhi et al. 2010; Eisenberg and Seidensticker 1976) in a 150m circular buffer around each camera. This buffer size was chosen based on the spacing of farms sampled to encompass the most variation. Habitat variables were extracted using Google Earth. We tested predictor variables using Pearson correlation coefficients (r), using r > 0.5 as the threshold for collinearity).

We used Generalised Linear Models (GLMs) in the ‘stats’ package in R statistical software (R Core Team 2018) to predict prey biomass. We used a Gamma distribution

48 since the response variable (prey RBI) was continuous, positive and right-skewed (Ng and Cribbie 2017), and a log link because the biomass values varied widely. Gamma is suitable when you assume a ‘fixed’ linkage between mean and variance (i.e. small biomass means small variance). We assumed that the number of animals in this area was the same between the two consecutive years sampled, and we did not find any evidence of spatial autocorrelation in model residuals using a Moran’s I test (Moran’s I statistic = -0.97, p = 0.81). We assessed model fit by checking the R2 value of the top model as well as the deviance explained. Using the ‘predict’ function, we extrapolated prey RBI values to each individual farm location given their respective habitat variables, standardised by the mean and standard deviation of the covariates obtained at each camera trap station. The extrapolated prey estimates for each farm were fitted values that were back-transformed into original units (kg/day).

3.2.5 Variation in reported depredation We modelled the number of livestock reported lost to depredation in the past 3 years (2015-2018) using a set of GLM models that included the global model with all predictor variables, and then all possible additive combinations (Appendix I). We used a negative binomial distribution to account for overdispersion in the response variable (Zuur, Hilbe, and Ieno 2013). Covariates were all standardised to a mean of 0 and a standard deviation of 1 in order for direct comparisons to be made. We examined for collinearity among all continuous predictor variables using variance inflation factors (Zuur, Ieno, and Elphick 2010); all variables were retained since VIF values were below 1.4. We tested for spatial autocorrelation in model residuals using a Moran’s I test (Moran’s I statistic = -0.81, p = 0.79; indicating non-significant spatial autocorrelation). We used an information theoretic approach based on Akaike’s Information Criterion, corrected for small sample size (AICc), to rank the top models and identify the most parsimonious model that best fit the data (Mazerolle 2006; Burnham and Anderson 2002). We assessed the adequacy of the top- ranked model using standardised residual diagnostics in the DHARMa package (Hartig 2019) and effect size stability (Heinze, Wallisch, and Dunkler 2018). Estimates of regression coefficients and their 95% confidence intervals are reported, and statistical significance was determined based on whether 95% confidence intervals overlapped with

49 0. We assessed model fit by checking the R2 value of the top-ranked model as well as the deviance explained. We analysed all data using R statistical software, version 3.5.1 (R Core Team 2018).

3.3 Results 3.3.1 Causes of livestock loss Reported leopard depredation accounted for 11.1% of livestock loss. This was the fourth highest cause of reported livestock loss, preceded by theft, disease and cattle wandering off into guarded areas (Table 3.2).

Table 3.2. Causes of livestock loss and the average percentage of livestock loss attributed to each cause, across the 61 survey respondents in Yala, Sri Lanka. Surveys were conducted from May-August 2018. Cause of livestock loss Average percentage Standard of livestock loss deviation attributed to cause Snake bite 1.1 0.035 Elephant 2.0 0.070 Crocodile 3.7 0.073 Dog 6.7 0.115 Leopard 11.1 0.153 Wandering off 13.8 1.156 Disease 21.1 0.167 Theft 29.8 0.298

3.3.2 Livestock husbandry practiced by respondents We observed four main forms of livestock enclosures in the Yala study site: steel chain-link enclosures with a fixed roof (Fig.3.3a), steel chain-link enclosures with no roof (Fig.3.3b), partial steel chain-link walls with segments of barbed wire and/or thorn brushes (Fig.3.3c), and enclosures made of primarily thorn brushes and lined with barbed wire (Fig.3.3d). The steel enclosures are around 10 feet wide x 15 feet long and 5.5 feet high, with a capacity for around 35 calves. The more enclosed and reinforced the enclosure was (i.e. steel chain-link walls with a fixed roof), the better this method served to reduce

50 reported depredation, with no reported livestock losses from leopard attacks if they were kept in these enclosures (Table 3.3). As the enclosure became dismantled (i.e. roof was taken off, or walls were supplemented with thorn bushes or barbed wire), the effectiveness in preventing depredation events decreased.

Fig.3.3a. Steel chain-link enclosure with a fixed roof Fig.3.3b. Steel chain-link enclosure without a roof

Fig.3.3c. Partial steel chain-link fencing, supplemented with thorns Fig.3.3d. Rudimentary thorn and barbed wire

Table 3.3. Livestock enclosures used by respondents in Yala (n = 61) surveyed May- August 2018, and proportion of livestock reported to have been lost by depredation from leopards in the past 3 years Livestock enclosure materials and method Percentage of Average percentage respondents using of reported livestock penning method depredation Steel chain-link enclosure with roof 3.3 0.0 Steel chain-link enclosure walls, no roof 11.5 2.8 Partial steel chain-link walls with segments of 19.7 8.1 barbed wire/thorn bushes/sticks Thorn bushes and barbed wire enclosures 65.6 13.1

3.3.3 Mitigations supported by respondents The mitigation techniques we discussed with pastoralists and the level of support they received (Fig.3.4) were (in increasing order of support): using trained livestock guarding dogs (6.56%), increasing human presence while livestock are grazing (34.43%), higher thorn fencing (36.07%), increasing human patrols during the night (when reported depredation incidents take place the most) (36.07%), relocating cattle to lower-risk areas (e.g. further from park boundary) (57.38%), PVC-lined enclosures (59.02%), light and sound distractions (73.77%) and using reinforced enclosures, a form of husbandry (98.36%).

Fig.3.4. Percentage of respondents in Yala (n=61) surveyed in May-August 2018 who supported the future testing of proposed conflict mitigation techniques.

3.3.4 Leopard prey species and biomass availability

Potential prey species for which adult animals were detected on camera traps included: black-naped hare (Lepus nigricollis), civets (Paradoxurus spp.), grey langur (Semnopithecus priam), Sri Lankan jackal (Canis aureus naria), jungle cat (Felis chaus), Sri Lankan jungle fowl (Gallus lafayettii), mongoose (Herpestes spp.), mouse deer (Moschiola meminna), Indian palm squirrel (Funambulus palmarum), Indian peafowl (Pavo cristatus), Indian crested porcupine (Hystrix indica), rat (Rattus spp., Bandicota spp., Srilankamys spp), sambar deer (Rusa unicolor unicolor), Ceylon spotted deer (Axis axis ceylonensis) and wild boar (Sus scrofa) (Table 3.4). There were 6307 independent detections, with the most commonly detected species being Ceylon spotted deer and the least common being mouse deer. Prey species for which juvenile animals were detected included grey langur, peafowl, sambar deer, spotted deer, water buffalo (Bubalus bubalis) and wild boar. Prey RBI for each camera station ranged between 0 kg/day to 822.4 kg/day for juvenile prey species, and 3.6 kg/day to 9251.9 kg/day for adult prey species. See Appendix G for body weights used and RBI estimates for each camera trap station, and Appendix H for example camera trap images.

Table 3.4. Independent detections (based on 30 min threshold for independence) of adult and juvenile leopard prey species detected by camera trap sampling (n=28 stations) in Yala, Sri Lanka, between May-August 2018 and June-July 2019. Species Scientific name Independent Independent detections of detections of juveniles adults Black-naped hare Lepus nigricollis NA 506 Ceylon spotted deer Axis axis ceylonensis 463 2123 Civet spp. Paradoxurus spp. NA 37 Grey langur Semnopithecus priam 81 794 Indian crested porcupine Hystrix indica NA 74 Indian palm squirrel Funambulus palmarum NA 7 Indian peafowl Pavo cristatus 9 485 Mongoose spp. Herpestes spp. NA 80 Mouse deer Moschiola meminna NA 2 Rat spp. Rattus spp., Bandicota NA 140 spp., Srilankamys spp Sambar deer Rusa unicolor unicolor 16 102 Sri Lankan jackal Canis aureus naria NA 5 Sri Lankan jungle fowl Gallus lafayettii NA 18 Water buffalo Bubalus bubalis 384 NA Wild boar Sus scrofa 341 640

3.3.5 Prey model Prey availability increased with distance from roads (0.37, 95% CI [-0.06, 0.81]), water sources (0.45, 95% CI [0.03, 0.87]) and in areas with lower forest cover (-0.34, 95% CI [- 0.77, 0.08]; Fig.3.5). The adjusted R2 value for this model was 0.34, with 26% of deviance explained.

Fig.3.5. Estimated effects of all predictor variables on factors associated with leopard prey availability (measured as Relative Biomass Index; kg/day) for each camera trap location (n = 28) in Yala, Sri Lanka. Predictor variables included distance to roads (km), distance to water (km) and percent forest cover within a 150m buffer. Coefficient estimates of the global generalised linear model are displayed with their 95% confidence intervals. Predictor variables have been standardised to a mean of 0 and standard deviation of 1 to allow for direct comparison.

3.3.6 Depredation model We found evidence to support two of our initial hypotheses: depredation levels decreased with improved livestock husbandry and increased with increasing livestock numbers (Fig.3.6.). All other covariate effects had confidence intervals that overlapped zero and are therefore not statistically significant. The top-ranked model included husbandry score (-0.607, 95% CI [-0.836, -0.390] and number of cattle (0.403, 95% CI [0.251, 0.564]. The adjusted R2 value for this model was 0.60, with 45% of deviance explained. The full model set with all possible additive combinations can be found in Appendix I.

57 Table 3.6. Top-ranked generalised linear models of factors relating to reported livestock depredation from 2015-2018 across respondents (n = 61) in Yala, Sri Lanka. Models within 4 ΔAICc units of the model with the lowest AICc, with associated degrees of freedom (df), AICc, ΔAICc and Akaike model weight are shown. Model df AICc Δ AICc AIC weight Husbandry score + number of cattle 4 311.7 0.00 0.248 Husbandry score + number of cattle + residency time 5 312.5 0.89 0.159 Husbandry score + number of cattle + distance to 5 313.5 1.86 0.098 national park Husbandry score + number of cattle + road density 5 313.7 2.02 0.090 Husbandry score + number of cattle + prey 5 314.0 2.37 0.076 availability Husbandry score + number of cattle + residency time 6 314.1 2.43 0.074 + distance to national park Husbandry score + number of cattle + residency time 6 314.9 3.28 0.048 + road density Husbandry score + number of cattle + prey 6 314.9 3.29 0.048 availability + residency time

3.4 Discussion

Our results indicated that the two most important factors underlying reported livestock depredation in Yala were livestock husbandry and the number of livestock owned. Depredation by leopards was the fourth-highest cause of reported livestock loss, and the most supported form of mitigation was using stronger cattle enclosures and light deterrents.

3.4.1 Factors involved in depredation incidents Our top model showed that as the number of cattle increased, incidents of depredation also increased, consistent with our hypothesis. There is some evidence to suggest that the Sri Lankan leopard may preferentially select larger prey species in Yala, such as sambar deer (Rusa unicolor unicolor), but their most commonly consumed prey species is the spotted deer (Axis axis ceylonensis), which is consumed proportionally to their availability (Kittle, Watson, and Fernando 2017). Dairy cattle are similar in size to these two deer species but lack the same defence mechanisms and escape abilities (Jackson and Nowell 1996), and may thus continue to be lost to depredation unless better protected.

58 Having large herds of livestock makes protecting them effectively more difficult, in terms of resources and space required. Larger herds may also attract predators, or have a higher probability of containing vulnerable or weak individuals (Bradley and Pletscher 2005), which may provide opportunistic predators such as leopards with a prey choice that requires less foraging, yet represents high profitability (Polisar et al. 2003). By extension, smaller farms with fewer livestock might not have enough animals to attract attention from predators (Bradley and Pletscher 2005), suggesting that a reduction in livestock numbers may lead to reduced predator attacks.

Our top model also showed that incidents of depredation decreased as cattle husbandry improved, consistent with our hypothesis. However, any changes in husbandry that are promoted should be appropriate to the local context, adaptable and compatible with local carnivores in order to facilitate long-term coexistence. Pastoralists who had a properly maintained steel chain-link enclosure with a roof reported no cattle depredation. Fortification of livestock enclosures is a globally observed method of reducing depredation events (Woodroffe et al. 2007; Jackson and Wangchuk 2004). Not all pastoralists in Yala possess such enclosures due mainly to the expense, with enclosures currently given only to select pastoralists by one luxury hotel in the region as part of a Corporate Social Responsibility (CSR) program. Only 3.3% of respondents had a complete enclosure, with 11.5% having walls without the roof, which therefore allows an entry point for leopards resulting in an average increase of 2.9% of livestock lost. Furthermore, pastoralists regularly dismantled the steel enclosures, inserting barbed wire and thorn bushes between panels in an effort to fit more cattle inside. This reduced the effectiveness of the enclosure, allowing leopard entry, with 8.1% of livestock reported lost to depredation when employing this system.

During interviews with pastoralists, 54% of respondents indicated that leopards regularly kill multiple individuals at a time when entering pens, but only take one carcass. The approximate cost of each pen ranges from Rs. 40,000-80,000, depending on the price of steel and the workmanship involved. We suggest that encouraging pastoralists to participate in sourcing the materials and constructing the enclosures will help increase their motivation to maintain it (e.g. using leftover motor oil to wipe the chain link to avoid rust

59 caused from cattle urine, a problem mentioned by 4 respondents), and allow them to be used for a longer period of time while reducing the overall cost of maintenance. This may lead to increased tolerance of leopards, similar to how community-led installation of livestock corrals improved predator tolerance in the trans-Himalayan region of Ladakh (Jackson and Wangchuk 2004).

3.4.2 Mitigations and husbandry techniques supported by the community Most pastoralists (70%) have tested simple mitigations by hanging a flashlight above their livestock pen which moves in the wind to cast irregular patterns. This has low reported efficacy. We suggested alternatives such as Foxlights® (Bexley North, Australia), that have worked to reduce depredation by pumas in Chile by changing the colour and frequencies of light emitted (Ohrens, Bonacic, and Treves 2019). The use of LED lights has also worked in reducing nocturnal depredation incidents in Kenya in areas adjacent to Nairobi National Park (Lesilau et al. 2018), and is one method that can be tested with relative ease and affordability in our Sri Lankan study system.

The mitigation technique supported by the most people (98.4%) was an improved form of livestock husbandry, specifically using an alternative fencing system. As opposed to using steel, which is prone to increased and fluctuating taxes in Sri Lanka, 59% of pastoralists supported the testing of PVC plastic sheets. In Hwange National Park, Zimbabwe, PVC plastic sheets lining livestock pens have been effective at limiting depredation, though this is used in conjunction with guarding systems (Loveridge et al. 2017). Pastoralists told us that using taller (around 3m) plastic sheets may be effective, and an easier method of enclosing their livestock. We also were advised that these pens may only work if kept in open areas, away from any trees that leopards can climb and attack from. An alternative material to PVC sheets is rice “gunny” bags, a cheap sack used to package rice, often made of burlap or other natural fibres. These are cheap, porous and likely easier to source in Sri Lanka. We suggest that providing pastoralists with an alternative form of livestock enclosure that they can easily create themselves and maintain would act to reduce depredation incidents and reduce their dependence on pens that are doled out infrequently. This alternative penning system, perhaps used in conjunction with affordable light mitigations such as Foxlights®, may serve as a more sustainable and

60 attainable way of protecting livestock, and can be done in conjunction with tourism sector CSR programs. The added benefit of such a system is that it involves non-permanent structures and therefore is not subject to potential issues with law enforcement, given the unclear definition of “buffer zone” and acceptable activities in this landscape. Conducting a future experiment that tests the effectiveness of these mitigations against experimental controls will help to contribute quantitative knowledge towards the effectiveness of mitigations, a field which is currently lacking (van Eeden et al. 2018).

