Behaviour and Ecology of an Apex Predator, the Dingo Canis Lupus Dingo, and an Introduced Mesopredator, the Feral Cat Felis Catus

ResearchOnline@JCU

This file is part of the following reference:

Brook, Leila Amy (2013) Predator guild interactions in northern Australia: behaviour and ecology of an apex predator, the dingo Canis lupus dingo, and an introduced mesopredator, the feral cat Felis catus. PhD thesis, James Cook University.

Access to this file is available from:

http://researchonline.jcu.edu.au/36291/

The author has certified to JCU that they have made a reasonable effort to gain permission and acknowledge the owner of any third party copyright material included in this document. If you believe that this is not the case, please contact [email protected] and quote http://researchonline.jcu.edu.au/36291/

Predator guild interactions in northern Australia: behaviour and ecology of an apex predator, the dingo Canis lupus dingo, and an introduced mesopredator, the feral cat Felis catus

Thesis submitted by Leila Amy Brook BSc (Hons) University of Melbourne December 2013

for the degree of Doctor of Philosophy in the School of Marine and Tropical Biology James Cook University

Top: Dingo Canis lupus dingo on Marion Downs Sanctuary in the Kimberley, Western Australia Bottom: Feral cat Felis catus on Holroyd River Station, Cape York Peninsula, Queensland

iii

Acknowledgments

I am very grateful to have worked on such a ripper of a project during my PhD. I feel lucky to have been able to pursue an interesting field of research in a way which is all- absorbing, and to still find it fascinating at the end. Studying large, free-ranging predators in remote areas across northern Australia was a challenge but an enjoying one, from the thrill of getting hands-on during trapping, to learning to set realistic fieldwork expectations! The only downside of working in such magnificent examples of tropical savanna is that other places now pale in comparison. I have a lot of appreciation for all the fantastic people who facilitated this project and contributed support along the way. I must first thank my supervisors Chris Johnson, Euan Ritchie and Lin Schwarzkopf. I am grateful to have worked with such a knowledgeable and diverse team during my PhD. Chris, thanks for taking me on for this project. Your rich insights into the role of the dingo, your inspired scientific thinking and thoughtful comments supported me throughout my candidature. Euan, thanks for your constant positive attitude and knowledge not only of dingoes, predator ecology and field research but of all things savanna. You have passed on a love for the north and a quite serious birdwatching obsession. Thank you Lin for your support in the later years of my project, for the reassuring discussions, practical advice and guidance on chapters and analysis. I would also like to thank Jenny Martin, who with Euan, introduced me to many of the great things about Townsville and provided encouragement before and during my candidature. Every chapter of this thesis involves research conducted on properties owned by the Australian Wildlife Conservancy, who not only permitted me to work on their sanctuaries, but provided much in-kind support, knowledge, time and hospitality during my fieldwork. Thanks to Sarah Legge (Mornington), John Kanowski, Jill MacNicol (Piccaninny Plains), Peter and Pam Hensler (Mt Zero-Taravale), Mick Blackman (Brooklyn) and all the staff and volunteers that assisted with my project in the Kimberley. I am very grateful to the property owners or managers who allowed me to run surveys on their properties: Darcy and Raylee Byrnes of Holroyd River station, James and Elaine Spreadborough of Spoonbill station, Evan, Kim and Phil Acton of Millungera station and Brian Furber of Hidden Valley. I thank the Department of Defence for permission to work on defence land, and I particularly appreciate the help of Allan McManus who facilitated approvals for my research on both the Townsville and Mount Stuart Field Training Areas. A special thanks to the lovely Laura Mitchell, who worked as a research assistant on the dingo project in my first year. In addition to sharing the joys of learning about camera trapping and dingo lures, you taught me many valuable field skills and made the early field trips a barrel of laughs. Hugh McGregor helped immensely, taking time to introduce me to the details of feral cat trapping and GPS telemetry, sharing his cat wisdom, spending long nights processing dingoes and kindly sharing some of his hard-earned data.

v Damian Morrant has been a great go-to guy for all things dingo. Thanks for the many enthusiastic discussions on all aspects of dingo research, for teaching me the tricks of trapping and especially for volunteering your time and efforts on one of the most challenging field trips ever! A huge thanks to Angus McNab for your untiring help on the trapping field trip. I never would have seen so many fantastic birds and reptiles, or been willing to go spotlighting in our few non-working hours if you hadn't been there! I'm very grateful to Devi Hermsen, who spent several months helping on the dingo project and worked on several field trips. Your enthusiasm and great work made the trips much easier and lots of fun. I couldn't have achieved all the remote camera field work without the help of many great volunteers, in particular a big thanks to Kate Stookey, Susan Campbell, Paul Mitrovski, Frank Bird, Mika Trombini, Teigan Allan, Emily Hynes, Katrin Schmidt, Delphi Ward, Karen Glynn, Jen Leech and Gus McNab for coming on the long camera survey trips. I apologise especially to those who had to sweat it out in the hazmat suits or be exposed to particularly vile dingo attractants! I am grateful to Julie Fedorniak, Sue Reilly, Ross Barrett, Rob Gegg and Janice Cran at JCU for helping me prepare for field trips with a huge amount of gear, vehicles and OHS logistics. Felicity Smout, Sarah Gill, Maria Bellei and the Aachilpa Vet Clinic provided veterinary advice and training which I greatly appreciate. Thanks to the Centre for Tropical Biodiversity and Climate Change (CTBCC), in particular Yvette Williams, for welcoming me to ATSIP when the Mammal lab dwindled to one, and for inviting me to lab retreats and meetings. I am also grateful to the JCU Herpetology lab for support and feedback during my time in Townsville, and to the members of the Deakin University Burwood ecology lab, which became my new uni base when I moved to Melbourne. I asked a lot of analysis questions during my candidature and am very grateful for the support and guidance I received. Thanks to Dale Nimmo, Shane Blowes, Justin Perry, Brett Abbott, Jeremy VanDerWal, Ross Alford, James Moloney, Sally Keith, Sean Connolly, Ben Phillips, Matt Symonds and Barbara Anderson. I'm also very grateful to Kerrie Mengersen, Jessie Wells and Tom Prowse for taking the time to help me with analyses that didn't make the final cut. I thank Simon Benhamou, Roger Powell, Hawthorne Beyer, Clément Calenge, Kamil Bartoń, Bram van Moorter and Eric Vander Wal for patient, informative responses to questions asked over email regarding R scripts and analysis techniques. I received helpful advice on dingoes and feral cats when I was first designing my research from Lee Allen, Brad Purcell, Malcolm Kennedy, Alex Diment, Mark Goullet, Dave Algar, Damian Byrne, Mike Johnston, Russell Palmer and James Speed. Thanks to Alex Kutt, who provided the opportunity to analyse and publish dingo dietary analysis, which provided a primer for my own diet studies. I really appreciate the many discussions and support from fellow students, particularly Brooke Bateman, April Reside, Shane Blowes, Takahiro Shimada, Shannon Troy (UTAS), Stewart MacDonald, Tammy Inkster, Collin Storlie, John Llewellyn, Mat Vickers, Anita

Heim and Rickard Abom. Thanks to the friends and housemates, especially Véronique Mocellin, who made Townsville home and a great place to live. I am extremely grateful to my parents Geoff and Theresa Brook, who instilled in me a love and respect for nature and animals in particular. You have always provided unconditional support and encouragement to pursue the things in life that I love and believe worthwhile. A huge thank you to my wonderful friend Morag Stewart, who provided so much wisdom, laughter and support in long-distance phone calls throughout my candidature. Sharing the PhD journey with you has been an emotional life-saver! The greatest thanks go to Sam Loy, who moved to Townsville, gave me love and support in many ways through the ups and downs, and inadvertently learnt a great deal about dingoes, cats and the challenges of fieldwork and analysis. Thank you for tolerating my science nerdiness and keeping me 'sane'. I couldn't have done it without you.

vii

viii

Statement on contributions of others

Research Funding

Australian Research Council Discovery grant DP0877398: New thinking on the relationship of dingo ecology to biodiversity conservation and sustainable cattle production Australian Research Council Linkage grant LP100100033: Mammal declines in northern Australia: science for conservation and recovery James Cook University (JCU)

In-kind support

Australian Wildlife Conservancy (AWC) School of Life and Environmental Sciences, Deakin University

Stipend

Australian Postgraduate Award JCU Graduate Research Scheme PhD write-up grant

Supervision (and editorial assistance)

Professor Christopher N Johnson, University of Tasmania (UTAS) Dr Euan G Ritchie, Deakin University Professor Lin Schwarzkopf, JCU

Statistical and analytical support

Dr Jeremy VanDerWal, JCU (Chapter 2-3) Dr Brooke Bateman, University of Wisconsin - Madison (Chapter 3) Dr Dale Nimmo, Deakin University (Chapter 3) Shane Blowes, JCU (Chapter 3) Professor Ross Alford, JCU (Chapter 5)

Data collection and processing

Dr Euan G Ritchie (Chapter 2, App A) Angus McNab (research assistant) (Chapter 2, App A) Laura Mitchell (research assistant) (Chapter 2-3, App A) Georgeanna Storey, Scatsabout (dietary analysis) (App A) Brooke Bateman (diet sample collection) (App A) Brian MacElvaine (diet sample collection) (App A) Hugh McGregor, AWC (Chapters 4-5)

ix Field research assistance

Laura Mitchell Angus McNab Damian Morrant Katrin Schmidt Jen Leech Delphi Ward Karen Glynn Devi Hermsen Paul Mitrovski Susan Campbell Frank Bird Michael Trombini Teigan Allen Emily Hynes Alex Ainscough Martin Aisthorpe Dimitris Bastian Jess Beaulieu Shane Blowes Rithika Fernandes Bianca Grover Rachel Heckathorn Sydney Jones Nina McLean Véronique Mocellin Alex Young Sam Loy Kate Stookey Jyoti Blencowe Hugh MacGregor AWC staff and volunteers

Ethics statement

The research presented and reported in this thesis was conducted in compliance with the National Health and Medical Research Council (NHMRC) Australian Code of Practice for the Care and Use of Animals for Scientific Purposes, 7th Edition, 2004 and the Queensland Animal Care and Protection Act 2001. Camera surveys were permitted by the James Cook University Animal Ethics Committee (JCU AEC) (no. A1423) and Department of Environment and Resource Management (Scientific Purposes Permit: WITK06493009). Animal trapping and handling methods were approved by the JCU AEC (no. A1613) and/or the Western Australian Department of Environment and Conservation (DEC) AEC (no. 2010/35 and 2011/03) and was permitted under the Wildlife Conservation Act 1950 (no. SF007564).

Publications and presentations

Peer-reviewed publications

Brook, L.A., Johnson, C.N. & Ritchie, E.G. (2012) Effects of predator control on behaviour of an apex predator and indirect consequences for mesopredator suppression. Journal of Applied Ecology 49:1278-1286.

Manuscripts in preparation

Brook, L. A., Johnson, C. N., Schwarzkopf, L. Nimmo, D. G. & Ritchie, E. G. (in prep) Hiding in plain sight: Do feral cats use complex habitats in north Queensland to avoid encounters with dingoes? To be submitted to Austral Ecology. Brook, L. A., Johnson, C. N., McGregor, H., Schwarzkopf, L., Ritchie, E. G. & Legge, S. (in prep) Seasonal space use of dingoes and feral cats: differences in patterns of intensity in northern Australian savanna. To be submitted to Journal of Mammalogy. Brook, L.A., Johnson, C.N., McGregor, H., Schwarzkopf, L., Legge, S. & Ritchie, E.G. (in prep) Spatial interactions between sympatric dingoes and feral cats in northern Australia: patterns of intraguild avoidance in a landscape of fear. To be submitted to Journal of Animal Ecology.

Conference presentations

Biology in the Tropics Postgraduate Research Student Conference (2012), James Cook University, Townsville, QLD, Australia (Spoken paper) Brook, L.A., Johnson, C.N., Ritchie, E.G. and Schwarzkopf, L. (2012) When the dingo’s away does the feral cat play? Indirect effects of predator control on predator behaviour and suppression.

Ecological Society of Australia Annual Conference (2011) Hobart, TAS, Australia (Spoken paper) Brook, L. A., Johnson, C. N., Ritchie, E. G., Schwarzkopf, L. (2011) The effects of dingo control on feral cats and their behaviour.

Australian Mammal Society Annual Conference (2010) Canberra, ACT, Australia (Poster) Brook, L. A., Johnson, C. N., Ritchie, E. G., Schwarzkopf, L. (2010) Do feral cats avoid interactions with an apex predator, the dingo in space or time?

xi Other publications during candidature

Brook, L. A., Kutt, A. S. (2011) The diet of the dingo (Canis lupus dingo) in north-eastern Australia with comments on its conservation implications. The Rangeland Journal 33(1): 79-85.

xii

Thesis Abstract

Predation can be an important force in ecosystem dynamics, imposing top-down effects on lower trophic guilds. Dominant apex predators can suppress smaller mesopredator populations, both directly via aggressive, sometimes lethal interactions, and indirectly through resource competition and/or changes to behaviour, as mesopredators attempt to reduce the risk of interactions. In the absence of apex predators, behavioural restrictions on mesopredators are lifted. This process of 'mesopredator release' can lead to increased mesopredator abundance and improved access to prey, and may culminate in stronger flow-on effects of mesopredators on prey. Understanding the mechanisms of intraguild interactions will clarify the role of apex predators in regulating trophic interactions, through the suppression of smaller predators.

In the tropical savannas of northern Australia, the introduced feral cat Felis catus co- occurs with the dingo Canis lupus dingo, a mammalian apex predator. Feral cats prey on native wildlife and are implicated in the recent declines of small mammals in northern Australia.

Further, feral cats are extremely difficult to control with current technologies. It has been suggested that dingoes could suppress feral cat populations in northern Australia, given recent evidence of negative impacts of dingoes on red foxes Vulpes vulpes, the interspecific killing of foxes and cats by dingoes, and international studies demonstrating mesopredator suppression.

In this thesis, I examine the ecological interactions between dingoes and feral cats in northern Australia. I investigate a) whether dingoes can numerically suppress feral cat populations; and b) how dingoes influence the behaviour of cats. I used two contrasting methods to address these aims. Firstly, I examined broad-scale trends in predator abundance and behaviour with remote camera surveys on multiple, paired rangeland properties, where one property practiced dingo control, and the other left dingoes alone. Secondly, I collected high- frequency telemetry data from sympatric dingoes and feral cats in the Kimberley in northwestern Australia, to investigate fine-scale patterns of space use and potential effects of dingoes on cat behaviour.

xiii Lethal control can affect both the abundance and behaviour of apex predators, which may influence top-down effects on mesopredators. I assessed how predator control affects abundance indices and activity patterns of dingoes and feral cats in nine areas across central and northern Australia, spanning more than 2,500 km. Predator control generally reduced dingo abundance indices. In areas with greater contrasts in dingo activity between paired sites, feral cat indices were higher on sites with lower dingo activity, indicating an inverse correlation between species. Feral cat visitation rates were lower at cameras used frequently by dingoes, with no cats recorded where dingo rates exceeded 1 night -1. This suggests that dingoes may potentially exclude feral cats at a local scale. Dingoes in areas without predator control were broadly crepuscular, similar to their major prey (kangaroos), but controlled populations were less active in the early evening. Reduced evening activity of dingoes in areas with predator control was associated with increased cat activity in the early evening, which is a time when many small native prey (e.g. rodents and geckos) emerge to feed. These results suggest that cats adjust their behaviour to avoid times or places with higher dingo activity.

I explored whether feral cats select habitat features to reduce the risk of interactions with dingoes in two regions of north Queensland, using generalised linear mixed models and remote camera data. In woodland habitats of Cape York Peninsula, both predators used roads, potentially for ease of movement. Where dingo activity was high, feral cats were recorded more frequently in open areas with less ground cover. Open habitats may enable feral cats to detect dingoes earlier, allowing them to reach nearby trees upon encounter. Dingo activity was low on the Gulf Plains, and cats were recorded more frequently in areas with relatively few refuges

(tree cover). Treeless habitats would be expected to increase risk associated with dingo encounters, but the low dingo risk in this region could be insufficient to counter benefits of using open areas. Habitats with sufficient escape routes, such as woodlands, could reduce the risk of interactions and facilitate predator co-existence.

Spatial and temporal characteristics of space use may influence predator-prey interactions. I estimated predator space use and core area intensity in the Kimberley region of

Western Australia. Dingoes used larger areas than cats, but focussed activity on relatively small

xiv

core areas. Cats used space more evenly, spreading activity over core areas that formed a larger proportion of their total range, than in the case of dingoes. Dingoes moved more over time, although both species exhibited low temporal overlap in high-use areas. Predation pressure on small prey may therefore be more continuous within feral cat core areas, while the contrast in dingo activity could create areas of both low and high risk for large herbivores, mesopredators and small prey.

Mesopredators should minimise the risk of intraguild interference when making space use decisions, in addition to maximising access to prey. I examined spatial interactions between sympatric dingoes and feral cats using telemetry data. Overlap between dingo and cat core areas was much lower than for overall space use. Use of shared space between the two predators was variable, ranging from avoidance to attraction. The number of occasions on which dingoes and feral cats were recorded within 1 km of each other was lower than expected, suggesting that cats either avoided close encounters with dingoes or responded to such encounters by moving away.

While dingoes and feral cats are sympatric, cats appear to minimise intraguild interactions by either limiting spatial overlap or maintaining a safe distance from dingoes.

In summary, feral cats may be excluded from areas used heavily by dingoes in northern

Australian savannas, potentially providing refuge for prey from cat predation. However, as dingoes are mobile, these interactions are likely to be spatially and temporally dynamic. The strength of top-down control may be context-dependent, influenced by factors such as predator control, which alters dingo behaviour, and habitat complexity and resource availability, which can affect the rate of encounter and degree of aggressive interactions. This study demonstrates that behavioural shifts may be a critical component of suppression of mesopredators by apex predators, as feral cats appear to alter behaviour to avoid dingoes. Behavioural effects on mesopredators should be considered when assessing the potential benefits of dingoes for lower trophic guilds.

xvi

Table of Contents

Acknowledgments ...... v

Statement on contributions of others ...... ix

Publications and presentations ...... xi

Thesis Abstract...... xiii

Table of Contents ...... xvii

List of Tables ...... xxi

List of Figures ...... xxv

List of Appendices...... xxxiii

Chapter 1 General Introduction...... 1

The trophic role of the apex predator ...... 1 Eliminating the competition: effects on mesopredators ...... 2 Factors influencing top-down effects ...... 5 Contention in trophic cascades ...... 6 Predator interactions in Australia...... 7 The feral cat...... 8 The dingo ...... 10 Potential benefits of trophic regulation in northern Australia ...... 16 Thesis aims and objectives ...... 18 Aim 1: Determine whether dingoes can numerically suppress feral cat populations...... 18 Aim 2: Further understanding of predator ecology ...... 19 Aim 3: Explore how dingoes potentially influence the behaviour of feral cats ...... 19 Thesis outline and structure...... 20

Chapter 2 Effects of predator control on behaviour of an apex predator and indirect consequences for mesopredator suppression ...... 23

Abstract ...... 23 Introduction ...... 24 Methods ...... 26 Study area and data collection ...... 26

xvii Analysis...... 28 Results ...... 30 Effect of control on indices of dingo abundance ...... 30 Relationships between indices of dingo and feral cat abundance ...... 31 Dingo behaviour...... 32 Feral cat behaviour ...... 33 Prey activity patterns ...... 35 Discussion ...... 36 Predator control and predator numbers...... 36 Avoidance in time ...... 37 Avoidance in space...... 38 How could behavioural shifts affect prey species? ...... 39

Chapter 3 Hiding in plain sight: Do feral cats use complex habitats in north Queensland to avoid encounters with dingoes? ...... 43

Abstract ...... 43 Introduction...... 44 Methods ...... 47 Study areas ...... 47 Vegetation ...... 48 Survey methods...... 50 Generalised Linear Mixed Modelling (GLMM) ...... 51 Results ...... 54 Cape York ...... 54 Gulf Plain ...... 58 Discussion ...... 60 Dingoes on Cape York Peninsula ...... 61 Feral cats on Cape York Peninsula...... 62 Feral cats on the Gulf Plains...... 63 Caveats and future research directions ...... 64 Are cats using habitat features to avoid dingoes? ...... 66 Conclusion...... 67

Chapter 4 Seasonal space use of dingoes and feral cats: differences in patterns of intensity in northern Australian savanna ...... 69

Abstract ...... 69 Introduction...... 70 Methods ...... 73

xviii

Study area...... 73 Trapping and processing ...... 75 Data screening ...... 77 Autocorrelation ...... 78 Analysis of space use ...... 79 Results ...... 85 Autocorrelation ...... 86 Kernel density estimation ...... 86 Movement kernel density estimation ...... 90 Intensity and recursion distributions ...... 90 Core intensity indices ...... 90 Activity patterns within the core area...... 92 Site fidelity ...... 92 Discussion...... 94

Chapter 5 Spatial interactions between sympatric dingoes and feral cats in the Kimberley: patterns of intraguild avoidance in a landscape of fear ...... 101

Abstract ...... 101 Introduction ...... 102 Methods ...... 106 Static overlap indices...... 107 Distance between animals ...... 108 Dynamic overlap indices ...... 109 Results ...... 112 Were there differences in static spatial overlap between and among predators? ...... 112 Were predators found closer to or further from each other than expected? ...... 115 Did predators use shared space differently?...... 118 Discussion...... 122 Dingo-cat interactions ...... 122 Dingo-dingo interactions ...... 124 Cat-cat interactions...... 125 Caveats and future research directions ...... 126 Implications for biodiversity conservation ...... 128 Conclusion ...... 128

Chapter 6 General discussion...... 131

Summary of major findings...... 132 Aim 1: Determine whether dingoes can numerically suppress feral cat populations...... 132

xix Aim 2: Further understanding of predator ecology ...... 133 Aim 3: Explore how dingoes potentially influence the behaviour of feral cats...... 134 Can dingoes effectively suppress feral cats in northern Australia? ...... 136 Behavioural games between clever predators? ...... 137 Habitat complexity ...... 137 Strength of competition...... 138 Predator control...... 138 Dynamic trophic interactions...... 139 Future research directions...... 139 Integrate prey and habitat information...... 139 Investigate prevalence of interspecific killing ...... 140 Experimental studies...... 140 Incorporate resource fluctuations ...... 141 Concluding remarks ...... 141

References...... 143

List of Tables

Table 3.1. Habitat complexity variables measured at each camera station...... 52

Table 3.2. Explanatory variables and interactions included in generalised linear mixed models in each area to address each of the research predictions. Some variables differed between areas due to differences in collinearity or better model fit. Habitat variables were also included as single factors to compare the importance of the interaction to the habitat variables alone...... 53

Table 3.3. Best supported models ( < 2) of predator activity on Cape York Peninsula (CYP) and Gulf Plains (GP), selected using the second-order Akaike Information Criterion (AICc). .. 55

Table 3.4. Relative importance of variables explaining predator activity on Cape York Peninsula

(CYP) and Gulf Plains (GP), drawn from model averaging. Variables with > 0.5 are considered to be more important (Ritchie et al. 2009)...... 58

Table 3.5. Number of nights on which predators were recorded, over seasons and treatments. Seasonal differences are compared using binomial tests, against a null hypothesis that the response is equal between seasons...... 59

Table 4.1. Fix rates for dingo Canis lupus dingo and feral cat Felis catus data (mean ± se) following removal of zeros and errors...... 85

Table 4.2. Kernel density estimates for dingoes Canis lupus dingo and cats Felis catus, including estimates of core area and intensity of use. Range is provided in parentheses. Appendix D.2 presents estimates for all individuals, including those that did not reach asymptote...... 86

Table 4.3. Movement kernel density estimates for the utilisation (UD), intensity (ID) and recursion (RD) distributions of dingoes Canis lupus dingo and feral cats Felis catus. The range is within parentheses...... 89

Table 4.4. Estimates of spatial overlap between short-term utilisation distributions, using the Bhattacharyya Index (BA), at different isopleths, for dingoes Canis lupus dingo and feral cats Felis catus. Pooled mean and standard error are presented...... 92

Table 5.1. Fix rates and specifications for GPS data used in this chapter. Fix rates are for full datasets prior to subsampling. Pooled mean percentages are shown in parentheses. One dingo and one feral cat were caught in both years but are treated as separate animals in these analyses...... 112

Table 5.2. Distance analysis results for each dyad type. The observed value is averaged across dyads, and the null mean is averaged first across each dyad in each replicate, and then across

xxi replicates. The P-value is for the pooled test, while the number of significant dyads indicates the number (and % in parentheses) of individual dyads for which the observed metric was significantly different from the expected value. Italicised P-values should be treated with caution, as the null distribution was not considered symmetrical using the MGG test (Miao et al. 2006)...... 117

Table 5.3. Number of significant spatial effects in relation to the overlap zone in dynamic overlap analysis of dingoes and feral cats. Likelihood coefficient values are presented below in parentheses (one value for single samples or mean ± se). For D-C dyads, dingo coefficient values are in plain text and cat coefficient values are italicised...... 119

Table 5.4. Summary of significant non-random temporal interactions in relation to the overlap zone in dynamic overlap analysis of dingoes and feral cats. Odds ratios considerably greater than 1 (>1.2, in bold) indicate a larger than expected frequency, while those smaller than 1 indicate lower than expected frequency...... 120

Table A.1. Summary of dietary items recorded in dingo scats in each area and in the total sample...... 185

Table A.2. Summary of dietary items recorded in feral cat scats in the total sample...... 190

Table C.1. Filtering rules applied to the test data...... 198

Table D.1. Summary of data collection and screening outcomes for all collared individuals. . 205

Table D.2. Utilisation distribution estimates (KDE) for all collared individuals...... 206

Table D.3. Utilisation distribution estimates (MKDE) for all collared individuals...... 207

Table D.4. Intensity and recursion distribution estimates (MKDE) for all collared individuals...... 208

Table D.5. Minimum Convex Polygon (MCP) estimates for all collared individuals using different proportions of the data...... 209

Table E.1. Summary of animals for who the density estimates for each distribution type reached sufficient asymptote...... 211

Table G.1. Mean (± se) overlap indices calculated using UDOI (plain) and PHR (italics), for different dyad types and distribution contours. Parentheses contain the number of dyads with positive overlap, although all dyads that overlapped at the 95% contour are included...... 253

Table G.2. Wilcoxon signed-rank test statistics and exact P-values for comparisons between overlap at different distributions for each dyad type, estimated using UDOI. No comparisons were made between distribution types for cat-cat dyads, as there was no significant difference

xxii

between groups (Asymptotic Friedman test: = 4.2, P = 0.24). Data sets that could not be log- transformed to ensure the distribution of differences was symmetrical are marked NA...... 254

Table G.3. Wilcoxon signed-rank test statistics and exact P-values for comparisons between overlap at different distributions for each dyad type, estimated using PHR. Data sets that could not be log-transformed to ensure the distribution of differences was symmetrical are marked NA...... 255

Table G.4. Distance analysis results for dingo-dingo and cat-cat dyads. P-values are from randomisation tests comparing the observed metric with the null distribution. P-values in italics should be treated with caution, as the null distribution was not considered symmetrical by the MGG test (Miao et al. 2006)...... 256

Table G.5. Distance analysis results for dingo-cat dyads. P-values are from randomisation tests comparing the observed metric with the null distribution. P-values in italics should be treated with caution, as the null distribution was not considered symmetrical by the MGG test (Miao et al. 2006)...... 258

Table G.6. Extended results of dynamic overlap analysis...... 260

Table H.1. Percentage of dyads for each dyad type that had less than 10% overlap for MCP95 and MCP100, estimated using HR...... 264

xxiii

xxiv

List of Figures

Figure 1.1. Proposed interactions between dingoes, feral cats and prey in northern Australian savanna ecosystems. Dingoes suppress populations of large prey, but impacts on small prey may be alleviated indirectly, due to suppression of feral cat predation and reduced removal of vegetation by larger prey such as macropods. Photo of stripe-faced dunnart Sminthopsis macroura © Greg Calvert...... 15

Figure 1.2. Thesis structure...... 21

Figure 2.1. Paired survey areas across Australia. Each pair consists of one site that controls dingoes, and one without dingo control...... 28

Figure 2.2. Abundance indices of dingoes and feral cats derived from camera records in each paired survey area. Black and light grey bars represent dingoes in sites without and with predator control, and dark grey and white bars represent feral cats in sites without and with predator control, respectively...... 31

Figure 2.3. Contrast in predator abundance indices (AI) in paired survey areas differing in dingo control (N = 9). Positive values represent higher abundance indices in the site without dingo control, and negative values represent higher abundance indices in site with control. Hence the bottom right quadrant represents survey areas with more dingoes in the site without control and more cats in the site with control...... 32

Figure 2.4. Predator trap rates at individual camera stations (N = 279 stations with at least one predator recorded). Regression lines, slope coefficients and significance values for quantiles are shown (50th quantile: dashed; 75th: gray; 95th: dot-dashed; 99th: solid)...... 33

Figure 2.5. Proportion of activity records a) with and b) without dingo control. Time has been scaled to a circular distribution with equal distance between sunset and sunrise. In 5a and b, black bars represent dingoes with (N = 85) and without (N = 249) control, and light grey bars represent cats with (N = 81) and without (N = 129) control...... 34

Figure 2.6. Proportion of activity records in the dusk/early evening (one hour pre-sunset to three hours post-sunset) (N = 13 sites with at least one dingo and one cat record). Data are square-root transformed...... 35

Figure 2.7. Proportion of a) dingo (dark grey) and large macropod (light grey) activity records in areas without predator control and b) feral cat (white) and small mammal (black) activity records in areas with predator control...... 36