Multiple studies have noted an association between pastoralists’ past experiences with livestock losses and their subsequent efforts to protect their livestock through improved husbandry methods (Miller, Jhala, and Schmitz 2016). Livestock husbandry can serve as a link between the social and ecological components of human-carnivore conflict, being influenced by human experiences and capacities while remaining open to behavioural changes that can reduce incidents of livestock depredation. Poor husbandry, particularly in the form of unattended livestock in enclosures that are not fully secure or well-lit, has been shown to be associated with increased levels of depredation (Kabir et al. 2014). Other studies have shown that simply fencing livestock rather than letting them wander through the habitat (or in fenced pastures) reduces incidents of depredation significantly (Swenson and Andrén 2005; Kissling, Fernández, and Paruelo 2009). An alternative to using solely chain-link enclosures is the “living wall” concept, where fast- growing thorny trees are used as fence posts and chain-link fencing is installed around them, proven to be 99.9% successful in preventing depredation during the night (Lichtenfeld, Trout, and Kisimir 2015). Various improvements to livestock husbandry therefore exist, however determining their applicability in Sri Lanka should rely on their evidence-based success.

3.4.3. Livestock loss due to other causes More livestock were reported lost to theft, to disease and from wandering away than depredation (Table 3.2). Theft is a frequent occurrence in this region, as few pastoralists have the ability to keep all of their livestock enclosed overnight. Due to the inability to develop permanent structures, most livestock are therefore kept in makeshift pens, or left untied outside overnight. Many pastoralists now mark their livestock with unique symbols

61 or words to discourage others from stealing them, particularly the females and calves. However, theft will be hard to address, with the best option likely being to enforce a reduction in herd sizes and to enclose them more securely.

The second highest cause of livestock loss was disease. This mirrors findings in other studies, in which a larger proportion of livestock are lost to disease than to depredation (Dickman et al. 2014; Dar et al. 2009). Often, perceived harm to livelihood or lives will result in retaliatory attacks against carnivores, even when realised harm is absent or minimal (Woodroffe, Thirgood, and Rabinowitz 2005). This is an important consideration, as retaliation against predators may not be proportional to the damage they cause, due to complex nature of how we assess risk and what specific threats we perceive to be within or outside of our control. Responses to different forms of livestock loss vary. Though numerically few livestock may be lost to leopard attacks, carnivore interactions may elicit stronger reactions in people due to the imbalance between real and perceived threats (Miller, Jhala, and Schmitz 2016). For example, though a pastoralist may lose more livestock to disease, he may still react more negatively towards leopards after even one depredation incident. This may be due to the fact that pastoralists may feel like they have more influence in controlling the situation (e.g. by setting snares and traps, or killing the creature that caused the damage), versus losing livestock to something like a disease which pastoralists expressed that they felt was out of their control.

One way to reduce the relative impact of livestock depredation, and to potentially increase tolerance of leopard-caused losses, is to address other forms of livestock loss, e.g. through increasing access to preventative vaccinations and veterinary services. However, lack of access to such services can be a significant barrier to improving livestock conditions and cause increased predation risk, especially as livestock health declines (Khorozyan, Soofi, et al. 2015). Through interviews with pastoralists, we learned about the rapid spread of “kuramuka rogaya”, a highly contagious viral disease that is characterised by lesions, sores, loss of appetite diarrhea and reduced milk production. Particularly when livestock are contained in high densities overnight in pens, they are susceptible to disease spread and many pastoralists can lose high proportions of their stock within a short period of time. For pastoralists whose livestock are already malnourished, this disease is a severe loss of

62 livelihood, and some are only able to produce 10% of the milk that they once used to. One respondent claimed to have lost 73% of his stock to this disease, likely due to the confined space the livestock are left in overnight where the disease can spread. Upon further investigation it was discovered that the Livestock Veterinary Board had not yet assessed the severity of this disease outbreak as of July 2019. We suggest first that they be contacted and send veterinarians to assess livestock conditions and health. This should ideally be followed by an analysis of the feasibility of a widespread vaccination program to reduce livestock loss to disease and hopefully increase tolerance of leopards amongst pastoralists as a result, by reducing the number of total livestock lost. Similar to the findings of Chapter 2, an opportunity exists for the participation of the tourism sector by financially supporting larger scale vaccination programs, perhaps conditionally dependent on a maximum herd size – in an effort to strengthen livestock and improve productivity and pastoralist tolerance towards any depredation incidents.

An average of 6.7% of livestock loss was due to attacks by dogs, which we did not anticipate. Respondents told us that these free-ranging dogs often form packs and target the smallest calves. Free-ranging dogs have been known to threaten livestock in various global contexts such as India (Suryawanshi et al. 2017), Chile (Villatoro et al. 2019), and Poland (Wierzbowska et al. 2016). This potentially shifting leopard prey base and its magnitude is a place to focus future research in this area in Sri Lanka, as leopard diets in India have included feral and free-ranging dogs (Athreya et al. 2015). Free-ranging dogs remains an important consideration in developing coexistence strategies, due to the close association dogs have with humans, their role as potential prey species, acting as potential disease vectors (Packer et al. 1999), as well as their potential to increasingly compete with native predators for livestock prey.

3.4.4 Reducing livestock densities and preventing overcrowding on the landscape An additional consideration with high livestock numbers in this landscape is the potential for overgrazing. The average number of cattle per pastoralist was 84 in the small (71km2) study area. Many pastoralists congregated in the same small areas while grazing and shared steel or thorn pens with other pastoralists. Overcrowding of livestock, especially in this ecologically sensitive region, likely leads to over-grazing, decreased vegetation availability and reduced carrying capacity of the landscape (Pimentel and Kounang 1998).

63 In central Kenya, compensation programs had incentivised pastoralists to increase their livestock, leading to increased conflict – with predators and wild ungulates – for grazing habitat (Romañach, Lindsey, and Woodroffe 2007). Therefore, by suggesting (and potentially enforcing) a reduced amount of livestock within a certain area, it is possible that this would increase the carrying capacity of native leopard prey, and act to reduce levels of livestock depredation while attempting to address the problem of overgrazing. This has a social component, as livestock rearing is a largely generational practice, so persuading pastoralists to reduce the quantity of livestock owned will likely require a large educational campaign, and may also require alternative forms of income, such as from eco- tourism. As the Yala study area is managed by the DWC, an option exists to set aside smaller areas to conserve native leopard prey species and prevent livestock grazing or occupation of these areas. These areas could be included in routine canvasses of the landscape conducted by the DWC in an effort to decrease depredation events.

3.4.5 Conclusion Our study is the first to attempt to quantify and evaluate drivers of depredation in Yala. Livestock depredation and the resulting retaliatory killing of leopards does occur in Sri Lanka, though the impact on leopard populations is not yet thought to be significant. Given the dairy industry’s growth across the country, there is potential for increased conflict to occur. Results from this study may help proactively guide dairy industry growth in Sri Lanka to inform human-leopard coexistence, ultimately ensuring viable leopard populations while also securing livelihoods for smallholder pastoralists. Despite their endangered status, Sri Lankan leopards have the potential to pose a genuine threat to human livelihoods (Thirgood, Woodroffe, and Rabinowitz 2005). Focused attention on human- leopard conflict in the Yala region is required. With limited resources available for conservation, directing mitigations towards the most effective methods (e.g. improving livestock enclosures) and towards at-risk pastoralists will help conservation strategies be more successful (Broekhuis, Cushman, and Elliot 2017). Factors such as feasibility and perception (belief of effectiveness) of potential mitigations are important to consider when promoting interventions in livestock husbandry (van Eeden et al. 2018),

64 In an effort to combine empirical data with local experiences (Treves and Karanth 2003), this study provides insight into the factors associated with reported levels of livestock depredation, and the mitigations supported by pastoralists in the region. Our models suggest that an improvement in livestock husbandry practices and a reduction in the number of cattle may lead to reduced incidents of depredation. Because law enforcement is not always consistent in this study area, conservation strategies that rely on individual capacities and self-enforcing mechanisms are likely to be the most successful. Given this context, we suggest testing the efficacy of reinforced enclosures (using materials such as PVC plastic or rice bags) and light mitigations – as these were the options most supported by pastoralists – along with a widespread vaccination program and reduction in stock sizes.

Leopards have been shown to exhibit behavioural differences in habitats where they do not compete with other carnivores, in terms of habitat selection, activity patterns and prey selection (Marker and Dickman 2005). Ecological influences are not the only factors affecting leopard ecology and behaviour; human land-use alteration and livestock farming practices have significantly affected leopard populations globally, leading to increased conflict and their fragmented spatial ecology (Marker and Dickman 2005). Although depredation was the fourth highest form of stock loss in our study (far less than to theft and disease), carnivore attacks generate intense hostility and lead to significant levels of retaliatory and preventative carnivore killing (Abade et al., 2014). Therefore, understanding factors associated with a higher risk of depredation is important for identifying the most effective future conflict mitigation. Though we were not sampling for leopard abundance, we obtained 5 independent detections of leopards, 4 of which were outside the park boundary but within a narrow 2km range where cattle rearing also occurs. As leopards do exist outside the national park boundary, management approaches that address activities both inside and outside protected area boundaries are essential in order to conserve their populations effectively (Balme, Slotow, and Hunter 2010). This will help inform coexistence strategies in this specific landscape. Spatial segregation methods such as improved livestock husbandry, or temporal segregation methods such as increased nighttime livestock guarding are potential strategies to consider (Carter et al. 2012). The

65 effectiveness of these strategies can help us strive towards proactive human-leopard coexistence in this tropical biodiversity hotspot.

Chapter 4: Conclusion

4.1 Synthesis and conclusions Large carnivores are of special conservation concern as their populations decline globally. Many large carnivores have the potential to be ‘umbrella species’ that act to conserve a suite of co-occurring species across the vast habitats and ranges they occupy (Roberge and Angelstam 2004). A leading cause of large carnivore decline is conflict with humans, specifically due to livestock depredation. Human-wildlife conflict is the direct result of behavioural interactions between humans and wildlife (Lischka et al. 2018). Though that alone is a simple concept, there are a multitude of ecological and social dimensions involved, requiring interdisciplinary research in order to better understand the complexities of human-carnivore conflict (Dickman 2010). There is growing consensus that integrating both ecological and social knowledge will lead to more sustainable and effective conservation outcomes (Pooley, Mendelsohn, and Milner-Gulland 2014). This thesis aimed to contribute towards this integration, focusing on a case study in two locations in Sri Lanka where leopards and livestock-rearing communities exist in shared landscapes.

In Chapter 2, we found determinants of attitudes towards leopards to vary between the two study sites. In Yala, attitudes were positively associated with increasing socio- demographics (i.e. age, number of dependents, number of years spent rearing livestock) and a greater desire to conserve wildlife species in general. Attitudes were negatively associated with pastoralists level of awareness of leopard ecology and leopard-related tourism for Sri Lanka. Analysis of qualitative results showed the main issues faced by pastoralists were the lack of ability to own land and therefore construct better enclosures for their cattle, the inability to guard their cattle at night and a tenuous relationship with the Department of Wildlife Conservation (DWC). These barriers suggest that the DWC should be involved to a greater extent in facilitating coexistence, one way of which is to improve lines of communication between the DWC and pastoralists regarding verification of depredation incidents and specifying grazing locations. Options such as group land ownership should be considered, acknowledging the complications with land ownership in

67 Sri Lanka and that the majority of cattle rearing occurs on state-owned land. Potential programs could include certain criteria such as maintaining a minimum distance to park boundary and enforcing a maximum size of cattle herds following the results of this study. In the Central Hills, pastoralist attitudes towards leopards were positively associated with the desire for increased government involvement. During interviews, respondents seemed to either be very hopeful or very pessimistic about the government’s interest in them and their struggles, with many hoping that this research will help illuminate some of these struggles. Although these communities do not experience direct leopard conflict over livestock, they do share space with leopards and therefore efforts made, by government and/or regional tourism companies, to facilitate livestock rearing would be well-received and perhaps help positively influence attitudes towards leopards.

In Chapter 3, we found that depredation levels reported by pastoralists in Yala were lower at farms with better cattle husbandry methods and higher at farms with greater cattle numbers. These results mirror findings in other studies that emphasize the impact of having larger herds of livestock in depredation events (Bradley and Pletscher 2005), and the importance of improving livestock security in order to reduce depredation incidents (Kabir et al. 2014; Ogada et al. 2003; Lichtenfeld, Trout, and Kisimir 2015). Nevertheless, how specifically to improve husbandry may vary between contexts. The results from our surveys and interviews suggest that the most supported husbandry improvements included creating larger enclosures with fencing from PVC plastic or rice bags (similar to burlap material), and the use of light mitigations such as Foxlights. These improvements are relatively inexpensive and appropriate for the local context, given the high cost of steel enclosures and difficulty in constructing them. Contextually appropriate mitigations have been successful in reducing depredation events in the trans-Himalayan region of Ladahk and in Kenya, following similar interdisciplinary research (Jackson and Wangchuk 2004; Lichtenfeld, Trout, and Kisimir 2015). Importantly, depredation was only the fourth highest cause of reported livestock loss in Yala, preceded by theft, disease and cattle wandering off. Therefore, it may be useful to consider implementing vaccination schemes to better strengthen livestock given the crowding that occurs in enclosures and subsequent rapid spread of disease. Linking back to Chapter 2, if such programs are funded by the

68 tourism industry, this may help reduce negative attitudes towards leopards and improve tolerance to leopard-related losses if overall livestock losses are reduced.

4.2 Research strengths and limitations Assessing both social and ecological aspects of human-leopard conflict at local scales gives us a more holistic understanding of this specific form of human-wildlife conflict, which was the original aim of this research. We drew upon literature and studies from both the social and ecological sciences to guide each chapter and inform this local example of human-wildlife conflict in a comprehensive, interdisciplinary manner. We found that speaking to pastoralists in larger groups as well as individually was an effective way for them to contribute their own experiences and respond to the experiences of others, allowing for open dialogue between pastoralists. In Chapter 2, we chose to assess only the attitudes towards leopards, as this was feasible given the ability to measure attitudes through a survey and quantify results to be used in subsequent regression models. However, to comprehensively understand the social histories and trends of human communities is a long-term, complex process that is beyond the scope of this thesis. There are many other aspects of these communities that were not included in this study, such as predator-specific behaviour (e.g. retaliatory killings of leopards), behavioural intentions (e.g. intent to kill, level of tolerance to livestock loss after which point they may consider retaliating), and inherent values. To better guide conservation and development, these aspects should be considered in combination with attitudes.

In Chapter 3, we set out remote camera traps in 28 locations in the Yala landscape to estimate leopard native prey availability. The design and ultimate locations of cameras set was opportunistic, as each location was contingent on the Department of Wildlife Conservation’s approval. Cameras were lost to theft, and locations were therefore guided by areas most likely to be avoided by humans. This resulted in a skewed sampling design, which is a reality of fieldwork, however may ultimately bias the results. Sampling was conducted during the dry season (May-August) in 2018 and 2019, and we therefore were unable to sample during the wet season, which is likely to cause fluctuations in prey availability, and therefore influence depredation. Additionally, we were unable to sample

69 for leopard detections specifically, and this is an area of future research that will complement the findings of this thesis.

4.3 Recommendations and directions for future research This thesis helped to identify factors involved in depredation incidents and mitigation measures that are supported by the communities, but follow-up work remains. We recommend testing the effectiveness of PVC or rice bag fencing combined with a LED light deterrent in reducing depredation incidents, as these were the options most supported by pastoralists, with many of them offering to participate in such testing. Carrying out the controlled testing of these mitigations and having experimental evidence that shows their effectiveness in reducing incidents of livestock depredation will help contribute knowledge towards a field which is lacking in research (van Eeden et al. 2018; Treves, Krofel, and McManus 2016).

Yala is an increasingly developing area, and as the dairy industry expands, we can presume that interactions between humans and leopards will likely increase. Often, reported and perceived conflict is greater than actual conflict experienced, likely due to people feeling unable to control or influence the threat of a large carnivore, compounded by the fact that these carnivores are protected by government laws (Madden 2004). For example, a pastoralist may lose more livestock to disease than to leopards, but the response is not proportional to the losses, as the perceived threat is weighted more towards the leopard. We recommend to first verify incidents of reported depredation to better estimate livestock lost to leopards. When only a small proportion of depredation is reported to wildlife officials, this can potentially impact conflict management policies (Loveridge et al. 2017) and allow issues of depredation to escalate. Verifying reported livestock losses is therefore crucial before moving forward with coexistence strategies. A database of verified depredation incidents can be valuable for future studies involving modelling or mapping of conflict risks across the landscape.

A consequence of the observed overgrazing and the malnourished state of many cattle in Yala is that pastoralists increase their stock sizes in order to produce more milk. However, there may be knock-on effects, as increased herds cause increased grazing

70 pressure which continues to decrease their individual productivity. Accordingly, we recommend that pastoralists increase efforts to diversify their local economy and find alternative sources of income, e.g. through working in the nearby city of Tissamaharama or through the tourism industry as there are numerous hotels and guest houses in the area. This will reduce their dependence on cattle for their livelihood, reduce the impact that livestock are causing through overgrazing in this buffer zone habitat, but also address a source of negative attitudes, i.e. not being involved in the tourism sector. Further studies that help to contextualise the impact that livestock has on native non-carnivore species (e.g. through grazing competition, reduction of habitat quality and potential disease transmission that may occur) are also needed to further inform the scale of livestock rearing that should be tolerated in this sensitive landscape.

Further community engagement and interviews should be followed up with questions regarding retaliatory measures taken and tolerance towards livestock losses. Attitudes and behavioural intentions are often uncorrelated with actual behaviour (Heberlein 2012; Romañach, Lindsey, and Woodroffe 2007) and therefore it is important to assess both attitude and actual behaviour at site-specific contexts. We also recommend that awareness and education programs be considered in the Yala region with stakeholders present from the DWC, tourism board and farming community. This will facilitate opening the lines of dialogue and co-creating coexistence options, through collaborative projects or proposals, such as changes in milk collection or pricing systems that can benefit the entire farming community.

We also need more data on the specific spatial and temporal patterns of leopards in this landscape in order to better plan for coexistence at finer spatial scales (Carter et al. 2012). Obtaining data on individual leopard behaviour and prey choice is important, as those leopards engaging in livestock depredation may represent a very small proportion of the entire population, or those individuals who are older, injured, sick or of a specific age/sex (Nyhus 2016). A thorough leopard scat analysis and further native prey abundance sampling will help better understand prey choice in leopards and whether they exhibit preferential choice towards livestock (Chase Grey, Bell, and Hill 2017). Prey catchability might not correspond to abundance or accessibility, as leopards may require a certain

71 amount of habitat heterogeneity to be able to successfully stalk and encounter prey (Balme, Hunter, and Slotow 2007; Rostro-Garcia, Kamler, and Hunter 2015), emphasising the need for more studies to be able to tease apart these ecological components of depredation. If this information is then communicated back to the pastoralists, they may potentially alter their attitudes of leopards and better acknowledge the behaviour of an individual leopard as opposed to the entire species. Increased knowledge of leopard distribution would also help identify areas at higher risk of conflict, and can help guide the DWC and hotels that are distributing pens to do so in a more informed way to reduce human-leopard interactions (Takahata et al. 2014). Strategies could involve increasing the distance pens are given from park boundaries to discourage livestock rearing near the national park, establishing a cap size on the number of cattle before allocating a pen, and initiating a registration program to discourage outside pastoralists from coming to establish in this ecologically sensitive region.

The goal of balancing economic development with biodiversity conservation is not a new one. However, the specific strategies and programs implemented in hopes of achieving this goal vary across contexts. Our case study in Sri Lanka demonstrates the importance of conducting small-scale studies to better understand the human communities, and how they differ. Even in a smaller country like Sri Lanka, attitudes towards leopards varied greatly between two communities. The United Nations Sustainable Development Goals may be separated by number, but in reality, addressing matters such as sustainable economic growth (SDG 8) and halting biodiversity loss (SDG 15) cannot be done in isolation, as one may influence the other. Ultimately, there is no ‘one size fits all’ approach, and comprehensive studies that acknowledge, consider and integrate multiple disciplines will have a better understanding of local systems, and therefore a better chance to harmonise wildlife conservation and rural development.

72 References

Allport, G.W. 1935. “Handbook of Social Psychology.” In , 798–844. Woster, MA: Clark University Press. Alpert, Peter. 1996. “Integrated Conservation and Development Projects.” BioScience 46 (11): 845–55. https://doi.org/10.2307/1312970. Arjunan, M, Christopher Holmes, Jean-philippe Puyravaud, and Priya Davidar. 2006. “Do Developmental Initiatives Influence Local Attitudes toward Conservation ? A Case Study from the Kalakad – Mundanthurai Tiger Reserve , India” 79: 188–97. https://doi.org/10.1016/j.jenvman.2005.06.007. Athreya, Vidya, Morten Odden, John D.C. Linnell, Jagdish Krishnaswamy, and K. Ullas Karanth. 2016. “A Cat among the Dogs: Leopard Panthera Pardus Diet in a Human- Dominated Landscape in Western Maharashtra, India.” Oryx 50 (1): 156–62. https://doi.org/10.1017/S0030605314000106. Athreya, Vidya, Arjun Srivathsa, Mahi Puri, Krithi K. Karanth, N. Samba Kumar, and K. Ullas Karanth. 2015. “Spotted in the News: Using Media Reports to Examine Leopard Distribution, Depredation, and Management Practices Outside Protected Areas in Southern India.” PLoS ONE 10 (11): 1–19. https://doi.org/10.1371/journal.pone.0142647. Bagchi, S., and C. Mishra. 2006. “Living with Large Carnivores: Predation on Livestock by the Snow Leopard (Uncia Uncia).” Journal of Zoology 268 (3): 217–24. https://doi.org/10.1111/j.1469-7998.2005.00030.x. Balme, G. A., R. Slotow, and L.T.B. Hunter. 2010. “Edge Effects and the Impact of Non- Protected Areas in Carnivore Conservation: Leopards in the Phinda-Mkhuze Complex, South Africa.” Animal Conservation 13 (3): 315–23. https://doi.org/10.1111/j.1469-1795.2009.00342.x. Balme, Guy A., Luke Hunter, and R O B Slotow. 2007. “Feeding Habitat Selection by Hunting Leopards Panthera Pardus in a Woodland Savanna : Prey Catchability versus Abundance,” 589–98. https://doi.org/10.1016/j.anbehav.2006.12.014. Balme, Guy A., Peter A. Lindsey, Lourens H. Swanepoel, and Luke T.B. Hunter. 2014. “Failure of Research to Address the Rangewide Conservation Needs of Large Carnivores: Leopards in South Africa as a Case Study.” Conservation Letters 7 (1): 3–11. https://doi.org/10.1111/conl.12028. Baral, Nabin, Marc J. Stern, and Joel T. Heinen. 2007. “Integrated Conservation and Development Project Life Cycles in the Annapurna Conservation Area, Nepal: Is Development Overpowering Conservation?” Biodiversity and Conservation 16 (10): 2903–17. https://doi.org/10.1007/s10531-006-9143-5. Barlow, S. 2011. “The Enter-Growth Project - Sri Lanka. Applying a Market Development Lens to an ILO Local Enterprise Development Project.” Bawa, K.S., A Das, J Krishnaswamy, U.K. Karanth, N.S. Kumar, and M Rao. 2007. “Western Ghats & Sri Lanka Biodiversity Hotspot.” Berger, Joel, Peter B. Stacey, Lori Bellis, and Matthew P. Johnson. 2001. “A Mammalian Predator-Prey Imbalance: Grizzly Bear and Wolf Extinction Affect Avian Neotropical Migrants.” Ecological Applications 11 (4): 947–60. https://doi.org/10.1890/1051-0761(2001)011[0947:AMPPIG]2.0.CO;2. Berkes, F., and C. Folke. 1998. “Linking Social and Ecological Systems: Management Practices and Social Mechanisms for Building Resilience.” Conservation Ecology 4

73 (2): 5. Berkes, F, J Colding, and C Folke, eds. 2002. Navigating Social-Ecological Systems: Building Resilience for Complexity and Change. Cambridge, United Kingdom: Cambridge University Press. Bhattacharjee, Anushree, and N Parthasarathy. 2013. “Coexisting With Large Carnivores : A Case Study From Western Duars , India.” Human Dimensions of Wildlife 18 (1): 20–31. https://doi.org/10.1080/10871209.2012.698403. Bland, J.M., and D.G. Altman. 1997. “Statistics Notes: Cronbach’s Alpha.” BMJ 314: 572. https://doi.org/https://doi.org/10.1136/bmj.314.7080.572. Bradley, Elizabeth H., Leslie A. Curry, and Kelly J. Devers. 2007. “Qualitative Data Analysis for Health Services Research: Developing Taxonomy, Themes, and Theory.” Health Services Research 42 (4): 1758–72. https://doi.org/10.1111/j.1475- 6773.2006.00684.x. Bradley, Elizabeth H., and Daniel H. Pletscher. 2005. “Assessing Factors Related to Wolf Depredation of Cattle in Fenced Pastures in Montana and Idaho.” Wildlife Society Bulletin 33 (4): 1256–65. https://doi.org/10.2193/0091- 7648(2005)33[1256:afrtwd]2.0.co;2. Brockington, Dan. 2004. “Community Conservation , Inequality and Injustice: Myths of Power in Protected Area Management.” Conservation and Society 2 (2): 411–32. Broekhuis, Femke, Samuel A. Cushman, and Nicholas B. Elliot. 2017. “Identification of Human–Carnivore Conflict Hotspots to Prioritize Mitigation Efforts.” Ecology and Evolution 7 (24): 10630–39. https://doi.org/10.1002/ece3.3565. Brooks, Thomas M, R A Mittermeier, G A da Fonseca, J Gerlach, M Hoffmann, J F Lamoreux, C G Mittermeier, J D Pilgrim, and A S L Rodrigues. 2006. “Global Biodiversity Conservation Priorities.” Science 313: 58–62. Burnham, Kenneth P, and David R Anderson. 2002. Model Selection and Multimodel Inference: A Practical Information-Theoretical Approach. Second Edition. New York: Springer. Cardillo, Marcel, Andy Purvis, Wes Sechrest, John L. Gittleman, Jon Bielby, and Georgina M. Mace. 2004. “Human Population Density and Extinction Risk in the World’s Carnivores.” PLoS Biology 2 (7): 909–14. https://doi.org/10.1371/journal.pbio.0020197. Carr, Deborah S., and Kathleen Halvorsen. 2001. “An Evaluation of Three Democratic, Community-Based Approaches to Citizen Participation: Surveys, Conversations with Community Groups, and Community Dinners.” Society and Natural Resources 14 (2): 107–26. https://doi.org/10.1080/089419201300000526. Carter, Neil H., and John D.C. Linnell. 2016. “Co-Adaptation Is Key to Coexisting with Large Carnivores.” Trends in Ecology and Evolution 31 (8): 575–78. https://doi.org/10.1016/j.tree.2016.05.006. Carter, Neil H., Binoj K. Shrestha, Jhamak B. Karki, Narendra Man Babu Pradhan, and Jianguo Liu. 2012. “Coexistence between Wildlife and Humans at Fine Spatial Scales.” Proceedings of the National Academy of Sciences of the United States of America 109 (38): 15360–65. https://doi.org/10.1073/pnas.1210490109. Cernea, Michael M., and Kai Schmidt-Soltau. 2006. “Poverty Risks and National Parks: Policy Issues in Conservation and Resettlement.” World Development 34 (10): 1808–30. https://doi.org/10.1016/j.worlddev.2006.02.008. Chapron, Guillaume, Dale G. Miquelle, Amaury Lambert, John M. Goodrich, Stéphane Legendre, and Jean Clobert. 2008. “The Impact on Tigers of Poaching versus Prey

74 Depletion.” Journal of Applied Ecology 45 (6): 1667–74. https://doi.org/10.1111/j.1365-2664.2008.01538.x. Charmaz, Kathy. 2006. “Coding in Grounded Theory Practice.” In Constructing Grounded Theory: A Practical Guide Thorugh Qualitative Analysis, 42–71. SAGE. Chase Grey, Julia N., Sandra Bell, and Russell A. Hill. 2017. “Leopard Diets and Landowner Perceptions of Human Wildlife Conflict in the Soutpansberg Mountains, South Africa.” Journal for Nature Conservation 37: 56–65. https://doi.org/10.1016/j.jnc.2017.03.002. Coen, A. Bernaards, and Robert I. Jennrich. 2005. “Gradient Projection Algorithms and Software for ArbitraryRotation Criteria in Factor Analysis.” http://www.stat.ucla.edu/research/gpa. Corbin, J, and A Strauss. 2008. “Basics of Qualitative Research: Techniques and Procedures for Developing Grounded Theory.” In Basics of Qualitative Research: Techniques and Procedures for Developing Grounded Theory. Los Angeles, CA: Sage Publications. Cowling, Richard M., Andrew T. Knight, Daniel P. Faith, Simon Ferrier, Amanda T. Lombard, Amanda Driver, Mathieu Rouget, Kristal Maze, and Philip G. Desmet. 2004. “Nature Conservation Requires More than a Passion for Species.” Conservation Biology 18 (6): 1674–76. https://doi.org/10.1111/j.1523- 1739.2004.00296.x. Crawley, Michael. 2007. The R Book. 1st ed. Chichester, West Sussex: Wiley Publishing. Creel, S., M. S. Becker, S. M. Durant, J. M’Soka, W. Matandiko, A. J. Dickman, D. Christianson, et al. 2013. “Conserving Large Populations of Lions - the Argument for Fences Has Holes.” Ecology Letters 16 (11): 3–6. https://doi.org/10.1111/ele.12145. Dar, Naeem Iftikhar, Riaz Aziz Minhas, Qamar Zaman, and Matthew Linkie. 2009. “Predicting the Patterns, Perceptions and Causes of Human-Carnivore Conflict in and around Machiara National Park, Pakistan.” Biological Conservation 142 (10): 2076–82. https://doi.org/10.1016/j.biocon.2009.04.003. Decker, Daniel J., S.J. Riley, and W.F. Siemer. 2012. Linking Social and Ecological Systems: Management Practices and Social Mechanisms for Building Resiliance. Human Dimensions of Wildlife Management. Deng, Lifang, and Wai Chan. 2017. “Testing the Difference Between Reliability Coefficients Alpha and Omega.” Educational and Psychological Measurement 77 (2): 185–203. https://doi.org/10.1177/0013164416658325. Development, United States Agency for International. 2013. “Supporting Livelihoods for Vulnerable Populations in Sri Lanka: Assessment of Findings and Opportunities.” Dhanwatey, Harshawardhan S., Joanne C. Crawford, Leandro A.S. Abade, Poonam H. Dhanwatey, Clayton K. Nielsen, and Claudio Sillero-Zubiri. 2013. “Large Carnivore Attacks on Humans in Central India: A Case Study from the Tadoba-Andhari Tiger Reserve.” Oryx 47 (2): 221–27. https://doi.org/10.1017/S0030605311001803. Dickman, Amy. 2010. “Complexities of Conflict: The Importance of Considering Social Factors for Effectively Resolving Human-Wildlife Conflict.” Animal Conservation 13 (5): 458–66. https://doi.org/10.1111/j.1469-1795.2010.00368.x. Dickman, Amy, Leela Hazzah, Chris Carbone, and Sarah M. Durant. 2014. “Carnivores, Culture and ‘Contagious Conflict’: Multiple Factors Influence Perceived Problems with Carnivores in Tanzania’s Ruaha Landscape.” Biological Conservation 178: 19– 27. https://doi.org/10.1016/j.biocon.2014.07.011.