Figure 3.1. Location of study sites in the Gulf Plains (fawn) and Cape York Peninsula (olive) bioregions in north Queensland, Australia...... 48

xxv Figure 3.2. Examples of habitat variability on the Gulf Plain (a, c, e, g) and Cape York Peninsula (b, d, f, h) ranging from complex ground cover and tree cover (a, b), tree cover with low ground cover (c, d), complex ground cover and minimal tree cover (e, f) and low ground cover and minimal tree cover (g, h) ...... 49

Figure 3.3. Model averaged parameter estimates (mean ± CI) influencing a) dingo and b) feral cat activity on Cape York Peninsula. Bar colour represents relative importance, from > 0.75 (dark grey), between 0.75 and 0.5 (medium grey), between 0.5 and 0.25 (light grey), to < 0.25 (white). UB: unbaited property not using predator control...... 57

Figure 3.4. Model averaged parameter estimates (mean ± CI) influencing feral cat activity on the Gulf Plains. Bar colour represents relative importance, from between 0.5 and 0.25 (light grey), to < 0.25 (white). UB: unbaited property not using predator control...... 60

Figure 4.1. The central Kimberley bioregion in north-western Australia (top left), showing the boundaries of the Mornington and Marion Downs Wildlife Sanctuaries and major vegetation groups (National Vegetation Information System) (bottom left). Trapping zones are highlighted in blue within each study area...... 74

Figure 4.2. Core delineation method showing scaled isopleth volume against proportion of home range area. The core is delineated (in red) where the slope of the curve = 1, or where the distance between the curve and line of random use is maximal. The solid black arrow indicates the maximum distance using the method adapted from Vander Wal & Rodgers (2012), which identifies the greatest difference in % area...... 82

Figure 4.3. Core area as a proportion of the space use area (95% contour), for two core delineation methods (maximum probability method of Seaman & Powell 1990) and isopleth volume method used here) and four distribution metrics. All Spearman's rank rho coefficients were significant (P < 0.005)...... 83

Figure 4.4. Kernel density estimates of 95% (hollow) and core (filled) isopleths for a) dingoes Canis lupus dingo and b) feral cats Felis catus at Mornington and c) dingoes at the Siddins Valley (Marion Downs) site. The 95% and core isopleths and trajectory of transient dingo DS_M3 are not shown. Property boundaries, study areas and the Gibb River Rd are provided inset. See Appendix F for individual figures...... 87

Figure 4.5. Kernel density estimates of 95% (hollow) and core (filled) isopleths for a) dingoes Canis lupus dingo and b) feral cats Felis catus at Mornington in 2012. Property boundaries, study areas and the Gibb River Rd are provided inset. See Appendix F for individual figures.. 88

Figure 4.6. Indices of relative intensity of core area use (% core isopleth/core area as % of 95% isopleth area) for dingoes Canis lupus dingo (grey) and feral cats Felis catus (white) using different distribution metrics (mean ± se)...... 91

xxvi

Figure 4.7. The mean proportion of feral cat Felis catus and dingo Canis lupus dingo fixes within animal-specific core areas estimated using MKDE. Blue panels cover the range of sunrise and sunset times during the study periods...... 91

Figure 4.8. Examples of temporal overlap between core isopleths of the KDE utilisation distribution for each two-week period for dingoes Canis lupus dingo a) DM_F5 and b) DS_M4 and feral cats Felis catus: c) CM_M2 and d) CM_M3...... 93

Figure 4.9. Ninety-five percent isopleths of utilisation distribution estimates for dingo Canis lupus dingo DM_M1 (2011: purple, 2012: blue) and cat Felis catus CM_M4 (2011: orange, 2012: red) with core areas shaded, showing overlap between years...... 94

Figure 4.10. Movements and space use of dingo Canis lupus dingo DS_F1: a) High-frequency trajectory over 1-s slope layer with hillshading, showing avoidance of steep slopes; b) MKDE utilisation distribution, with 95% and core isopleths; c) MKDE intensity distribution with core isopleth and d) MKDE recursion distribution with core isopleth...... 95

Figure 5.1. The study site, Mornington Wildlife Sanctuary, located within the central Kimberley bioregion (light green) in north-western Australia (inset). Dingoes and feral cats were collared near Mornington’s Wildlifelink Research Station, in the area shaded blue, north of the Fitzroy River...... 106

Figure 5.2. Contingency table of probabilities and number of simultaneous fixes for each possible space use outcome (adapted from Minta 1992). The number of simultaneous sightings of species and in the overlap zone (OZ) is represented by . Similarly, there were occasions when neither individual were in the OZ. The probability of both animals present in the OZ is , while the probability of species being in the OZ independent of is . ....110

Figure 5.3. Mean (± se) overlap values for a) UDOI and b) PHR indices for different dyads and distribution contours. Dingo-cat columns refer to the probability of a dingo being located within a feral cat contour, while cat-dingo columns represent the probability of a feral cat being located within a dingo contour...... 113

Figure 5.4. Feral cat locations collected at 15-min intervals over both years. CM_M1-5 were collared in 2011. CM_M4, CM_M6-8 and CM_F1 were collared in 2012...... 114

Figure 5.5. Percent frequency distributions for distances between pooled synchronous fixes and pooled asynchronous fixes from the randomised null distribution for a) dingo-cat dyads, to a maximum of 2,000 m, b) dingo-dingo dyads to a maximum of 5,000 m and c) cat-cat dyads to a maximum of 2,000 m...... 116

Figure 5.6. Examples of overlap between KDEs for a) and b) two pairs of female dingoes, c) a female dingo and male cat and d) a male dingo and male cat. Animals in a) and c) have significant symmetrical and simultaneous attraction to the OZ. Animals in d) both avoid the OZ.

xxvii Animals in b) were not analysed for dynamic overlap, and appear to represent dingoes in adjacent packs...... 121

Figure A.1. Examples of habitat from a) Cape York Peninsula, b) Einasleigh Uplands, c) Gulf Plain and d) the Kimberley...... 183

Figure A.2. Percent frequency of occurrence of different prey categories for each survey area and total sample...... 188

Figure A.3. Volume of different prey categories in the dietary samples for each area and the total sample...... 189

Figure C.1.Distribution of all errors (a, c, e, g) and errors < 1 km (b, d, f, h) dependent on fix type (a, b), dilution of precision (c, d), number of satellites used (e, f) and satellite sensor value (g, h)...... 199

Figure C.2. Relationship between the difference in altitude reading and a) all errors, b) errors < 1 km and c) errors < 100 m. Fixes in c) are also restricted to those with < 200 m difference in altitude reading, and show the cut-off used in screening for altitude reading, and show the cut- off used in screening for altitude difference at 100 m (red dashed line). The trade-off between mean error (thick line) and data loss (thin line) with altitude value cut-off is shown in 2d. .... 200

Figure C.3. The mean (weighted) and standard deviation of error in fixes remaining once different screening methods were applied. The mean number of errors indicates levels of data loss. Filtering rule abbreviations are given in full in Table C.1...... 202

Figure C.4. The mean (weighted) and standard deviation of error in fixes following screening methods that removed fixes with differences in altitude ≥ 100 m...... 203

Figure E.1. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug- in method) and b) MKDE utilisation distributions for DS_M1, a male dingo collared in 2011 in Siddins Valley. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line)...... 212

Figure E.2. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug- in method) and b) MKDE utilisation distributions for DS_M2, a male dingo collared in 2011 in Siddins Valley. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line)...... 213

Figure E.3. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug- in method) and b) MKDE utilisation distributions for DS_M3, a male dingo collared in 2011 in Siddins Valley. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line)...... 214

xxviii

Figure E.4. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug- in method) and b) MKDE utilisation distributions for DS_M4, a male dingo collared in 2011 in Siddins Valley. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line)...... 215

Figure E.5. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug- in method) and b) MKDE utilisation distributions for DS_F1, a female dingo collared in 2011 in Siddins Valley. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line)...... 216

Figure E.6. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug- in method) and b) MKDE utilisation distributions for DS_F2, a female dingo collared in 2011 in Siddins Valley. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line)...... 217

Figure E.7. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug- in method) and b) MKDE utilisation distributions using data from 2011 for DM_M1, a male dingo collared in both years at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line)...... 218

Figure E.8. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug- in method) and b) MKDE utilisation distributions using data from 2012 for DM_M1, a male dingo collared in both years at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line)...... 219

Figure E.9. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug- in method) and b) MKDE utilisation distributions using data from both years for DM_M1, a male dingo collared in both years at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line)...... 220

Figure E.10. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug- in method) and b) MKDE utilisation distributions for DM_M2, a male dingo collared in 2011 at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line)...... 221

Figure E.11. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug- in method) and b) MKDE utilisation distributions for DM_F1, a female dingo collared in 2011 at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line)...... 222

Figure E.12. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug- in method) and b) MKDE utilisation distributions for DM_F2, a female dingo collared in 2011 at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line)...... 223

xxix Figure E.13. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug- in method) and b) MKDE utilisation distributions for DM_F3, a female dingo collared in 2012 at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line)...... 224

Figure E.14. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug- in method) and b) MKDE utilisation distributions for DM_F4, a female dingo collared in 2012 at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line)...... 225

Figure E.15. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug- in method) and b) MKDE utilisation distributions for DM_F5, a female dingo collared in 2012 at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line)...... 226

Figure E.16. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug- in method) and b) MKDE utilisation distributions for DM_F6, a female dingo collared in 2012 at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line)...... 227

Figure E.17. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug- in method) and b) MKDE utilisation distributions for CM_M1, a male cat collared in 2011 at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line)...... 228

Figure E.18. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug- in method) and b) MKDE utilisation distributions for CM_M2, a male cat collared in 2011 at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line)...... 229

Figure E.19. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug- in method) and b) MKDE utilisation distributions for CM_M3, a male cat collared in 2011 at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line)...... 230

Figure E.20. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug- in method) and b) MKDE utilisation distributions using data from 2011 for CM_M4, a male cat collared in both years at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line)...... 231

Figure E.21. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug- in method) and b) MKDE utilisation distributions using data from 2012 for CM_M4, a male cat collared in both years at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line)...... 232

xxx

Figure E.22. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug- in method) and b) MKDE utilisation distributions using data from both years for CM_M4, a male cat collared in both years at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line)...... 233

Figure E.23. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug- in method) and b) MKDE utilisation distributions for CM_M6, a male cat collared in 2012 at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line)...... 234

Figure E.24. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug- in method) and b) MKDE utilisation distributions for CM_M7, a male cat collared in 2012 at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line)...... 235

Figure E.25. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug- in method) and b) MKDE utilisation distributions for CM_M8, a male cat collared in 2012 at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line)...... 236

Figure E.26. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug- in method) and b) MKDE utilisation distributions for CM_F1, a female cat collared in 2012 at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line)...... 237

Figure F.1. Ninety-five percent and core isopleths for the a) KDE and b) MKDE utilisation distributions and c) fixes and trajectory of DS_M1, and the d) KDE and e) MKDE utilisation distributions and f) fixes and trajectory of DS_M2, both male dingoes collared in 2011 in Siddins Valley...... 239

Figure F.2. Ninety-five percent and core isopleths for the a) KDE and b) MKDE utilisation distributions and c) fixes and trajectory of DS_M3, a male dingo, and the d) KDE and e) MKDE utilisation distributions and f) fixes and trajectory of DS_F1, a female dingo, both collared in 2011 in Siddins Valley...... 240

Figure F.3. Ninety-five percent and core isopleths for the a) KDE and b) MKDE utilisation distributions and c) fixes and trajectory of DS_M4, a male dingo, and the d) KDE and e) MKDE utilisation distributions and f) fixes and trajectory of DS_F2, a female dingo, both collared in 2011 in Siddins Valley...... 241

Figure F.4. Ninety-five percent and core isopleths for the a) KDE and b) MKDE utilisation distributions and c) fixes and trajectory of DM_M1, and the d) KDE and e) MKDE utilisation distributions and f) fixes and trajectory of DM_M2, both male dingoes collared in 2011 at Mornington...... 242

xxxi Figure F.5. Ninety-five percent and core isopleths for the a) KDE and b) MKDE utilisation distributions and c) fixes and trajectory of DM_F1, and the d) KDE and e) MKDE utilisation distributions and f) fixes and trajectory of DM_F2, both male dingoes collared in 2011 at Mornington...... 243

Figure F.6. Ninety-five percent and core isopleths for the a) KDE and b) MKDE utilisation distributions and c) fixes and trajectory of DM_F3, and the d) KDE and e) MKDE utilisation distributions and f) fixes and trajectory of DM_F4, both female dingoes collared in 2012 at Mornington...... 244

Figure F.7. Ninety-five percent and core isopleths for the a) KDE and b) MKDE utilisation distributions and c) fixes and trajectory of DM_F5, and the d) KDE and e) MKDE utilisation distributions and f) fixes and trajectory of DM_F6, both female dingoes collared in 2012 at Mornington...... 245

Figure F.8. Ninety-five percent and core isopleths for the a) KDE and b) MKDE utilisation distributions and c) fixes and trajectory of CM_M1, and the d) KDE and e) MKDE utilisation distributions and f) fixes and trajectory of CM_M2, both male cats collared in 2011 at Mornington...... 246

Figure F.9. Ninety-five percent and core isopleths for the a) KDE and b) MKDE utilisation distributions and c) fixes and trajectory of CM_M3, a male cat collared in 2011, and the d) KDE and e) MKDE utilisation distributions and f) fixes and trajectory of CM_M4, a male cat collared in both years at Mornington...... 247

Figure F.10. Ninety-five percent and core isopleths for the a) KDE and b) MKDE utilisation distributions and c) fixes and trajectory of CM_M6, and the d) KDE and e) MKDE utilisation distributions and f) fixes and trajectory of CM_M7, both male cats collared in 2012 at Mornington...... 248

Figure F.11. Ninety-five percent and core isopleths for the a) KDE and b) MKDE utilisation distributions and c) fixes and trajectory of CM_M8, a male cat collared in 2012, and the d) KDE and e) MKDE utilisation distributions and f) fixes and trajectory of CM_F1, a female cat collared in 2012 at Mornington...... 249

Figure G.1. Boxplots comparing a) UDOI and b) PHR between dyad types at the 95% and core contours for the MKDE utilisation distribution...... 252

xxxii

List of Appendices

Appendix A: Notes on prey consumed by dingoes and feral cats in northern Australia ...... 179

Introduction ...... 179 Methods ...... 182 Results ...... 184 Discussion...... 190

Appendix B: Sun-time calibration of activity data ...... 193

Appendix C: The effect of screening methods on bias and data loss in GPS fixes ...... 195

Introduction ...... 195 Methods ...... 196 Results ...... 198 Discussion...... 203

Appendix D: Additional tables from Chapter 4 ...... 205

Appendix E: Summary of Incremental Area Analysis...... 211

Appendix F: Figures of density estimates for dingoes and feral cats ...... 239

Appendix G: Additional plots and tables from Chapter 5 ...... 251

Appendix H: Overlap in MCP as a measure of territoriality ...... 263

Methods ...... 263 Results ...... 264

xxxiii

xxxiv Chapter 1 General Introduction

The trophic role of the apex predator

Predators characteristically occupy the top spot in the food chain, but when several predators co-occur, aggressive competition for resources can result. Interactions between apex predators, the dominant, largest-bodied predators in an ecosystem, and smaller-bodied species known as mesopredators, may be just as strong as interactions between predators and prey

(Ritchie & Johnson 2009). Further, by restricting the abundance or behaviour of mesopredators

(Prugh et al. 2009; Berger 2010), intraguild interactions can have flow-on effects on lower trophic guilds (Estes et al. 2011).

Apex predators are often considered 'keystone' species (Paine 1969; Soulé et al. 2005), as they can impose ecosystem-wide effects via trophic cascades, by interacting both directly and indirectly with prey and other predators (Hebblewhite et al. 2005; Terborgh et al. 2010; Ripple

& Beschta 2012). Apex predators can enhance diversity, by limiting competitively superior species in lower trophic guilds (Sergio et al. 2007). Some predators may selectively take weak, diseased or old individuals (Moore 2002; Husseman et al. 2003), maintaining the health of prey populations (Packer et al. 2003). Remaining carcasses can support populations of scavengers, raptors, smaller predators and invertebrates (Melis et al. 2004; Selva et al. 2005; Wikenros et al.

2013), and facilitate nutrient recycling (Towne 2000).

Hairston, Smith and Slobodkin (HSS, 1960) proposed the 'Green World Hypothesis', that predators can structure ecosystems by limiting herbivores and ensuring vegetation communities are not depleted. Predators influence the population size and behaviour of large herbivore prey, both through direct killing and fear of predation, also known as nonconsumptive or risk effects (Lima 1998; Creel & Christianson 2008; Berger 2010). Risk effects that alter habitat use or foraging behaviour (Creel et al. 2005; Preisser & Bolnick 2008) can

1 reduce the fitness (Preisser et al. 2005) or reproductive output of prey (Ruxton & Lima 1997;

Creel et al. 2007; but see Middleton et al. 2013). Fear-driven risk effects may impose behaviourally-mediated trophic cascades (Schmitz et al. 1997), potentially influencing prey demography as significantly as direct predation (Preisser et al. 2005; Creel & Christianson

2008). Areas or habitats where prey are vulnerable carry a greater risk of predation compared to areas of refuge or low predator density. These 'hills' and 'valleys' of varying predation risk form a spatial and temporal 'landscape of fear' (Laundré et al. 2001). The decisions of prey are expected to be guided by fear associated with high-risk areas (Arias-Del Razo et al. 2012).

The ecological benefits of large predators are perhaps most evident once they disappear.

As human populations have expanded in recent centuries, increasing contact and conflict of people with predators, anthropogenic forces such as loss of habitat and prey, and direct persecution have reduced the diversity and distribution of large predators (Treves & Karanth

2003; Laliberte & Ripple 2004). Without the effects of predation and predation risk, herbivore populations can expand and increase demands on resources (Côté et al. 2004). This can lead to reduced plant growth or diversity, or changes to vegetation structure or communities (McLaren

& Peterson 1994; Estes & Duggins 1995; Schmitz et al. 2000; Soulé et al. 2003; Beschta &

Ripple 2009), prevention of forest regeneration (Ripple & Larsen 2000), and declines in other habitat-dependent taxa such as birds, amphibians, reptiles and invertebrates (Berger et al. 2001;

Ripple & Beschta 2006). The cascading damage following the loss of apex predators can be seen through multiple trophic guilds (Terborgh et al. 2001; Estes et al. 2011) and may potentially shift ecosystems to alternative stable states (Scheffer et al. 2001; Terborgh et al.

2010).

Eliminating the competition: effects on mesopredators

Intraguild competition can have negative impacts on subordinate predators. This competition can take several forms, two of which are discussed in this thesis:

 Exploitative competition occurs when predators compete for finite shared resources,

and is stronger when dietary overlap is high (Schoener 1983). Exploitative

competition may be indirect, whereby consumption by apex predators limits the

resources available for smaller predators (Petren & Case 1996), or direct, when

dominant predators steal kills of smaller predators, known as kleptoparasitism

(Gorman et al. 1998).

 Interference competition includes aggressive or antagonistic interactions between

sympatric predators (Lourenço et al. 2014). Interference competition can result in

death or interspecific killing (Palomares & Caro 1999), and occasionally

consumption of the losing individual, when it is known as intraguild predation

(Polis 1981). When intraguild predation is motivated by hunger, it can be termed

predatory aggression, while interference-driven mortality, frequently without

consumption, is also known as territorial aggression (Polis et al. 1989). Intraguild

predation can be a major source of mortality for smaller predators

(Moehrenschlager et al. 2007).

Mesopredators are usually proficient predators: they frequently occur at higher densities and focus predation more intensely and at a finer spatial scale, on a broader, more generalised range of species compared to apex predators (Prugh et al. 2009; Brashares et al. 2010), although apex predators also have access to larger-bodied prey (Radloff & du Toit 2004). Theory suggests that for predators to coexist, the smaller predator must be better at exploiting resources, to persist despite interference and predation from larger species (Polis & Holt 1992). Intraguild predation may actually facilitate coexistence of different-sized predators, as the increased consumption of resources by the smaller predator is limited by suppression from larger, less- competitive apex predators (Rosenzweig 1966). Where predators are similar in size however, the risk of injury from aggressive interactions increases (Donadio & Buskirk 2006), and interference may be less frequent due to mutual avoidance (e.g. wolves Canis lupus and lynx

Lynx lynx, Schmidt et al. 2009; Wikenros et al. 2010; although see Sunquist & Sunquist 2002).

Intraguild risk effects imposed by apex predators may be strong drivers of mesopredator behaviour (Ritchie & Johnson 2009). Unlike prey species, mesopredators have not necessarily evolved successful adaptations to avoid apex predators, hence their risk of capture, and the

3 'peaks' in their landscape of fear, may be particularly high (Ritchie & Johnson 2009). Further, mesopredators may face more consistent risk upon encountering an apex predator, as interference competition can occur independent of hunger (Polis et al. 1989). Mesopredators may adjust behaviour to avoid areas where larger predators are abundant or frequently active

(Berger & Gese 2007; Hamel et al. 2013), and minimise use of habitats favoured by apex predators (Linnell & Strand 2000; Wirsing & Heithaus 2009) or where vulnerability to predation is higher (Heithaus & Dill 2006; Sergio et al. 2007; Salo et al. 2008). If both species use the same areas, mesopredators may adjust their activity times (Kortello et al. 2007;

Harrington et al. 2009; Hayward & Slotow 2009) to reduce the probability of encountering a larger predator. Intraguild risk effects may reduce the fitness of mesopredators, for example, by reducing breeding success or population growth rate, leading to population declines (Linnell &

Strand 2000; Sergio & Hiraldo 2008; Ritchie & Johnson 2009).

The potential for aggressive interactions, and therefore the strength and form of intraguild risk effects, is complex and not completely understood. Apex predator density (Sergio et al. 2007), dietary overlap (Polis et al. 1989; Donadio & Buskirk 2006), prey availability

(Creel 2001; Wilson et al. 2010) and overall habitat complexity (Creel 2001; Janssen et al.

2007) can all mediate intraguild interactions. Furthermore, larger predators will sometimes display both aggressive and tolerant behaviours (Gese et al. 1996), suggesting intraguild interactions vary depending on ecological context.

The removal of apex predators can also relax competition with mesopredators, in a process known as 'mesopredator release' (Soulé et al. 1988), where mesopredator populations expand (Henke & Bryant 1999; Prugh et al. 2009) and individuals exhibit a broader range of behaviours (Ritchie & Johnson 2009). Given that mesopredators hunt more efficiently than larger competitors, mesopredator release can cause increased or more effective predation pressure on prey (Palomares et al. 1995), leading to declines in diversity or abundance of prey populations (Brashares et al. 2010). Crooks and Soulé (1999) measured bird diversity in remnant sage scrub patches in California, some of which were large enough to sustain coyotes

Canis latrans. In patches where coyotes were rare or absent, mesopredator release had occurred,

with increased abundances of native and introduced mesopredators such as feral cats Felis catus and a decline in bird diversity. Further, mesopredators adjusted their behaviour to avoid times or areas where coyotes were recorded. Thus, even the presence of a larger predator can limit the abundance and behaviour of mesopredators and provide benefits to prey communities.

Factors influencing top-down effects

Ecosystem diversity

If interactions between species are weak, the removal of an apex predator may have negligible effects on other trophic guilds (e.g. mesopredator release, Brashares et al. 2010).

Ecosystems or guilds with high species diversity may be resilient to the loss of one predator species, as complex interactions within the food web can absorb or compensate for the effects

(Polis & Strong 1996; Pace et al. 1999; Roemer et al. 2009; but see Borer et al. 2005).

Alternatively, habitat heterogeneity can offer protection and moderate the effects of predation

(Polis & Strong 1996). Hence, the loss or reintroduction of apex predators will not necessarily trigger community-wide trophic cascades (Sergio et al. 2008), although in some cases it may affect certain species within different guilds (Schmitz et al. 2000).

Hunting style

A predator's hunting mode may influence the strength with which risk effects can impinge on prey fitness (Middleton et al. 2013) and mediate indirect effects on lower trophic guilds (Schmitz 2008). Mobile, coursing predators such as canids use long chases and group- hunting strategies to exhaust prey, while ambush predators rely on stealth to surprise prey at short distances. Compared to cues from coursing predators, cues from ambush predators, such as scent, may correlate better with current predation risk, and be useful to lower guilds to reduce surprise and hence predation risk (Preisser et al. 2007). Mesopredators and prey may therefore have more to gain by adjusting behaviour in response to ambush predators rather than coursing predators (Thaker et al. 2011), whose hunting success depends more on predator group size and prey condition (e.g. Lycaon pictus, Fanshawe & Fitzgibbon 1993), and whose cues do not

5 reliably inform predation risk (Schmitz 2008). As a result, risk effects in response to cues from ambush predators are likely to be stronger, and lead to larger indirect effects on prey or vegetation (Schmitz 2008).

Bottom-up processes

Oksanen et al. (1981) presented the Exploitation Ecosystems Hypothesis, which proposed that ecosystems with low primary productivity are limited by bottom-up processes, while more resource-rich ecosystems can support more herbivores, which are in turn limited by predation. Elmhagen and Rushton (2007) demonstrated the link between mesopredator release and bottom-up processes, showing that density of red foxes Vulpes vulpes was highest following the removal of Eurasian lynx in more productive areas, but that the effect was damped in areas of low primary productivity. However other studies suggest that productivity is not a good indicator of cascade strength (Borer et al. 2005), as some plants defend against herbivory (Polis

& Strong 1996), while in other cases predation may have a stronger effect in resource-poor areas (Melis et al. 2009). Ultimately, the influence of productivity on the role of apex predators is complex, and likely to vary between ecosystems (Oksanen & Oksanen 2000) depending on other factors such as habitat modification (Ritchie et al. 2012; Norbury et al. 2013).

Contention in trophic cascades

Predator-entrained trophic cascades in terrestrial systems are not universally accepted in the literature. Some studies have examined the reliability of evidence (Mech 2012) while others question whether the simplicity of top-down effects provides sufficient explanation for the complexity of ecological communities (Polis & Strong 1996). The majority of studies demonstrating trophic cascades or mesopredator release have been correlative, and may be confounded by other effects such as land use (Brashares et al. 2010). Causal, experimental evidence is particularly valuable, but large-scale, long-term, controlled studies are difficult to implement. The reintroduction of wolves to Yellowstone National Park (YNP) provided an opportunity to observe potential effects of apex predator restoration; however this example also highlights the challenges of documenting top-down processes in terrestrial systems.

The reintroduction of wolves to YNP in 1995 led to effects such as reduced elk Cervus canadensis density or behaviour, increased tree recruitment (e.g. Ripple & Beschta 2004), and changes to behaviour and abundance of coyotes (Crabtree & Sheldon 1999). However, the return of apex predators alone may not restore an ecosystem to its previous condition if a state shift has occurred (Wolf et al. 2007; Scheffer 2010). In YNP, the hydrological changes engineered by beavers Castor canadensis may also be needed to revert to the previous ecosystem state (Marshall et al. 2013).

Others have questioned reports of behaviourally-mediated trophic cascades (Kauffman et al. 2010; Kauffman et al. 2013), suggesting that effects of wolves on elk density rather than behaviour are responsible for reduced browsing (Winnie Jr 2012), or that tree recruitment has not increased (Kimble et al. 2011). Indeed, behavioural responses can be unpredictable; elk may even increase willow browsing when wolves are present, as willow thickets provide protective cover (Creel & Christianson 2009). Although studies outside YNP have recorded that wolves indirectly favour pronghorn Antilocapra americana survival by suppressing coyotes (Berger et al. 2008), coyote numbers within the park have returned to pre-wolf densities, albeit in smaller groups (Mech 2012). Further, most studies supporting the role of wolves in restoring trophic cascades are from protected areas, which form only a small part of the wolf's distribution (Mech

2012). Outside these areas, the behaviour and abundance of large predators will be affected by humans (Ciucci et al. 1997), potentially interfering with trophic effects imposed by wolves

(Wallach et al. 2010; Choquenot & Forsyth 2013).

Predator interactions in Australia

Australia supports a diverse and unique flora and fauna, but since European arrival, 20 mammals have been lost and a further seven are now confined to islands (Department of the

Environment 2009). Predation from introduced predators, the red fox and feral cat, has contributed to the decline of many terrestrial marsupials and rodents (Johnson 2006), particularly those that fall within the Critical Weight Range (35 - 5,500 g) (CWR) (Burbidge &

7 McKenzie 1989; Burbidge & Manly 2002). Persistence of CWR marsupials correlates with retention of Australia's mammalian apex predator, the dingo Canis lupus dingo, suggesting mesopredator release may have contributed to recent extinctions (Johnson et al. 2007).

The feral cat

Cats are thought to have been domesticated from the African/Middle Eastern wildcat

Felis sylvestris lybica (Serpell 2000; Sunquist & Sunquist 2002), perhaps concurrently with the development of agricultural societies in the Near East (Driscoll et al. 2007) as early as 9,500 years ago (Vigne et al. 2004). However, domestic cats retain predatory behaviours and can successfully revert to self-sufficiency (Bradshaw et al. 1996), facilitating establishment in the wild where they can exploit naïve prey and have enduring negative impacts on wildlife (Paton

1993; Dickman 1996; Duffy & Capece 2012; Loss et al. 2013; Nogales et al. 2013). In this thesis, I define feral cats as those that survive independently from humans in self-sustaining populations (Moodie 1995 in Denny & Dickman 2010). Domestic or stray cats dependent on human refuse are not considered.