75 Dickman, Amy, Ewan Macdonald, and David Macdonald. 2011. “A Review of Financial Instruments to Pay for Predator Conservation and Encourage Human-Carnivore Coexistence.” Proceedings of the National Academy of Sciences 108 (34): 13937– 44. https://doi.org/10.1073/pnas.1012972108. Dietz, Thomas, Amy Fitzgerald, and Rachael Shwom. 2005. “Environmental Values.” Annual Review of Environment and Resources 30 (1): 335–72. https://doi.org/10.1146/annurev.energy.30.050504.144444. Dinerstein, E, M Rijal, B Bookbinder, and A Rajuria. 1999. “Tigers as Neighbors: Efforts to Promote Local Guardianship of Endangered Species in Lowland Nepal.” In Riding the Tiger: Tiger Conservation in Human-Dominated Landscapes, edited by John Seidensticker, S Christie, and P Jackson, 316–33. Cambridge, United Kingdom: Cambridge University Press. Dinerstein, E, K Varma, E Wikramanayake, G Powell, S Lumpkin, R Naidoo, M Korchinsky, et al. 2013. “Enhancing Conservation, Ecosystem Services, and Local Livelihoods through a Wildlife Premium Mechanism.” Conservation Biology 27 (1): 14–23. https://doi.org/10.1111/j.1523-1739.2012.01959.x. DiStefano, Christine, Min Zhu, and Diana Mîndrila. 2009. “Understanding and Using Factor Scores : Considerations for the Applied Researcher.” Practical Assessment, Research & Evaluation 14 (20): 1–11. Dolrenry, Stephanie, Leela Hazzah, and Laurence G. Frank. 2016. “Conservation and Monitoring of a Persecuted African Lion Population by Maasai Warriors.” Conservation Biology 30 (3): 467–75. https://doi.org/10.1111/cobi.12703. Dunlap, Riley E., Kent D. Van Liere, Angela G. Mertig, and Robert Emmet Jones. 2000. “New Trends in Measuring Environmental Attitudes: Measuring Endorsement of the New Ecological Paradigm: A Revised NEP Scale.” Journal of Social Issues 56 (3): 425–42. https://doi.org/10.1111/0022-4537.00176. Dunning, Jr., J. 2007. CRC Handbook of Avian Body Masses. 2nd ed. Boca Raton, FL: Taylor & Francis. Eeden, Lily M van, Mathew S Crowther, Christopher R Dickman, Thomas M Newsome, Jeannine McManus, Tara K Meyer, Oswald J Schmitz, et al. 2018. “Carnivore Conservation Needs Evidence-Based Livestock Protection.” PLoS Biology 16 (9): 1–8. http://10.0.5.91/journal.pbio.2005577%0Ahttp://esc- web.lib.cbs.dk/login?url=http://search.ebscohost.com/login.aspx?direct=true&db=af h&AN=131828110&site=ehost-live. Eisenberg, J.F., and John Seidensticker. 1976. “Ungulates in Southern Asia: A Consideration of Biomass Estimates for Selected Habitats.” Biological Conservation 10 (4): 293–308. Estes, J. A., M. T. Tinker, T. M. Williams, and D. F. Doak. 1998. “Killer Whale Predation on Sea Otters Linking Oceanic and Nearshore Ecosystems.” Science 282 (5388): 473–76. https://doi.org/10.1126/science.282.5388.473. Fabricius, Christo, Eddie Koch, Hector Magome, and Stephen Turner, eds. 2004. Rights Resources and Rural Development: Community-Based Natural Resource Management in Southern Africa. Rights Resources and Rural Development. https://doi.org/10.4324/9781849772433. Fahrig, Lenore. 2010. “Effects of Habitat Fragmentation on Biodiversity.” Review Literature And Arts Of The Americas 34 (2003): 487–515. https://doi.org/10.1146/132419. Fernando, Sanjiv. 2016. “Evaluation of Alternative Strategies to Prevent Leopard

76 Predation on Livestock around Yala National Park , Sri Lanka.” Clark University. Fishbein, M, and I Ajzen. 2010. Predicting and Changing Behaviour: The Reasoned Action Approach. New York: Psychology Press, Taylor & Francis. Fiske, S.T., and S.E. Taylor. 2013. Social Cognition-from Brains to Culture. London: SAGE. Fox, Helen E., Caroline Christian, J. Cully Nordby, Oliver R W Pergams, Garry D. Peterson, and Christopher R. Pyke. 2006. “Perceived Barriers to Integrating Social Science and Conservation.” Conservation Biology 20 (6): 1817–20. https://doi.org/10.1111/j.1523-1739.2006.00598.x. Fox, John. 2016. “Polycor: Polychoric and Polyserial Correlations.” https://cran.r- project.org/package=polycor. Fricker, Ronald D., Walter W. Kulzy, and Jeffrey A. Appleget. 2012. “From Data to Information: Using Factor Analysis with Survey Data.” Phalanx, 30–34. Fulton, David C., Michael J. Manfredo, and James Lipscomb. 1996. “Wildlife Value Orientations: A Conceptual and Measurement Approach.” Human Dimensions of Wildlife 1 (2): 24–47. https://doi.org/10.1080/10871209609359060. Gálvez, Nicolás, Gurutzeta Guillera-Arroita, Freya A. V. St. John, Elke Schüttler, David W. Macdonald, and Zoe G. Davies. 2018. “A Spatially Integrated Framework for Assessing Socioecological Drivers of Carnivore Decline.” Journal of Applied Ecology, no. December 2017: 1–13. https://doi.org/10.1111/1365-2664.13072. Gebresenbet, Fikirte, Brhane Baraki, Gidey Tirga, Claudio Sillero-Zubiri, and Hans Bauer. 2018. “A Culture of Tolerance: Coexisting with Large Carnivores in the Kafa Highlands, Ethiopia.” Oryx 52 (4): 751–60. https://doi.org/10.1017/S0030605316001356. Ghoddousi, Arash, Mahmood Soofi, Amirhossein Kh Hamidi, Tanja Lumetsberger, Lukas Egli, Igor Khorozyan, Bahram H. Kiabi, and Matthias Waltert. 2016. “Assessing the Role of Livestock in Big Cat Prey Choice Using Spatiotemporal Availability Patterns.” PLoS ONE 11 (4): 1–16. https://doi.org/10.1371/journal.pone.0153439. Gore, Meredith L, Barbara A Knuth, Paul D Curtis, James E Shanahan, Meredith L Gore, Barbara A Knuth, Paul D Curtis, and James E Shanahan. 2016. “Education Programs for Reducing American Black Bear-Human Conflict : Indicators of Success ? Published by : International Association for Bear Research and Management Stable URL : Http://Www.Jstor.Org/Stable/3873050 REFERENCES Linked References Are Availa” 17 (1): 75–80. Greenberg, Saul. 2018. “Timelapse: An Image Analyser for Camera Traps.” Griethuijsen, Ralf A.L.F. van, Michiel W. van Eijck, Helen Haste, Perry J. den Brok, Nigel C. Skinner, Nasser Mansour, Ayse Savran Gencer, and Saouma BouJaoude. 2015. “Global Patterns in Students’ Views of Science and Interest in Science.” Research in Science Education 45 (4): 581–603. https://doi.org/10.1007/s11165- 014-9438-6. Harrison, Xavier A., Lynda Donaldson, Maria Eugenia Correa-Cano, Julian Evans, David N. Fisher, Cecily E.D. Goodwin, Beth S. Robinson, David J. Hodgson, and Richard Inger. 2018. “A Brief Introduction to Mixed Effects Modelling and Multi-Model Inference in Ecology.” PeerJ 6: e4794. https://doi.org/10.7717/peerj.4794. Hartig, Florian. 2019. “DHARMa: Residual Diagnostics for Hierarchical (Multi-Level / Mixed) Regression Models. R Package Version 0.2.4.” https://cran.r- project.org/package=DHARMa.

77 Hayward, M. W., P. Henschel, J. O’Brien, M. Hofmeyr, G. Balme, and G. I.H. Kerley. 2006. “Prey Preferences of the Leopard (Panthera Pardus).” Journal of Zoology 270 (2): 298–313. https://doi.org/10.1111/j.1469-7998.2006.00139.x. Hazzah, Leela, Monique Borgerhoff Mulder, and Laurence Frank. 2009. “Lions and Warriors: Social Factors Underlying Declining African Lion Populations and the Effect of Incentive-Based Management in Kenya.” Biological Conservation 142 (11): 2428–37. https://doi.org/10.1016/j.biocon.2009.06.006. Hazzah, Leela, Stephanie Dolrenry, Lisa Naughton, Charles T.T. Edwards, Ogeto Mwebi, Fiachra Kearney, and Laurence Frank. 2014. “Efficacy of Two Lion Conservation Programs in Maasailand, Kenya.” Conservation Biology 28 (3): 851–60. https://doi.org/10.1111/cobi.12244. Heberlein, Tom. 2012. “Navigating Environmental Attitudes” 26 (4): 583–85. https://doi.org/10.1111/j.1523-1739.2012.01892.x. Heinze, Georg, Christine Wallisch, and Daniela Dunkler. 2018. “Variable Selection – A Review and Recommendations for the Practicing Statistician.” Biometrical Journal 60 (3): 431–49. https://doi.org/10.1002/bimj.201700067. Hemson, Graham, Seamus Maclennan, Gus Mills, Paul Johnson, and David Macdonald. 2009. “Community, Lions, Livestock and Money: A Spatial and Social Analysis of Attitudes to Wildlife and the Conservation Value of Tourism in a Human-Carnivore Conflict in Botswana.” Biological Conservation 142 (11): 2718–25. https://doi.org/10.1016/j.biocon.2009.06.024. Hill, Catherine M. 2004. “Farmers’ Perspectives of Conflict at the Wildlife–Agriculture Boundary: Some Lessons Learned from African Subsistence Farmers.” Human Dimensions of Wildlife 9 (4): 279–86. https://doi.org/10.1080/10871200490505710. Holmern, Tomas, Julius Nyahongo, and Eivin Røskaft. 2007. “Livestock Loss Caused by Predators Outside the Serengeti National Park, Tanzania.” Biological Conservation 135: 534–42. https://doi.org/10.1016/j.biocon.2006.10.049. Inskip, Chloe, and Alexandra Zimmermann. 2009. “Human-Felid Conflict: A Review of Patterns and Priorities Worldwide.” Oryx 43 (01): 18. https://doi.org/10.1017/S003060530899030X. IUCN. 1990. “IUCN Directory of South Asian Protected Areas.” IUCN Directory of South Asian Protected Areas. Gland, Switzerland and Cambridge, UK. https://doi.org/10.5962/bhl.title.45280. Jackson, P, and K Nowell. 1996. “Wild Cats: Status Survey and Conservation Action Plan.” Jackson, Rodney, and Rinchen Wangchuk. 2004. “A Community-Based Approach to Mitigating Livestock Depredation by Snow Leopards.” Human Dimensions of Wildlife 9 (4): 1–16. https://doi.org/10.1080/10871200490505756. Jacobson, Andrew P., Peter Gerngross, Joseph R. Lemeris Jr., Rebecca F. Schoonover, Corey Anco, Christine Breitenmoser-Würsten, Sarah M. Durant, et al. 2016. “Leopard ( Panthera Pardus ) Status, Distribution, and the Research Efforts across Its Range.” PeerJ 4: e1974. https://doi.org/10.7717/peerj.1974. Kabir, Muhammad, Arash Ghoddousi, Muhammad Sadique Awan, and Muhammad Naeem Awan. 2014. “Assessment of Human – Leopard Conflict in Machiara National Park, Azad Jammu and Kashmir, Pakistan.” European Journal of Wildlife Research 60: 291–96. https://doi.org/10.1007/s10344-013-0782-z. Kaczensky, Petra. 1999. “Large Carnivore Depredation on Livestock in Europe.” Ursus 11 (April 1998): 59–71.