History in Australia

Most evidence suggests that feral cats became established across Australia through the

1800s, following the settlement of Europeans, rather than with earlier visits of Asian traders or

Dutch explorers (Abbott 2002). In northern Australia, feral cats were observed in the latter half of the 19th century (Abbott 2008). Thousands of cats were released to curb exploding rabbit

Oryctolagus cuniculus populations, despite cats being rather ineffective at rabbit control (Rolls

1969). At least 18 million feral cats (Pimentel et al. 2001) are now distributed across the entire mainland, Tasmania and some offshore islands (Denny 2008).

The impacts of feral cats on wildlife were recognised early in the 20th century (Short &

Calaby 2001; reviewed in Denny & Dickman 2010), although their role in species declines has been considered less important than that of the fox (Short 1998; Kinnear et al. 2002), due to chronological differences between cat arrival and species declines, and the ir lack of impact in high-rainfall areas such as Tasmania (Abbott 2002). However, some arid-zone disappearances

corresponded with the displacement of indigenous Australians, which suggests that removal of hunting pressure on cats (Burbidge et al. 1988) may have led to increased cat numbers and impacts on vulnerable species (Johnson 2006). Feral cats are responsible for local extinctions of

CWR mammals from offshore islands (Algar et al. 2002; Woinarski et al. 2011b) and are associated with the decline of small rodents and other mammals, particularly in areas of low rainfall (Smith & Quin 1996; Burbidge & Manly 2002). Some efforts to reintroduce CWR mammals have also failed entirely or in part due to feral cat predation (Gibson et al. 1994;

Priddel & Wheeler 2004; Moseby et al. 2011). In the Gibson desert, no burrowing bettongs

Bettongia lesueur could be detected 60 days following reintroduction, after presumably being killed by feral cats, whose numbers increased following poison baiting for dingoes and foxes

(Christensen & Burrows 1994). Feral cats may also impact on fauna through disease transmission. Cats are the primary host for the parasite Toxoplasma gondii, which also infects native wildlife (Smith & Munday 1965). In addition to direct mortality (Portas 2010), toxoplasmosis may cause increased activity and relaxed anti-predator behaviour in small mammals, increasing the risk of cat predation (Berdoy et al. 2000)

A supreme predator

Feral cats possess flexible physiological, dietary and behavioural traits, making them an effective and resilient predator in the Australian landscape. Feral cats can adapt to a broad range of habitats and climates, having successfully established on sub-Antarctic islands (Jones 1977), in montane forests (Smucker et al. 2000), and in deserts (Mahon et al. 1998). As obligate carnivores, cats prefer live prey (Scott 1968 in Bradshaw et al. 1996), and can generally meet water requirements by consuming animal tissue (Prentiss et al. 1959; Macdonald et al. 1984).

They hunt prey ranging from invertebrates up to medium-sized mammals (Spencer 1991;

Dickman 1996; Kutt 2011), and prefer several small meals each day rather than gorging on larger prey (Macdonald et al. 1984; Bradshaw 2006), potentially explaining their disproportionate impacts on small wildlife.

9 Feral cats can use versatile hunting strategies, including stalking and ambush (Corbett

1979; Fitzgerald & Turner 2000; Read & Bowen 2001) to optimise prey acquisition. They adapt relatively quickly when a new prey source becomes available or current prey decline. For example, rabbit declines in central Australia led to a crash in fox numbers, while feral cats persisted, presumably by switching prey to smaller vertebrates (Read & Bowen 2001). Hunting flexibility is beneficial in Australia’s variable ecology. Feral cats may increase in number following prey irruptions and then switch to a less abundant food source when irruptive prey subside. Surges in predator density can lead to 'hyperpredation', where less abundant, alternative prey face unsustainable pressure from expanded predator populations (Taylor 1979; Smith &

Quin 1996). Feral cats can also become individually focussed on one prey source (Fitzgerald &

Turner 2000), making it difficult to predict their relationship with vulnerable prey. In some cases feral or domestic cats have hunted island vertebrates to extinction. One cat is thought responsible for the local extirpation of an endemic deer mouse from Estanque island in Mexico

(Vázquez-Domínguez et al. 2004), and feral cats most likely exterminated several bird species from Stephens Island in New Zealand, including an endemic flightless wren (Medway 2004). A reintroduction of rufous hare-wallabies Lagorchestes hirsutus in the Tanami desert demonstrated the potential impacts of prey specialisation in feral cats. The hare-wallabies survived for several months before predation rapidly reduced their population. This was suspected to have been due to one cat that may have temporarily specialised in hunting hare- wallabies (Gibson et al. 1994).

The dingo

The dingo is descended from the wolf, during a long history of semi-domestication through East Asia (Corbett 2001; Oskarsson et al. 2012 ). Dingoes are thought to have arrived in Australia with south-east Asian seafarers between 3,500-5,000 years ago (Savolainen et al.

2004). They quickly spread across the mainland, replacing the largest marsupial predators the thylacine Thylacinus cynocephalus and Tasmanian devil Sarcophilus harrisii, which declined due either to intensified human activity (Johnson & Wroe 2003; Johnson & Brook 2011), or

competition with the larger canid (Fillios et al. 2012). Dingoes appear to have continued a semi- domestic relationship with humans, potentially assisting with hunting, but maintaining connectivity with wild populations (Corbett 2001; Johnson 2006).

As with many other apex predators, the dingo is both admired and maligned (Fleming et al. 2001). With the advent of agriculture and pastoralism following European settlement, dingo numbers likely increased due to provision of artificial water points, and increased prey in the form of introduced rabbits and livestock carrion, which facilitated their spread through the semi- arid zone (Corbett 2001; Johnson 2006). However, from the early days of settlement, dingoes were persecuted and (where possible) eradicated for attacking livestock, particularly sheep

(Rolls 1969). Bounties, poison baiting, trapping and the construction of the Dingo Barrier Fence from South Australia to the Queensland coast have all been deployed in an attempt to reduce the abundance or distribution of dingoes, with variable levels of success (Allen & Sparkes 2001;

Corbett 2001). Persecution continues in pastoral areas across northern and central Australia to reduce calf mortality, although with questionable effectiveness (Allen in press). While lethal control can reduce dingo abundance, it may also disrupt the ir social structure and behaviour

(Wallach et al. 2009b), potentially influencing interactions with other species (Wallach et al.

2010).

Dingoes have been threatened since European settlement by hybridisation with feral or free-roaming domestic dogs (Newsome & Corbett 1982; Fleming et al. 2001). Estimating the extent of hybridisation is difficult, as physical features are not reliable indicators of genetic status, skull measurements may reveal hybrids but require that animals be killed, and the purity of type-dingoes in genetic tests is uncertain (Newsome & Corbett 1985; Jones 1990; Elledge et al. 2008; Newsome et al. 2013c). The population structure of dingoes apparently differs from that of domestic dogs, although breeding patterns of feral dogs are unknown. Breeding in dingo packs is largely restricted to the alpha pair (Corbett 1988) in a single annual breeding event

(Catling et al. 1992; Thomson 1992b), which can potentially regulate population growth

(Corbett 2001; but see Newsome et al. 2013c). It is unknown how hybridisation will affect dingo behaviour and biology, and therefore trophic interactions with other fauna (Elledge et al.

11 2006; Claridge & Hunt 2008). In this thesis I define dingoes as wild, self-sustaining populations of Canis lupus dingo, with no commensal relationship with humans.

Do dingoes enforce top-down cascades?

As the mammalian apex predator on the Australian mainland, the dingo is thought to exert both direct and indirect effects on lower trophic guilds (Glen & Dickman 2005; Glen et al.

2007; Visser et al. 2009; Letnic et al. 2012). There is evidence that dingo removal releases large herbivores from predation, as higher abundances of kangaroos Macropus spp. have been recorded ‘inside’ the Dingo Barrier Fence where dingoes are removed (Caughley et al. 1980;

Pople et al. 2000; Letnic et al. 2009b: but see Newsome et al. 2001).

It has been suggested that dingo reintroduction in degraded rangelands may benefit

CWR mammals by suppressing cats and foxes (Dickman et al. 2009; Letnic et al. 2009b). Early studies recommended retaining the dingo to protect prey populations, after noting negative correlations between dingoes and mesopredators (Lundie-Jenkins et al. 1993). While dingoes generally suppress larger mammals (5-100 kg, e.g. macropods), they appear to benefit smaller species (< 1 kg, e.g. small mammals) by restricting mesopredators and herbivores (Letnic et al.

2009b; Letnic et al. 2012). Relationships with prey of intermediate sizes are not consistent, and may depend on other ecosystem interactions (Letnic et al. 2012). Dingoes are associated with the persistence of ground-dwelling marsupials (Johnson et al. 2007), native rodents (Smith &

Quin 1996) and threatened macropods (Wallach et al. 2009a), as well as small mammal abundance and richness (Letnic et al. 2009b). This suggests that the removal of dingoes releases mesopredator populations and provides them with increased access to native prey. A study comparing ecological communities on either side of the Dingo Barrier Fence found higher abundance of dusky hopping mouse Notomys fuscus in areas with more dingo activity, suggesting dingoes provide indirect benefits (Letnic et al. 2009a). However, removal of dingoes is also associated with increased pastoralism and herbivory and potential reduction in ground cover, which may also reduce abundance of small mammals (Letnic & Dworjanyn 2011; Kutt &

Gordon 2012).

While dingoes may suppress foxes in the arid zone (Letnic & Dworjanyn 2011), they may also exclude foxes from areas of high dingo abundance elsewhere (Johnson & VanDerWal

2009; Letnic et al. 2011). Further, dingo activity may lead to fine-scale partitioning by foxes to avoid interference competition (Mitchell & Banks 2005; Brawata & Neeman 2011), and potential for intraguild predation (Cupples et al. 2011; Glen et al. 2011; Letnic & Dworjanyn

2011). Dietary overlap between dingoes and foxes suggests exploitative competition may be considerable in some ecological contexts (Paltridge 2002; Pavey et al. 2008; Letnic &

Dworjanyn 2011), although dingoes tend to take larger mammals (Mitchell & Banks 2005;

Letnic et al. 2009b; Cupples et al. 2011).

Letnic et al. (2012) reviewed the effect of primary productivity on the dingo's ability to affect lower guilds. They suggested that resource pulses caused by sporadic rainfall in arid

Australia weaken top-down effects of dingoes, as dingoes may preferentially prey on abundant small mammals during flush periods (Corbett & Newsome 1987), effectively releasing large herbivores. Further, macropods and mesopredators can respond numerically to increased resource availability and access, and disperse across a more uniform landscape. Increased predation pressure from mesopredators can then cause declines in small mammals (Letnic et al.

2005). In the savannas of northern Australia, where the annual wet season is followed by a prolonged dry season, Letnic et al. (2012) predict that dingoes will exert stronger top-down effects on mesopredators in the late dry season when resources are limited, therefore providing temporary benefits to small mammal prey.

Support for top-down suppression of feral cats

Despite mounting evidence that dingoes can suppress foxes, the relationship between dingoes and feral cats remains ambiguous (Letnic et al. 2012). To assess the potential for suppression of feral cats, I examined current evidence and determined whether there is motivation for competitive interactions according to ecological theory. An average dingo (14 kg, Corbett 2008) is 3.1 times larger than an average feral cat (4.55 kg, Denny 2008), within the range where intraguild predation is most likely to occur (Donadio & Buskirk 2006). Extensive

13 dietary overlap has also been recorded during resource pulses in the arid zone (Pavey et al.

2008), but dingoes consume larger prey than do cats (Paltridge 2002; Brook & Kutt 2011; Kutt

2012). Predation on cats by dingoes has been recorded, albeit in small amounts, in the arid zone

(Whitehouse 1977; Marsack & Campbell 1990; Lundie-Jenkins et al. 1993; Paltridge 2002;

Letnic et al. 2009b) in northern Australia (Allen et al. 2012) and in eastern forests (Pascoe et al.

2012). Perhaps the most conclusive evidence of interference competition comes from an arid- zone study where a pair of dingoes were introduced to a large, 37-km2 enclosure (Moseby et al.

2012). After several months, wild foxes and feral cats from the area were released into the enclosure and intraguild interactions were monitored. At least half the cats and all the foxes were killed but not consumed by dingoes, suggesting an aggressive rather than a predatory motive for the attacks (Polis et al. 1989). Hence interspecific killing may be more widespread than expected from dietary studies recording intraguild predation.

A suppressed mesopredator may relax behaviourally or increase numerically when top- down pressure is released (Ritchie & Johnson 2009). Complex habitats (Edwards et al. 2002) or different activity peaks (Wang & Fisher 2012), might be used by cats to avoid dingoes, although these risk effects require further investigation. Anecdotal evidence from central Queensland

(QLD) recorded an explosion in male cats following the removal of dingoes (Pettigrew 1993).

Feral cats have also increased in abundance following poison baiting for canids in both the

Gibson (Christensen & Burrows 1994) and Tanami deserts (Lundie-Jenkins et al. 1993).

However, this evidence conflicts with studies that found no relationship (Letnic et al. 2009b), or even a positive correlation (Smith & Quin 1996; Letnic et al. 2009a; Letnic & Koch 2010), between dingoes and feral cat activity. The interaction may be complicated by the presence of foxes, which are intermediate in size between dingoes and cats. By suppressing foxes, dingoes might indirectly release feral cats from interference with a closer competitor (Letnic et al.

2012).

In northern Australia above 18° S, foxes are rare or absent (DEWHA 2008), hence dingoes are more likely to interact directly with feral cats. Kennedy et al. (2012) found a negative correlation between dingo and feral cat activity in northern Australian savannas, and

suggested that dingoes suppress feral cat numbers rather than modify their behaviour, as cat activity did not increase following dingo removal. Hence, there appears to be sufficient grounds to expect that dingoes could suppress feral cats in northern Australia. When resources are limited in the late dry season, exploitative and interference competition between dingoes and feral cats is likely to increase, leading to cats altering their activity to avoid interactions.

Dingoes could theoretically provide indirect benefits to small prey, by limiting feral cat predation (Figure 1.1).

Negative direct effects on Positive indirect prey and mesopredators effects on prey via predation and risk via mesopredator effects suppression - - - + - -

15 Potential benefits of trophic regulation in northern Australia

Current limitations to cat management

The cryptic behaviour of feral cats may have contributed to the late realisation of their impacts on Australian wildlife. Monitoring can be difficult, in part due to low detectability

(O'Connell Jr et al. 2006; Robley et al. 2010b). Extensive effort is required to estimate absolute abundance and changes in population size (Forsyth et al. 2005). Novel techniques and noninvasive technology such as remote cameras may improve the reliability of feral cat monitoring with less survey effort (Bengsen et al. 2011).

Feral cats are also difficult to control. They have been removed from islands (Algar et al. 2002), or from fenced areas (Moseby et al. 2009a), but eradication is not feasible for large- scale management in the rangeland savannas of northern Australia (Sharp & Saunders 2005).

Bomford and O'Brien (1995) established three requirements for successful species eradication: immigration must be prevented, removal must exceed population growth, and all reproductive animals must be susceptible to control, each of which is virtually impossible for cats with current methods. Feral cats do not consistently take poison baits and uptake is low when live prey are available (Risbey et al. 1997; Algar & Burrows 2004; Algar et al. 2007). Labour- intensive shooting and trapping programs need to be sustained in the long term to be effective, and offtake may be limited where cats occur at low densities (Sharp & Saunders 2005).

Dingoes and feral cats exert different predation pressures

Despite claims that dingo predation poses a threat to native fauna (e.g. Allen & Fleming

2012), there is greater evidence for other factors such as habitat loss being implicated in species’ declines (e.g. Kerle et al. 1992). Small and/or isolated populations of endangered species, such as northern hairy-nosed wombats Lasiorhinus krefftii and bridled nailtail wallabies

Onychogalea fraenata, are threatened by dingoes, but dingo predation alone did not drive these species to their current status (Fisher et al. 2001; Horsup 2004). Three species (thylacine,

Tasmanian devil and native hen Gallinula mortierii) have disappeared from the mainland since

dingo arrival, however evidence that dingoes caused the declines is equivocal (Johnson & Wroe

2003). Although dingoes occasionally prey on threatened species (Lundie-Jenkins et al. 1993;

Vernes 2000; Paltridge 2002), their preference for more available prey (Corbett & Newsome

1987) should generally place less predation pressure on smaller threatened species compared to mesopredators (e.g. the red fox, Claridge et al. 2010).

There is strong evidence that introduced predators have a more severe effect on native wildlife than native predators, particularly in Australia (Salo et al. 2007). Naïve prey may lack co-evolved antipredator behaviours to counter recently arrived predators (Cox & Lima 2006;

Banks & Dickman 2007; reviewed in Carthey & Banks 2014). As mesopredators, feral cats are also expected to exploit prey more effectively than dingoes, allowing them to coexist with the larger predator (Polis & Holt 1992). Feral cats have been described as 'resident' predators

(Fitzgerald & Turner 2000) that will systematically move across their range when hunting.

Their ability to switch prey types and exploit both terrestrial and arboreal prey means they could apply more consistent predation pressure over their home range than apex predators (Brashares et al. 2010). In contrast, dingoes are mobile predators with flexible hunting strategies (Corbett

1995a). They consume prey from a broader size range over a wider geographic area, potentially imposing lower or transitory predation pressure on local prey populations.

A new wave of mammal extinctions

The arid zone mammal extinctions that occurred last century are now being echoed in northern Australia (Woinarski et al. 2001; Ziembicki et al. 2013). Long-term studies in Kakadu

National Park have recorded drastic declines in small mammal abundance and diversity

(Woinarski et al. 2010), and there appears to be a similar trend across the north (Woinarski et al. 2011a). Declines could be due to additive or interactive effects of inappropriate fire regimes, grazing by introduced herbivores, cane toads, disease, or predation by feral cats (Woinarski et al. 2011a). However, the declining species are mostly small (~ 1 kg), savanna-based mammals, suggesting that feral cat predation, enhanced by reduced ground cover, is the probable driver

(Fisher et al. in press).

17 Thesis aims and objectives

For the reasons outlined above, dingoes could provide a cost-effective method to limit the impacts of feral cat predation on native wildlife, particularly in northern Australia. The tropical savanna of northern Australia is an ideal location to examine interactions between dingoes and feral cats and explore the mechanisms by which dingoes may exert top-down control, as foxes are rare and unlikely to influence the relationship between the study species.

The goal of this research was to enhance understanding of the ecological interactions between the dingo and the feral cat in northern Australia. To address this goal, my aims were to:

1. Determine whether dingoes can numerically suppress feral cat populations, and

2. Further understanding of predator ecology, to

3. Explore how dingoes potentially influence the behaviour of feral cats.

I structured my research to achieve these aims using both non-invasive and direct survey methods. Objectives to address each of the aims are listed below.

Aim 1: Determine whether dingoes can numerically suppress feral cat populations

Objective 1: Assess the effect of baiting on dingo abundance indices and consequent effects on feral cat indices

In addition to exploring the relationship between dingo and feral cat abundance, this objective examined how predator control affects the abundance of dingoes, and whether it allows populations of feral cats to expand following dingo removal (Chapter 2).

Objective 2: Assess levels of dietary overlap between dingoes and feral cats and determine the extent of intraguild predation

Dietary analysis can confirm both resource and interference competition between sympatric predators by demonstrating dietary overlap or intraguild predation. Dietary samples

collected from study locations were analysed to identify species consumed by each predator, and to provide evidence of competition between dingoes and cats at the sites (Appendix A).

Aim 2: Further understanding of predator ecology

Objective 3: Compare space use patterns of dingoes and feral cats

Patterns of space use by predators may influence how they interact with prey as well as with other predators. I compared a range of space use metrics for sympatric dingoes and feral cats during the dry season in northern Western Australia (Chapter 4). I also assessed the potential for social interactions within species by comparing spatial and temporal overlap between sympatric conspecifics (Chapter 5).

Aim 3: Explore how dingoes potentially influence the behaviour of feral cats

Objective 4: Determine whether feral cats alter their temporal activity to avoid dingoes

Mesopredators may avoid encounters with apex predators by shifting activity patterns to avoid times when the larger predator is most active. I compared temporal distributions of dingoes and feral cats in areas with and without predator control to determine whether predator control influenced the temporal patterns of dingoes, and how feral cat activity patterns differed from those of sympatric dingoes (Chapter 2).

Objective 5: Determine whether feral cats alter their use of habitat to avoid dingoes

Mesopredators may avoid habitats favoured by apex predators, or use habitat features that reduce predation risk. I explored whether feral cats alter their use of habitat in areas where or when dingoes are more common, to test predictions on the role of complex habitat in intraguild interactions and the facilitation of species coexistence (Chapter 3).

19 Objective 6: Determine whether feral cats alter their use of space to avoid dingoes

Areas used by apex predators are likely to carry a higher risk for mesopredators, and may be avoided to mitigate interference competition. I compared camera trap rates for feral cats and dingoes, to detect any patterns of negative correlation in local space use (Chapter 2). Using telemetry data, I also estimated spatial overlap between and within species and compared patterns of avoidance or attraction where space use overlapped. Finally, I determined whether feral cats avoid close encounters with dingoes over time (Chapter 5).

Thesis outline and structure

The four data chapters of this thesis draw on the outcomes of two separate field studies differing in scope and research methods. The first two data chapters address the objectives using data at a landscape scale, while the last two data chapters examine the behaviour of individual predators in sympatry to contribute to Aims 2 and 3. Figure 1.2 presents how the data chapters are structured to address the research aims of the thesis.

Chapter 1 introduces the theory and previous research that informs this study, outlining the role of apex predators in trophic interactions and mesopredator release. Here I also review current understanding of predator interactions in Australia, and provide rationale for this thesis.

Chapters 2 and 3 use data from remote camera surveys on large, rangeland properties across northern and central Australia, paired to compare the effects of predator control on the abundance and behaviour of dingoes and feral cats. These chapters explore broad landscape trends in predator abundance, behaviour and habitat use. Chapters 4 and 5 use high-frequency telemetry data from sympatric predators in the Kimberley in northern Western Australia (WA).

These chapters compare space use characteristics and investigate how feral cats might adjust their behaviour, both spatially and temporally, to avoid interactions with dingoes.

Chapter 1 General introduction

Aim 1 Appendix A Determine whether Prey consumed by dingoes can suppress Chapter 2 dingoes and feral cats feral cat populations Effects of predator control in northern Australia on activity patterns and abundance indices

Chapter 3 Aims 2 & 3 Habitat use and seasonal Further understanding trends of dingoes and feral of predator ecology cats in north Queensland

Explore the influence Chapter 4 of dingoes on feral cat Appendix B behaviour Seasonal space use patterns GPS screening of dingoes and feral cats in methods the Kimberley

Chapter 5 Spatial interactions between dingoes and feral cats in the Kimberley

Chapter 6 General discussion

Figure 1 .2. Thesis structure.

In Chapter 6, I summarise all results in light of the aims and objectives. I then discuss how the research findings support or raise issues concerning the potential for dingoes to suppress feral cats and provide conservation benefits in northern Australia. Finally, I provide suggestions for future research. Appendix A presents the results of dietary analysis from study sites. Appendix B outlines the GPS screening methods used in Chapters 4 and 5. Further appendices contain supporting information to accompany the data chapters.

Chapter 2 Effects of predator control on behaviour of an apex predator and indirect consequences for mesopredator suppression 1

Abstract

Apex predators can benefit ecosystems through top-down control of mesopredators and herbivores. However, apex predators are often subject to lethal control aimed at minimising attacks on livestock. Lethal control can affect both the abundance and behaviour of apex predators. These changes could in turn influence the abundance and behaviour of mesopredators. We used remote camera surveys at nine pairs of large Australian rangeland properties, comparing properties that controlled dingoes Canis lupus dingo with properties that did not, to test the effects of predator control on dingo activity and to evaluate the responses of a mesopredator, the feral cat Felis catus. Indices of dingo abundance were generally reduced on properties that practiced dingo control, in comparison with paired properties that did not, although the effect size of control was variable. Dingoes in uncontrolled populations were crepuscular, similar to major prey. In populations subject to control, dingoes became less active around dusk and activity was concentrated in the period shortly before dawn. Shifts in feral cat abundance indices between properties with and without dingo control were inversely related to corresponding shifts in indices of dingo abundance. There was also a negative relationship between predator visitation rates at individual camera stations, suggesting cats avoided areas where dingoes were locally common. Reduced activity by dingoes at dusk was associated with higher activity of cats at dusk. Our results suggest that effective dingo control not only leads to higher abundance of feral cats, but allows them to optimise hunting behaviour when dingoes are less active. This double effect could amplify the impacts of dingo control on prey species selected by cats. In areas managed for conservation, stable dingo populations may thus contribute to management objectives by restricting feral cat access to prey populations. Predator control not only reduces indices of apex predator abundance but can also modify their behaviour. Hence, indicators other than abundance, such as behavioural patterns, should be considered when estimating a predator's capacity to effectively interact with lower trophic guilds. Changes to apex predator behaviour may relax limitations on the behaviour of mesopredators, providing enhanced access to resources and prey.

1 This is the Accepted Version of a paper published in the Journal of Applied Ecology, with minor formatting changes: Brook, L.A., Johnson, C.N. & Ritchie, E.G. (2012) Effects of predator control on behaviour of an apex predator and indirect consequences for mesopredator suppression. Journal of Applied Ecology 49: 1278- 1286.

23 Chapter 2: Abundance indices and activity patterns

Introduction

When predators occur in sympatry, smaller mesopredators may face a reduction in fitness due to competition from larger apex predators (Creel & Creel 1996). This can take the form of exploitation competition among species that consume the same prey species, or interference competition via harassment, killing and (sometimes) consumption of smaller mesopredators by apex predators (Polis et al. 1989; Polis & Holt 1992). These two forms of competition can directly affect mesopredators by reducing their abundance, and causing modifications of behaviour that allow them to avoid encounters with their larger enemies

(Durant 2000; Hunter & Caro 2008; Salo et al. 2008; Ritchie & Johnson 2009; Elmhagen et al.

2010).

Mesopredator avoidance of apex predators can occur in both space and time, but most research has focussed on spatial shifts (Kronfeld-Schor et al. 2001). Although some studies have recorded temporal partitioning between sympatric, similar-sized felids (Di Bitetti et al.

2010; Romero-Muñoz et al. 2010), few have directly explored temporal shifts involving apex- and mesopredators (but see Hayward & Slotow 2009). Patterns of temporal activity are driven primarily by circadian stimuli such as light, but animals may shift activity in response to other factors such as predation risk (Rasmussen & Macdonald 2011) or competition (Kronfeld-Schor et al. 2001). These shifts may incur costs such as reduced foraging opportunities (Rasmussen &

Macdonald 2011). Larger predators are not immune: their activity patterns can be influenced by the risk of encounter with humans (Theuerkauf 2009), particularly when facing persecution

(Ordiz et al. 2012). Rasmussen & McDonald (2011) found that African wild dogs Lycaon pictus shifted hunting activity to moonlit periods, which would probably reduce the success rate of hunting, to avoid encounters with humans.

Persecution by humans contributes to the decline of apex predator populations (Treves

& Karanth 2003) and consequent ecological disruptions via trophic cascades (Estes et al. 2011).

Removal of apex predators can ‘release’ mesopredators from top-down pressure (Soulé et al.

24 Chapter 2: Abundance indices and activity patterns

1988; Ritchie & Johnson 2009), lifting previous constraints on abundance and behaviour and allowing their populations to expand. As a consequence, abundance and diversity of prey may decline (Crooks & Soulé 1999; Berger et al. 2008). Maintaining apex predators may indirectly protect vulnerable prey and sustain biodiversity (Sergio et al. 2006).

Since its arrival in Australia 3,500–5,000 years ago (Savolainen et al. 2004), the dingo

Canis lupus dingo (L.) has been the largest mammalian predator on the continent. Widespread control programs, using poison baiting, trapping and shooting, aim to reduce the dingo’s impacts on livestock (Fleming et al. 2001). There is evidence that dingoes can suppress abundance of the invasive red fox Vulpes vulpes and thereby have positive effects on species preyed upon by foxes (Johnson et al. 2007; Johnson & VanDerWal 2009; Letnic et al. 2009b;

Wallach et al. 2010). Dingoes may also suppress a smaller mesopredator, the feral cat Felis catus (L.), but there is less evidence for this (Letnic et al. 2012; but see Moseby et al. 2012).

Feral cats arrived with European settlers in the 1800s (Abbott 2002) and now occur across the entire continent (Denny 2008). They have contributed to the decline and extinction of native mammals, reptiles and birds (Johnson 2006), thwarted reintroduction programs for threatened species (Gibson et al. 1994; Priddel & Wheeler 2004) and may be partly responsible for current mammal declines across northern Australia (Woinarski et al. 2011a). Unfortunately feral cats are difficult to control or monitor. Control efforts aimed at dingoes and foxes usually involve distribution of baits laced with the poison sodium fluoroacetate (known as 1080), but these are ineffective against cats because cats are disinclined to take baits.

Both dingoes and feral cats are mostly active during crepuscular and nocturnal periods in warm climates (Jones & Coman 1982; Thomson 1992b). They have physiological adaptations for crepuscular and nocturnal activity, such as optimal vision in low-light conditions and highly- developed olfactory (dingo) and auditory (cat) capabilities (Kavanau & Ramos 1975; Kitchener et al. 2010). In addition, predators may adjust their activity schedules to match periods when their prey are most active or vulnerable (Ferguson et al. 1988). We would expect crepuscular periods, particularly dusk and early evening, to provide optimal foraging conditions for both predators, because they coincide with activity of preferred prey: nocturnal reptiles such as

25 Chapter 2: Abundance indices and activity patterns geckoes are most active in the hours following dusk (Bustard 1967) while diurnal reptiles are retreating, and mammals such as kangaroos and small marsupials (Coulson 1996), rodents

(Breed & Ford 2007) and rabbits (Williams et al. 1995) tend to be crepuscular or nocturnal.

However, we might expect feral cats to underutilise these time periods if they trade off foraging benefits against the higher risk of encountering dingoes.

In this study we explored the effects of predator control on interactions between dingoes and feral cats. We worked on nine pairs of large properties across Australia, where each pair consisted of a site that controlled dingoes (mainly by 1080 baiting, but also by opportunistic shooting) and a similar site in the same environment that did not. We examine (a) how predator control affected indices of abundance and activity schedules of dingoes, (b) whether predator control resulted in shifts in spatio-temporal activity by feral cats and c) whether predator control and/or dingo removal led to increased abundance of feral cats.

Methods

Study area and data collection

We surveyed dingoes and feral cats on eighteen properties spread across north and central Australia, spanning tropical to arid climates, in habitats varying from open forest and woodlands to native grasslands (Figure 2.1). The properties were arranged in pairs, each consisting of one property on which dingoes were controlled and a matched property with no dingo control. Properties varied in size from 7,850 ha to 705,496 ha and ranged from being adjacent to up to 153 km apart. Paired properties were selected to minimise differences in habitat, climate and management. Most properties were working cattle stations, with three exceptions: the TFTA, owned by the Department of Defence, and Mt Zero-Taravale, owned by the Australian Wildlife Conservancy (AWC) and managed for conservation, both in the

Einasleigh Uplands, and Piccaninny Plains on the Cape York Peninsula (CYP), jointly owned by the AWC and WildlifeLink. The TFTA was paired with Mt Zero-Taravale. Neither property

26 Chapter 2: Abundance indices and activity patterns controls dingoes, but we selectively surveyed sections of the TFTA along the property boundary with cattle stations that do control dingoes to measure the effect of that baiting and obtain a contrast with Mt Zero-Taravale. Piccaninny Plains, which has a herd contained behind wire, and is grazed at low levels by feral cattle and horses outside these paddocks, was paired with a cattle station with patchy grazing pressure and broad areas of ungrazed woodland.

We surveyed dingoes and feral cats (and other wildlife) using infrared remote movement-triggered cameras. We used either I-60 Game Spy (Moultrie, EBSCO Industries,

Birmingham, USA) or DLC Covert II (DLC Trading Co., Lewisburg, USA) cameras. Cameras were distributed in pairs along transects, with a spacing of 2–5 km to avoid correlation between pairs (Sargeant et al. 1998). We used minor, unsealed vehicle tracks as transects and each camera pair consisted of one camera placed 1–5 m from the track and the other 50–200 m away, to allow for fine-scale differences in predator activity due to the presence of the track. Cameras were baited with different combinations of attractants such as chicken, the synthetic fermented egg spray FeralMone™ (Animal Control Technologies, Somerton, Australia), Felid Attracting

Phonics (Westcare Industries, Bassendean, Australia), bird seed or wild cat urine (Outfoxed Pest

Control, Ivanhoe, Australia). In some paired areas we surveyed for prey prior to predator surveys, using small mammal bait (rolled oats, vanilla and peanut butter) and positioning cameras 100 m off the road midway between camera pairs. We used 20–40 cameras on each property depending on property size and available tracks, and operated the cameras for five to eight days.

Surveys were generally run consecutively or concurrently on the properties within each pair, except on two occasions where surveys were up to two months apart but still within the same season. Camera type, number and spacing, survey duration and lure combinations were consistent within paired study areas. In the three most northern areas (areas 7, 8 and 9; Figure

2.1), surveys were repeated in the early and late dry season, with a maximum of three surveys on CYP over three years. For this analysis, repeat surveys in the same area were pooled. Other pairs of properties were surveyed once, between March and November. Cameras were

27 Chapter 2: Abundance indices and activity patterns

programmed to record 5-second videos at night and 20- or 5-second videos during the day.

Time and date were recorded with each video.

16.0° S 3

2.0° S Tropic of Capricorn

32.0° S

1 1: Murchison (MUR) 6: Longreach (LGR) 6.0°2: SFinke (FIN) 7: Gulf Plain (GP) 3: Burt Plain (BP) 8: Einasleigh Uplands (EU) 4: Mitchell (MIT) 9: Cape York Peninsula (CYP) 5: Channel Country (CC)

Figure 2.1. Paired survey areas across Australia. Each pair consists of one site that controls dingoes, and one without dingo control.

Analysis

Abundance indices

To distinguish repeat ‘captures’ of the same individual on the same camera and night,

we plotted histograms of times elapsed between consecutive nightly records for each predator

species. These showed a distinct peak for elapsed times of less than 10 minutes, which we

assumed were repeats. To avoid these, we considered records as being independent only when

separated by 30 minutes or more, unless individuals were distinguishable. We then calculated an

28 Chapter 2: Abundance indices and activity patterns abundance index (AI) for each species at each property (individuals trap night-1) (Rovero &

Marshall 2009), which accounted for the number of cameras, survey length and camera failure in each survey. Abundance indices derived from camera trap rates have successfully detected reductions in feral cat abundance (Bengsen et al. 2011). Cumulative indices (such as our AI) are better than proportional indices at detecting changes in density, and the relationship between AI and true abundance is likely to be linear (MacFarland & Van Deelen 2011).

Contrasts in abundance indices

To contrast differences in dingo and feral cat AI within paired areas, we calculated ratios of the AI on properties without predator control over properties with control. The ratios were then converted to natural logs so that their values were centred on zero, and symmetrical about zero. To allow use of zero measures of the raw AI scores in the ratio calculation, we added 0.003 to each index record, a value smaller than the minimum non-zero recorded index

(0.004).

Station-level activity rates

We examined the potential for local separation by comparing dingo and feral cat trap rates at camera stations, using quantile regression implemented in the quantreg package version

4.76 (Koenker 2011) in R version 2.14.1 (R Development Core Team 2011). We calculated regressions for the 50th, 75th, 95th and 99th quantiles. Least-squares regression was unsuitable for several reasons: the data distributions were triangular and not normally distributed, and we were interested in a limiting effect of dingoes on feral cat activity rather than the average correlation

(Cade et al. 1999). Standard errors were estimated using bootstrapping. Camera stations with no predators recorded were excluded.

Temporal data

We surveyed over a wide longitudinal range and at different times of year, introducing variation to the relationship between clock time, day and night length and the timing of sunrise and sunset. We therefore re-scaled clock time for each survey to a standard unit range (from 0 to

29 Chapter 2: Abundance indices and activity patterns

1) with equal spacing between sunset (at 0.5) and sunrise (at 0 and 1) (Appendix B). Temporal distributions were analysed in the program Oriana 4 (Kovach Computing Services, Wales, UK), using the non-parametric Mardia-Watson-Wheeler test to detect differences in the mean angle or angular variance of circular data (Batschelet 1981). This test assumes no repeat data, so identical records were altered by one second in the raw data. Each independent (>30 min interval) capture of a species was considered one time record, regardless of the number of individuals detected.

Prey activity

We analysed the temporal patterns of mammal prey recorded incidentally in the camera surveys. Species were categorised depending on the prey preferences of the two predators. The large macropod category included wallabies (>7 kg) and kangaroos hunted only by dingoes. The small mammal category (≤ 6.65 kg) included species such as rodents, possums and small wallabies, which are known to be preyed upon by cats and also potentially by dingoes.

Results

We recorded 398 independent dingo records (334 time records) and 211 independent cat records (210 time records) over 5309 trap nights.

Effect of control on indices of dingo abundance

Dingo abundance index (AI) varied widely between different properties, ranging from zero at MIT to a maximum of 0.27 at Piccaninny Plains on CYP (Figure 2.2). Abundance indices were generally lower on properties where dingoes were controlled than on matched properties without control: the natural log of the ratio of dingo AIs in unbaited vs. baited areas was larger than zero (one-sided one sample t test: t = 1.95, d.f. = 8, P = 0.044), demonstrating a significant reduction attributable to control. However, the effectiveness of control was variable

(Figure 2.2), and in one case dingo AI was actually higher on the property with dingo control.

30 Chapter 2: Abundance indices and activity patterns

0.3

)

1 -

0.25

0.2

0.15

0.1

0.05

Abundanceindex (individualstrap night 0 CYP MUR BP FIN CC EU GP LGR MIT Survey areas

Relationships between indices of dingo and feral cat abundance

We detected a shift in abundance indices of feral cats inverse to dingoes within paired

2 sites (R = 0.70, F1,7 = 16.23, P = 0.005, Figure 2.3). As the ratio of dingo AI in the unbaited to baited sites increased, cats were more likely to show the inverse trend, that is to occur at a higher rate on the baited site than on the unbaited site. The x-intercept at 0.75, indicating a dingo AI ratio of 2.12, suggests that once dingo indices were reduced by more than half, feral cat indices tended to be higher on the baited property. Feral cat abundance indices did not increase consistently with dingo control when tested across all areas; the natural log of the ratio of AIs in unbaited vs. baited areas was not less than zero (one-sided one sample t test: t = 0.03, d.f. = 8, P = 0.51).

Trap rates at individual camera stations also suggested a limiting effect of dingoes on feral cats. We found a threshold relationship between the trap rates of dingoes and feral cats

31 Chapter 2: Abundance indices and activity patterns

(Figure 2.4). Where dingoes were rare or not recorded, feral cat trap rates ranged from 0.125–1 cats night -1. As dingo trap rates increased, feral cat records declined. No feral cats were recorded at stations where dingoes were recorded more than once per night. All quantile regression slopes were significant.

-1

-2 Ratio of cat vs.control Ratio AI without dingo with

-2 -1 0 1 2 Ratio of dingo AI without vs. with dingo control

Dingo behaviour

Activity patterns of dingoes were significantly different on properties with and without dingo control (Mardia-Watson-Wheeler test: W = 22.95, P < 0.001; Figure 2.5). Crepuscular activity in the early evening was reduced in areas with predator control, and the activity pattern shifted towards a single peak before sunrise.

32 Chapter 2: Abundance indices and activity patterns

Tau τ Slope Standard error P 0.5 -0.13 0.02 < 0.001 0.75 -0.11 0.02 < 0.001

1.0 0.95 -0.19 0.02 < 0.001

0.99 -0.37 0.10 < 0.001

1 0.8

0.6

l 0.4

0.2 0.0 0.0 0.5 1.0 1.5 2.0 Dingoes night 1

Feral cat behaviour

Feral cat activity patterns differed on properties with and without predator control (W =

5.87, P = 0.053), and were significantly different from dingoes in areas both with (W = 23.78, P

< 0.001; Figure 2.5a) and without predator control (W = 25.50, P < 0.001; Figure 2.5b). In areas where dingoes were baited, feral cat activity peaked in the early evening, when nocturnal dingo activity was lowest.

There was a negative relationship in the proportion of activity concentrated around dusk

(that is, during the hour before and three hours after sunset) between the two species across survey areas: feral cats were more likely to be active at dusk when dingoes were less active at

2 dusk (R = 0.38, F1,11= 6.82, P = 0.024; Figure 2.6).

33 Chapter 2: Abundance indices and activity patterns

0.2 a) Predator control 0.2 b) No predator control

0.16 0.16

0.12 0.12

0.08 0.08

0.04 0.04 Proportionof activity records 0 0

Figure 2.5. Proportion of activity records a) with and b) without dingo control. Time has been scaled to a circular distribution with equal distance between sunset and sunrise. In 5a a nd b, black bars represent dingoes with (N = 85) and without (N = 249) control, and light grey bars represent cats with (N = 81) and without (N = 129) control.

The shift in predator activity patterns under dingo control was most evident in areas

where baiting was more effective in reducing dingo AI. In BP, EU, GP, LGR and MIT, where

the difference in dingo AI with control was weaker, the contrasting activity patterns between

dingoes and feral cats disappeared in baited areas (W = 1.17, P = 0.56, N = 16 dingoes, 36 cats),

while in CYP, MUR, FIN and CC, sites with a larger contrast in dingo AI, the inverse

relationship in crepuscular activity remained strong (W = 23.78, P < 0.001, N = 69 dingoes, 45

cats).

Chapter 2: Abundance indices and activity patterns

1.0

0.8

0.6

Feralcats

0.4

0.2 0.0

0.0 0.2 0.4 0.6 0.8 1.0 Dingoes

Figure 2.6. Proportion of activity records in the dusk/early evening (one hour pre -sunset to three hours post-sunset) (N = 13 sites with at least one dingo and one cat record). Data are square-root transformed.

Prey activity patterns

Activity patterns of dingoes were not significantly different from those of their large macropod prey (number of prey records N = 82) in unbaited areas (W = 5.03, P = 0.081; Figure

2.7a), but the reduction in dusk activity in areas with predator control shifted dingo activity away from the crepuscular peaks of large mammals (N = 157) (W = 13.07, P = 0.001). In contrast, feral cat activity patterns were closer to the nocturnal activity of small mammals (N =

56) in areas with dingo control (W = 3.28, P = 0.19; Figure 2.7b) than without (W = 5.37, P =

0.068, small mammal N = 196).

35 Chapter 2: Abundance indices and activity patterns

0.18 a) 0.18 b)

0.15 0.15

0.12 0.12

0.09 0.09

0.06 0.06

Proportionof activity records 0.03 0.03

0 0

Discussion

Our study confirms that predator control can influence not only abundance but also the behaviour of large predators. The effects of predator control on dingoes and their behaviour may provide opportunities in the spatial and temporal landscape for increased feral cat activity, by reducing the encounter rate between predators and lowering risk for feral cats (Laundré et al.

2001).

Predator control and predator numbers

Sites with dingo control had significantly lower indices of dingo abundance than sites without control, but not in all cases. The exceptions may be due to a) ineffective control, b) control effects flowing into neighbouring properties or c) low detection rates. The effectiveness of dingo control is dependent on numerous factors. Dingo density may actually increase following baiting (Wallach et al. 2009b), as young dingoes can colonise vacant territories at high density if control is not coordinated over sufficiently large areas (Allen & Gonzalez 1998).

Poison baits may not be as accessible to dingoes in complex habitats and landscapes, and

36 Chapter 2: Abundance indices and activity patterns abundant prey can also reduce bait uptake (Allen & Sparkes 2001). Dingo density on unbaited sites may be reduced if dingoes visit adjacent baited properties and are killed. Dingoes probably traversed property boundaries despite our large survey areas, as dingo home ranges can be extensive (up to 22,622 ha in south-east Australia) and they can also make long-distance forays up to 60 km (Claridge et al. 2009). That detectability was imperfect was evident in one area

(MIT), where dingo tracks were observed on roads, but no dingoes were recorded on camera (A.

McNab, pers. obs.).

Shifts in indices of dingo abundance due to control were associated with inverse shifts in feral cat abundance, suggesting a negative relationship between abundances of the two species. Although this shift was significant when comparing paired sites, we did not find an overall increase in feral cat indices in direct response to predator control. However, we would not expect feral cats to respond to predator control per se, but to consequent changes in the dingo population, which were variable. In some areas, both feral cat and dingo abundance indices were higher on unbaited properties, suggesting competition may have been minimised by prey densities that could sustain both predators, or by habitat features. Habitat complexity can mediate interference competition between predators (Janssen et al. 2007) by reducing the encounter rate. Additionally, feral cats can climb trees to avoid dingoes. Hence in areas with considerable tree cover, such as CYP, feral cats may be able to occur at relatively high densities due to the protection provided by complex habitats (Lima & Dill 1990; Ritchie & Johnson

2009).

Avoidance in time

Temporal activity of dingoes in unbaited areas was similar to previous observations, with bimodal crepuscular peaks, frequent activity during the night and sporadic activity during the day (Harden 1985; Thomson 1992b; Robley et al. 2010a). However, in areas where dingoes were controlled, activity at dusk was reduced and shifted to a peak before sunrise. This could allow dingoes to avoid potentially lethal encounters with people, given that although poison baiting is the primary method of lethal control, landholders also opportunistically shoot dingoes.

37 Chapter 2: Abundance indices and activity patterns

Canids may trade off crepuscular or diurnal hunting activity to minimise the risk of encountering people (Theuerkauf 2009), particularly when faced with persecution (Kitchen et al. 2000).

Feral cat activity was inversely related to differences in dingo activity due to control of dingoes. This relationship was evident in areas where dingo control was more effective, and was not apparent in areas where dingo control had less effect, providing further support for the inference that feral cats were responding to reduced dingo presence. Feral cats may inherently have a bimodal circadian rhythm (Randall et al. 1987), but they can exhibit crepuscular and nocturnal patterns (Jones & Coman 1982; Burrows et al. 2003) or irregular cathemeral activity

(Molsher et al. 2005; Moseby et al. 2009b), to adapt to different needs such as predator avoidance (Langham 1992).

Temporal partitioning is probably due to interference rather than exploitation competition, as it allows competitors to reduce aggressive encounters (Carothers & Jaksić

1984). Harrington et al. (2009) found the invasive American mink Neovison vison exhibited a diel shift from nocturnal to diurnal activity, without reducing abundance, following recolonisation by native competitors in the UK. Apart from a few such examples, competitors are unlikely to drastically shift activity from their circadian rhythms, which are entrained to environmental cues (Kronfeld-Schor & Dayan 2003; but see Gutman & Dayan 2005).

Competitors that evolve under similar ecological conditions may develop similar activity schedules, a further limitation to diel-scale partitioning (Kronfeld-Schor & Dayan 2003).

Almost all small mammals in Australia, a substantial prey resource for feral cats, are crepuscular or nocturnal (Van Dyck & Strahan 2008), hence the activity shifts we observed in feral cats, which still provide some prey overlap, may be a more likely response to interference competition than a complete diel shift.

Avoidance in space

Species exposed to predation (or interference competition) also avoid high-risk areas in space, forgoing potential foraging gains from using those areas (Brown et al. 1999; Laundré et

38 Chapter 2: Abundance indices and activity patterns al. 2001; Wirsing et al. 2008). For feral cats, hotspots of dingo activity may carry a greater risk of potentially lethal encounters. Our results suggest that space use by dingoes restricts feral cats at a patch level, although factors such as habitat and prey availability will also influence their abundance. Patterns of mesopredators avoiding dingoes have been observed in other studies. In arid Australia, dingoes are common around water points and feral cats are rare, in areas without predator control. However, where dingoes are poisoned, feral cats use areas near water more often (Brawata & Neeman 2011). Studies analysing field data (Johnson & VanDerWal 2009) and historical bounty data (Letnic et al. 2011) both found significant effects of dingo activity indices on the upper range of fox indices, suggesting dingoes can limit fox activity.

How could behavioural shifts affect prey species?

Predators are expected to optimise their activity by matching it to that of their prey. The reduced activity of dingoes around dusk decoupled their activity patterns from those of macropod prey. An indirect consequence of this behavioural shift may be reduced hunting pressure from dingoes and the demographic release of herbivores such as kangaroos or rabbits.

Herbivores trade off optimal foraging conditions with the perceived threat associated with obtaining those resources (Lima & Dill 1990). Herbivores can more effectively exploit preferred foraging areas if predator presence is reduced, potentially leading to population growth and impacts on vegetation.

Reduced dingo activity at dusk may also provide a window for feral cats to hunt with less interference. Mesopredators released from top-down control can exert more predation pressure on prey than apex predators (Prugh et al. 2009). They tend to be more effective hunters, allowing them to coexist with apex predators (Polis & Holt 1992). Many Australian animals are active at dusk, and some may be particularly vulnerable to predation by feral cats at this time: reptiles become slower as temperature declines (Bennett 1983), and mammals must forage or hunt to satisfy energy demands and are likely to show a surge of activity early in the night. In northern Australia where native mammals are in rapid decline (Woinarski et al.

39 Chapter 2: Abundance indices and activity patterns

2011a), increased hunting success by feral cats in the absence of dingoes could accelerate the extinction trajectories of these vulnerable species.

Our study shows that mesopredators can coexist with apex predators by concentrating their use of space and time to avoid encounters. Control measures not only reduce the abundance of apex predators, but can lead to behavioural changes that may relax top-down pressure on mesopredators, potentially allowing them to shift to more prey-rich areas or time periods, facilitating an increase in mesopredator abundance and predation pressure on prey. If predator control is used to reduce apex predator abundance, but maintain a population to retain ecological functions such as suppression of herbivores and mesopredators (Soulé et al. 2003), it is important to consider the effects of control on behaviour as well as abundance (Ritchie et al.

2012). Hence in areas managed primarily for conservation, predator control should be reconsidered in light of potential risks to wildlife. The presence of dingoes may restrict feral cats to sub-optimal niches and thus provide refuge from cat predation for prey.

Author contributions

LB, CJ and ER conceived the study, LB collected field data, performed analyses and wrote the paper, CJ assisted with analysis and editing, ER collected field data and assisted with editing.

Chapter 3 Hiding in plain sight: Do feral cats use complex habitats in north Queensland to avoid encounters with dingoes? 2

Abstract

Mesopredators should adjust their behaviour to avoid potentially lethal interactions with more aggressive and dominant apex predators, and the structural complexity of habitats may be a key determinant of the risk of detection and attack by apex predators. Complex habitats can provide protective cover and escape opportunities, while open habitats may increase vulnerability once detected. In northern Australia, an invasive mesopredator, the feral cat Felis catus, co-occurs with the larger apex predator, the dingo Canis lupus dingo. Despite the ubiquitous distribution of these species in Australian ecosystems and their effects on other biodiversity, it is unknown how habitat complexity may mediate cat-dingo interactions. I surveyed large, paired properties in two bioregions of north Queensland with remote cameras, in both the early and late dry seasons. One property in each pair practised predator control to reduce dingo populations, while the other property left dingoes alone. I used generalised linear mixed models to explore how habitat complexity, season and predator control influenced the frequency with which predators were recorded. In a woodland matrix on Cape York Peninsula, where dingo activity was high, feral cats were recorded more frequently in open areas with less ground cover. By using open habitats, cats may increase early detection of dingoes, potentially allowing them to escape to trees and avoid lethal interactions. I recorded dingoes more frequently in the early dry season, corresponding to their mating season, but feral cats were active all year, and did not appear to avoid interaction with dingoes by decreasing activity during dingo mating season. Although roads were strongly favoured by dingoes, feral cats were also recorded frequently on roads. In the relatively more open habitats of the Gulf Plains, where dingo activity was low, feral cats tended to use areas with less tree cover on a broad scale, which potentially increased vulnerability to dingo capture. This suggests that the probability of dingo encounter in these risky habitats was not sufficient to outweigh the associated benefits of their use. Habitats that provide escape routes, such as woodlands, may therefore allow feral cats to coexist with dingoes, even when using areas of lower complexity. Feral cats do not appear to favour low-risk habitats in areas where dingoes are rare, but further research should investigate whether habitat use shifts as dingo activity increases.

2 Brook, L. A., Johnson, C. N., Schwarzkopf, L. Nimmo, D. G. & Ritchie, E. G. Hiding in plain sight: Do feral cats use complex habitats in north Queensland to avoid encounters with dingoes? To be submitted to Austral Ecology.

43 Chapter 3: Habitat use

Introduction

Predators should use habitat and space to optimise access to resources, by selecting areas where prey are abundant (Lima 2002) or vulnerable (Laundré et al. 2009; Arias-Del Razo et al. 2012), as well as securing access to shelter (Rosalino et al. 2005) or mates (Liberg et al.

2000). Subordinate mesopredators however, also need to account for the risk of antagonistic and potentially lethal interactions with larger predators (Ritchie & Johnson 2009). Fear of apex predators may therefore influence the habitat decisions of mesopredators.

Different habitats or areas may vary in their level of associated risk, creating a

'landscape of fear' for animals that avoid predators (Laundré et al. 2001). Mesopredators may select habitats to minimise intraguild aggression in several ways. They may reduce the risk of encounter by using habitats that larger predators use less frequently (Silva-Rodriguez et al.

2010), or that provide protection by reducing visibility for the apex predator (Janssen et al.

2007). For example, complex habitats that provide considerable cover can minimise interference and allow predators to coexist (Finke & Denno 2002). Alternatively, mesopredators can reduce the likelihood that encounters will result in capture and death (Heithaus et al. 2009), by using areas or habitats that provide escape routes following encounter (Heithaus & Dill 2006), or that allow them to detect predators first (Lima & Dill 1990; Arias-Del Razo et al. 2012). Risk effects incurred to avoid predator interactions, such as changes to habitat use, can impose limitations on both individuals and populations (Creel & Christianson 2008), as mesopredators forgo access to resource-rich, but high-risk, areas.

Similarly, habitat use of apex predators may be influenced by humans (Martin et al.

2010; Davis et al. 2011), particularly in areas where predators face persecution. Brown bears

Ursus arctos shifted to nocturnal activity when persecution increased in the hunting season

(Ordiz et al. 2012), suggesting apex predators respond to risk effects from humans. In Québec, where up to 35% of the wolf Canis lupus population is harvested annually through hunting and trapping (Larivière et al. 2000), wolves avoid areas surrounding cabins, particularly when human visitation, and therefore, the likelihood of encounter, is high (Lesmerises et al. 2012).

44 Chapter 3: Habitat use

Exclusion of apex predators, or changes in their behaviour due to human activity, can weaken their ability to regulate lower trophic guilds (Hebblewhite et al. 2005).

In northern Australia, the introduced mesopredator, the feral cat Felis catus, occurs in sympatry with an apex predator, the dingo Canis lupus dingo, without additional interference from another introduced mesopredator, the red fox Vulpes vulpes, which is widespread in subtropical and temperate Australia. Dingoes are thought to suppress the abundance and/or behaviour of feral cats, either through direct mortality (Moseby et al. 2012), or by effecting shifts in behaviour (Johnson 2006; Brook et al. 2012). However, at present there is little mechanistic understanding of how dingoes and feral cats use habitat in sympatry, and in particular how variation in habitat complexity may affect interactions. Feral cats may use habitat features to reduce risk associated with dingoes, potentially favouring complex forested habitats (Edwards et al. 2002), as feral cats can climb trees to escape dingoes in pursuit. In forests of south-east Australia, feral cats were found in proximity to creek lines, where complex vegetation may provide protection from larger predators (Buckmaster 2011).

Feral cats are adaptable, ubiquitous predators; they occur in habitats as varied as sub-

Antarctic tussock grasslands (Brothers et al. 1985), arid deserts (Mahon et al. 1998) and tropical forests (Thornton et al. 2001). The feral cat's broad dietary habits (Kutt 2011), non-reliance on free water (Prentiss et al. 1959), flexible hunting styles (Corbett 1979; Fitzgerald & Turner

2000) and ability to switch prey types (Read & Bowen 2001) may facilitate their success in such diverse habitats. Therefore, it is not surprising that, following their introduction and spread through the 1800s, feral cats have adapted to most Australian ecosystems and are now distributed across the entire continent (Abbott 2002). Although feral cats are found across northern Australia, I am unaware of any previous assessments of their habitat use in savanna ecosystems. The tropical savannas of northern Australia cover around 2 million km2 (Williams et al. 2005) and represent one of the most extensive, intact savannas remaining globally

(Woinarski et al. 2007). They support diverse, endemic flora and fauna (Woinarski et al. 2007) and are relatively unmodified compared to southern ecosystems (Garnett et al. 2010). However, mammal fauna in northern Australia have recently declined in abundance and diversity

45 Chapter 3: Habitat use

(Fitzsimons et al. 2010; Woinarski et al. 2011a), and feral cats are thought to be contributing to these declines (Fisher et al. in press).

The dingo arrived in Australia at least 3,500 years ago with south-east Asian traders, and rapidly spread across the mainland (Corbett 1995a; Savolainen et al. 2004). Dingoes are found in most habitats, from arid zones (Thomson 1992a; Wallach et al. 2010) to the wet-dry tropics (Corbett 1995b; Vernes et al. 2001) and eastern forests (Catling & Burt 1997; Robley et al. 2010a). However their distribution has declined since European arrival due to the effects of fencing and predator control to prevent impacts on livestock (Fleming et al. 2001). They are restricted by water availability (Wallach & O'Neill 2009), and have benefited from rangeland development, and the installation of artificial water sources, in semi-arid Australia. Dingoes frequently move along unsealed vehicle tracks (Mahon et al. 1998) which may facilitate movement, but also provide shared spaces, such as scent posts at intersections, for communicating with conspecifics (Corbett 1995a). Previous studies have found dingoes using habitat relative to availability (Edwards et al. 2002), while others report dingoes selecting relatively open habitats, that might facilitate group hunting, in complex environments (Robley et al. 2010a). Alternatively, dingoes may use areas that offer reliable access to food or water

(Thomson 1992d; Newsome et al. 2013a). Higher dingo activity has been recorded prior to or during the breeding season (occurring between approximately March to June) (Thomson 1992b;

Purcell 2010), when individuals may increase patrolling and communication. Dingoes tend to retreat to dens to whelp and raise litters later in the dry season, ranging from June and

November (Thomson 1992d; Corbett 1995a).

I investigated the habitat use of dingoes and feral cats in two case study areas in northern Queensland that differed in habitat type and bioregion. In each study region there were two paired sites, one with and one without predator control (poison baiting and shooting). I examined whether predator control affected dingo habitat use, and how the habitat use of feral cats varied correspondingly and with variation in landscape complexity. I predicted that dingoes:

1. are more active in areas without predator control;

46 Chapter 3: Habitat use

2. use complex habitats more frequently in areas with predator control, at both local and

broad scales;

3. use roads less frequently in areas with predator control; and

4. are more active in the early dry season.