78 Kansky, Ruth, Martin Kidd, and Andrew T. Knight. 2016. “A Wildlife Tolerance Model and Case Study for Understanding Human Wildlife Conflicts.” Biological Conservation 201: 137–45. https://doi.org/10.1016/j.biocon.2016.07.002. Kansky, Ruth, and Andrew T. Knight. 2014. “Key Factors Driving Attitudes towards Large Mammals in Conflict with Humans.” Biological Conservation 179: 93–105. https://doi.org/10.1016/j.biocon.2014.09.008. Karanth, K. Ullas, Melvin E. Sunquist, and K.M Chinnappa. 1999. “Long-Term Monitoring of Indian Tigers.” In Riding the Tiger: Tiger Conservation in Human- Dominated Landscapes, 114–22. Cambridge: Cambridge University Press. Karanth, K Ullas, James D Nichols, N Samba Kumar, William A Link, and James E Hines. 2004. “Tigers and Their Prey : Predicting Carnivore Densities from Prey Abundance.” PNAS 101 (14): 4854–58. Khorozyan, Igor, Arash Ghoddousi, Mahmood Soofi, and Matthias Waltert. 2015. “Big Cats Kill More Livestock When Wild Prey Reaches a Minimum Threshold.” Biological Conservation 192: 268–75. https://doi.org/10.1016/j.biocon.2015.09.031. Khorozyan, Igor, Tanja Lumetsberger, Arash Ghoddousi, Mahmood Soofi, and Matthias Waltert. 2017. “Global Patterns in Biomass Models Describing Prey Consumption by Big Cats.” Mammal Review 47 (2): 124–32. https://doi.org/10.1111/mam.12084. Khorozyan, Igor, Mahmood Soofi, Amirhossein Khaleghi Hamidi, Arash Ghoddousi, and Matthias Waltert. 2015. “Dissatisfaction with Veterinary Services Is Associated with Leopard (Panthera Pardus) Predation on Domestic Animals.” PLoS ONE 10 (6): 1– 10. https://doi.org/10.1371/journal.pone.0129221. Khumalo, KathrynElizabeth, and LaurieAnn Yung. 2015. “Women, Human-Wildlife Conflict, and CBNRM: Hidden Impacts and Vulnerabilities in Kwandu Conservancy, Namibia.” Conservation and Society 13 (3): 232. https://doi.org/10.4103/0972-4923.170395. Kissling, Daniel W., Néstor Fernández, and José M. Paruelo. 2009. “Spatial Risk Assessment of Livestock Exposure to Pumas in Patagonia, Argentina.” Ecography 32 (5): 807–17. https://doi.org/10.1111/j.1600-0587.2009.05781.x. Kissui, Bernard M. 2008. “Livestock Predation by Lions, Leopards, Spotted Hyenas, and Their Vulnerability to Retaliatory Killing in the Maasai Steppe, Tanzania.” Animal Conservation 11 (5): 422–32. https://doi.org/10.1111/j.1469-1795.2008.00199.x. Kitchener, A. 1991. The Natural History of the World’s Cats. Ithaca, NY: Cornell University Press. Kittle, Andrew M., Anjali C. Watson, Samuel A. Cushman, and D. W. Macdonald. 2017. “Forest Cover and Level of Protection Influence the Island-Wide Distribution of an Apex Carnivore and Umbrella Species, the Sri Lankan Leopard (Panthera Pardus Kotiya).” Biodiversity and Conservation, 1–29. https://doi.org/10.1007/s10531-017- 1431-8. Kittle, Andrew M., Anjali C. Watson, and T. Saminda P. Fernando. 2017. “The Ecology and Behaviour of a Protected Area Sri Lankan Leopard (Panthera Pardus Kotiya) Population.” Tropical Ecology 58 (1): 71–86. Kittle, Andrew M., Anjali C. Watson, Chanaka P. H. Kumara, and Nimalka H. K. Sanjeewani. 2012. “Status and Distribution of the Leopard in the Central Hills of Sri Lanka.” Cat News 56: 28–31. Kittle_et_al_2012_Status_and_distribution_of_the_leopard_in_Sri_Lanka.pdf. Klöckner, C.A. 2013. “A Comprehensive Model of the Psychology of Environmental Behaviour-a Meta-Analysis.” Global Environmental Change 23: 1028–38.

79 Kolowski, J. M., and K. E. Holekamp. 2006. “Spatial, Temporal, and Physical Characteristics of Livestock Depredations by Large Carnivores along a Kenyan Reserve Border.” Biological Conservation 128 (4): 529–41. https://doi.org/10.1016/j.biocon.2005.10.021. Lesilau, Francis, Myrthe Fonck, Maria Gatta, Charles Musyoki, Maarten Van, Gerard A Persoon, Kees C J M Musters, Geert R De Snoo, and Hans H De Iongh. 2018. “Effectiveness of a LED Flashlight Technique in Reducing Livestock Depredation by Lions (Panthera Leo ) around Nairobi National Park , Kenya.” PLoS ONE 13 (1): 1–18. https://doi.org/10.1371/journal.pone.0190898. Lichtenfeld, Laly L., Charles Trout, and Elvis L. Kisimir. 2015. “Evidence-Based Conservation: Predator-Proof Bomas Protect Livestock and Lions.” Biodiversity and Conservation 24 (3): 483–91. https://doi.org/10.1007/s10531-014-0828-x. Lindsey, Peter A., Johan T. Du Toit, and M. G.L. Mills. 2005. “Attitudes of Ranchers towards African Wild Dogs Lycaon Pictus: Conservation Implications on Private Land.” Biological Conservation 125 (1): 113–21. https://doi.org/10.1016/j.biocon.2005.03.015. Lischka, Stacy A, Tara L Teel, Heather E Johnson, Sarah E Reed, Stewart Breck, Andrew Don, and Kevin R Crooks. 2018. “A Conceptual Model for the Integration of Social and Ecological Information to Understand Human-Wildlife Interactions.” Biological Conservation 225: 80–87. https://doi.org/10.1016/j.biocon.2018.06.020. Loveridge, Andrew J., Timothy Kuiper, Roger H. Parry, Lovemore Sibanda, Jane Hunt Hunt, Brent Stapelkamp, Lovelater Sebele, and David W. Macdonald. 2017. “Bells, Bomas and Beefsteak: Complex Patterns of Human-Predator Conflict at the Wildlife-Agropastoral Interface in Zimbabwe.” PeerJ 5: e2898. https://doi.org/10.7717/peerj.2898. Loveridge, Andrew J, Sonam W Wang, Laurence G Frank, and John Seidensticker. 2010. “People and Wild Felids : Conservation of Cats and Management of Conflicts.” In Biology and Conservation of Wild Felids, 161–95. Oxford: Oxford University Press. Maclennan, Seamus D., Rosemary J. Groom, David W. Macdonald, and Laurence G. Frank. 2009. “Evaluation of a Compensation Scheme to Bring about Pastoralist Tolerance of Lions.” Biological Conservation 142 (11): 2419–27. https://doi.org/10.1016/j.biocon.2008.12.003. Madden, Francine. 2004. “Creating Coexistence between Humans and Wildlife: Global Perspectives on Local Efforts to Address Human–Wildlife Conflict.” Human Dimensions of Wildlife 9 (4): 247–57. https://doi.org/10.1080/10871200490505675. Marchini, Silvio, and David W Macdonald. 2012. “Predicting Ranchers ’ Intention to Kill Jaguars : Case Studies in Amazonia and Pantanal.” Biological Conservation 147 (1): 213–21. https://doi.org/10.1016/j.biocon.2012.01.002. Marker, L. L., M. G.L. Mills, and D. W. Macdonald. 2003. “Factors Influencing Perceptions of Conflict and Tolerance toward Cheetahs on Namibian Farmlands.” Conservation Biology 17 (5): 1290–98. https://doi.org/10.1046/j.1523- 1739.2003.02077.x. Marker, L.L, and A.J Dickman. 2005. “Factors Affecting Leopard (Panthera Pardus) Spatial Ecology, with Particular Reference to Namibian Farmlands.” South African Journal of Wildlife Research 35 (2, October): 105–115. http://www.cheetah.org/ama/orig/leopard.pdf. Mazerolle, M.J. 2006. “Improving Data Analysis in Herpetology: Using Akaike’s Information Criterion (AIC) to Assess the Strength of Biological Hypotheses.”

80 Amphibia-Reptilia 27 (2): 169–80. McManus, J. S., A. J. Dickman, D. Gaynor, B. H. Smuts, and D. W. Macdonald. 2015. “Dead or Alive? Comparing Costs and Benefits of Lethal and Non-Lethal Human– Wildlife Conflict Mitigation on Livestock Farms.” Oryx 49 (04): 687–95. https://doi.org/10.1017/S0030605313001610. Michalski, Fernanda, R. L P Boulhosa, A. Faria, and C. A. Peres. 2006. “Human-Wildlife Conflicts in a Fragmented Amazonian Forest Landscape: Determinants of Large Felid Depredation on Livestock.” Animal Conservation 9 (2): 179–88. https://doi.org/10.1111/j.1469-1795.2006.00025.x. Miller, Jennifer R.B., Yadvendradev V. Jhala, and Oswald J. Schmitz. 2016. “Human Perceptions Mirror Realities of Carnivore Attack Risk for Livestock: Implications for Mitigating Human-Carnivore Conflict.” PLoS ONE 11 (9): 1–15. https://doi.org/10.1371/journal.pone.0162685. Minin, Enrico Di, Rob Slotow, Luke T.B. Hunter, Federico Montesino Pouzols, Tuuli Toivonen, Peter H. Verburg, Nigel Leader-Williams, Lisanne Petracca, and Atte Moilanen. 2016. “Global Priorities for National Carnivore Conservation under Land Use Change.” Scientific Reports 6 (April): 1–9. https://doi.org/10.1038/srep23814. Mishra, Charudutt. 1997. “Livestock Depredation by Large Carnivores in the Indian Trans-Himalaya: Conflict Perceptions and Conservation Prospects.” Environmental Conservation 24 (4): 338–43. https://doi.org/10.1017/S0376892997000441. Mishra, Charudutt, Priscilla Allen, Tom Mccarthy, M. D. Madhusudan, Agvaantserengiin Bayarjargal, and Herbert H T Prins. 2003. “The Role of Incentive Programs in Conserving the Snow Leopard The Role of Incentive Programs in Conserving the Snow Leopard.” Conservation Biology 17 (6): 1512–20. https://doi.org/10.1111/j.1523-1739.2003.00092.x. Miththapala, Sriyanie, John Seidensticker, and Stephen J. O’Brien. 1996. “Phylogeographic Subspecies Recognition in Leopards (Panthera Pardus): Molecular Genetic Variation.” Conservation Biology 10 (4): 1115–32. https://doi.org/10.1046/j.1523-1739.1996.10041115.x. Mkonyi, Felix J., Anna B. Estes, Maurus J. Msuha, Laly L. Lichtenfeld, and Sarah M. Durant. 2017. “Socio-Economic Correlates and Management Implications of Livestock Depredation by Large Carnivores in the Tarangire Ecosystem, Northern Tanzania.” International Journal of Biodiversity Science, Ecosystem Services and Management 13 (1): 248–63. https://doi.org/10.1080/21513732.2017.1339734. Mueller-Domboi, D. 1972. “Crown Distortion and Elephant Distribution in the Woody Vegetations of Ruhuna National Park, Ceylon.” Ecology 53: 208–26. Myers, N, R a Mittermeier, C G Mittermeier, G a da Fonseca, and J Kent. 2000. “Biodiversity Hotspots for Conservation Priorities.” Nature 403 (6772): 853–58. https://doi.org/10.1038/35002501. Naidoo, Robin, L. Chris Weaver, Richard W. Diggle, Greenwell Matongo, Greg Stuart- Hill, and Chris Thouless. 2016. “Complementary Benefits of Tourism and Hunting to Communal Conservancies in Namibia.” Conservation Biology 30 (3): 628–38. https://doi.org/10.1111/cobi.12643. Naughton-treves, Lisa, Jose Luis Mena, Adrian Treves, Nora Alvarez, and Volker Christian Radeloff. 2003. “Naughton-Treves et Al. - 2003 - Wildlife Survival Beyond Park Boundaries the Impact of Slash-and-Burn Agriculture and Hunting on Mammals in Tambopata , Peru.Pdf” 17 (4): 1106–17. Ng, Victoria K.Y., and Robert A. Cribbie. 2017. “Using the Gamma Generalized Linear

81 Model for Modeling Continuous, Skewed and Heteroscedastic Outcomes in Psychology.” Current Psychology 36 (2): 225–35. https://doi.org/10.1007/s12144- 015-9404-0. Nyhus, Philip J. 2016. Human–Wildlife Conflict and Coexistence. Annual Review of Environment and Resources. Vol. 41. https://doi.org/10.1146/annurev-environ- 110615-085634. O’Brien, Timothy G., Margaret F. Kinnaird, and Hariyo T. Wibisono. 2003. “Crouching Tigers, Hidden Prey: Sumatran Tiger and Prey Populations in a Tropical Forest Landscape.” Animal Conservation 6 (2): 131–39. https://doi.org/10.1017/S1367943003003172. Ogada, Mordecai O, Rosie Woodroffe, Nicholas O Oguge, and Laurence G Frank. 2003. “Limiting Depredation by African Carnivores: The Role of Livestock Husbandry.” Conservation Biology 17 (6): 1521–30. Ogutu, Joseph O., Robin S. Reid, Hans Peter Piepho, Nicholas T. Hobbs, Michael E. Rainy, Russell L. Kruska, Jeffrey S. Worden, and Meshack Nyabenge. 2014. “Large Herbivore Responses to Surface Water and Land Use in an East African Savanna: Implications for Conservation and Human-Wildlife Conflicts.” Biodiversity and Conservation 23 (3): 573–96. https://doi.org/10.1007/s10531-013-0617-y. Ohrens, Omar, Cristian Bonacic, and Adrian Treves. 2019. “Non-Lethal Defense of Livestock against Predators: Flashing Lights Deter Puma Attacks in Chile.” Frontiers in Ecology and the Environment 17 (1): 32–38. https://doi.org/10.1002/fee.1952. Packer, C., S. Altizer, M. Appel, E. Brown, J. Martenson, S. J. O’Brien, M. Roelke- Parker, R. Hofmann-Lehmann, and H. Lutz. 1999. “Viruses of the Serengeti: Patterns of Infection and Mortality in African Lions.” Journal of Animal Ecology 68 (6): 1161–78. https://doi.org/10.1046/j.1365-2656.1999.00360.x. Packer, C., D. Ikanda, B. Kissui, and H. Kushmir. 2005. “Lion Attacks on Humans in Tanzania.” Nature 463 (7053): 927–28. https://doi.org/http://dx.doi.org/10.1038/436927a. Packer, C., A. Loveridge, S. Canney, T. Caro, S. T. Garnett, M. Pfeifer, K. K. Zander, et al. 2013. “Conserving Large Carnivores: Dollars and Fence.” Ecology Letters 16 (5): 635–41. https://doi.org/10.1111/ele.12091. Pailler, Sharon, Robin Naidoo, Neil D. Burgess, Olivia E. Freeman, and Brendan Fisher. 2015. “Impacts of Community-Based Natural Resource Management on Wealth, Food Security and Child Health in Tanzania.” PLoS ONE 10 (7): 1–22. https://doi.org/10.1371/journal.pone.0133252. Patterson, Bruce D, Samuel M Kasiki, Edwin Selempo, and Roland W Kays. 2004. “Livestock Predation by Lions ( Panthera Leo ) and Other Carnivores on Ranches Neighboring Tsavo National Parks , Kenya.” Biological Conservation 119: 507–16. https://doi.org/10.1016/j.biocon.2004.01.013. Perera, B M a O, and M C N Jayasuriya. 2008. “The Dairy Industry in Sri Lanka : Current Status and Future Directions for a Greater Role in National Development.” Veterinary Medicine, no. February: 115–26. https://doi.org/10.4038/jnsfsr.v36i0.8050. Phillips, W.W.A. 1980. Manual of the Mammals of Sri Lanka. 2nd ed. Colombo: Wildlife and Nature Protection Society of Sri Lanka. Pimentel, D., and N. Kounang. 1998. “Ecology of Soil Erosion in Ecosystems.” Ecosystems 1 (5): 416–26.