In response, I expect that feral cats:

5. are more active in areas where, or at times of year when dingoes are less active;

6. use more complex habitats in areas without predator control, to provide concealment

and escape routes, at both local and broad scales; and

7. avoid roads to reduce encounters with dingoes, particularly in areas without predator

control.

Methods

Study areas

I surveyed a pair of properties in each of two study areas in north Queensland,

Australia: Spoonbill (141°5' E, 19°58' S) and Millungera (141°34' E, 19°51' S) stations near

Julia Creek in the Gulf Plains bioregion, and Piccaninny Plains (142°37' E, 13°10' S) and

Holroyd River (142°37' E, 14°7' S) on the Cape York Peninsula (CYP) (Figure 3.1). Spoonbill,

Millungera and Holroyd River are active cattle stations, and Piccaninny Plains, owned by the

Australian Wildlife Conservancy and WildlifeLink, is managed for conservation. Within each area, one property (Millungera, Holroyd River) used 1080 poison baits to reduce dingo numbers, as well as occasional shooting. Within each pair of properties, grazing pressure was similar. Although Piccaninny Plains is primarily a sanctuary, grazing by feral cattle and horses occurs at low levels.

Both areas experience a seasonal climate with most rainfall and vegetation growth in the summer 'wet season', followed by an extended dry season (Woinarski et al. 2005). The early dry season is characterised by sporadic rainfall and mild temperatures, compared to the hotter

47 Chapter 3: Habitat use late dry season. Ground cover may be reduced during the dry season, due to grazing and fires, which increase in intensity and extent as the dry season progresses. Mean annual rainfall in the study areas was 464 mm on the Gulf Plains (Julia Creek Post Office), to 1,205 mm on Cape

York Peninsula (Coen Airport) (http://www.bom.gov.au).

Figure 3.1. Location of study sites in the Gulf Plains (fawn) and Cape York Peninsula (olive) bioregions in north Queensland, Australia.

Vegetation

The Gulf Plains survey area included woodlands and open woodlands (Figure 3.2a) featuring coolibah Eucalyptus microtheca, bloodwood Corymbia polycarpa, gidgee Acacia cambagei (Figure 3.2c), bauhinia Lysiphyllum cunninghamii and whitewood Atalaya hemiglauca, grasslands dominated by Mitchell Astrebla spp., feathertop Aristida spp. and

Flinders grasses Iseilima spp. (Figure 3.2e), and floodplains of Queensland bluebush

Chenopodium auricomum.

48 Chapter 3: Habitat use

a. b.

c. d.

e. f.

g. h.

Figure 3.2. Examples of habitat variability on the Gulf Plain (a, c, e, g) and Cape York Peninsula (b, d, f, h) ranging from complex ground cover and tree cover (a, b), tree cover with low ground cover (c, d), complex ground cover and minimal tree cover (e, f) and low ground cover and minimal tree cover (g, h)

49 Chapter 3: Habitat use

Dominant vegetation types on Cape York Peninsula included open-forest and woodlands of box E. leptophleba, bloodwoods C. clarksonia and E. nesophila, Darwin stringybark E. tetradonta and Cooktown ironwood Erythrophleum chlorostachys (Figure 3.2b) and open woodlands of eucalypts and/or broad-leaved paperbark Melaleuca viridiflora (Figure

3.2f, h). Understoreys were dominated by juvenile trees and native grasses Themeda triandra,

Heteropogon spp. and Schizachyrium spp. (Figure 3.2b, f).

Survey methods

Surveys were repeated in the early and late dry seasons to capture information related to seasonal fluctuations in the study species’ behaviour and population size , and differences in vegetation cover. Survey timing depended on local climate of the study area, and ranged from

May to August for early dry surveys, and from September to October for late dry surveys.

I used infra-red remote cameras (DLC Covert II (DLC Trading Co., Lewisburg, USA) to survey predators. Cameras were distributed in pairs along unsealed road transects, with one camera 1-5 m from the track and the other 200 m away, to allow for local differences in predator activity due to the variation in track use. Camera pairs were spaced 5 km apart, to reduce spatial autocorrelation between pairs (Sargeant et al. 1998).

Surveys were run concurrently with a two-day delay on the Gulf Plains, and consecutively on CYP, due to logistical difficulties in running surveys at the same time, given the distance between the two properties (Figure 3.1). In each survey, thirty-two cameras were operated for eight days on each property on the Gulf Plains, while forty cameras were operated for six days on each property on CYP. I programmed cameras to record 5-second videos with a

1-minute interval. Time and date were recorded with each video.

I used species-specific lures to improve the detection rate of predators. Different lures were used within each camera pair, and were alternated midway through the survey, to minimise differences in detectability at each camera. I used a combination of an auditory call lure (Felid Attracting Phonics (Westcare Industries, Bassendean, Australia)) and an olfactory attractant (wild cat urine (Outfoxed Pest Control, Ivanhoe, Australia)) for feral cats. Chicken

50 Chapter 3: Habitat use was presented in bait holders to lure dingoes, as more specific lures such as urine or gland-based mixtures could influence both dingo (Sillero-Zubiri & Macdonald 1998) and cat behaviour

(Kats & Dill 1998). A call lure of chicken and aniseed oil was tied nearby at a height of 2 m.

Aniseed oil has been found to attract dingoes in far north QLD (A. Bengsen pers. comm.) and call lures are useful to disperse odours more effectively at longer distances (Schlexer 2008).

Gloves were worn at all times when handling equipment to avoid depositing human odour, as previous antagonistic interactions (e.g. shooting) may reduce detectability of dingoes on baited properties.

Generalised Linear Mixed Modelling (GLMM)

Predator indices

To distinguish repeat ‘captures’ of the same individual on the same camera and night, I plotted histograms of times elapsed between consecutive nightly records. These showed a peak for elapsed times of less than 10 minutes, which I assumed represented repeats. To avoid these repeats, I considered records as being independent only if they were separated by 30 minutes or more, unless individuals were distinguishable. I calculated a proportional index as the proportion of survey nights at each camera where each species was recorded. This response variable is a compromise between using count data and binary data. It reduces overdispersion and outlier frequency, but provides more information than a presence/absence index.

Predictor variables

I collected quantitative and qualitative measurements of habitat complexity in a 100-m2 circular area (5.7-m diameter) around the lure at each camera station. I estimated broad-scale habitat cover using Google Earth (Image © 2012 GeoEye, Image © DigitalGlobe, ©2012

CNES/Spot Image, ©2012 Whereis® Sensis Pty Ltd), the majority of which had a 2.5-m resolution. To estimate short-term habitat choice by feral cats, % tree, shrub and grass cover were estimated both in a 100-m radius circular measurement area around each camera, and in a

51 Chapter 3: Habitat use

3-km2 circular area centred at the midpoint between each camera pair. Each area was divided into equal quadrants and estimated separately, then summed over the measurement area.

The set of habitat variables was reduced to remove those unlikely to influence predator space use (e.g. canopy height) or which varied little within study sites (e.g. burn class). The remaining variables were standardised by centering to a mean of zero and dividing by two standard deviations (Gelman 2008) using the arm package (Gelman & Su 2013). Highly correlated variables (Spearman correlation coefficient rho > 0.5 or rho < -0.5) were removed, leaving four variables potentially representing habitat at local and broad scales (Table 3.1).

Table 3.1. Habitat complexity variables measured at each camera station.

Variables Measurement technique Local habitat: measured at camera Tree density Number of stems with DBH > 70 mm within 100-m2 area

Ground cover Mean % visible obstruction of 1-m wide view from each of two equidistant measurement points 5.7 m from lure, back towards lure, at 0.5 m height parallel to ground, measured in categories as follows: 0: 0%; 1: 1-20%; 2: 21-40%; 3: 41-60%; 4: 61-80%; 5: 81- 100% Broad habitat: estimated remotely Broad ground cover Proportion low habitat (grass cover) in a 100-m radius circular measurement area around each camera station.

2 Broad tree cover Proportion thick habitat (tree cover) in a 3 km circular measurement area around the midpoint between each pair of camera stations.

I graphically explored relationships between the response and each predictor variable, and applied log-transformations where necessary to improve linear fit. I inspected boxplots to ensure habitat variables did not differ considerably between properties. I then estimated multicollinearity by analysing the variance inflation factors (VIFs) of all predictors (Fox &

Weisberg 2011) when combined in a generalised linear model (R Development Core Team

2013), to ensure no VIFs were above 3.

52 Chapter 3: Habitat use

Model structure and selection

I used a semi-parametric mixed model framework in the gamlss package (Rigby &

Stasinopoulos 2005) to model the proportion of nights at each camera where each predator was recorded. This package can fit response variables approximating a range of distributions and model multiple distribution parameters such as variance, as a function of explanatory variables.

I formulated a global model for each species in each study area including explanatory variables and interaction terms to address each of the specific a priori predictions (Table 3.2). I included camera pair as a random effect in all models, to account for repeat surveys in each season (Bolker et al. 2009) and specified one target degree of freedom for the random effect in every model. Standard errors were calculated using the qr method, which does not account for correlation between the distribution parameters (Stasinopoulos et al. 2008). I selected the most suitable distribution (e.g. beta-binomial, binomial) for the global model, and assessed whether a variance structure and/or smoothing functions improved fit, using both AIC and graphical exploration of residuals.

Explanatory variable Prediction CYP GP Dingoes more active without predator control (PC) Predator control Predator control Feral cats more active with predator control (PC) (PC)

Dingoes are more active in the early dry season Season Season Feral cats are more active in the late dry season

Dingoes use roads more frequently Position (pos) Position (pos) Feral cats use roads less frequently Dingoes use roads more frequently in the unbaited area PC * pos PC * pos Feral cats use roads more frequently in the baited area

Dingoes use more complex habitat with PC PC * tree density PC * ground Feral cats use more complex habitat without PC PC * ground cover cover

PC * log (broad PC * log (broad low cover) tree cover)

53 Chapter 3: Habitat use

I then used an all-subsets approach on the global model to compare the support for all possible parameter combinations (69 models). This approach is suited to this study, which is largely exploratory and testing multiple predictions in a poorly understood system (Symonds &

Moussalli 2011). I used the second-order information criterion AICc for model selection, with an adjusted penalty term to correct for small sample bias (Burnham & Anderson 2002). The random effect and any variance structure were held constant within the model set.

I used model selection processes to determine the relative support (wi) for each model.

The parameter estimates were averaged without shrinkage, to explore how predators are influenced by the explanatory variables, rather than build a predictive model (Symonds &

Moussalli 2011). I calculated unconditional standard errors for averaged parameter estimates, which accommodate uncertainty surrounding model selection, and report unconditional confidence intervals as mean ± 2 standard errors (Burnham & Anderson 2002; Grueber et al.

2011). I include models with < 2 in the candidate set for the Kullback-Liebler best supported model (Burnham & Anderson 2002). I also report a generalised coefficient of determination R2

(Nagelkerke 1991) to indicate explained deviance for each model, comparing the variation explained by each fitted model against a null model with no random effect. Finally, the relative importance (w+) of the predictor variables was calculated, with more support for predictors with w+ > 0.5 (Ritchie et al. 2009). Finally, I used binomial tests to run post-hoc comparisons on the effect of season on predator activity with and without predator control. Analyses were run in R version 3.0.1 (R Development Core Team 2013).

Results

Cape York

I recorded 168 independent observations of dingoes on 114 nights at 58 cameras, and 58 independent observations of feral cats on 53 nights at 48 cameras. Surveys ran over 917 trap nights, over the two seasons. I removed an outlying data point, which prevented feral cat models

54 Chapter 3: Habitat use with random effects from converging, representing five cats recorded over four nights at a camera station with no predator control. Data from the corresponding camera at the baited site were also removed. In the late-dry survey in the area without predator control, one camera pair was shifted approximately 500 m down the track, to avoid structures built near the previous camera pair between surveys.

Table 3.3. Best supported models ( < 2) of predator activity on Cape York Peninsula (CYP) and Gulf Plains (GP), selected using the second -order Akaike Information Criterion (AICc).

2 CYP dingo models K AICc R 1 PC SSN POS LH 9.00 290.55 0 0.30 43.27% 2 PC SSN POS LH PC:POS 10.00 292.47 1.93 0.11 43.39% CYP feral cat models 1 PC POS LH PC:LH 7.0 1 217. 19 0 0.1 6 12.9 0% 2 PC POS LH BH PC:LH PC:BH 9.00 217.45 0.26 0.14 15.18% 3 PC POS LH BH PC:LH 8.00 219.14 1.95 0.06 13.05%

GP feral cat models 1 PC BH PC:BH 15.65 253.43 0 0.10 37.53 % 2 PC SSN BH PC:BH 17.15 253.97 0.54 0.07 39.21% 3 PC BH 17.09 254.01 0.58 0.07 39.11% 4 PC SSN BH 18.44 254.50 1.08 0.06 40.62% 5 PC 18.71 254.59 1.16 0.05 40.94% 6 PC SSN 19.80 254.94 1.52 0.05 42.18% 7 PC SSN LH BH PC:BH 17.68 255.11 1.68 0.04 39.35%

Effective degrees of freedom, K; distance from the best supported model, ; ; generalised coefficient R2. Predictors include predator control, PC; season, SSN; camera position, POS; local habitat, LH; broad habitat, BH; and intera ctions between predator control and camera position, PC:POS; local habitat, PC:LH; and broad habitat, PC:BH.

Dingoes

The proportion of survey nights recording dingoes was modelled using a beta-binomial distribution, with a variance structure including season and broad-scale cover. Of the 69 models assessed, two had substantial support ( < 2) for explaining variation in dingo visitation (Table

55 Chapter 3: Habitat use

3.3). The most important variables influencing nightly dingo visitation, from those included in the models, were season and camera position (Table 3.4, Figure 3.3a). Dingoes were recorded on more nights in the early dry season (N = 76) compared to the late dry (N = 38), and on cameras situated on tracks (N = 105) rather than 200 m off road (N = 9). Camera position had the largest effect on dingo frequency, with over 90% of dingo records on tracks. Habitat complexity at the camera station was also relatively important, with dingoes recorded more frequently at cameras with higher tree density. Predator control was less important and not as strong an explanatory variable as expected. However, it still had a negative effect on dingo activity, with dingoes recorded on more nights at sites without predator control (N = 72) compared to sites with poison baiting (N = 42). Interactions between predator control and camera position or habitat complexity were less important. In a post-hoc comparison, the effect of season was stronger at the unbaited site; dingoes were recorded on 54 nights in the early dry compared to 18 nights in the late dry (Table 3.5). At the site implementing predator control, nights recording dingoes were almost identical in the early dry (N = 22) versus the late dry season (N = 20).

Feral cats

Three binomial models of feral cat activity were included in the candidate set (Table

3.3), with camera position being the most important variable assessed (Table 3.4). Similarly to dingoes, feral cats were recorded on more nights on roads (N = 37) compared to off roads (N =

16), but the magnitude of the effect was weaker. An interaction between predator control and habitat complexity at the camera had the strongest effect on feral cat frequency (Figure 3.3b).

However, the effect was not as expected: in areas without predator control, cats were detected on more nights as habitat openness increased. Although not as important as the aforementioned variables, the interaction between predator control and broad-scale low cover also influenced cat activity. The mean proportion of nights where cats were recorded tended to increase as broad levels of low cover increased on the unbaited property, while there was a declining trend on the baited property. Season, predator control and habitat complexity as single factors were not

56 Chapter 3: Habitat use

important predictors of feral cat visitation. In particular, there was no difference in the number

of nights with feral cats recorded at the unbaited (N = 27) versus baited site (N = 27).

a. 5 4 3 2 1 0 -1 -2 -3 -4

b. 5 4 3 2 1 0 -1 -2 -3 -4

57 Chapter 3: Habitat use

Gulf Plain

I recorded 17 independent observations of dingoes on 11 nights at 10 stations, and 87 independent observations of feral cats on 78 nights at 55 stations, during 969 trap nights over the two seasons. All cameras collected data on at least one night, except for one camera at the baited site in the early dry season, where the camera failed with both lures. I could not include a local habitat explanatory variable for tree density due to collinearity with broad habitat variables. To compromise, local habitat complexity is represented solely by ground cover complexity while % tree cover in a 3 km2 area around the camera pair represented broad habitat complexity.

Table 3.4. Relative importance of variables explaining predator activity on Cape York Peninsula (CYP) and Gulf Plains (GP), drawn from model averaging. Variables with > 0.5 are considered to be more important (Ritchie et al. 2009).

w+ Explanatory variable CYP Dingo CYP Feral cat GP Feral cat predator control 0.44 0.04 0.48 season 1.00 0.27 0.50 camera position 0.77 0.76 0.28 local habitat complexity 0.74 0.15 0.32 broad habitat complexity 0.24 0.17 0.33 predator control * position 0.23 0.22 0.08 predator control * local habitat 0.18 0.68 0.08 predator control * broad habitat 0.07 0.40 0.40

Dingoes

Due to the low number of dingoes recorded at either site, I could not adequately assess dingo activity using generalised linear modelling. Dingoes were recorded on 7 nights in the unbaited area compared to 4 nights in the baited area. There appeared to be a seasonal trend, largely driven by differences in the area without predator control (Table 3.5). It also appeared that dingoes may have been more active on roads, with dingoes recorded at road cameras on 8

58 Chapter 3: Habitat use

nights, compared to 3 nights on cameras off roads. However I could not assess the strength of

these apparent effects given the low sample size, although these trends are broadly similar to

those measured on CYP.

Cape York Peninsula Gulf Plains Species Treatment Early dry Late dry Binomial test Early dry Late dry Binomial test Dingo No predator control 54 18 P < 0.001 7 0 P = 0.02 Predator control 22 20 P = 0.88 3 1 P = 0.63

Feral cat No predator control 10 17 P = 0.25 22 36 P = 0.09 Predator control 15 11 P = 0.56 9 11 P = 0.82

Feral cats

The selected binomial models were less capable of explaining cat activity on the Gulf

Plains, with seven models included in the candidate set, and the model with most support having

a relatively low model weight of 0.10 (Table 3.3). No model-averaged parameter estimates had

a relative importance greater than 0.5, suggesting they were not particularly strong predictors of

cat activity, and that the considerable explained variation may be due to the random effect

(Table 3.3).

The most important explanatory variables were season and predator control, with cats

recorded on more nights in the late dry season (N = 47) and in the area without predator control

(N = 58), compared to the early dry (N = 31) and the baited area (N = 20). I recorded cats at high

rates, on average over 12% of nights, on the unbaited property, compared to an average 4% of

nights on the baited property. The confidence interval for season encompassed zero, hence the

direction and magnitude of this variable were not very strong (Figure 3.4). There was a trend for

cats to be recorded more frequently in areas with less broad tree cover, especially in the

59 Chapter 3: Habitat use unbaited area, where cats were recorded at high rates on the almost-treeless floodplains. Camera position was far less important as a predictor of feral cat activity on the Gulf Plains compared to

Cape York Peninsula; cats were recorded on 44 nights on road cameras compared to 34 nights off road.

-1

-2

-3

-4

Discussion

Habitat composition and complexity can influence the distribution and abundance of species, providing not only resources, but also protection from predation or interference. Hence, habitat may have an important role in mediating interactions between humans and predators, and within the predator guild. Here I examined whether dingoes use habitats that provide protection from humans, and whether feral cats use habitats that minimise risky encounters with

60 Chapter 3: Habitat use dingoes. Both species used a broad range of habitats in the study areas. Rather than avoiding dingoes by steering clear of the habitat they use, or by hiding in complex vegetation, feral cats appear to use similar habitat features to dingoes, such as roads, and more open habitats in the presence of dingoes. This suggests that at the scale of this study, feral cats do not appear to use habitats to reduce encounter rates with dingoes. Further, the avoidance tactics feral cats use may depend on the habitat types available across the landscape.

Dingoes on Cape York Peninsula

Unsealed roads or tracks appeared to act as main thoroughfares for dingoes on Cape

York Peninsula. Camera position was a relatively important variable and had the strongest effect when modelling the frequency at which dingoes were recorded. Although offroad cameras were only positioned 200 m from the track or road, less than 10% of dingo records came from these stations. Roads may provide dingoes with easy access through a more complex habitat matrix to move between points of interest. They may also represent space shared by multiple individuals or packs, that communicate via scent posts or scat deposition sites at intersections or on roadsides (L. Brook, pers. obs.; Corbett 1995a; Fleming et al. 2001).

Dingoes were recorded more frequently at cameras in areas of higher tree density, but in other respects, habitat complexity was not very important. Higher activity in areas with more tree cover may occur because dingoes were recorded in open forest or rainforest, even though these habitat types were less common than more open woodland areas. Previous studies have not recorded consistent results in terms of habitat use in forest or woodlands, and results range from more records in complex habitats (Catling & Burt 1995), to a preference for more open understoreys (Robley et al. 2010a). As dingo ranges are large enough to encompass multiple vegetation types, and dingo prey preferences and hunting tactics are flexible (Corbett 1995a), they are likely to use many different habitats, depending on when and where resources are accessible. For example, in the wet-dry tropics of the Northern Territory, dingoes are restricted to forested areas in the wet season when floodplains are submerged, and mainly consume medium-sized mammals. However, dingoes move onto the floodplain in the dry season to eat

61 Chapter 3: Habitat use young geese and rodents (Corbett 1995a). Although I expected dingoes to use more complex habitats on baited properties to avoid encounters with humans, these interaction terms were not important. It may be that complex habitat did not provide additional protection from predator control, or that encounters with people were too intermittent or inconsistent to warrant avoidance of risky habitats. Predator control had a small, negative effect on the number of nights in which dingoes were recorded on Cape York Peninsula. The effects of predator control may be mitigated by dispersal into the area following baiting (Allen & Miller 2009), while in unbaited areas, dingo populations may be restrained through social regulation (Wallach et al.

2009b).

Dingoes may be less active in the late dry season, as they retreat from areas of human activity to birth and raise pups (Allen & Byrne 2008). Dingoes on Cape York Peninsula were also less active in the late dry season, although this trend was apparently driven by differences in activity at the unbaited site. In the early dry season, dingoes may be more active as they seek mating opportunities (Thomson 1992b), which appears to have occurred in the area without predator control. The reduced seasonal influence on dingo activity in the baited area may be due to predator control removing active animals during the early dry; alternatively, dingo patrolling behaviour may be dampened by control activities (Wallach et al. 2009b) or in an attempt to avoid encounters with people (Wallach et al. 2009a). As this result was not predicted, further research should examine whether the effect of predator control on seasonal differences in activity is widespread.

Feral cats on Cape York Peninsula

My expectations that feral cats would use habitats that provided the most cover as protection from dingoes, or more trees as escape routes, were not realised. To the contrary, feral cats used relatively open habitats in the unbaited areas where dingo activity was higher, while habitat complexity variables in themselves were not important. Further, feral cats were recorded more frequently on roads compared to off-road, suggesting they frequently use this habitat feature even though dingoes do as well. Our findings contrast those of previous studies where

62 Chapter 3: Habitat use cats did not use roads (Mahon et al. 1998) and were found in denser habitats (Edwards et al.

2002), but are more consistent with the findings of Molsher et al. (2005), in which cats generally used habitat relative to its availability, with some preference for grasslands within their range.

One would expect feral cats to also use a range of habitats, as they are found across

Australia, presumably in diverse landscapes (Denny 2008). However, habitat use is likely to be influenced by availability. In a study on Kangaroo Island, feral cats preferred more complex woodland habitats over paddocks in a pastoral matrix, but in bushland areas where habitat complexity was presumably higher, preferences were less consistent (Bengsen et al. 2012). All cameras at CYP were located at most 100 m from trees, and almost all were within 5 m of a tree or potential escape route. Hence, relatively open areas may still be close enough to trees for cats to retreat to should they encounter dingoes. Relatively open habitats, such as burnt areas, may provide other benefits for feral cats such as ease of movement and increased visibility, both to identify approaching dingoes and locate prey. Improved visibility and movement may benefit cats using a mobile hunting strategy (Corbett 1979). Alternatively, feral cats may avoid areas of high ground cover complexity, where vision and hearing, both important senses (Fitzgerald &

Turner 2000), may be impeded. Hence, use of open habitats in areas with higher dingo activity may potentially allow feral cats to perceive dingoes from further away, providing more time to retreat to safety. The more complex habitat features on CYP may facilitate coexistence of both predators (Janssen et al. 2007), by allowing feral cats to successfully avoid dingoes, potentially mitigating negative effects on abundance.

Feral cats on the Gulf Plains

Feral cats on the Gulf Plains did not appear to use habitat in the same way as feral cats on Cape York. Cat activity was not as strongly influenced by habitat, season or predator control in this region, as no variables were particularly strong or important predictors of camera records. Much of the explained variation in the R2 may be due to spatial variation across the landscape, included in the model as the random effect of camera pair.

63 Chapter 3: Habitat use

I recorded more frequent activity of feral cats in the unbaited area, especially in the late dry season, when I did not record any dingoes. This property may have high prey abundances to support a larger cat population, but further studies would be needed to elaborate on why feral cats were more active in this area. In contrast to feral cats on CYP, camera position was not an important explanatory variable for cat activity on the Gulf Plains. The grasslands and open habitats that dominate this landscape may mean feral cats do not need to use roads to avoid complex habitats that impede movement

At a broader spatial scale, feral cats were recorded more frequently in areas with relatively less tree cover. This trend was particularly noticeable on the unbaited property where

I recorded high levels of activity on bluebush floodplains, which had very low levels of tree cover. This suggests that feral cats did not rely on trees for protection in this area. It may be that dingo activity was so low that the benefits of using these areas, perhaps in terms of prey abundance, may outweigh any risks of being encountered in a vulnerable context. In Finland, female American mink Mustela vison swam shorter distances between island habitat patches when closer to nests of white-tailed sea eagles Haliaeetus albicilla, or when eagle observations were more frequent (Salo et al. 2008). Female mink may therefore be assessing the risk associated with swimming, which makes them vulnerable to predation, dependent on the probability of encountering an eagle, and adjusting their activity in response.

Caveats and future research directions

Here I examined whether predator activity was influenced by risks associated with certain habitats, however space use decisions of hunters are likely to be primarily focused on prey acquisition (Sih 1984; Arias-Del Razo et al. 2012). This potentially accounts for the relatively low explanatory power of the models, and the lack of important explanatory variables for cat activity on the Gulf Plains. Modelling rare species or carnivores can result in low deviance or explanatory power (May et al. 2008; Mannocci et al. in press), particularly if variables are indirect (Austin 2002). Future studies would, therefore, benefit from measurement of prey populations across and between treatment areas, to also explore the effect of prey

64 Chapter 3: Habitat use abundance on dingo and feral cat activity. These data may also reveal how habitat decisions of predators are influenced by prey abundance versus improved hunting success (Arias-Del Razo et al. 2012), and for mesopredators, predation risk.

Insufficient dingo records from the Gulf Plains prevented a comparison of activity and habitat use between dingoes in different regions, or between intraguild predators in this landscape. Similarly, I could not examine how feral cat habitat use related to trends in dingo activity in this area. Dingoes can travel long distances (Robley et al. 2010a), hence the dingoes in the area without predator control may have been affected by poison baiting on neighbouring properties.

Although I attempted to minimise the effect of roads on habitat measurements, ground cover at roadsides cameras was still lower than at off-road cameras, as was tree density on road cameras on CYP. This may have been due to removal of trees along road edges, movement of cattle along roads, road grading or other edge effects. However, Spearman correlation coefficient values were all lower than 0.4, hence camera position is unlikely to be considerably confounding the effect of habitat complexity or vice versa.

Ecological studies such as this often calculate detection probabilities, to estimate occupancy across the landscape, but I did not include detection probability, for several reasons.

I designed my study to minimise differences in detection probability, by using species-specific lures for equal periods at each camera, and by sampling in different seasons to encompass species-specific seasonal trends. The larger body size of dingoes, which may improve detectability using remote cameras, was potentially offset by their greater speed (Rowcliffe et al. 2011). Occupancy modelling can be unsuitable for assessing rare or cryptic species (Welsh et al. 2013), which often have low detection probabilities (O'Connell Jr et al. 2006). Further, results can be biased when abundance affects detection probability (Welsh et al. 2013).

Although dingo individuals were difficult to identify, feral cats could often be differentiated by variations in pelage. I frequently noted different cats visiting camera stations, sometimes on the same night, which suggests that repeat visits may be correlated with abundance. I suspect that once feral cats have investigated a camera lure and satisfied their curiosity, they are rarely

65 Chapter 3: Habitat use compelled to revisit. Ultimately, dingoes and feral cats are likely to occupy most of the landscape. In this study I was primarily interested in identifying the areas or habitat features that are used more frequently by each predator rather than their general distributions. Occupancy modelling cannot provide this level of information (Welsh et al. 2013).

Are cats using habitat features to avoid dingoes?

In this study, feral cats did not appear to use complex habitats that provided protection, nor did they appear to exhibit niche partitioning with dingoes. Spatial or habitat partitioning between predators has been demonstrated in some contexts (Gosselink et al. 2003; Kozlowski et al. 2008; Wirsing & Heithaus 2009), however it may not always represent the best mechanism for mesopredators to avoid encounters, nor a viable trade-off in terms of reduced access to resources. Prey may be more abundant in areas or habitats used by an apex predator, hence when sympatric predators overlap in diet, mesopredators may use a range of avoidance tactics to reduce risk of harm while minimising habitat partitioning and maintain their access to resource- rich areas. In a telemetry study on sympatric predators in South Africa, Vanak et al. (2013) found that lions Panthera leo overlapped with the prey of subordinate predators, particularly in poor seasons. Rather than shifting to areas with fewer prey, leopards P. pardus, avoided areas recently used by lions. African wild dogs Lycaon pictus however, who face strong intraguild interference, avoided spatial overlap with larger predators, particularly in the dry season when kleptoparasitism increased. In Botswana, cheetahs used similar habitat to lions, but avoided interactions by maintaining a safe distance when using the same landscape (Broekhuis et al.