82 Polisar, J., I.E. Maxit, D. Scognamillo, L. Farrell, M. Sunquist, and J.F. Eisenberg. 2003. “Jaguars, Pumas, Their Prey Base, and Cattle Range: Ecological Interpretation of a Management Problem.” Biological Conservation 109: 297–310. Pooley, Simon P., J. Andrew Mendelsohn, and E. J. Milner-Gulland. 2014. “Hunting down the Chimera of Multiple Disciplinarity in Conservation Science.” Conservation Biology 28 (1): 22–32. https://doi.org/10.1111/cobi.12183. Prasannath, Vanitha. 2015. “Trends and Developments in Sri Lanka ’ s Livestock Industry” 01 (04): 46–55. R Core Team. 2018. “R: A Language and Environment for Statistical Computing.” Vienna, Austria: R Foundation for Statistical Computing. https://www.r-project.org/. Ranaweera, N.F.C. 2009. “Sri Lanka: Opportunities for Dairy Sector Growth.” Food and Agriculture Organisation 2003 (Table 1): 93–109. ftp://ftp.fao.org/docrep/fao/011/i0588e. Ranaweera, N.F.C. 2015. “Sri Lanka_ Opportunities for Dairy Sector Growth.” Smallholder Dairy Development: Lessons Learned in Asia. Bangkok: APHCA/FAO, 93109. Rasmussen, G S A. 1999. “Livestock Predation by the Painted Hunting Dog Lycaon Pictus in a Cattle Ranching Region of Zimbabwe: A Case Study.” Biologi 88: 133– 39. Redpath, Steve M., Juliette Young, Anna Evely, William M. Adams, William J. Sutherland, Andrew Whitehouse, Arjun Amar, et al. 2013. “Understanding and Managing Conservation Conflicts.” Trends in Ecology and Evolution 28 (2): 100– 109. https://doi.org/10.1016/j.tree.2012.08.021. Revelle, W. 2018. “Psych: Procedures for Personality and Psychological Research.” Evanston, Illinois: Northwestern University. https://cran.r- project.org/package=psych. Ripple, W. J., J. A. Estes, R. L. Beschta, C. C. Wilmers, E. G. Ritchie, M. Hebblewhite, J. Berger, et al. 2014. “Status and Ecological Effects of the World’s Largest Carnivores.” Science 343 (6167): 1241484–1241484. https://doi.org/10.1126/science.1241484. Robbins, Paul, Kendra Mcsweeney, Thomas Waite, and Jennifer Rice. 2006. “Even Conservation Rules Are Made to Be Broken: Implications for Biodiversity.” Environmental Management 37 (2): 162–69. https://doi.org/10.1007/s00267-005- 0009-5. Roberge, Jean Michel, and Per Angelstam. 2004. “Usefulness of the Umbrella Species Concept as a Conservation Tool.” Conservation Biology 18 (1): 76–85. https://doi.org/10.1111/j.1523-1739.2004.00450.x. Romañach, Stephanie S., Peter A. Lindsey, and Rosie Woodroffe. 2007. “Determinants of Attitudes towards Predators in Central Kenya and Suggestions for Increasing Tolerance in Livestock Dominated Landscapes.” Oryx 41 (2): 185–95. https://doi.org/10.1017/S0030605307001779. Rostro-Garcia, Susana, Jan F. Kamler, and Luke T B Hunter. 2015. “To Kill, Stay or Flee: The Effects of Lions and Landscape Factors on Habitat and Kill Site Selection of Cheetahs in South Africa.” PLoS ONE 10 (2): 1–20. https://doi.org/10.1371/journal.pone.0117743. Rostro-García, Susana, Lhendup Tharchen, Leandro Abade, Christos Astaras, Samuel A. Cushman, and David W. Macdonald. 2016. “Scale Dependence of Felid Predation Risk: Identifying Predictors of Livestock Kills by Tiger and Leopard in Bhutan.”

83 Landscape Ecology 31 (6): 1277–98. https://doi.org/10.1007/s10980-015-0335-9. Saldana, Johnny. 2012. “An Introduction to Codes and Coding.” The Coding Manual for Qualitative Researchers, no. 2006: 1–8. https://doi.org/10.1519/JSC.0b013e3181ddfd0a. Salerno, Jonathan, Monique Borgerhoff Mulder, Mark N. Grote, Margherita Ghiselli, and Craig Packer. 2016. “Household Livelihoods and Conflict with Wildlife in Community-Based Conservation Areas across Northern Tanzania.” Oryx 50 (4): 702–12. https://doi.org/10.1017/S0030605315000393. Samarkoon, P.A.G.V. 1999. “Conserving the Leopard in Sri Lanka: The Need for a Management Strategy.” Loris 22: 14–16. Sanderson, Steven E., and Kent H. Redford. 2003. “Contested Relationships between Biodiversity Conservation and Poverty Alleviation.” Oryx 37 (4): 389–90. https://doi.org/10.1017/S003060530300070X. Sandker, Marieke, Bruce M. Campbell, Zacharie Nzooh, Terry Sunderland, Victor Amougou, Louis Defo, and Jeffrey Sayer. 2009. “Exploring the Effectiveness of Integrated Conservation and Development Interventions in a Central African Forest Landscape.” Biodiversity and Conservation 18 (11): 2875–92. https://doi.org/10.1007/s10531-009-9613-7. Sangay, Tiger, and Karl Vernes. 2008. “Human-Wildlife Conflict in the Kingdom of Bhutan: Patterns of Livestock Predation by Large Mammalian Carnivores.” Biological Conservation 141 (5): 1272–82. https://doi.org/10.1016/j.biocon.2008.02.027. Santiapillai, C., M.R. Chambers, and C. Jayawardene. 1981. “Observations on the Sambar Cervus Unicolor Kerr, 1792 (Mammalia: Cervidae) in the Ruhuna National Park, Sri Lanka.” Ceylon Journal of Science (Biological Sciences) 14: 193–204. Schaller, G.B. 1967. The Deer and the Tiger: A Study of Wildlife in India. Chicago: University of Chicago Press. Sekhar, Nagothu Udaya. 2003. “Local People’s Attitudes towards Conservation and Wildlife Tourism around Sariska Tiger Reserve, India.” Journal of Environmental Management 69 (4): 339–47. https://doi.org/10.1016/j.jenvman.2003.09.002. Shehzad, Wasim, Muhammad Ali Nawaz, François Pompanon, Eric Coissac, Tiayyba Riaz, Safdar Ali Shah, and Pierre Taberlet. 2015. “Forest without Prey: Livestock Sustain a Leopard Panthera Pardus Population in Pakistan.” Oryx 49 (2): 248–53. https://doi.org/10.1017/S0030605313001026. Sodhi, Navjot S., Mary Rose C Posa, Tien Ming Lee, David Bickford, Lian Pin Koh, and Barry W. Brook. 2010. “The State and Conservation of Southeast Asian Biodiversity.” Biodiversity and Conservation 19 (2): 317–28. https://doi.org/10.1007/s10531-009-9607-5. Soh, Yi Hui, Luis Roman Carrasco, Dale G. Miquelle, Jinsong Jiang, Jun Yang, Emma J. Stokes, Jirong Tang, Aili Kang, Peiqi Liu, and Madhu Rao. 2014. “Spatial Correlates of Livestock Depredation by Amur Tigers in Hunchun, China: Relevance of Prey Density and Implications for Protected Area Management.” Biological Conservation 169: 117–27. https://doi.org/10.1016/j.biocon.2013.10.011. Sri Lanka Department of Census and Statistics. 2012. “Census of Population and Housing of Sri Lanka , 2012.” http://www.statistics.gov.lk/PopHouSat/CPH2011/Pages/Activities/Reports/District/ Kurunegala/A8.pdf. Sri Lanka Tourism Development Authority. 2017. “2017 Annual Statistical Report.”

84 Colombo, Sri Lanka. http://www.sltda.lk/sites/default/files/annual-statical-report- 2017.pdf. Stein, A. B., V. Athreya, P. Gerngross, G. Balme, P. Henschel, U. Karanth, D. Miquelle, et al. 2016. “The IUCN Red List of Threatened Species: Panthera Pardus” 8235. http://www.iucnredlist.org/details/15954/0. Stone, Suzanne A., Stewart W. Breck, Jesse Timberlake, Peter M. Haswell, Fernando Najera, Brian S. Bean, and Daniel J. Thornhill. 2017. “Adaptive Use of Nonlethal Strategies for Minimizing Wolf-Sheep Conflict in Idaho.” Journal of Mammalogy 98 (1): 33–44. https://doi.org/10.1093/jmammal/gyw188. Störmer, Nuria, L. Chris Weaver, Greg Stuart-Hill, Richard W. Diggle, and Robin Naidoo. 2019. “Investigating the Effects of Community-Based Conservation on Attitudes towards Wildlife in Namibia.” Biological Conservation 233 (March 2018): 193–200. https://doi.org/10.1016/j.biocon.2019.02.033. Suryawanshi, Kulbhushansingh R., Yash Veer Bhatnagar, Ranjana Pal, Rishi Kumar Sharma, Abi Tamim Vanak, and Chandrima Home. 2017. “Commensal in Conflict: Livestock Depredation Patterns by Free-Ranging Domestic Dogs in the Upper Spiti Landscape, Himachal Pradesh, India.” Ambio 46 (6): 655–66. https://doi.org/10.1007/s13280-016-0858-6. Swenson, J.E., and H Andrén. 2005. “A Tale of Two Countries: Large Carnivore Depredation and Compensation Schemes in Sweden and Norway.” In People and Wildlife: Conflict or Coexistence?, edited by Rosie Woodroffe, Simon Thirgood, and Alan Rabinowitz, 323–39. New York: Cambridge University Press. Takahata, Chihiro, Scott Eric Nielsen, Akiko Takii, and Shigeyuki Izumiyama. 2014. “Habitat Selection of a Large Carnivore along Human-Wildlife Boundaries in a Highly Modified Landscape.” PLoS ONE 9 (1). https://doi.org/10.1371/journal.pone.0086181. Talha, Mowdudul Hasan, Mohan Mia, Jannatul Mawa Momu, Mamun Ur Rahman, Mustaq Ahmad, Maffuz Rahman Adnan, Alimul Razee, Nazmul Haque, and Bashudeb Paul. 2018. “Morphometric , Productive and Reproductive Traits of Indian Peafowl (Pavo Cristatus) in Bangladesh.” International Journal of Development Research 8 (2): 19039–43. Taylor, Russell. 2009. “Community Based Natural Resource Management in Zimbabwe: The Experience of CAMPFIRE.” Biodiversity and Conservation 18 (10): 2563–83. https://doi.org/10.1007/s10531-009-9612-8. Thirgood, Simon, Rosie Woodroffe, and Alan Rabinowitz. 2005. “The Impact of Human- Wildlife Conflict on Humans Lives and Livelihoods.” People and Wildlife: Conflict or Coexistence?, 13–26. https://doi.org/http://dx.doi.org/10.1017/CBO9780511614774. Thorn, Michelle, Matthew Green, Kelly Marnewick, and Dawn M. Scott. 2015. “Determinants of Attitudes to Carnivores: Implications for Mitigating Human– Carnivore Conflict on South African Farmland.” Oryx 49 (02): 270–77. https://doi.org/10.1017/S0030605313000744. Treves, Adrian, and K. Ullas Karanth. 2003. “Human-Carnivore Conflict and Perspectives on Carnivore Management Worldwide.” Conservation Biology 17 (6): 1491–99. https://doi.org/10.1111/j.1523-1739.2003.00059.x. Treves, Adrian, Miha Krofel, and Jeannine McManus. 2016. “Predator Control Should Not Be a Shot in the Dark.” Frontiers in Ecology and the Environment 14 (7): 380– 88. https://doi.org/10.1002/fee.1312.

85 Treves, Adrian, Lisa Naughton-Treves, Elizabeth K. Harper, David J. Mladenoff, Robert A. Rose, Theodore A. Sickley, and Adrian P. Wydeven. 2004. “Predicting Human- Carnivore Conflict: A Spatial Model Derived from 25 Years of Data on Wolf Predation on Livestock.” Conservation Biology 18 (1): 114–25. https://doi.org/10.1111/j.1523-1739.2004.00189.x. Treves, Adrian, Robert B. Wallace, Lisa Naughton-Treves, and Andrea Morales. 2006. “Co-Managing Human–Wildlife Conflicts: A Review.” Human Dimensions of Wildlife 11 (6): 383–96. https://doi.org/10.1080/10871200600984265. Turner, Alan. 1997. The Big Cats and Their Fossil Relatives: An Illustrated Guide to Their Evolution and Natural History. New York: Columbia University Press. https://doi.org/10.2307/1313168. United Nations. 2015. “Transforming Our World: The 2030 Agenda for Sustainable Development.” In Transforming Our World: The 2030 Agenda for Sustainable Development, 1–29. https://doi.org/10.1017/CBO9781107415324.004. Vernooij, Adriaan, Wim Houwers, and Jelle Zijlstra. 2015. “Old Friends - New Trends,” no. January. Villatoro, Federico J., Lisa Naughton-Treves, Maximiliano A. Sepúlveda, Paulina Stowhas, Fernando O. Mardones, and Eduardo A. Silva-Rodríguez. 2019. “When Free-Ranging Dogs Threaten Wildlife: Public Attitudes toward Management Strategies in Southern Chile.” Journal of Environmental Management 229 (December 2017): 67–75. https://doi.org/10.1016/j.jenvman.2018.06.035. Wang, S W, and D W Macdonald. 2006. “Livestock Predation by Carnivores in Jigme Singye Wangchuck National Park , Bhutan.” Biological Conservation 129: 558–65. https://doi.org/10.1016/j.biocon.2005.11.024. Wierzbowska, Izabela A., Magdalena Hędrzak, Barłomiej Popczyk, Henryk Okarma, and Kevin R. Crooks. 2016. “Predation of Wildlife by Free-Ranging Domestic Dogs in Polish Hunting Grounds and Potential Competition with the Grey Wolf.” Biological Conservation 201: 1–9. https://doi.org/10.1016/j.biocon.2016.06.016. Wijeyeratne, G. 2008. A Photographic Guide to Mammals of Sri Lanka. London, UK: Bloomsbury Publishing PLC. ———. 2016. Mammals of Sri Lanka. London, UK: Bloomsbury Natural History. Woodroffe, Rosie, Laurence G. Frank, Peter A. Lindsey, Symon M.K. Ole Ranah, and Stephanie Romañach. 2007. “Livestock Husbandry as a Tool for Carnivore Conservation in Africa’s Community Rangelands: A Case-Control Study.” Biodiversity and Conservation 16 (4): 1245–60. https://doi.org/10.1007/s10531-006- 9124-8. Woodroffe, Rosie, and Joshua R. Ginsberg. 1998. “Edge Effects and the Extinction of Populations inside Protected Areas.” Science 280 (5372): 2126–28. https://doi.org/10.1126/science.280.5372.2126. Woodroffe, Rosie, Peter Lindsey, Stephanie Romañach, Andrew Stein, and Symon M.K. Ole Ranah. 2005. “Livestock Predation by Endangered African Wild Dogs (Lycaon Pictus) in Northern Kenya.” Biological Conservation 124 (2): 225–34. https://doi.org/10.1016/j.biocon.2005.01.028. Woodroffe, Rosie, Simon Thirgood, and Alan Rabinowitz. 2005. “The Impact of Human- Wildlife Conflict on Natural Systems.” People and Wildlife, Conflict or Coexistence?, 1–12. https://doi.org/10.1017/S0030605306000202. Zimmermann, A., M.J. Walpole, and N. Leader-Williams. 2005. “Cattle Ranchers’ Attitudes to Conflicts with Jaguar Panthera Onca in the Pantanal of Brazil.” Oryx 39

86 (04): 406. https://doi.org/10.1017/S0030605305000992. Zuur, A, H Hilbe, and E Ieno. 2013. A Beginner’s Guide to GLM and GLMM with R : A Frequentist and Bayesian Perspective for Ecologists. Nweburgh, UK: Highland Statistics Ltd. Zuur, A, E Ieno, and C Elphick. 2010. “A Protocol for Data Exploration to Avoid Common Statistical Problems.” Methods in Ecology and Evolution 1 (1): 3–14. https://doi.org/10.1111/j.2041-210X.2009.00001.x.

87 Appendix A: Survey protocol

Administrative Information Interviewer: ______Date: ______Study site: ______GPS: ______CODE: ______Oral consent: ______, initials:____

1. Background On this farm, you are the o owner o manager o employee o other______

Age: ______Gender: _____

What religion do you consider yourself to be? ______What ethnicity do you consider yourself to be? ______

How long have you worked at this farm? ______years What is the size of this farm? ______perches/acres How many people work at this farm? ______

Is dairy farming your primary source of income? o Yes, around ______% of income is derived o No, around ______% of income is derived from from cattle cattle

How many people directly depend on this farm for their income? ______

If there are other sources of income, tick all that apply: o tourism o farming (crops) o other______

2. Cattle How many adult cattle are there on this farm? Bulls:_____ Cows:______How many calves are there on this farm? ______How many pregnant cows are there on this farm? ______Do you have more cattle now than 3 years ago? o Same (0% change) o Many fewer (>50% less) o Fewer (0-50% less) o More (0-50% more) o Many more (>50% more)

Do you have more calves now than 3 years ago? o Same (0% change) o Many fewer (>50% less) o Fewer (0-50% less) o More (0-50% more) o Many more (>50% more)

88 3. Cattle husbandry and mitigations

How do you keep and graze your cattle? (Husbandry techniques practiced) o Cinnamon Wild pen – all cattle o Daytime adult grazing, calves penned o Cinnamon Wild pen – calves o Stick and thorn pen – calves o Cinnamon Wild pen with roof - calves o Stick and thorn pen – adults + calves o No pen – all cattle left out o Adults + calves tied and left outside o Daytime free grazing adults + calves o Only calves tied

Please give details on whether these methods occur at certain times of year, during certain times of day, how far apart any shelters are from each other, or any other relevant details:

______

What methods have you used to try and protect your cattle?

Mitigation Tried? Comments Willing to try Increased human presence (grazing) Light and sound distractions Relocating cattle to lower-risk areas Higher thorn walls Trained guard dogs Human night patrols Reinforced enclosures (using metal walls and roofing)

Would you be open to adjusting your husbandry practices if it would better safeguard your cattle? o yes o maybe o no

What’s the biggest limiting factor in changing husbandry practices? o Price of equipment o maintenance o labour o other______

Please explain and estimate the approximate cost and maintenance fees: Protecting cattle in pens: ______Disease prevention: ______Veterinary expenses: ______

Do you shift grazing locations? How often do you graze in the same location? Explain: ______

4. Wildlife (Leopards) & Depredation Is there a change in the amount of wildlife seen? o yes o no

Less/more Level: 1 2 3 4 5 (1=low, 5=high)

89 Animals (photos will be used to help identify)

Wildcats: Leopard Rusty spotted jungle cat fishing cat Primates: grey langur macaque loris Deer: spotted deer barking deer sambar mouse deer Bear: yes no Elephant: yes no Small mammals: pangolin hare giant squirrel Other species: jackal wild boar other:______

What are the issues you encounter raising cattle in this landscape? (monkeys, elephants, leopards)

______

Do wildlife sightings around your farm or the community make you feel concerned for your safety, or the safety of your cattle? Please explain:

______

Do you see more or less leopards (or signs of leopard) than 3 years ago? o many more o more o same o fewer o many fewer

Have there been any losses of cattle at this farm within the last 3 years? o yes, ______cattle lost o no

If yes, please describe the incidents and the time of year most incidents occurred, if possible:

______

What was the cause of cattle loss? ______

o don’t know o snake bite, ~_____ o disease, ~ ____ o theft, ~ ____ o wandering off, ~____ o leopard predation, ~ ____

IF leopard predation, how did you know this was a leopard? What were the signs?

______

Is number of cattle lost to leopard predation more or less than 3 years ago (new government)? o more now o less now o same o don’t know

What is your course of action when an animal is killed? (allow them to answer first) o Report to DWC o Report to police o Ignore

90 o Move cattle to new location

What if you lose multiple cattle? ______

If cattle are lost to predation, what is the biggest cost for you (direct, indirect)?

______

If predation occurs, which animals are most often lost? o Strong/healthy animals o Older/weaker animals o Pregnant animals o Calves

Have you noticed any patterns in these losses in terms of time and season? a) time o always night o always daytime o mostly night o mostly day o no pattern o other______b) season o mostly wet season o mostly dry season o all year round o other ______

Have you had any successes on your farm in safeguarding your cattle? If so, what was the method and how long has it been implemented for?

Have you had less sightings of leopards since these measures were implemented? o yes o no o don’t know

What is the approximate cost to you (Rs) of a) Loss of 1 female calf ______Rs b) Loss of 1 pregnant cow ______Rs c) Loss of 1 lactating cow ______Rs d) Loss of 1 male stud ______Rs

How much does the loss of cattle to leopards worry you? o very much o some o not much o not at all

Do you feel the government should play a role in this issue? o Yes o no o don’t know/maybe

If yes, could you suggest how? ______

Have you benefitted from any tourism or development projects focused on the leopard? o yes o no

5. Conservation Is Sri Lankan wildlife (elephants, primates, leopards, amphibians, birds) important to you?

Elephants o yes o somewhat o no Deer o yes o somewhat o no Primates o yes o somewhat o no Leopards o yes o somewhat o no Reptiles and amphibians o yes o somewhat o no Birds o yes o somewhat o no

Do you think there are too many/too few/the right amount of wildlife in this area?

Elephant o too many o too few o right amount Deer o too many o too few o right amount Primates o too many o too few o right amount Leopards o too many o too few o right amount Reptiles and amphibians o too many o too few o right amount Birds o too many o too few o right amount

6. Awareness Do you know that the Sri Lankan leopard is endangered, and a very important species for the larger ecosystems they live in?

o yes o no

Do you know you graze your cattle very close to a national park where leopards and other animals live? o yes o no Do you know that these protected national parks provide the best environment for leopards to live, across Sri Lanka? o yes o no

92 Do you know that leopards do exist outside of these protected national parks, and can live and reproduce in smaller forest patches? o yes o no

Do you know that leopards in Sri Lanka like to eat animals like deer, langurs, buffalo, hares and wild boar? o yes o no

Do you know that dairy cattle often have reduced defence instincts and may be easier for older or younger leopards to capture and eat? o yes o no

Do you know that leopards produce economic benefits both in this area and across Sri Lanka? o yes o no

Do you know that Yala National Park is the most visited national park in Sri Lanka? o yes o no

Do you know that Yala National Park generates over 600 million rupees each year? o yes o no

Do you know that a nearby hotel has been giving out steel cattle enclosures as part of a program to market increased coexistence between cattle farmers and leopards in this area? o yes o no

Attitude Statements The following questions will be answered on a 5-point Likert scale (1 = strongly disagree, 5 = strongly agree)

+The nature and wildlife of Sri Lanka is a national treasure and should be conserved +I respect leopards for their economic value they bring to the country through wildlife tourism -My livelihood is more important than current leopard populations +It does not matter if a leopard kills a few of my cattle, they are wild animals just trying to survive -At this farm we cannot tolerate leopards killing any cattle at any time -I would be happier if there were fewer leopards around where I live and raise my cattle -I do not want to kill leopards, but if they kill my cattle I might have to +I would like to communicate and work together with scientists, government staff and organizations to find a solution to this problem

93 Appendix B: Semi-structured interview protocol

Could you tell me what you know about the Sri Lankan leopard? How do you consider them, as something you are sharing the land with? • Endangered? • Good/bad? Threat/nice animal? • Conflict experienced? Negative feeling? Increased feeling of risk following an incident? More violence? • Economic value of leopard tourism – do you benefit from the tourism in the area?

How do you farm your cattle? Do you think that might affect leopards in the area? • Defensive/prevention (mitigation) methods employed? How successful were they? • Biggest threat to cattle in your farm? Leopard? Drought? • How many cattle are you able to lose before taking action against leopards?

Would you be open to adjusting your farming practices if it would reduce the chance of leopard attacks? • Are you interested in testing some of these proposed mitigations and monitor the results in the short-term future? • Are you interested in maintaining a balance with the leopards, as long as they don’t significantly harm your cattle or personal safety? • Are you willing to cooperate with researchers and stakeholders to develop and try practical options? • What is your relationship with the DWC? Do you think they are open to cooperating with livestock owners? Do you think they are doing their job?

What’s the biggest issue or concern you have raising cattle here? • Has this changed over time? When did it change (was there a turning point)? • What assurances do you need? What do you think should be done?

Is there anything we haven’t discussed that you would like to mention? • Stay in touch re: progress

94 Appendix C: Respondent characteristics

Table C.1. Socio-demographic, socio-economic and livestock characteristics of respondents in Yala (n=61) and the Central Hills (n=52) who rear cattle in lands adjacent to national parks or protected sanctuaries. Surveys were conducted May-August 2019. Yala Central Hills Socio-demographic and socio-economic characteristics Proportion of men 1.00 0.94 Proportion of women 0.00 0.06 Average age (years) 38.34 41.67 Average number of dependents 3.36 5.44 Average expenditure per month (Rs) 21000 26200 Average time spent rearing livestock (years) 18.08 15.51 Average number of workers at each farm 1.69 1.83 Proportion who solely depend on livestock for 0.74 0.21 livelihood Livestock characteristics Average number of cattle (total) 84.33 3.56 Average number of calves 26.95 1.15 Livestock loss Proportion having lost livestock to leopards 0.90 0.00 Proportion having lost livestock to disease 0.80 0.52 Proportion having lost livestock to theft 0.36 0.00 Proportion having lost livestock to miscarriage 0.00 0.40

95 Appendix D: Exploratory Factor Analysis (EFA)

Exploratory Factor Analysis (EFA) is a form of dimension reduction when there are multiple variables measuring the same latent trait. Different dimensions (e.g. questions responses) are combined through a factor analysis approach to give a final (continuous) score for each respondent, that will be used in later regression modeling.

Extracting Factors First, we calculated the correlation matrix for each round of EFA depending on the variable type, as follows: • Pearson’s correlations for numeric data • Polychoric correlations for ordered variables • Tetrachoric correlations for dichotomous variables

We used a maximum likelihood factoring method due to its clearer interpretability. Maximum likelihood factoring method gives us the best estimate by maximising the empirical probability function. We used an oblique rotation when performing EFA as it allows for correlations to exist in the data, which it likely does when pertaining to social science data. We tested the two most commonly used and flexible rotations, ‘promax’ and ‘oblimin’, and found that the proportion of variance explained did not differ between rotations, for any factor used in this analysis. We chose to use the ‘promax’ oblique rotation.

We then produced a scree plot to determine the number of factors to use. The scree plot displays the factors and their eigenvalues using the correlation matrix. The number of factors that are strongly present in the data was determined visually at the point where the plot suddenly changes direction. When the eigenvalues indicated two factors be used, the decision to use two or one came down to the proportion of additional variance explained by including the second factor, the interpretability of each factor (i.e. can I understand and explain how and why these two factors are separate?) and the number of predictor variables the relatively small sample size allowed for (leaving a minimum of 10 observations per variable included). A general rule of thumb used was that each identified factor has at least 3 variables with high factor loadings (e.g. above 0.7), and that each variable loads highly on only one factor.

Factor loadings and factor scores The output of EFA is a table of factor loadings. The loading of each item (e.g. each attitude statement, or each column used in EFA) represents the strength and directionality of the relationship between each item and the underlying factor, from -1 to 1. If all loadings are very similar to each other, we can calculate linear combinations of items and ignore their weights, through a simple sum or average. However this is rare in reality, and if loadings vary we want an index to reflect that each item has an unequal association with the factor. We did this by combining the items through calculating an index variable through an optimally-weighted linear combination called factor scores. Each item’s weight is its factor loading, so each item’s contribution to the

96 factor score depends on how strongly it relates to the factor. We used Bartlett scores as opposed to Thurstone or Anderson-Rubin scores as this method uses maximum likelihood estimation to produce unbiased estimates of factor scores, and allows for correlations in the data which does exist in this context. Measurement error was not considered during this analysis due to sample size limitations. Factor scores were then created for each respondent, providing a continuous value of each latent phenomenon measured, relative to others surveyed, to be used in subsequent regression models.

R packages used • psych (Revelle 2018) • polycor (J. Fox 2016) • GPArotation (Coen and Jennrich 2005)

97 Appendix E: Complete model set representing all possible additive combinations to identify determinants of attitudes

CONS COST ECOL TOURISM LOST_LEO SOCIO AICc delta AICc 0.154 -3.411 -0.381 0.044 236.6 0.00 0.162 0.038 -3.143 -0.347 0.051 236.7 0.14 0.140 -3.644 -0.427 237.0 0.39 0.166 0.045 0.060 237.8 1.26 0.145 0.029 -3.463 -0.406 238.0 1.43 0.159 0.054 238.7 2.17 0.154 -3.528 -0.396 -0.010 0.042 239.0 2.40 0.141 -3.818 -0.449 -0.017 239.0 2.42 0.162 0.037 -3.209 -0.356 -0.006 0.050 239.3 2.76 0.172 0.046 -0.117 0.064 240.1 3.55 0.165 0.046 0.008 0.060 240.2 3.62 0.146 0.028 -3.615 -0.425 -0.014 240.3 3.69 0.165 0.045 -0.002 0.059 240.3 3.73 0.153 241.0 4.47 0.162 -0.080 0.057 241.0 4.48 0.155 -0.008 0.051 241.1 4.50 0.158 0.005 0.055 241.1 4.52 0.158 0.037 241.2 4.69 0.139 -0.032 242.1 5.57 0.172 0.047 -0.140 0.011 0.066 242.5 5.93 0.145 0.036 -0.031 242.5 5.95 0.165 0.046 0.000 0.008 0.060 242.8 6.19 0.146 0.154 243.0 6.44 0.151 0.037 0.147 243.3 6.76

98 CONS COST ECOL TOURISM LOST_LEO SOCIO AICc delta AICc 0.152 0.002 243.3 6.76 0.162 -0.094 0.007 0.058 243.4 6.88 0.155 -0.007 0.004 0.052 243.5 6.95 0.157 0.038 0.005 243.6 7.04 0.140 -0.033 -0.003 244.5 7.94 0.145 0.036 -0.031 0.000 245.0 8.43 0.146 0.154 0.000 245.4 8.83 0.151 0.037 0.143 0.003 245.8 9.23 -3.603 -0.475 257.3 20.75 0.010 -3.540 -0.468 259.6 23.02 -3.678 -0.485 -0.007 259.6 23.09 -3.587 -0.473 0.003 259.7 23.13 -0.086 260.0 23.47 0.017 -0.086 262.0 25.45 0.009 -3.606 -0.477 -0.006 262.0 25.45 0.011 -3.514 -0.464 0.004 262.0 25.47 0.675 262.1 25.53 -3.663 -0.482 -0.007 0.002 262.1 25.55 -0.083 0.010 262.2 25.67 -0.085 0.007 262.3 25.73 0.018 0.672 264.0 27.45 0.624 0.016 264.1 27.57 0.018 -0.081 0.012 264.2 27.69 0.657 0.012 264.3 27.70 0.018 -0.084 0.008 264.3 27.77 -0.081 0.008 0.011 264.6 28.00 0.010 -3.579 -0.472 -0.006 0.004 264.6 28.01 264.8 28.25 0.037 265.7 29.12 0.020 0.612 0.018 266.1 29.50 0.019 0.651 0.013 266.2 29.67 0.600 0.013 0.017 266.4 29.80 0.023 266.6 29.99 0.019 -0.079 0.010 0.013 266.6 30.07 0.019 266.6 30.08 0.023 0.039 267.4 30.85 0.024 0.038 267.4 30.88 0.022 0.584 0.015 0.020 268.3 31.76

99 CONS COST ECOL TOURISM LOST_LEO SOCIO AICc delta AICc 0.021 0.025 268.4 31.82 0.025 0.026 0.040 269.2 32.60

Table E.2. All possible additive combinations to explain determinants of attitudes towards leopards in the Central Hills, Sri Lanka, measured from 52 surveys in May-August 2018 using generalised linear models. Variables include the desire for increased government involvement (GOV), the level of worry that leopard sightings invoke (SIGHTINGS), importance of conservation (CONS), awareness of leopard-related tourism benefits (TOURISM), and the frequency of observing leopards or signs of leopards in the past 3 years (LEO_3). Df is degrees of freedom of the model, AICc is the AIC value adjusted for small sample size, delta AICc is the difference in AICc scores from the top model.