2013). Hence feral cats may be able to minimise risky encounters with dingoes, while using the same areas or habitats, by using areas at different times (Chapter 2), by using more fine-scale avoidance strategies, or by using more open habitats that provide advance visual warning of approaching dingoes.

Heithaus and Dill (2006), in a study in Shark Bay, Western Australia, found that when tiger sharks Galeocerdo cuvier were present, bottlenose dolphins Tursiops aduncus avoided resource-rich shallow areas, which were further from the protection of deep water. But when

66 Chapter 3: Habitat use they did use shallow areas, dolphins stuck to edge habitats where they had a greater chance of escape, despite these edges having a high density of sharks. As feral cats are unlikely to outrun a dingo, early detection could provide a better means of minimising risk (Lima & Dill 1990), providing more time to escape to nearby cover or trees. Further studies in habitats at a greater distance from tree cover would clarify whether these areas are considered more risky by feral cats, due to their lower probability of escape, and are therefore used less frequently. Similarly, higher levels of dingo activity on the Gulf Plains could enhance the risk associated with open habitats, increasing the likelihood of a lethal outcome if feral cats were encountered, and reducing the frequency at which feral cats use these habitats.

Conclusion

This study has confirmed the adaptability of both dingoes and feral cats to a range of habitats in northern Australia. The use of similar habitat features, such as roads, suggests that feral cats may use other mechanisms such as temporal avoidance to minimise encounters. Feral cats appear to use relatively open habitats, either for ease of movement, for improved hunting success, or to improve detection of both predators and prey. Rather than simply avoiding detection, feral cats may use open habitats to detect dingoes first and escape, reducing the risk of capture. In more open landscapes, higher levels of dingo activity may make use of open habitats more risky for feral cats.

Author contributions

LB conceived the study, CJ and ER assisted with design, LB collected field data, analysed the data and wrote the paper, DN assisted with analysis, CJ, LS and ER assisted with editing.

67 Chapter 3: Habitat use

Chapter 4 Seasonal space use of dingoes and feral cats: differences in patterns of intensity in northern Australian savanna 3

Abstract

Patterns and dimensions of predator space use vary through time, and may influence interactions with lower trophic guilds. The dingo, the largest mammalian carnivore in Australian ecosystems, is a wide-ranging coursing predator consuming both small- and large- bodied prey. Co-occurring with dingoes are feral cats, ubiquitous ambush mesopredators. Few studies have explored the space use patterns of these species in sympatry. I used GPS collars to collect movement data on dingoes (N = 10) and feral cats (N = 8) across two dry seasons in the Kimberley, in northern Western Australia. I described seasonal space use and delineated core isopleths using both traditional kernel density estimation, and movement kernel density estimation, which interpolates the path between consecutive fixes. I also examined how individuals shifted space use during the study period. Dingoes used larger areas than feral cats, but they concentrated their activity on more intense core areas. Feral cats showed more contiguous space use, with activity spread more evenly over core areas that were larger relative to the total area used. Both species left their core areas mainly during crepuscular periods. Dingoes shifted space use over time more than feral cats on broad- and core-area scales, but both predators had low levels of temporal overlap in areas of most concentrated use. Results suggest that predation pressure imposed by each predator differs across the landscape: while small prey may face more consistent pressure within feral cat core areas, the gradient of dingo activity may create pockets of relative safety for large herbivores, small prey and mesopredators, in addition to areas of high risk.

3 Brook, L. A., Johnson, C. N., McGregor, H., Schwarzkopf, L., Ritchie, E. G. & Legge, S. Seasonal space use of dingoes and feral cats: differences in patterns of intensity in northern Australian savanna. To be submitted to Journal of Mammalogy.

69 Chapter 4: Seasonal space use

Introduction

Understanding patterns and dimensions of space use provides valuable information on predator ecology and interactions within predator guilds. Animals use components of the landscape to meet their biological needs, such as habitats that provide shelter for reproduction

(Walters et al. 1988), protection from predators (Lima & Dill 1990) or that improve hunting success (Laundré et al. 2009), and they take advantage of conspicuous features such as roads or scent posts to demarcate a territory or communicate with conspecifics (Harden 1985; Purcell

2010). Some areas may be used rarely, while others, such as prey-rich locations or thoroughfares, are used intensively or repeatedly. While predators that hunt using ambush may have more stable or consistent space use patterns, coursing predators are more mobile, with space use that changes over time (Preisser et al. 2007; Thaker et al. 2011). Patterns of space use can thus reflect hunting behaviours and habitat requirements of a predator, which may influence its relationship with other predators and provide insight into their potential impacts on sympatric prey. Predation risk also varies spatially, depending on predator activity and prey vulnerability, creating a ‘landscape of fear’, in response to which prey adjust their behaviour (Laundré et al.

2001). Greater variation in intensity of space use by predators may result in areas of relative safety offset by areas of frequent predator use. These areas of low predator use may be valuable to prey populations, allowing them to recover from predation. Space-use estimates are also valuable for management of predators, to define the scale and quality of areas required to sustain vulnerable species (e.g. Linnell et al. 2001) or control pest populations (Thomson

1992d; Moller & Alterio 1999).

Burt (1943) described the home range (HR) as the "area traversed by the individual in its normal activities of food gathering, mating and caring for young," excluding exploratory movements outside this stable area. However, he also conceded that home ranges shift over time. It may thus be difficult to discriminate between exploratory movements and ventures into a known, but rarely-visited, part of the home range, using this definition (Kie et al. 2010).

Recently, the home range concept has become more dynamic, representing the spatial extent

70 Chapter 4: Seasonal space use over which an individual interacts with its environment during a defined period (Börger et al.

2008; Fieberg & Börger 2012), and reflecting the "cognitive map" of an animal, shaped by its behaviour and understanding of the landscape (Powell & Mitchell 2012).

Animals generally do not use space randomly, but concentrate activity in areas that contain resources necessary for feeding and shelter (Ewer 1968). Information on core areas can reveal which resources and landscape features are most important to each individual (Samuel et al. 1985). Space use can be delineated into core areas, where activity is focussed, and less- frequently used peripheral areas (Seaman & Powell 1990). Core areas may occupy a small or large proportion of the overall home range, and the proportion of time spent in a core area may also fluctuate, depending on intensity of space use (Vander Wal & Rodgers 2012). Therefore, individuals and species will use core areas at differing levels of intensity. The intensity with which activity is concentrated in core areas, and the frequency and duration of visits to core areas, could therefore influence direct and indirect interactions between predators and their prey.

The size and shape of estimated home ranges and core areas will differ depending on the methods applied by researchers to quantify them (Samuel et al. 1985). The range of tools available has expanded recently, with a shift from methods initially applied to smaller datasets primarily collected using VHF telemetry, such as minimum convex polygons (MCP) (Mohr

1947), to those that maximise the value of high-frequency data collected using Global

Positioning Systems (GPS), such as movement-based kernel density estimation (MKDE)

(Benhamou 2011). Kernel density estimation (KDE) methods (Worton 1989), which produce a utilisation distribution (UD) or probabilistic model of space use, provide more accurate estimates of the UD than methods which rely on convex polygons (Lichti & Swihart 2011;

Cumming & Cornélis 2012: although see Getz et al. 2007). However, kernel-based estimates can vary broadly depending on the choice of input variables such as bandwidth (Gitzen et al.

2006), which will influence the level of error (Seaman & Powell 1996; Fieberg 2007).

Ultimately, all methods can only approximate the actual space use of an animal (Powell &

Mitchell 2012).

71 Chapter 4: Seasonal space use

The dingo Canis lupus dingo arrived in Australia between 3,500-5,000 years ago

(Savolainen et al. 2004) and is currently the continent's largest mammalian predator. Dingoes are found in a diversity of habitats across Australia and consume a range of prey using variable hunting strategies (Thomson 1992c; Corbett 1995a). Differences in resource availability and social structure are likely to be associated with considerable variation in space use. Home range estimates vary from 3,420 ha (90% KDE) (Purcell 2010) and 9,923 ha (95% KDE) in southeast

Australian forests (Claridge et al. 2009) to up to 99,900 ha (85% KDE) in the arid Tanami desert (Newsome et al. 2013b). Several studies have reported animals, particularly sub-adults, making long-distance movements (e.g. 230 km in 9 days, Robley et al. 2010a). Core areas were estimated at 590 ha (50% KDE) in eastern Australian forests (Purcell 2010), but core use has not previously been investigated in tropical, northern Australia.

The feral cat Felis catus had spread across northern Australia by the late 1800s, following its introduction by British settlers (Abbott 2002). Feral cats consume a wide range of native fauna (Kutt 2011), and are suspected to have contributed to the declines of many native mammals (Dickman 1996; Smith & Quin 1996; Fisher et al. in press). Feral cats exhibit marked variation in density, sociality and space use (Macdonald et al. 2010). Home ranges may be quite small in resource-rich areas such as rubbish tips where cats reach high densities (e.g. Hutchings

2003), but generally expand in natural environments, where they are reliant upon hunting.

Within and between Australian studies, feral cats have demonstrated great variation in home range size, from 50-13,200 ha (mean 1,741 ha) (95% MCP) in the arid centre (Moseby et al.

2009b), to 248 ha (95% MCP) in pastoral woodlands of eastern Australia (Molsher et al. 2005), and 976 ha (95% MCP) on semi-desert Dirk Hartog Is (Hilmer 2010). Core areas are also highly variable, from 560 ha (60% KDE) in arid Australia (Moseby et al. 2009b), to 64 ha and 73 ha

(50% MCP) in woodlands (Molsher et al. 2005) and tall forests (Buckmaster 2011) respectively.

Feral cats may also use temporary 'focal points' or den sites rather than occupying a permanent den site or core area, potentially due to territory patrol or patchy resources (Edwards et al. 2001;

Moseby et al. 2009b).

72 Chapter 4: Seasonal space use

Here, I characterise and compare the seasonal space use of sympatric apex- and mesopredators, dingoes and feral cats, and discuss how variability in intensity of space use may influence each predator's relationship with prey communities. I achieve this by estimating the size, layout and intensity of use of the UD and core areas, and exploring differences in space use between the two predator species and over time. Although I use home-range estimation methods, I consider the results 'seasonal space use' estimates (SSUs) rather than home ranges, because this study was conducted in the dry season only, and it is quite possible that patterns of space use differed in the wet season (Otis & White 1999; Fieberg 2007).

Methods

Study area

I collected data on dingoes and feral cats on two adjacent properties owned and managed by the Australian Wildlife Conservancy (AWC) for biodiversity conservation, in the central Kimberley bioregion of northern Western Australia (WA) (Figure 4.1). Mornington

(17.5°S, 126.1°E) includes 313,273 hectares of mainly tropical savanna, and includes a 40,300- ha fenced, introduced herbivore-free area. Marion Downs (17.1°S, 126.8°E) is a 249,862-ha property abutting the northern boundary of Mornington and is lightly stocked with cattle.

On Marion Downs, I caught and collared dingoes in the Siddins Valley area (-17.1°S,

126.8°E) in 2011 (Figure 4.1). Siddins Valley is traversed by Siddins Creek and its tributaries, and bordered to the northeast by sandstone escarpments and hills, and to the south by a ridge of undulating sandstone hills. Habitat consisted mainly of savanna woodlands, grasslands, riparian forest and sandstone escarpment, with red sandy soils and cracking black clays. Woodlands were dominated by cabbage gum Eucalyptus grandifolia, Darwin box E. tectifica, and bloodwoods Corymbia spp. interspersed with beefwood Grevillea striata, bauhinia Bauhinia cunninghamii, terminalia Terminalia spp. and boab Adansonia gregorii. Acacia spp. and tea tree

Melaleuca minutifolia formed a closed to open shrub layer over grasslands dominated by

73 Chapter 4: Seasonal space use whitegrass Sehima nervosum, kangaroo grass Themeda triandra, black speargrass Heteropogon contortus and annual native sorghum Sorghum stipoideum or spinifex Triodia spp.

Figure 4.1. The central Kimberley bioregion in north -western Australia (top left), showing the boundaries of the Mornington and Marion Downs Wildlife Sanctua ries and major vegetation groups (National Vegetation Information System) (bottom left). Trapping zones are highlighted in blue within each study area.

At Mornington, we caught and collared both dingoes and feral cats within the introduced herbivore-free area (17.5°S, 126.2°E), in a landscape interspersed with rocky hills and escarpments including Mt Brennan, and bounded by the Fitzroy River to the south (Figure

74 Chapter 4: Seasonal space use

4.1). The Annie and Roy Creeks bisect the area north to south and their tributaries formed a network of dry riverbeds of rock, sand or gravel during the study period. Habitat at the

Mornington site was mainly grasslands and woodlands similar to the Marion Downs site. The area features considerable riparian vegetation around Roy Creek, Annie Creek and the Bluebush wetlands.

Trapping and processing

Dingoes and cats were trapped using soft-jaw traps (Oneida Victor® Soft Catch® traps, size 1.5 for feral cats, and size 3 for dingoes, Oneida Victor Inc. Ltd., Euclid, USA; coyote

Jake™ traps for dingoes, J.C. Conner Ltd., Newcomerstown, OH, USA) set along transects of frequent predator movement such as ungraded roads or dry riverbeds. Offset pairs or individual traps were baited with species-specific visual or auditory lures (Westcare Industries,

Bassendean, Australia) plus cat urine for feral cats, and fermented egg spray (Feralmone®,

Animal Control Technologies, Somerton, Australia) and meat for dingoes. Dingo traps were also set at cattle carcasses. We reduced the risk of bycatch by opening traps in the late afternoon and closing within three hours of dawn, and by building bowers of wood and brush around traps, which also facilitated predator movement toward the lure. We usually checked traps at

Siddins Valley near midnight, either directly, or by checking the status of trap alerts (Sirtrack,

Hawkes Bay, New Zealand, or custom-made), and again at dawn. Trap alerts were fitted with

VHF transmitters that increased in pulse rate when the trap was disturbed. At Mornington, traps were checked every 4-6 hours and most traps were fitted with trap alerts.

Trapped dingoes were sedated with either an intramuscular (IM) injection of medetomidine hydrochloride (Domitor ®, Pfizer Australia, West Ryde, Australia) at 0.05-0.08 mg/kg bodyweight or zolezapan/tiletamine (Zoletil ®, Virbac Australia, Milperra, Australia) at

5 mg/kg bodyweight. Upon sedation, we immobilised the dingo on a restraint board and fitted a muzzle to prevent potential injury or escape in the case of spontaneous arousal. Trapped feral cats were sedated with an IM injection of medetomidine hydrochloride at 0.04-0.1 mg/kg bodyweight, or zolezapan/tiletamine at 5 mg/kg bodyweight. The trapped foot was soaked in

75 Chapter 4: Seasonal space use chlorhexidine and massaged and an antiflammatory painkiller (Metacam®, Boehringer

Ingelheim Vetmedica GmbH, Germany) was injected subcutaneously at 0.2 mg/kg bodyweight to reduce the risk of necrosis (L. Allen, pers. comm.) (2011 animals only). We inspected the animals for any injuries, either pre-existing or incurred from the trap. Any wounds were treated with 100% ethanol, chlorhexidine disinfectant and/or betadine and an antiseptic and insecticide cream if necessary. Dingoes trapped in 2011 also received an intramuscular antibiotic

(Betamox®, Norbrook Laboratories, Newry, Ireland) at 15 mg/kg bodyweight to prevent illness associated with capture trauma. Where possible, we weighed and measured each animal, estimated age and noted condition and reproductive status.

Due to low trap success for feral cats, we also used a dog trained to locate and immobilise (e.g. by flushing the cat up a tree) feral cats in the field. Feral cats caught using dogs were either not sedated and restrained by hand in a canvas bag, or were given an IM injection of zolezapan/tiletamine at the aforementioned dose administered by hand or using an X-caliber dart rifle (Pneu-Dart, Williamsport, USA). We first inspected the cat for any injuries sustained during the chase and these were treated with chlorhexidine and/or betadine and an antiseptic was applied if necessary.

Dingoes were fitted with a GPS radio-collar weighing 280 g (Quantum 4000 LS26500,

Telemetry Solutions, Concorde USA) equipped with a drop-off mechanism and both VHF and

GPS capabilities. Feral cats were fitted with a smaller collar weighing approximately 100 g

(Quantum 4000 LS17500), with the same specifications but no drop-off mechanism.

Animals were released at the site of capture once processing was complete. Those sedated with medetomidine received an intramuscular dose of atipamezole (Antisedan®, Pfizer

Australia, West Ryde Australia) at 0.25-0.40 mg/kg (dingo) or 0.10-0.25 mg/kg (cat) bodyweight to reverse the sedative. The animal was monitored as it recovered and left the area.

We collected data either via remote download, recapture of feral cats using the trained dog, or by retrieving dingo collars once they had dropped off as programmed. Data used in this chapter were equally sampled across the diel period. Collar schedules were set to collect simultaneous fixes (within 3 min) of both species at high frequency. In 2011, collars recorded

76 Chapter 4: Seasonal space use fixes on a four-day cycle, starting at midday and ending at 11:45, with 48 hours of data collection (a "burst") followed by 48 hours of no attempted fixes. In the first 24 hours, dingo collars recorded fixes every 5 min, while in the second, the fix rate slowed to every 15 min.

Feral cat collars recorded fixes every 15 min for the entire burst. In 2012, the cycle was extended to seven days, including two 48-hour bursts of data collection, beginning and ending at midday, separated by one or two days of no fixes. In 2012, feral cat collars partially followed this schedule, but occasionally data from some bursts were collected at irregular schedules. In these cases, data from irregular bursts were restricted to one 24-hour period from 00:00-23:45.

Data screening

GPS data require assessment and (potentially) screening prior to analysis, to identify and remove imprecise and inaccurate fixes (Frair et al. 2010), and movements which may not reflect normal animal behaviour (Hurford 2009). I assessed the trade-off between increased precision and data loss using a series of screening methods to determine the most appropriate methods for my data (Appendix C), and then applied the optimal methods.

First I applied the non-movement method of Bjørneraas et al. (2010), which removes fixes representing unrealistic movements for the specific study species, such as movements which exceed the maximum possible distance travelled, or 'spikes': single fixes set apart from the general trajectory with a small inner angle. This method selectively removes spurious fixes with very little data loss (Bjørneraas et al. 2010). In the first round, fixes at an impossible speed

( ) from the median position of a moving window of 21 fixes, then those at a maximum speed

( ) from the mean position, were identified. Speed values were obtained from the literature, with set to 60,000 m h-1 for dingoes (Lycaon pictus gallop (Creel & Creel 2002)) and 7,200 m h-1 for feral cats (2 m s-1 maximum running trot (Bishop et al. 2008)). Maximum speed was set to 15,948 m h-1 (C. lupus familiaris gallop speed (Deban et al. 2012)) for dingoes and 3,600 m h-1 (1 m s-1, approximate fastest walk speed (Bishop et al. 2008)) for feral cats. I visually screened the first round errors, as regular fixes were occasionally misidentified. In the second round, I removed all spikes ≥ 100 m from surrounding fixes by specifying the travel speed ( )

77 Chapter 4: Seasonal space use and the turning angle ( ), where cos( ) was less than -0.97, equal to a turning angle of 166-194°

(Bjørneraas et al. 2010). Some additional spike errors, highlighted once more outlying fixes were omitted, were subsequently removed.

Finally, I removed fixes that recorded altitude values ± 100 m from reference elevation data recorded in a 1-second (approx. 30 m) Digital Elevation Model (Source: Geoscience

Australia). Data collected within 24 h of release, 1 h of remote download or 1.5 h of recapture using dogs, were also discarded to reduce effects of collar acclimation or human interference on behaviour (Dennis & Shah 2012).

Autocorrelation

Traditional kernel density methods assume that sequential data are independent and not autocorrelated (van Winkle 1975; Legendre 1993). This assumption is easily violated when data are collected at high frequency. However, animal movement is inherently a dependent process.

We may also expect that high-frequency, dependent data can better describe the behavioural decisions that underpin space use (de Solla et al. 1999). Autocorrelation can be managed by a) using a statistical approach that accommodates autocorrelation (Legendre 1993), b) subsampling data until independence is reached, incurring data losses, or c) collecting data at regular time intervals over a sufficient timeframe, using a representative sampling design rather than unbalanced bursts, which may reduce bias more than eliminating autocorrelation (de Solla et al.

1999; Otis & White 1999). I sampled data using a balanced burst design, by sampling all time periods equally, on a schedule (24-48 h) that should be sufficient to encompass any cyclical behaviours.

I examined positive autocorrelation using Schoener's ratio in the adehabitat package

(Swihart & Slade 1985; Calenge 2006). Schoener's ratio compares the mean squared distance between consecutive fixes to the mean squared distance between fixes and the barycentre, with a value of 2 indicating independence. These analyses were run for each burst of each collared animal with data at 15-min intervals.

78 Chapter 4: Seasonal space use

Analysis of space use

I used two different home-range methods to characterise the utilisation distribution: the classic kernel density estimation method (KDE) to provide comparisons with previous studies, and the movement-based kernel density estimation method (MKDE), which is designed for autocorrelated data. KDE generally uses a larger bandwidth, resulting in a smoother contour with less variability but increased bias (Fieberg 2007). KDE may provide a more generalised snapshot of space use beyond the study period, and may represent the broader area 'known' to the animal. MKDE uses a smaller, species-specific bandwidth that provides a more explicit description of space use during data collection. The model assumes an animal moves between fixes following a biased random walk, and creates a bridge of interpolated points along this path. The bandwidth is largest at the midpoint between GPS fixes, where location uncertainty peaks (Benhamou 2011). MKDE estimates the movement path of an animal, and can identify whether areas were traversed. The MKDE utilisation distribution can be partitioned into distributions of intensity (residence time per visit) and recursion (frequency of repeat visits) to identify areas of prolonged or frequent activity (Benhamou & Riotte-Lambert 2012). MKDE is a parsimonious method, in that it minimises type-I errors from oversmoothing (Cumming &

Cornélis 2012). I considered the SSU for both KDE and MKDE to be delineated at the 95% isopleth, to be consistent with the general trend in home range studies (Laver & Kelly 2008).

Distributions were calculated over a grid twice the range of the data. Cell size

(resolution) was set to 40 m for all analyses, which encompasses the mean 95% quantile of error between screened consecutive fixes (pooled: 19.82 ± 3.01 m), calculated in stationary tests. The mean 95% quantile of error measured from feral cat collars was higher than that for dingoes

(feral cat: 30.17 ± 1.37 m; dingo: 17.44 ± 3.37 m; Welch two-sample t-test: t = 4.07, d.f. =

13.32, P = 0.001), but may be associated with a lower sample size (cat N = 3; dingo N = 13). I used distances between consecutive fixes to estimate error associated with the movement process, rather than calculating distance from the true GPS position, which estimates location error around the fix itself (Cornélis et al. 2011). Movement error is more appropriate for use with MKDE, which characterises movement between fixes to estimate space use. It was also a

79 Chapter 4: Seasonal space use more conservative measure, being almost twice as large as location error (distance from the median). I did not consider the Fitzroy River, which lies to the south of the Mornington study area, a physical boundary to movement, as at least one animal was recorded crossing the river.

Kernel density estimation

I estimated the fixed kernel density, using a bivariate plug-in bandwidth matrix. Plug-in bandwidth selection estimates a pilot bandwidth, in this case that which minimises the sum of asymptotic mean square error (Duong & Hazelton 2003), and this estimate is 'plugged in' to the formula for optimal bandwidth that minimises error (Wand & Jones 1995). Plug-in methods find an appropriate balance between bias and variance (Jones et al. 1996) and are recommended for species that range widely but are not central-place foragers (Gitzen et al. 2006). To determine whether data were sufficient to adequately describe the seasonal space use, I used asymptote analysis (Laver & Kelly 2008), adding data sequentially (Harris et al. 1990) at 5% intervals from 10-100% of the dataset. I considered datasets to be sufficient when the majority of SSUs using 75-100% of the data were within 10% of the estimate using all data.

Bandwidth and kernel probabilities were estimated in the ks package, then the isopleths and SSU areas were estimated using adehabitatHR. The density probabilities of the ks distribution often amounted to less than 1 (range: 0.981-0.999), and adehabitatHR appeared to compensate for the discrepancy by slightly extending the utilisation distribution at higher isopleths. There was a strong linear correlation between the probability sum and the % difference between the KDE 95% isopleth calculated in adehabitatHR and the area estimated from grid cells with probability of use (Spearman's rho = 0.81, P < 0.001).

Movement-based kernel density estimation

MKDE benefits from autocorrelated data, hence all available 15-min interval fixes were used in this analysis. Although MKDE can filter out short movements to focus on periods when an animal is active, I included all fixes in the utilisation distribution, as feral cats may be actively hunting at slow speeds or when stationary (Benhamou 2011). Asymptote analysis was repeated using the methods outlined above.

80 Chapter 4: Seasonal space use

MKDE requires a range of input parameters specific to the GPS capabilities of the collar and study species:

 GPS measurement error (Lmin) was set at 20 m to reflect the 95th quantile of

distances between consecutive fixes from stationary collar tests.

 The diffusion coefficient (D) was estimated using the plug-in method, and ranged

from 1.04-3.45 m2 s-1 for cats and 12.13-46.67 m2 s-1 for dingoes.

 The interpolation time interval between recorded fixes (Tau τ) was set to 180 sec

for feral cats, creating 4 interpolated fixes between consecutive actual fixes, and to

100 sec for dingoes, creating 8 interpolated fixes between consecutive actual fixes.

 The minimum bandwidth (hmin), applied at recorded fixes, needs to exceed the

standard deviation of GPS errors, but also include variability related to the animal's

behaviour and mobility, which influence the range of its potential position within

the local habitat (Benhamou 2011). Minimum bandwidth was set to 50 m for feral

cats and 100 m for dingoes. The distance between interpolated fixes was less than

two hmin for 99.93% and 99.95% of cat and dingo fixes respectively (Benhamou &

Cornélis 2010). Maximum bandwidth (hmax) was calculated as

and reached 201.55 ± 9.46 m for dingoes and 72.32 ±

2.58 m for feral cats.

 MKDE only interpolates points between fixes recorded within the maximum time

interval (Tmax), hence fixes within this interval should be correlated (Benhamou &

Cornélis 2010). Tmax was set to 45 min for both species.

 The maximum time (Maxtime) an animal can spend outside an area before it is

considered to have left was set to 35 min for both species.

 The radius of the area within which to calculate visit duration and frequency was set

to 3 hmin (Benhamou & Riotte-Lambert 2012).

81 Chapter 4: Seasonal space use

Core areas and intensity of use

Individuals use space differently, hence an arbitrary core contour may under- or overestimate the core area, or falsely identify a core when space use is random (Powell 2012). Core areas were identified separately for each animal and each space use metric by plotting area as a proportion of the total space use (95% isopleth) associated with isopleth volumes (IV) up to

95% at fine-scale intervals, similar to Vander Wal & Rodgers (2012).

82 Chapter 4: Seasonal space use

I scaled IV from 0-1 to match the scale of % space use. The core was delineated by identifying the point on the kernel area curve where slope = 1, at the maximum distance between the animal's recorded space use and random use (straight line with slope = 1) (Figure

4.2). Core isopleths were then identified by back-transforming the scaled IV to the original range. This method produced core areas that, as a percentage of the 95% isopleth, were within

1% of those derived by plotting space use area against maximum probability of use, in all but one case (Seaman & Powell 1990; Powell 2012) (Figure 4.3).

Figure 4.3. Core area as a proportion of the space use area (95% contour), for two core delineation methods (maximum probability method of Seaman & Powell 1990) and isopleth volume method used here) and four d istribution metrics. All Spearman's rank rho coefficients were significant (P < 0.005).

83 Chapter 4: Seasonal space use

I used an index of relative intensity of use (% use or core isopleth value/ % home range in core area) (Samuel et al. 1985; Vander Wal & Rodgers 2012) to confirm that space use differed from random, and to determine whether species use their core areas at different intensities relative to the 95% isopleth. A value greater than 1 indicates non-random use.

I then examined the times dingoes and feral cats used their core areas during the study period, using MKD estimates and 15-min data. I calculated the proportion of fixes recorded in the core area at hourly intervals for each animal that reached asymptote for MKDE, and averaged the percentage for each species.

Site fidelity

To compare variation in space use over time, I used the conditional Bhattacharyya’s affinity (BA) index, which estimates the similarity of different utilisation distributions (Fieberg

& Kochanny 2005), and which varies from 0 (no overlap in contours) to 1 (identical utilisation distribution contours). I divided the dataset for each animal into fortnightly periods, with any remainder greater than a week constituting an additional period. I estimated the utilisation distribution for each fortnightly period using the fixed kernel and plug-in bandwidth. I then calculated the BA index for each dyad of successive periods at the 95 % isopleth, and at the core isopleth calculated from the full utilisation distribution. I also estimated overlap in particularly high-use areas at the 20% isopleth, previously used to identify feral cat focal points (Moseby et al. 2009b). BA indices were averaged within individuals prior to analysis. I also compared site fidelity between years at the 95% and core isopleths, using data from the feral cat and dingo we collared in multiple years. Core areas were delineated by taking the mean of the core isopleths for each year.

Analyses were mostly run in R version 3.0.1 (R Development Core Team 2013). Plug- in bandwidths were calculated in ks (Duong 2012), and all other HR analyses including isopleth calculation and core analysis were run using the adehabitatLT and adehabitatHR packages

(Calenge 2006). Seasonal space use estimates were visualised using R or ArcGIS 10.0 (ESRI

2010). To compare space use estimates from different groups, I used either Welch's t tests, after

84 Chapter 4: Seasonal space use confirming normality using the Shapiro-Wilk test, or Wilcoxon Mann-Whitney rank sum tests with exact p-value calculation, using the coin package (Hothorn et al. 2008). I ensured similar variance using the Brown-Forsythe tests in the lawstat package (Gastwirth et al. 2013). If necessary, I took the natural log of estimates to approximate normality. Results are presented as mean ± standard error unless specified otherwise.

Results

We caught sixteen dingoes and twelve cats over two years. Due to collar breakdown or failure to relocate the animal, we collected data on six dingoes (2 ♀, 4 ♂) in Siddins Valley and four dingoes (2 ♀, 2 ♂) and four cats (4 ♂) at Mornington in 2011, and five dingoes (4 ♀, 1 ♂) and five cats (1 ♀, 4 ♂) at Mornington in 2012. One dingo and one cat were collared in both years, and for most space use analyses, their data were combined across years. Dingoes were generally caught in June or July and collars collected data until August to October, depending on collar failure. Feral cats were caught from May to September, with data collected until

August to October. Fix rates were generally high: 1.73 ± 1.47% and 2.30 ± 0.65% of original dingo and feral cat datasets consisting of zeros, and screening removing on average 1.45 ±

0.13% and 2.01 ± 0.36% of dingo and feral cat fixes respectively (Table 4.1, Appendix D.1).

Table 4.1. Fix rates for dingo Canis lupus dingo and feral cat Felis catus data (mean ± se) following removal of zeros and errors.

Dingo Feral cat No. animals a 14 8 No. days of high-freq. data b 34.36 ± 4.17 57 ± 6.65 No. high-freq. fixes (screened) c 6,342 ± 773 No. 15-min fixes (screened) 3,210 ± 394 5,197 ± 582 3D fix rate (screened) 98.52 ± 0.46% 92.36 ± 2.51% Fix rate (screened) 96.81 ± 1.53% 95.69 ± 0.92%

a One dingo and one cat caught in both years. b Mean number of operating days, not including days in between bursts. c All high-frequency fixes includes those at 5-min frequency for dingoes.

85 Chapter 4: Seasonal space use

Autocorrelation

When data were subsampled to a regular 15-min interval, all dingo (N = 248) and cat (N

= 244) bursts contained significant positive autocorrelation, with a pooled mean Schoener's ratio of 0.07 ± 0.01 and 0.02 ± 0.001 respectively. These data were used for both analyses.

Kernel density estimation

The KD estimates of 10 dingoes and all cats reached an asymptote, and were used in analysis (Appendix D.2, Appendix E). Some individuals exhibited a declining, rather than increasing, asymptote towards the 95% isopleth. Of the dingoes that did not reach an asymptote, one individual (DS_M3) could be described as transient, covering 71,863 ha (95% isopleth) with a large proportion of what appeared to be exploratory movements, rather than repeated use of a defined area (Appendix D.2, Appendix D.3, Appendix F.2a-c). One dingo (DS_M1) was also excluded due to early collar failure.

Dingoes Feral cats No. individual s 10 8

No. fixes 3,216 ± 295 5,196.88 ± 581.74 (1,530-4,173) (2,094-7,702)

95% isopleth area 2,698.42 ± 781.72 736.82 ± 147.54 (ha) (715.52-9,405.28) (127.52-1,321.6)

Core isopleth 70.34 ± 1.03% 66.94 ± 0.77% value (65.40-75.80%) (63.60-69.40%)

Core area (ha) 715.47 ± 236.51 232.24 ± 53.42 (86.56-2,723.36) (41.92-479.04)

Core as % of SSU 24.39 ± 2.13% 30.55 ± 1.63% (12.10-32.71%) (22.05-36.25%)

Intensity index 3.20 ± 0.41 2.25 ± 0.15 (2.00-6.03) (1.80-3.14)

86 Chapter 4: Seasonal space use a. b.

The mean male dingo 95% contour (4,128.88 ± 1,764.76 ha, N = 4) was more than

twice as large as that of females (1,744.77 ± 364.39 ha, N = 6), but medians were closer (male:

2,545.76 ha, female: 1,577.76 ha) and the difference was not significant (t = -1.84, d.f. = 5.10, P

= 0.12). There was no significant difference in 95% isopleth estimates between years for

dingoes (t = 0.76, d.f. = 7.57, P = 0.47) or cats (t = -0.31, d.f. = 4.26, P = 0.77).

c Chapter 4: Seasonal space use

As expected, dingoes used more space than feral cats at the 95% isopleth (t = -3.64, d.f.

= 14.17, P = 0.003). Core isopleths occurred at a higher probability for dingoes compared to feral cats (t = -2.65, d.f. = 15.61, P = 0.02); however the core areas of feral cats occupied a larger percentage of the 95% isopleth (t = 2.29, d.f. = 15.69, P = 0.04) (Table 4.2).

Feral cats appeared to use space in a more compact and uniform manner compared to dingoes (Figures 4.4 & 4.5): their 95% isopleths were more contiguous and broke into fewer

88 Chapter 4: Seasonal space use

(7.13 ± 1.37) areas, whereas dingoes appeared to scatter their use of space in multiple (29.1 ±

11.38) areas spread over a larger scale (t = -2.61, d.f. = 14.5, P = 0.02). However, there was no significant difference in number of core areas between species (dingoes: 15.2 ± 2.09, cats: 11.5

± 1.45, Wilcoxon Mann-Whitney Rank Sum test; Z = -0.72, P = 0.50), due to the same median

(13.5), although numbers of core areas for each species were skewed in opposing directions.

Dingoes Feral cats No. Individuals 8 8

UD 95% isopleth area 1,649.22 ± 150.71 547.98 ± 91.44 (ha) (1,083.36-2,254.88) (132.96-960.64)

Core isopleth 71.84 ± 1.05% 66.73 ± 0.52% (68.00-76.70%) (64.90-68.80%)

Core area (ha) 373.44 ± 56.10 151.66 ± 24.83 (132.80-605.28) (45.44-245.76)

Core as % of SSU 22.03 ± 2.29% 28.24 ± 1.26% (12.26-30.70%) (22.87-34.18%)

Core intensity 3.58 ± 0.45 2.40 ± 0.12 (2.23-5.73) (1.95-3.01)

ID Core isopleth 68.79 ± 0.66% 62.44 ± 0.83% (66.40-71.50%) (59.10-66.60%)

Core as % of SSU 19.85 ± 1.31% 28.10 ± 1.10% (15.16-26.53%) (23.06-33.14%)

Core intensity 3.58 ± 0.25 2.25 ± 0.11 (2.50-4.55) (1.78-2.74)

RD Core isopleth 62.88 ± 0.65% 62.68 ± 1.09% (59.00-65.50%) (59.60-68.40%)

Core as % of SSU 29.57 ± 1.70% 35.84 ± 0.77% (23.62-38.39%) (32.87-39.26%)

Core intensity 2.17 ± 0.10 1.75 ± 0.04 (1.71-2.58) (1.59-1.91)

89 Chapter 4: Seasonal space use

Movement kernel density estimation

The MKD estimates of 8 dingoes and 8 cats reached an asymptote (Appendix E) and were used to describe space use during the data collection period (Table 4.3, Appendix D.3).

The data of DS_M3 and DS_M1 were excluded from analysis due to transient behaviour and collar failure respectively. Dingoes had a higher probability of occurring in core areas compared to feral cats (t = -4.35, d.f. = 10.25, P = 0.001), despite encompassing a smaller proportion of the 95% isopleth (t = 2.38, d.f. = 10.88, P = 0.04) (Table 4.3).

Intensity and recursion distributions

Core isopleths for the intensity distribution were larger for dingoes than for feral cats (t

= -5.97, d.f. = 13.32, P < 0.001) (Table 4.3, Appendix D.4). The size of the core area as a proportion of the 95% isopleth area suggests that the areas in which dingoes spend longer periods of time are smaller relative to overall space use, compared to feral cats (t = 4.83, d.f. =

13.59, P < 0.001).

Core isopleths for the recursion distribution were similar for dingoes and cats (t = -0.16, d.f. = 11.47, P = 0.88), but as with other distributions, the proportion of the 95% isopleth area that dingoes revisited most heavily was smaller than that for feral cats (t = 3.36, d.f. = 9.77, P =

0.007).

Core intensity indices

All core calculations resulted in intensity indices greater than one, indicating non- random use of the SSU. The core intensity indices suggest that dingoes use smaller core areas more intensively, compared to feral cats (Tables 4.2 & 4.3, Figure 4.6).

This outcome is consistent in terms of general use (KDE utilisation distribution: t = -

2.37, d.f. = 13.69, P = 0.03; MKDE utilisation distribution: t = -2.75, d.f. = 9.22, P = 0.02), and the length of time or frequency of use in which concentrated activity occurs (MKDE intensity distribution: t = -4.82, d.f. = 9.41, P < 0.001; MKDE recursion distribution: t = -3.76, d.f. =

8.79, P = 0.005).

90 Chapter 4: Seasonal space use

4.5

3.5 3 2.5 2 1.5

1 Index of relative intensity relative Indexof 0.5 0 KDE utilisation MKDE utilisation MKDE intensity MKDE recursion

91 Chapter 4: Seasonal space use

Activity patterns within the core area

Activity patterns within animal-specific core areas appeared quite similar for both dingoes and feral cats. Use of core areas was highest during the day and in the night hours following midnight. The probability of being in a core area was lowest for both species following sunrise and in the hour prior to sunset (Figure 4.7).

Site fidelity

Feral cats had greater overlap in short-term space use at the 95% isopleth (t = 4.20, d.f.

= 15.71, P = 0.001) and at the mean core isopleth (t = 2.70, d.f. = 14.30, P = 0.02). There was no difference however in the level of short-term overlap of areas of highest use (t = 1.86, d.f. =

15.76, P = 0.08): both species appeared to have very little successive overlap at the 20% isopleth when space use was estimated every fortnight (Table 4.4). However, individuals from both species returned to previous high-use areas later in the study period (Figure 4.8).

Table 4.4. Estimates of spatial overlap between short -term utilisation distributions, using the Bhattacharyya Index (BA), at different isopleths, for dingoes Canis lupus dingo and feral cats Felis catus. Pooled mean and standard error are presented.

Dingo Feral cat No. animals with sufficient data 10 8 No. short-term UDs 46 62 No. UD overlaps 36 52 BA 95% isopleth 0.42 ± 0.02 0.59 ± 0.02 BA mean core isopleth 0.21 ± 0.02 0.30 ± 0.02 BA 20% isopleth 0.03 ± 0.01 0.05 ± 0.01

The dingo tracked in both years (DM_M1) showed considerable site fidelity between years, but with lower levels of overlap in core areas of use (95% contour: BA = 0.42; mean core

(71%) contour: BA = 0.18). The 2012 seasonal space use estimate was almost twice as large as

92 Chapter 4: Seasonal space use in the previous year, despite the shorter sampling period (2011: 1,437.60 ha from 3,773 fixes;

2012: 2,728.48 ha from 2,140 fixes).

a. b.

c. d.

Only the 2012 dataset reached an asymptote for this animal, hence space use may not have been fully described. Similar levels of overlap were recorded at the 95% isopleth (BA =

0.56) but overlap was higher at the core (64%) contour (BA = 0.30) for the feral cat tracked over both years (CM_M4). The seasonal space use estimate was larger in the second year (2011:

93 Chapter 4: Seasonal space use

740.16 ha; 2012: 1,094.40 ha), when over three times as many fixes were recorded (2011: 1,566 fixes; 2012: 5,075 fixes) (Figure 4.9).

Discussion

This study provided estimates of dry season space use by dingoes and feral cats in an intact northern Australian savanna. Although dingoes and feral cats have similar space use patterns in some respects, using comparable numbers of core areas that they leave at similar times of day, dingoes appear to use core areas more intensely than feral cats independent of the utilisation distribution being estimated. This may have implications for how each predator interacts with prey communities at the home-range scale.

94 Chapter 4: Seasonal space use

Space use estimates for dingoes were smaller than previous estimates in similar environments (Thomson 1992d), comparable to some estimates from more forested landscapes

(Purcell 2010) and much smaller than estimates from arid environments (Newsome et al.

2013b). This may be due to methodology, for example the shorter duration of this study, fix frequency, choice of bandwidth or inclusion versus removal of transient dingoes from calculations, or it may indicate that the environment of this study was productive or stable

(Corbett 1995a; Newsome et al. 2013b).

a. b.

c. d.

Figure 4.10. Movements and space use of dingo Canis lupus dingo DS_F1: a) High -frequency trajectory over 1-s slope layer with hillshading, showing avoidance of steep slopes; b) MKDE utilisation distribution, with 95% and core isopleths; c) MKDE intensity distribution with core isopleth and d) MKDE recursion distribution with core isopleth.

95 Chapter 4: Seasonal space use

Estimates of space use by feral cats in this study were intermediate relative to previous estimates from woodland or forest (Molsher et al. 2005; Buckmaster 2011; Bengsen et al. 2012) and those from arid environments (Edwards et al. 2001; Moseby et al. 2009b; Hilmer 2010). It is possible that average space use of feral cats in my population was somewhat lower than it appears, as my sample was biased towards large, male cats. Space use by female cats depends on the availability and distribution of resources, namely food and shelter for breeding, whereas large, breeding males also require access to females (Liberg et al. 2000). Therefore, several studies have recorded more extensive movements for male cats (Liberg et al. 2000; Buckmaster

2011), although others recorded no sex-difference (Molsher et al. 2005; Bengsen et al. 2012), suggesting that there may be real differences in space use between populations.

Dingoes tended to occupy multiple, smaller core areas relative to the total SSU, at similar or higher probability than feral cats, resulting in higher core intensity indices for all distributions. Core areas potentially represent both resting sites and resources such as water points (Brawata & Neeman 2011) or large carcasses (Thomson 1992c; Byrne 2009).

Conversely, feral cats had larger core areas relative to the SSU, suggesting they spread their activity more evenly through their range. Although both cats and dingoes are likely to hunt within their core areas, core activity patterns show that both predators venture into less- frequented areas around dusk and dawn. Many prey species are active during crepuscular periods (Van Dyck & Strahan 2008), hence these forays may represent hunting activity.

Alternatively, dingoes may be using these forays after dusk and at dawn to communicate with conspecifics (Corbett 1995a), either by interacting with pack members, patrolling territorial boundaries or visiting water points. Some female dingoes had much larger indices of intensity of use of core areas, suggesting they may have been breeding during the study period and spending large amounts of time at den sites.

The core areas of the intensity (areas visited for long periods) and recursion (areas used frequently) distributions revealed important locations used by dingoes and feral cats during the study period (Figure 4.10). The intensity distribution cores, often small, scattered areas, potentially represent temporary or permanent shelters or kill sites. The recursion distribution

96 Chapter 4: Seasonal space use cores contain the 'highways' frequently used by the predators as they move through the landscape (Figure 4.10d). They tended to occur in more central, contiguous areas within the space use estimate, compared to the intensity distribution cores, and are likely to connect important resources such as water points, shelters and carcasses, or represent areas an animal patrols frequently.

Both species used different areas over time, particularly by shifting high-use areas, suggesting that neither species exclusively occupied permanent den sites during the study period. Dingoes in particular demonstrated 'drifting ranges' at the broader scale, potentially as a response to conspecifics or fluctuations in resources (Doncaster & Macdonald 1991). Dingoes may occupy a larger home range over time, but use different patches with seasonal shifts in behaviour, or to optimise local resource availability. Feral cats however shifted less in terms of overall space use, suggesting their long-term space use may be less fluid compared to the larger predator. Although we had data from only one cat and one dingo over both years, they had similar levels of annual overlap in dry season space use, suggesting that both species are capable of maintaining relatively stable home ranges in the long term.

Failure to reach asymptote indicated that the SSUs for some individuals, including the transient dingo DS_M4, could not be adequately characterised. However, asymptotes were particularly affected by the timing of forays outside frequently-used areas, as demonstrated by a dingo and cat using a similar area. The cat (CM_M1) made a small foray at the start of data collection and its space use, following an initial peak, eventually reached an asymptote. The dingo (DM_F1) however, made a similar foray but at a later period, creating a large peak in space use, disrupting the asymptote and excluding the data from analysis (Appendix D.2,3,

Appendix E). When animals do venture outside their stable area of use, it is difficult to predict how space use estimates may fluctuate with further sampling time. However in most cases, they may be expected to increase with additional space use outside the stable area.

Estimates of space use are only models for how an animal interacts with their environment (Powell & Mitchell 2012), and will depend upon the methods and parameters used.

I attempted to use parameters and methods that were data-derived or species-specific, to avoid

97 Chapter 4: Seasonal space use arbitrary assumptions on species behaviour and to allow comparisons between space use characteristics of two different predators. Estimates are also constrained by the representativeness of the data. This study ran over a relatively short timeframe, hence it is conservative to expect home ranges to be larger than my estimates, and for some areas of non- use to be sites of predator activity at other times of year. Site fidelity analyses demonstrate that both species are likely to use larger and potentially different areas at different times of year, such as the wet season, or for specific life-history events such as breeding. Further, as it is logistically very difficult to collar all individuals in an area (but see Vanak et al. 2013), we cannot know how the space use of other conspecifics may alter the probability of predator use across the landscape.

If a predator remains at a location for long periods, or returns frequently, it is likely to impose higher predation pressure on local prey. This may have a direct effect, if the predator is hunting, or indirect, if its presence alters prey behaviour (Creel & Christianson 2008). The difference in space use characteristics of dingoes and feral cats could thus influence how they interact with prey. In this study, I found feral cats spread their activity more evenly, over a larger proportion of their more contiguous and smaller range. This may create more uniform predation pressure from feral cats on local prey, and may intensify impacts on prey populations, if refugia from predation are limited (Dickman 1996). Dingo activity however, appears to be concentrated in more ephemeral hotspots through their range, which is likely to create a more variable landscape of fear (Laundré et al. 2001). This may impose high predation pressure temporarily in small areas, but lower levels across a larger proportion of their range. The behaviour of dingoes may thus provide time for prey populations to recover from predation, and spatial refugia where predation pressure is low.

This study used a range of methods to describe different aspects of dry season space use patterns of dingoes and feral cats living in sympatry. Dingoes used more space overall compared to feral cats, but the majority of their activity was concentrated in core areas that were smaller relative to their overall range, and which shifted over time. Feral cats used more contiguous areas, with activity spread more evenly over larger core areas. Further studies could

98 Chapter 4: Seasonal space use investigate how differences in predator space use patterns may directly or indirectly affect prey by creating different levels of predation risk across space and time, potentially affecting the long-term stability of local prey communities.

Author contributions

LB conceived the study, LB and HM collected field data, LB analysed the data and wrote the paper, CJ, LS, SL and ER assisted with editing.

99 Chapter 4: Seasonal space use

100

Chapter 5 Spatial interactions between sympatric dingoes and feral cats in the Kimberley: patterns of intraguild avoidance in a landscape of fear 4

Abstract

1. Space use in predators may be primarily influenced by the distribution of prey, but in mesopredators, which often face direct threat and interference from larger predators, decisions on space use must trade off prey acquisition against survival. Mesopredators may therefore change their behaviour to reduce spatial overlap and minimise interactions with apex predators. 2. I investigated spatial interactions between sympatric dingoes and feral cats in intact northwestern Australian savanna. I examined levels of static overlap between dingoes and cats and within each species, at both the 95% and core contours of their utilisation distributions, and compared shared space use by characterising dynamic spatial and temporal interactions. I also used a null model approach to assess the distance between pairs of individual predators through time, to determine whether feral cats avoid close encounters with dingoes. 3. Feral cats were recorded less often than expected within 1 km of dingoes. Both predators exhibited variable responses to shared space, but simultaneous use was rare. There were no significant differences in mean static overlap between or within species, although overlap in high-use areas was much lower than at the 95% contour. While some dingoes exhibited extensive spatial and temporal overlap and frequently occurred close to each other, others avoided shared areas, suggesting that overlap between adjacent packs may be minimal. Static and dynamic overlap between male feral cats was low, suggesting they occupy exclusive home ranges. 4. These results suggest that while the two predators use similar areas, feral cats reduce risk by avoiding areas visited repeatedly by dingoes, or by moving to a safe distance when they find themselves in close proximity to the larger predator. Dingoes could therefore restrict feral cat access to areas they use intensively or regularly, but these areas may shift over time depending on dingo activity.

4 Brook, L.A., Johnson, C.N., McGregor, H., Schwarzkopf, L., Legge, S. & Ritchie, E.G. (in prep) Spatial interactions between sympatric dingoes and feral cats in northern Australia: patterns of intraguild avoidance in a landscape of fear. To be submitted to Journal of Animal Ecology.

101 Chapter 5: Spatial overlap

Introduction

Mesopredators must balance resource requirements with the risk of potentially lethal encounters with apex predators (Heithaus 2001), just as any potential prey organism must avoid predators (Sih 1980). Reducing spatial and temporal overlap can minimise the likelihood of interference competition and may be achieved in a number of ways. Mesopredators might avoid areas or habitats where they are more likely to encounter an apex predator (Durant 1998;

Janssen et al. 2007), be active at different times (Arjo & Pletscher 1999; Harrington et al. 2009;

Brook et al. 2012), avoid apex predator activity hotspots such as den sites (Berger & Gese

2007), or monitor approaches by apex predators and use short-term avoidance to maintain separation (Broekhuis et al. 2013). In this way, mesopredators can avoid places and times with higher perceived risk: the hills in a 'landscape of fear' (Laundré et al. 2001). Understanding patterns of avoidance and overlap between predators can reveal the influence of apex predator interference on space use of mesopredators, and explore the potential for apex predators to exclude mesopredators and therefore influence trophic interactions.

The level of spatial overlap between predators may be influenced by prey abundance and dietary overlap, as these factors can shift the trade off between conflicting requirements of food and survival. If resources are plentiful, competition may be reduced, or mesopredators may be able to acquire prey in areas used less frequently by apex predators. Wilson et al. (2010) found that bobcats Lynx rufus avoided areas heavily used by a larger predator, the coyote Canis latrans, where prey were abundant. However, spatial overlap increased where resource availability was low. This may explain why other studies sometimes record extensive overlap between these two predators (Major & Sherburne 1987), although core overlap may still be avoided (Thornton et al. 2004). Spatial overlap may also be high if sympatric predators specialise on different prey, minimising exploitation competition. In eastern Poland, where sympatric wolves Canis lupus and Eurasian lynx Lynx lynx hunt different cervid species, the two predators overlap in both 95% contours and core areas and show no avoidance patterns

(Schmidt et al. 2009). However, territorial conflict for space rights may also be sufficient

102 Chapter 5: Spatial overlap grounds for interference. In southern Africa, where dietary overlap between black-backed jackals Canis mesomelas and the smaller Cape fox Vulpes chama and bat-eared fox Otocyon megalotis is quite low, both fox species were killed by jackals, and avoided core area overlap with the larger canid (Kamler et al. 2012).

Spatial overlap or avoidance may be estimated in various ways. Doncaster (1990) defined two types of spatial interaction: static interaction is overlap in long-term utilisation distributions, while dynamic interaction describes the instantaneous influence of the location of one animal on the movement decisions of another. In addition to investigating interactions between species, spatial overlap analyses may also shed light on the intraspecific spatial structure of predator populations (e.g. Samson & Huot 2001). Carnivores display diverse social structures, from the generally solitary felids (Macdonald et al. 2010), to the frequently social canids (MacDonald et al. 2004). Lack of spatial overlap provides evidence for territoriality; a territory being defined as 'any defended area' (Noble 1939 in Burt 1943), while extensive overlap suggests a social group structure (Doncaster & Macdonald 1991).

In the tropical savannas of northern Australia, predation by the introduced feral cat Felis catus is thought to be contributing to a series of small mammal declines (Fisher et al. in press).

However, movements and space use of feral cats are potentially influenced by the likelihood of encounter with the dingo Canis lupus dingo, a sympatric apex predator. The dingo is a medium- sized canid, brought to Australia 3,500-5,000 years ago from south east Asia (Oskarsson et al.

2012 ). It is thought to exert a positive impact on native fauna by suppressing introduced mesopredators, the red fox Vulpes vulpes and feral cat (Johnson 2006; Letnic et al. 2012).

Dingoes either live alone, or in social packs of up to 23 individuals (Thomson 1992d). Pack formation and size may depend on prey type, abundance and resource stability (Corbett 1995a;

Newsome et al. 2013b) and the extent of artificial predator control (Allen et al. 2000; Wallach et al. 2009b). Spatial overlap between individuals has been recorded in forests (Harden 1985;

Robley et al. 2010a) and arid environments (Thomson 1992d; Newsome et al. 2013b), but pack members may also spend time alone and overlap little in areas of core use (Purcell 2010), depending on whether hunting is cooperative or independent (Corbett 2001). Although pack

103 Chapter 5: Spatial overlap

members overlap, adjacent packs are thought to occupy relatively exclusive territories

(Thomson 1992d; Purcell 2010).

Feral cats are adaptable predators, which aided their success following introduction to

Australia in the 1800s (Abbott 2002). Extraordinary variation in density or individual overlap has been recorded in cats (Liberg et al. 2000). Male feral cats may overlap extensively in stable, high-resource areas (e.g. Denny et al. 2002), maintain 'partial territoriality' in areas with lower resources (Konecny 1987) or occupy exclusive ranges when resources are poor (Edwards et al.

2001). Buckmaster (2011) recorded overlap between female cats, and complete overlap of one female range by a male cat in forested south-east Australia, while feral cats on Stewart Island in

New Zealand overlapped both within and between sexes (Harper 2004). Even when conspecifics overlap, they may use shared areas at different times to avoid encounters

(Leyhausen & Wolff 1959 in Ewer 1968).

Several studies using sand pads or camera surveys to detect presence of predators have recorded dingoes and feral cats at the same location (Chapter 2; Kennedy et al. 2012; Wang &

Fisher 2012), but the extent of spatial and temporal overlap between individuals is unknown.

The size difference between dingoes and feral cats suggests there is potential for aggressive interactions, as they are just close enough in size to overlap in diet, yet dingoes are unlikely to avoid encounters, as feral cats are too small to pose a threat (Donadio & Buskirk 2006; Ritchie

& Johnson 2009). Intraguild killing of cats by dingoes has been observed (Moseby et al. 2012), which should provide an incentive for feral cats to avoid encounters with the larger predator.

Dietary overlap has also been recorded (Paltridge 2002), however, the extent may vary depending on regional and seasonal differences in prey, leading to fluctuations in the strength of competition, and thus aggressive encounters, between the two predators. Hence, patterns of avoidance may not be consistent across areas of co-occurrence.

Here I estimate patterns of static and dynamic overlap in the utilisation distributions of sympatric dingoes and feral cats, estimated during the dry season in a savanna ecosystem of north-western Australia. I also calculate the distance between simultaneous fixes for each pair of animals, as overlap does not necessarily imply that animals are in close contact (Thomson et

104 Chapter 5: Spatial overlap al. 1992a). I hypothesise that feral cats should adjust their space use to minimise encounters with the larger predator, hence avoiding areas and times of high risk on the landscape of fear. I make the following predictions, given this hypothesis, the evidence of dietary overlap and interference competition between the two predators, and current understanding of their social organisation:

 dingo-cat dyads should have low levels of spatial overlap, particularly at the core

isopleth, as feral cats would avoid areas perceived as risky due to prolonged or

frequent dingo activity;

 dingo-cat dyads should have moderate use of shared areas but few cases of close

proximity, given that feral cats are expected to avoid overlap zones and close

occurrences, but that dingoes are unlikely to modify space use patterns to either

avoid or increase overlap with a smaller predator;

 dingo-dingo dyads should exhibit a dichotomous response - either high or very low

levels of spatial and temporal overlap, in both the 95% and core isopleths, and

occasional proximity to conspecifics, given their social and pack-territorial

behaviour;

 cat-cat dyads should have moderate levels of overlap at the 95% isopleth, but lower

shared use of core areas, given that prey resources were likely to be dispersed

through a range of habitats (Legge et al. 2011). If resources were clumped,

increased overlap could be expected (Liberg et al. 2000). Shared areas should be

used alone, as both individuals will act to avoid proximity with conspecifics.

105 Chapter 5: Spatial overlap

Methods

Sympatric dingoes and feral cats were caught, fitted with GPS telemetry collars

(Telemetry Solutions, Concorde USA) and tracked at high frequency at Mornington Wildlife

Sanctuary in the central Kimberley bioregion of northern Western Australia (WA) in the dry seasons of 2011 and 2012 (Figure 5.1). Details of the study site, capture methods, data collection and screening, and home range and core area estimation are found in Chapter 4.

Figure 5.1. The study site, Mornington Wildlife Sanctuary, located within the central Kimberley bioregion (light green) in north -western Australia (inset). Dingoes and feral cats were collared near Mornington’s Wildlifelink Research Station, in the area shaded blue, north of the Fitzroy River.

Data schedules for some animals did not always collect a sample representative of the full diel period. In Chapter 4, where the aim was to estimate space use, I restricted the data to ensure an unbiased sample of the 24-hour period, and restricted the analysis to samples that reached an asymptote. Here I used all available data, to maximise the number of simultaneous

106 Chapter 5: Spatial overlap fixes. I also included data from all animals, regardless of whether their space use reached an asymptote within the study period, as this analysis is focussed on areas of overlap rather than accurate estimation of total space use. I included fixes when animals were stationary, rather than restricting the dataset to 'active' fixes (Benhamou 2011), as cats likely associate elevated risk with proximity to dingoes, independent of whether either species is active.