GOV SIGHTINGS CONS TOURISM LEO_3 df AICc delta AICc 0.314 3 214.7 0.00 0.326 0.037 4 215.1 0.39 0.333 -0.059 0.042 5 216.4 1.76 0.317 -0.040 4 216.5 1.85 0.349 -0.028 0.047 5 216.6 1.90 0.314 -0.006 4 216.9 2.19 0.320 -0.009 4 216.9 2.25 0.327 0.038 -0.008 5 217.2 2.50 0.357 -0.062 -0.030 0.054 6 217.9 3.24 0.351 -0.030 0.050 -0.010 6 218.7 4.00 0.334 -0.058 0.044 -0.008 6 218.7 4.04 0.317 -0.039 -0.005 5 218.9 4.19 0.322 -0.040 -0.008 5 218.9 4.21 0.320 -0.009 -0.006 5 219.2 4.53 0.358 -0.060 -0.031 0.056 -0.009 7 220.2 5.54 0.322 -0.038 -0.008 -0.005 6 221.3 6.65 2 224.4 9.75 0.027 3 225.8 11.17 -0.030 3 226.5 11.78 -0.006 3 226.5 11.87

100 GOV SIGHTINGS CONS TOURISM LEO_3 df AICc delta AICc 0.010 3 226.6 11.88 -0.042 0.030 4 227.8 13.10 0.028 -0.008 4 228.0 13.30 0.001 0.026 4 228.2 13.52 -0.031 0.011 4 228.7 14.00 -0.029 -0.005 4 228.7 14.02 0.010 -0.006 4 228.8 14.11 -0.041 0.031 -0.007 5 230.0 15.35 -0.042 0.000 0.030 5 230.2 15.55 -0.001 0.028 -0.008 5 230.4 15.75 -0.029 0.010 -0.005 5 231.0 16.35 -0.041 -0.001 0.032 -0.007 6 232.6 17.91

101 Appendix F: Camera trap deployment data sheet

CAMERA STATION DATA COLLECTION SHEET Site ID: Visit Date: (MM-DD-YYYY) Deployment Time (24 hr):

Crew: GPS Coordinates: Test photos

taken?

Camera Serial No:______

Camera active: Yes ☐ No ☐ Camera replaced: Yes ☐ No ☐ Camera moved: Yes ☐* No ☐ Camera repositioned: Yes ☐ No ☐ Batteries replaced SD card replaced # Motion-triggered photos % battery remaining ______% SD card full ______

Camera Site Description: Camera height (cm) :

Camera direction (Ex: NW) *If camera moved, note tree species camera is now attached to, if top of tree was sawed off, nearby landmarks, etc. Surrounding Habitat: Dom Tree Sp.: Dom Shrub Sp.:

Notes about visibility (+ potential barriers):

Water?

Comments:

Note presence of wildlife (e.g., scat, footprints) and human activity (e.g., litter, tracks).

102 How to complete the Camera Station Data Collection Sheet: Site ID: Name of site (e.g., Yala01) Visit Date: Date visited site (e.g., 04-19-2017) Deployment Time: Time at site (e.g., 13:30) Crew: Initials of crew (e.g., AU/PP) Test photos taken: Were whiteboard photos taken once camera was set, prior to leaving? Pls include Site ID and Date Camera active: Was the camera still operating/active when the crew arrived at the site? Camera replaced: Did the camera need to be replaced? Check if a new camera was deployed. Camera moved: Was the camera moved to a new location/tree at the same site? If yes, add info to Camera Site Description section. Camera repositioned: Did the camera need to be repositioned? For example, a bear might have slightly displaced, but not damaged, the camera, necessitating the camera to be repositioned on the same tree. Batteries replaced: Were fresh batteries put into the camera? SD card replaced: Was a new SD card put into the camera? % battery remaining: Record % battery remaining, displayed as ‘% lithium’ when camera is opened % SD card full: Record % of SD full, displayed as ‘% full’ when camera is opened # Motion-triggered photos: Record number of pictures taken by motion-triggered events, displayed as ‘# pictures’ when camera is opened If camera is being deployed for first time, write a brief site description. Include GPS coordinates in UTM and elevation. Camera height: height (in cm) to base of camera. Camera direction: which of the eight directions is the camera facing: N, NE, E, SE, S, SW, W, NW? Dom Tree Sp.:

Dom Shrub Sp.:

Comments: any comments about the site, including presence of wildlife and human activity.

103 Appendix G: Additional information on body mass values used for leopard prey species detected on camera traps and Relative Biomass Index (RBI) in kg/day for adult and juvenile prey

Table G.1. Sri Lankan leopard (Panthera pardus kotiya) ˆprey species detected at camera trap stations (n=28) during May-August 2018 and June-August 2019, and body weights (kg) used prior to calculating Relative Biomass Index (RBI) estimates. Detections were identified to the species and sex level as far as possible. Species Average bodyweight (kg) Source Black-naped hare (Lepus 2.5 (Wijeyeratne 2008) nigricollis) Ceylon spotted deer (Axis 75.5 (male), 49.5 (female), (Phillips 1980; Schaller 1967) axis ceylonensis) 22.7 (juvenile) Civet spp. (Paradoxurus 2.5 (Phillips 1980) spp.) Grey langur 12.5 (Wijeyeratne 2016) (Semnopithecus priam) Indian crested porcupine 13.05 (Phillips 1980) (Hystrix indica) Indian palm squirrel 0.112 Phillips 1980 (Funambulus palmarum) Indian peafowl (Pavo 5.66 (male), 3.52 (female), (Talha et al. 2018) cristatus) 1.70 (juvenile) Mongoose spp. 1.35 (Phillips 1980) (Herpestes spp.) Mouse deer (Moschiola 3.15 (Phillips 1980) meminna) Rat spp. (Rattus spp., 0.127g (avg of all) (Wijeyeratne 2016) Bandicota spp., Srilankamys spp) Sambar deer (Rusa 215 (male), 163 (female), (Santiapillai, Chambers, and unicolor unicolor) 45.5 (juvenile) Jayawardene 1981; Schaller 1967) Sri Lankan jackal (Canis 11 (Wijeyeratne 2016) aureus naria) Sri Lankan jungle fowl 0.915 (Dunning, Jr. 2007) (Gallus lafayettii) Water buffalo (Bubalus 100 (juvenile) (Phillips 1980) bubalis) Wild boar (Sus scrofa) 110 (male), 55 (female), 20 (Phillips 1980) (juvenile)

104 Table G.2. Camera trap stations in Yala, Sri Lanka (n=28), camera trap active days and the Relative Biomass Index (RBI) of adult and juvenile leopard prey species, measured in terms of kg/day. RBI values were obtained by calculating Relative Abundance Index (RAI) for each species at each station and multiplying this by the average bodyweight (see Table G.1). As far as possible, species were identified as adult males, adult females and juveniles, and the respective bodyweights were used. Camera station Camera trap RBI for juvenile prey RBI for adult prey active days species (kg/day) species (kg//day) Yala01 19 822.39 912.05 Yala02 60 6.08 100.21 Yala03 8 2.06 584.76 Yala04 33 123.41 250.07 Yala05 55 264.49 621.35 Yala06 19 1.05 94.65 Yala07 46 27.39 491.52 Yala08 50 1.50 74.72 Yala09 42 1.90 29.39 Yala10 30 1.33 43.50 Yala11 28 0.00 8.46 Yala12 14 0.00 6.96 Yala13 34 0.04 15.98 Yala14 37 53.11 100.90 Yala15 36 144.76 284.02 Yala16 14 33.57 407.74 Yala17 35 104.54 2058.01 Yala18 14 9.32 420.61 Yala19 21 164.40 373.96 Yala20 20 107.40 235.83 Yala21 20 563.75 617.62 Yala22 20 12.28 604.68 Yala23 13 0.00 3.56 Yala24 13 0.00 6.74 Yala25 13 0.00 0.32

105 Camera station Camera trap RBI for juvenile prey RBI for adult prey active days species (kg/day) species (kg//day) Yala26 10 392.20 9251.93 Yala27 10 0.50 59.80 Yala28 10 6.45 82.24

106 Appendix H: Camera trap photos

Fig.H.1. Sri Lankan leopard (Panthera pardus kotiya)

Fig.H.2. Grey langurs (Semnopithecus priam)

107

Fig.H.3. Ceylon spotted deer (Axis axis ceylonensis)

Fig.H.4. Ruddy mongoose (Herpestes smithii)

108

Fig.H.5. Wild boar (Sus scrofa)

Fig.H.6. Indian crested porcupine (Hystrix indica)

109

Fig.H.7. Ceylon spotted deer (Axis axis ceylonensis)

Fig.H.8. Black-naped hare (Lepus nigricollis)

110 Appendix I: Complete model set representing all possible additive combinations to identify factors associated with incidents of livestock depredation in Yala, Sri Lanka

Table I.1. All possible additive combinations to identify factors associated with incidents of livestock depredation in the past 3 years, reported by pastoralists in Yala, Sri Lanka, using generalised linear models with a negative binomial distribution. Variables include leopard native prey availability (PREY; kg/day), distance to national park boundary (DIST_NP;km), cattle husbandry (HUS), number of cattle (CATTLE), pastoralist residency time (RESIDENCY; years), road density within a 150m buffer (ROAD; km/km2) Df is degrees of freedom of the model, AICc is the AIC value adjusted for small sample size, delta AICc is the difference in AICc scores from the top model.

PREY DIST_NP HUS CATTLE RESIDENCY ROAD df AICc deltaAICc -0.607 0.403 4 311.7 0.00 -0.601 0.426 -0.107 5 312.5 0.89 -0.071 -0.608 0.391 5 313.5 1.86 -0.592 0.394 0.053 5 313.7 2.02 0.006 -0.607 0.403 5 314.0 2.37 -0.096 -0.602 0.413 -0.122 6 314.1 2.43 -0.594 0.420 -0.101 0.025 6 314.9 3.28 0.021 -0.600 0.426 -0.110 6 314.9 3.29 -0.068 -0.594 0.383 0.050 6 315.7 4.01 -0.026 -0.085 -0.609 0.389 6 315.9 4.25 0.005 -0.592 0.394 0.053 6 316.1 4.49 -0.018 -0.105 -0.603 0.412 -0.121 7 316.6 4.95 -0.094 -0.598 0.410 -0.118 0.016 7 316.6 4.96 0.019 -0.594 0.421 -0.104 0.024 7 317.4 5.79 -0.026 -0.082 -0.595 0.382 0.050 7 318.2 6.50 -0.018 -0.103 -0.599 0.408 -0.116 0.016 8 319.2 7.58 -0.189 -0.562 4 332.4 20.71 -0.552 3 332.6 20.92 -0.515 0.127 4 333.4 21.77 -0.175 -0.528 0.112 5 333.6 21.93 -0.067 -0.222 -0.567 5 334.4 22.77 -0.192 -0.562 -0.012 5 334.7 23.07

111 PREY DIST_NP HUS CATTLE RESIDENCY ROAD df AICc deltaAICc -0.553 0.022 4 334.8 23.17 0.021 -0.551 4 334.8 23.18 -0.516 0.047 0.135 5 335.6 23.95 -0.068 -0.209 -0.533 0.113 6 335.7 24.05 0.014 -0.514 0.126 5 335.8 24.13 -0.172 -0.528 0.011 0.115 6 336.0 24.38 0.352 3 336.3 24.61 0.324 0.147 4 336.7 25.02 -0.066 -0.223 -0.566 -0.007 6 336.9 25.23 0.382 -0.126 4 337.1 25.49 0.018 -0.553 0.019 5 337.2 25.52 0.005 -0.516 0.046 0.135 6 338.1 26.41 -0.070 -0.205 -0.533 0.018 0.117 7 338.2 26.58 -0.060 0.342 4 338.3 26.66 0.350 -0.093 0.121 5 338.3 26.67 0.017 0.351 4 338.5 26.88 -0.054 0.316 0.146 5 338.8 27.19 -0.087 0.371 -0.139 5 339.0 27.36 0.015 0.324 0.147 5 339.0 27.37 0.036 0.382 -0.131 5 339.4 27.75 -0.076 0.342 -0.106 0.116 6 340.4 28.73 -0.007 -0.064 0.341 5 340.7 29.03 0.030 0.351 -0.098 0.119 6 340.7 29.05 -0.007 -0.058 0.315 0.146 6 341.3 29.65 0.005 -0.085 0.371 -0.139 6 341.5 29.82 0.002 -0.075 0.342 -0.106 0.116 7 342.9 31.29 0.219 3 345.6 33.90 -0.145 0.212 4 346.6 34.96 2 347.0 35.34 0.034 0.224 4 347.8 36.10 -0.164 3 347.8 36.11 0.023 0.218 4 347.8 36.15 -0.040 -0.165 0.213 5 348.9 37.24 -0.144 0.003 0.212 5 349.0 37.34 0.034 3 349.1 37.48 -0.001 3 349.2 37.55 -0.174 -0.034 4 350.0 38.32 -0.036 -0.182 4 350.0 38.34

112 PREY DIST_NP HUS CATTLE RESIDENCY ROAD df AICc deltaAICc 0.017 0.030 0.223 5 350.1 38.46 -0.041 -0.163 0.007 0.214 6 351.4 39.70 0.036 -0.008 4 351.4 39.76 -0.032 -0.189 -0.031 5 352.3 40.64

113