I assessed interactions between each possible pair (dyad) of dingo-dingo (D-D), dingo- cat (D-C) and cat-cat (C-C) individuals that were collared in the same timeframe. For all analyses, only dyads with some level of overlap at the 95% utilisation distribution contour, given the fix interval, were considered.

Static overlap indices

I calculated static overlap in space use estimates derived using the movement-based kernel density estimate (MKDE) from data at 15-min intervals. To ensure equal sample effort, I limited datasets to the same timeframe when both collars were deployed, although some collars had higher fix rates than others. Space use estimates were calculated over the maximum range of a dyad's data with a 5-km buffer. I estimated the MKDE for the utilisation (UD), intensity

(ID) and recursion distributions (RD) and delineated core isopleths for each individual in a dyad. The dyad's mean core isopleth was taken as the average of the two isopleth values.

Additional MKDE and core isopleth delineation methods are outlined in Chapter 4. Animals that had no overlap at the 95% contour of the utilisation distribution were excluded from further overlap analyses of the core areas of the utilisation, intensity or recursion distributions.

Overlap indices were estimated for the UD 95% isopleth and the mean core isopleths of the UD, ID and RD. Conditional estimation restricted the UD to the contour being measured. I used two different overlap indices with different properties to examine static interactions.

PHR calculates the probability that animal will be found within the utilisation distribution of animal (Fieberg & Kochanny 2005):

107 Chapter 5: Spatial overlap

where is the home range area of animal and is the estimated UD of animal .

PHR is a directional index, hence it differs for each animal within a dyad, depending on the ratio of the overlap zone area to their total space use area.

The utilisation distribution overlap index (UDOI) measures "the amount of overlap relative to two individuals using the same space uniformly" (Fieberg & Kochanny 2005):

where is the overlapping area of the two UDs. The UDOI ranges from 0 (no overlap) to 1 (uniform use and complete overlap), but can exceed 1 if the UDs overlap and are non-uniform. UDOI may be a suitable correlate for contact rates between individuals (Robert et al. 2012).

I then calculated the mean PHR and UDOI for each dyad type to compare rates of overlap between dyad types for the different forms of space use. I compared different groups using a range of parametric and non-parametric tests, as not all datasets could be log- transformed to normal. To compare mean overlap between dyad types, I used the asymptotic

Kruskall-Wallis test. I used an asymptotic Friedman test to compare mean overlap levels between dependent distributions (UD 95% and UD, ID and RD cores), within each dyad type.

Significant differences within dyad types (e.g. between UD 95% and UD core) were further investigated using exact Wilcoxon signed-rank tests. Symmetry in differences between pairs for

Wilcoxon signed-rank tests was assessed using the MGG test (Miao et al. 2006) and homoscedasticity was confirmed using the Brown-Forsyth test.

Distance between animals

I used a null model approach to compare the frequency at which individuals came within close proximity of each other (Gotelli 2001). Null models have recently been used to assess spatial interactions between sympatric herbivores (Richard et al. 2013). For each dyad, I calculated the Euclidean distance between individuals at each synchronous fix, using data at 15- min intervals. I then created 1000 asynchronous replicates, by randomising the chronology of

108 Chapter 5: Spatial overlap fixes for one animal (dingo if applicable) and recalculating the distance between fixes. I measured several distribution metrics (mean distance between individuals, number of fixes within different distances) for both the synchronous fix distribution and for each randomised data set, and used a two-sided randomisation test in the ade4 package (Dray & Dufour 2007) to compare each observed metric to its null distribution calculated using the asynchronous data.

To assess the mean response of each dyad type, I averaged each metric within each dyad type, and compared the mean observed metrics to null distributions of the mean metric for each of the 1000 randomised samples. Figures show the % frequency of pooled observed data against pooled null data for each dyad type.

Dynamic overlap indices

I also applied dynamic interaction tests to determine whether pairs of animals used their spatial overlap zones independently of one another in space and time (Minta 1992). Minta's interaction indices compare the fix locations of two individuals relative to the overlap zone

(OZ) of their space use, at simultaneous points in time. Data are arranged as a contingency table

(Figure 5.2) to compare the number of simultaneous fixes where both, one or no animals are present in the OZ. The observed frequencies are then compared to the expected relative space use of the OZ and 95% contour for each animal, given they use the OZ independently relative to the overall size of the 95% contour. The χ2 statistic for the contingency table is partitioned into three separate components to compare the two 'main effects' or spatial hypotheses and a temporal interaction:

a) whether species A uses the OZ independently ;

b) whether species B uses the OZ independently ; and

c) whether the species use the OZ independently in time (Minta 1992).

The coefficients resulting from the two spatial tests ( and ) indicate whether animals avoid the OZ (<0), are attracted to the OZ (>0) or whether use of space is random (near

0). Responses within a pair can be symmetrical, in that they both show the same pattern,

109 Chapter 5: Spatial overlap

asymmetrical (different responses) or singular, when only one animal deviates from expected space use (Mace & Waller 1997). Symmetrical avoidance of the OZ provides support for territorial behaviour (Minta 1992). Understanding the coefficient from the interaction test ( ) is more complex, as there are several possible outcomes. Positive values indicate more simultaneous use of the shared space, while negative values indicate increased solitary use, and values close to zero suggest use of the OZ is random. A significant negative value for can be interpreted using the odds ratios from the contingency table to determine whether space use of each cell is as expected (Minta 1992). In this study, only significant interactions were interpreted. Partitioned values were considered significant when P < 0.1 (Minta 1993), to improve detection of potential dynamic interactions and reduce type II errors. Animals were considered to be attracted to or avoiding the OZ when and were ≥ 0.25 or ≤ -0.25 respectively. Similarly, if was ≥ 0.25, the response was likely to be simultaneous, while a value ≤ -0.25 suggested solitary behaviour.

Both present in OZ Only B present in OZ B present in OZ

Only A present in OZ Both absent from OZ B absent from OZ

A present in OZ A absent from OZ

110 Chapter 5: Spatial overlap

For this analysis, I used UDs derived using a fixed kernel density estimate with a plug- in bandwidth (KDE), and three-hourly fixes sampled representatively across the diel period, which produces smoother contours and larger areas of space use than MKDE. This analysis assumes that fixes are independent, hence three-hourly fixes, with lower levels of autocorrelation, were more appropriate than high-frequency data. Although these datasets are still largely autocorrelated (24.82% and 5.54% of bursts independent with mean Schoener's ratios of 0.92 ± 0.05 and 0.50 ± 0.02 for dingoes and feral cats respectively), the 3-hourly subsampling represents a trade-off between data retention and achieving independence. Dingoes and feral cats are potentially active throughout the diel period (Brook et al. 2012), hence sampling separate activity shifts to achieve biological independence (Minta 1992) was not viable. Methods for KDE are outlined in Chapter 4. I estimated overlap at the 95% contour, due to insufficient levels of overlap at the core isopleth. To ensure equal sample effort, I limited datasets to the same timeframe for each dyad, and included fixes only if they were recorded within three min of each other. This reduced the sample size for most dyads, but all pairs in the final dataset retained over 50 simultaneous fixes (range: 54-388), as recommended by Seaman et al. (1999) for kernel estimation.

Large sample sizes (observed or expected frequencies) are recommended in contingency tables, to ensure the test-statistic adequately approximates the χ2 distribution

(Agresti 2007). To prevent unreliable estimates, I removed a) dyads with no fixes in the OZ, b) dyads with ≤ 5 observations for each of the cells, and , correlated with a small OZ that both individuals rarely used and c) dyads with more than one cell having an expected frequency < 5. I also calculated the exact probabilities of the observed outcomes of each dyad, using a goodness-of-fit test for discrete multivariate data from the EMT package (Menzel 2013).

I then compared the exact P-value to that approximated by the χ2 distribution.

Analyses were mostly run in R version 3.0.1 (R Development Core Team 2013) using the adehabitatLT (analysis of animal movements) and adehabitatHR (home range estimation) packages (Calenge 2006). Plug-in bandwidths were estimated in ks (Duong 2012). Non- parametric tests were run using the coin package (Hothorn et al. 2008).

111 Chapter 5: Spatial overlap

Results

We caught nine dingoes and nine cats over two years, with one dingo and two cats collared in both years. Due to collar failure or inability to relocate the animal, we collected suitable data on four dingoes (2 ♀, 2 ♂) and five cats (5 ♂) at Mornington in 2011, and five dingoes (4 ♀, 1 ♂) and five cats (1 ♀, 4 ♂) in 2012. Data from one 2011 cat (CM_M5) were only suitable for 3-hourly (dynamic overlap) analysis. Each study animal exhibited overlap with at least one other at the 95% KDE contour. Fix rates were generally high (Table 5.1), with 4.16% and 4.24% of fixes removed from the dingo and feral cat GPS datasets respectively.

Dingo Feral cat

No. animals 9 10 No. days of data collection 30 ± 4.25 51.2 ± 6.81 No. 15-min fixes 2800 ± 402 4501 ± 671 No. 3-h fixes 240 ± 34 419 ± 49 3D fix rate 98.46 ± 0.73% 93.72 ±- 2.06% (98.94%) (92.51%) Fix rate 95.84 ± 2.36% 95.76 ± 0.95% (97.38%) (95.13%)

Were there differences in static spatial overlap between and among predators?

Overlap was detected at the 95% MKDE UD contour for 14 dingo-dingo dyads, 22 dingo-cat dyads and 3 cat-cat dyads (out of a possible 16, 41 and 16 dyad combinations respectively). Frequently dyads did not overlap at the core.

112 Chapter 5: Spatial overlap

Utilisation distribution overlap index (UDOI)

Levels of overlap in utilisation distribution were highest between dingoes at the 95% contour, almost seven times higher than overlap between dingoes and feral cats (Figure 5.3a) although there was no significant difference between dyad types (asymptotic Kruskal-Wallis

test: = 0.105, P = 0.95) and the result may be unduly influenced by outliers (Appendix G.1).

There was a more than five-fold decline in mean overlap between dingoes from the 95% isopleth to the core use area (exact Wilcoxon signed-rank test: Z = -3.30, P < 0.001, Appendix

G.2), compared to an even stronger ten-fold decline in mean overlap between dingoes and cats from 95% to core (Z = -4.11, P < 0.001).

a. 0.4 UD 95% isopleth 0.35 UD core isopleth

0.3 ID core isopleth

0.25 RD core isopleth 0.2

0.15 UDOI value UDOI 0.1

0.05

0 dingo-dingo dingo-cat cat-cat b. 0.3

0.25

0.2

0.15

PHR value PHR 0.1

0.05

0 dingo-dingo dingo-cat cat-dingo cat-cat

Figure 5.3. Mean (± se) overlap values fo r a) UDOI and b) PHR indices for different dyads and distribution contours. Dingo -cat columns refer to the probability of a dingo being located within a feral cat contour, while cat -dingo columns represent the probability of a feral cat being located within a dingo contour.

113 Chapter 5: Spatial overlap

Figure 5.4. Feral cat locations collected at 15-min intervals over both years. CM_M1-5 were collared in 2011. CM_M4, CM_M6-8 and CM_F1 were collared in 2012.

Despite apparent differences in overlap at the UD core level, there was no significant

difference between dyad types (dingo-dingo, cat-cat, dingo-cat) ( = 0.85, P = 0.65). Further, the appearance of higher levels of mean overlap in areas used regularly (recursion distribution core), compared to those visited for long periods (intensity distribution core), for dingo-dingo and dingo-cat dyads were not significant (D-D: Z = 0.94, P = 0.37, D-C: Z = 0.75, P = 0.47), nor

was there any difference in overlap for either distribution between dyad-types (ID core: =

0.03, P = 0.99, RD core: = 1.26, P = 0.53). Overlap between feral cats was rare, occurring in only 3 dyads, and was extremely low across all distribution types (asymptotic Friedman test:

= 4.2, P = 0.24, Figure 5.3a). The highest UDOI for cat-cat dyads was 0.003 between two males. Collared feral cats appeared to be arranged adjacently rather than overlapping (Figure

5.4). Overlap was particularly low for the recursion distribution (Appendix G.2). Within-group variances were close to being rejected as different in some Kruskal-Wallis tests, but as this would increase the type-1 error, it should not affect the results presented (Zar 2010).

114 Chapter 5: Spatial overlap

PHR overlap index

PHR overlap indices followed a similar trend to UDOI, albeit overlap levels were higher overall compared to UDOI calculations (Figure 5.3b, Appendix G.3). There were

significant differences between overlap at different contours for each dyad type (D-D: =

57.90, P < 0.001, D-C: = 28.77, P < 0.001, C-D: = 38.89, P < 0.001, C-C: = 9.80, P = 0.02, Appendix G.3). Overlap at the 95% contour was higher than at the utilisation distribution core for all dyad types (D-D: Z = -4.62, P < 0.001, D-C: Z = -4.11, P < 0.001, C-D: Z = -4.11, P

< 0.001, C-C: Z = -2.20, P = 0.03).

At the 95% contour, there were relatively high probabilities that both other dingoes and feral cats would be found in a dingo contour, and the mean probability of a dingo occurring in a cat contour was lower than the inverse probability, but the within-group difference was not quite

significant ( = 7.46, P = 0.06). There were also no significant differences in core overlap

between dyads for any core distribution (UD core: = 2.17, P = 0.54; ID core: = 6.12, P =

0.11; RD core: = 4.62, P = 0.20). Within-group variances for Kruskal-Wallis tests were again fairly different, but given the null hypotheses were accepted, possible sources of type-1 error are not of concern.

Were predators found closer to or further from each other than expected?

Cats were recorded within 100 m of dingoes on only three occasions, or in less than

0.01% of fixes (Appendix G.5). Fourteen dingo-cat dyads came within 500 m of each other, but at a lower frequency than expected compared to the null distribution when data were pooled

(Figure 5.5a). However, only a third of dyads recorded a significant difference, potentially due to statistical difficulties distinguishing between the observed from random data when no fixes were observed (Table 5.2, Appendix G.5).

115 Chapter 5: Spatial overlap

a. 1.2

1.0

0.8

0.6

0.2

b. 5.0

4.0

3.0

2.0

1.0

c. 1.5

1.0

0.5

Figure 5.5. Percent frequency distributions for distances between pooled synchronous fixes and pooled asynchronous fixes from the randomised null distribution for a) dingo -cat dyads, to a maximum of 2,000 m, b) dingo -dingo dyads to a maximum of 5,000 m and c) cat -cat dyads to a maximum of 2,000 m.

116 Chapter 5: Spatial overlap

Table 5.2. Distance analysis results for each dyad type. The observed value is ave raged across dyads, and the null mean is averaged first across each dyad in each replicate, and then across replicates. The P-value is for the pooled test, while the number of significant dyads indicates the number (and % in parentheses) of individual dyads for which the observed metric was significantly different from the expected value. Italicised P-values should be treated with caution, as the null distribution was not considered symmetrical using the MGG test (Miao et al. 2006).

Dyad type Dingo-dingo Dingo-cat Cat-cat

N 14 24 4

Observed mean 5704.19 5290.44 3470.09 Null mean 5967.04 5253.48 3489.66 Variance 23.07 7.49 6.16 P 0.001 0.001 0.001 No. significant dyads 12 (86%) 23 (96%) 4 (100%)

Observed no. fixes under 500 m 137.14 11.13 8.75 Mean null number under 500 m 26.16 17.20 14.02 Variance 1.63 0.59 3.20 P 0.001 0.001 0.003 No. significant dyads 3 (21%) 8 (33%) 2 (50%)

Observed no. fixes under 1000 m 193.86 44.67 61 Mean null number under 1000 m 74.05 59.97 63.58 Variance 3.70 1.76 11.66 P 0.001 0.001 0.49 No. significant dyads 7 (50%) 18 (75%) 0 (0%)

Observed no. fixes under 1500 m 272.86 129 232.25 Mean null number under 1500 m 142.29 134.49 200.81 Variance 5.25 3.00 26.47 P 0.001 0.001 0.001 No. significant dyads 12 (86%) 13 (54%) 3 (75%)

Observed no. fixes under 2000 m 361.43 231.58 490 Mean null number under 2000 m 233.14 241.34 441.02 Variance 6.64 3.77 46.24 P 0.001 0.001 0.001 No. significant dyads 11 (79%) 16 (67%) 4 (100%)

In over half the dyads, there was a significantly lower than expected number of fixes within 1,000 m. Distances between feral cats and dingoes appear to be lower than expected when in close proximity, up to a distance of about 1,200 m, after which there is a peak in the distribution of pooled data (Figure 5.5a).

117 Chapter 5: Spatial overlap

Three dingoes appeared to belong to a pack, as they were often together. This led to higher than expected frequencies of close contact between dingoes (Figure 5.5b, Table 5.2). A male, DM_M1 and two females DM_F3 and DM_F5 were frequently recorded within 100 m of each other (Appendix G.4). Other dingo dyads were generally located close to each other as frequently as expected or less than expected compared to randomisations (Appendix G.4). The mean distance between individual dingoes was still over 5 km, slightly lower than expected

(Table 5.2).

Male feral cats were never recorded within 250 m of each other (Figure 5.5c) and two dyads exhibited a lower than expected number of fixes within 500 m (Appendix G.4). Collared cats with overlapping ranges remained, on average, almost 3.5 km from neighbouring conspecifics (Table 5.2).

Did predators use shared space differently?

Seven dingo-dingo, sixteen dingo-cat and four cat-cat dyads had sufficient overlap in the 95% KDE for analysis. Periods of overlap in GPS tracking schedules ranged from 47 ± 5, 47

± 5 and 69 ± 12 days and included 204 ± 18, 195 ± 20 and 279 ± 44 fixes, respectively. UDOI between the UDs ranged from 0.0005-1.15 (D-D), 0.0003-0.51 (D-C) and 0.0007-0.01 (C-C).

There were no discrepancies between the χ2 test and the exact test that altered significance.

Dingo-cat

Ten out of sixteen dingo-cat dyads contained significant spatial dependence with a range of outcomes (Table 5.3). There were five cases of singular avoidance or attraction to the overlap zone, one (cat attraction) which occurred because a dingo never entered the overlap zone, preventing calculation of .

There were five cases of feral cats avoiding the overlap zone, compared to three where cats used it more than expected and seven where movement could not be distinguished from random. There were almost equal cases of dingoes attracted to (N = 4) and avoiding (N = 3) the overlap zone, compared to seven dyads where dingoes exhibited non-significant or apparently random use, and two dyads where the dingo did not use the shared space.

118 Chapter 5: Spatial overlap

However, there was no single interaction type observed more frequently than others

(Appendix G.6); five dyads exhibited singular responses, three were symmetrical, where each animal behaved similarly, with two dyads avoiding the overlap zone (Figure 5.6c) and one pair attracted to it (Figure 5.6d). This dyad had particularly high levels of overlap, resulting in an overlap zone that covered the majority of the cat's 95% contour. Two pairs were asymmetrical, where dingoes were attracted to the overlap zone, but cats avoided it.

Dyad type Spatial effect Dingo-dingo Dingo-cat Cat-cat Singular Attraction by dingo 1 (1.06) Attraction by cat 2 (1.16 ± 0.42) Avoidance by dingo 2 1 (-0.88 ± 0.23) -1.04 Avoidance by cat 1 1 -0.28 (-0.99) Asymmetrical Attraction by dingo 2 (avoidance by cat) (0.92 ± 0.14) (-0.42 ± 0.04) Attraction by cat (avoidance by dingo)

Symmetrical Attraction 4 1 (0.97 ± 0.06) (0.8) (1.01) Avoidance 1 2 1 (-0.76 ± 0.27) (-0.92 ± 0.40) (-0.67 ± 0.41) (-0.57 ± 0.26) Random use or NS result for both dyads 6 2

No. dyads 7 16 4

119 Chapter 5: Spatial overlap

The space use of the single collared female cat was entirely within that of a female

dingo (DM_F6-CM_F1, Appendix G.6), which prevented any analysis apart from determining

whether the female dingo used the overlap zone independently. Although she appeared to

slightly avoid the cat's range ( = -0.47), the effect was not significant (P = 0.42).

There were similarly variable significant temporal interactions observed in three dingo-

cat dyads (Table 5.4). The dyad with considerable spatial overlap recorded simultaneous

attraction of the dingo and feral cat to the overlap zone. While the temporal interactions for

dingo-dingo dyads were relatively straightforward when examining the odds ratios, there was

some ambiguity in the final two dingo-cat dyads. Dingoes used the overlap zone alone more

than expected, with some simultaneous avoidance in one and attraction in the second, to a lesser

extent.

Odds ratio

Dyad type P Interpretation Dingo-dingo DM_M1-DM_M2 1.08 <0.01 2.48 0 1.01 0.4 9 Simultaneous attraction DM_F3-DM_M1 0.68 0.03 1.78 0.40 0.62 0.21 Simultaneous attraction DM_F3-DM_F5 0.55 0.01 1.64 0.46 0.70 0.37 Simultaneous attraction

Dingo -cat DM_F5-CM_M6 0.81 <0.01 1.81 0.65 0.39 0.52 Simultaneous attraction DM_F3-CM_M6 -0.45 0.01 1.33 0.66 2.03 0.39 Solitary use by A, some simultaneous use DM_F6-CM_M7 -0.62 0.06 0.58 1.03 1.83 0.95 Solitary use by A, some simultaneous non-use

120 Chapter 5: Spatial overlap a. b.

c. d.

121 Chapter 5: Spatial overlap

Dingo-dingo

All dingo dyads (N = 7) differed significantly (P < 0.1) from spatial independence

(Table 5.3, Appendix G.6). Symmetrical attraction, where both individuals were attracted to the overlap zone (Figure 5.6a), was the most common spatial interaction. However, one pair of females exhibited symmetrical avoidance, and other dyads had insufficient overlap for analysis

(Figure 5.6b). Three pairs with symmetrical attraction also had a significant temporal interaction (Table 5.4, Appendix G.6), where they were simultaneously attracted to the overlap zone.

Cat-cat

Two of the four cat-cat dyads had significant spatial responses to the overlap zone

(Table 5.3). Both cases involved avoidance; one when the second cat used the shared space randomly, and the second where both cats avoided it in a symmetrical relationship. There were no significant temporal interactions between feral cat dyads.

Discussion

The study of intraguild spatial interactions can shed light on how species perceive and respond to each other across the landscape. Shifts in behaviour to avoid encounters may affect how predators interact not only with sympatric predators, but with lower trophic levels. Here I show that spatial interactions can vary not only between conspecifics, but between members of a predator guild, from attraction between members of apparent social groups, to avoidance of apex predators by mesopredators. Where feral cats and dingoes co-occur, cats appear to use multiple means to minimise encounters with the larger predator. These potential limitations on cat behaviour may affect the way feral cats interact with prey communities.

Dingo-cat interactions

It appears that feral cats can co-occur with dingoes in space, by avoiding high-use areas and minimising the risk of encounter through time. As expected, static overlap between dingoes

122 Chapter 5: Spatial overlap and feral cats was moderate, but there were no significant differences detected between dyad types. The appearance of different means may be due to some dyad-specific behaviour: one female dingo had very high static overlap with a male cat at both the 95% and core contour levels (Appendix G.1). PHR indices also did not detect any significant differences in static overlap between dyad types. Strangely, the mean probability of feral cats occurring within dingo contours was higher than for dingoes occurring within feral cat contours, although the difference was not significant. However, as dingoes generally occupied larger areas than feral cats (Chapter 4), the size of the overlap zone relative to their overall space use is likely to be much lower than for feral cats, potentially explaining this result.

The prediction that overlap in core areas would be lower than at 95% contours, was supported by my observations, as core areas would carry higher risk of encounter. However, despite the fact that half the significant dynamic overlap interactions with dingoes consisted of feral cats avoiding the overlap zone as expected, almost a third involved attraction to shared space, making it difficult to identify a consistent response. Furthermore, dingoes exhibited avoidance of or attraction to the overlap zone at similar levels, suggesting their response is just as variable. The landscape of fear should have higher 'peaks' for mesopredators, as they may lack the anti-predator behaviours required to evade predation (Ritchie & Johnson 2009). Most felids are not adapted to cursorial hunting (Kitchener et al. 2010) and feral cats do not have the stamina to outrun a canid. Hence feral cats should avoid areas in the landscape with higher risk of lethal encounter, such as core areas. However, as the two predators still occur in sympatry, and dingoes are unlikely to avoid areas of shared use with feral cats, overlap remains higher in the 'valleys' of low-use areas.

The distance analysis potentially explains how feral cats can overlap with dingoes in space, even using shared areas more than expected in some cases, while continuing to minimise interference competition. The low probability of recording a dingo and feral cat within approximately 1,200 m of each other, and particularly within about 300 m, suggests that feral cats reduce the frequency of close encounters, either by avoiding an area when a dingo is present, or by moving to what they consider a safe distance, upon encountering a dingo.

123 Chapter 5: Spatial overlap

Broekhuis et al. (2013) found that cheetahs Acinonyx jubatus, which are well-known to be persecuted by lions Panthera leo and spotted hyaenas Crocuta crocuta, did not avoid areas of high-use by the larger predators. Rather, cheetahs were found further than expected from lions, potentially judging the immediate risk associated with an area, rather than avoiding risky areas altogether. This suggests that predators may adopt 'reactive' avoidance mechanisms such as avoiding close encounters, that carry lower costs, as opposed to 'predictive' mechanisms such as spatial avoidance (Broekhuis et al. 2013), for sharing areas with apex predators may be unavoidable to access prey and other resources. My results suggest that feral cats may be flexible in their use of reactive and predictive mechanisms to avoid dingoes, as I observed both spatial and temporal avoidance. While one dingo (DM_F5) and cat (CM_M6) were simultaneously attracted to their shared space (Figure 5.6c), they were rarely close together

(Appendix G.5), suggesting this cat may have used reactive avoidance to limit encounters.

Dingo-dingo interactions

The extent of overlap between dingoes varied, consistent with the prediction that they often engage in social behaviour. Some dingoes exhibited relatively high levels of static overlap, and symmetrical, simultaneous attraction to the overlap zone, and frequent proximity, suggesting these three individuals were part of a social pack (Corbett 1995a). On the other hand, some pairs had little static overlap, avoided the overlap zone, and were rarely recorded close together, suggesting that individuals from several packs were collared, and that overlap between adjacent packs was comparatively low. Robley et al. (2010a) found similar high levels of overlap between dingoes, assumed to be members of the same pack, and less overlap with neighbouring individuals or packs. Territoriality, whereby an area is defended against non-pack individuals, is a common behaviour in canids (e.g. Ethiopian wolf, Sillero-Zubiri & Macdonald

1998) and has been previously recorded in dingoes (Thomson et al. 1992a; Corbett 2001;

Purcell 2010). Minta (1992) suggested that symmetrical avoidance, with no temporal interaction, which was recorded in one dingo pair, may be interpreted as classic territoriality.

124 Chapter 5: Spatial overlap

This could suggest that packs in our study area maintain territories excluding neighbouring dingoes.

Spatial overlap alone, however, does not necessarily indicate that the dyads involved are members of a social group. Despite some spatial overlap, Thomson et al. (1992a) observed only three occasions where dingoes from overlapping packs were actually in close proximity. Two of the male dingoes that I tracked in 2011 exhibited significant symmetrical, simultaneous attraction, but examination of their distance distributions told a different story. The space use of

DM_M2 was much broader and absorbed much of the space use of the DM_M1. However, they were recorded within 1,000 m of each other for less than 2% of fixes, perhaps because they were both drawn to valuable resources, such as carcasses, within the disjunct overlap zone at different times. After 40 days, DM_M2 relocated approximately 20 km north and remained there for the last month of data collection. This individual may have been a transient animal, temporarily passing through while attempting to locate a vacant territory. Alternatively, he may represent an 'outcast' that faced aggression from the pack (Corbett 1988) or a 'loner' with infrequent contact with conspecifics, that finally dispersed to a new area (Thomson et al.

1992b). Previous studies have found that males are more likely to disperse, to similar distances as this individual (Thomson et al. 1992b).

Cat-cat interactions

There was very little overlap between cat dyads, even at the 95% contour. The data suggest that male cats, when they do overlap in this area, only do so to a small extent. They did not come within 250 m of each other, tended to avoid using the overlap zone, and rarely overlapped in high-use areas (Ewer 1968). These patterns of space use are consistent with male feral cats avoiding potential encounters with conspecifics. A recent study of male cats on the

Kerguelen archipelago suggested that individuals avoided interactions in time, despite considerable spatial overlap (Martin et al. 2013a). Most of the collared feral cats in this study used space in a mosaic fashion, adjacent to other males, but rarely overlapping (Figure 5.4).

Similar results, in which male cats did not share space, have been recorded in semi-arid

125 Chapter 5: Spatial overlap

environments in Victoria (Jones & Coman 1982) and central Australia (Edwards et al. 2001), however contrary results from Dirk Hartog Is in WA observed considerable overlap between individuals (Hilmer 2010). Given that extensive overlap frequently occurs in areas with a supply of plentiful or concentrated resources (Liberg et al. 2000; Denny et al. 2002), the lower levels of overlap I observed may suggest resources were more evenly dispersed.

Although our results suggest range exclusivity or even territoriality between feral cats, these states are difficult to confirm, as they require data on all individuals in the population to rule out potential overlap. Liberg et al. (2000) suggested cats maintain exclusive ranges when mean overlap in convex polygons was less than 10%; mean overlap between male feral cats in this study was just over 10% for minimum convex polygons including all data (Appendix H).

However, other feral cats were recorded within the home ranges of our collared cats (H.

McGregor, unpubl. data), and as our study was mostly limited to large male cats, we cannot rule out possible overlap between sexes or size classes. Hence it may be that a) large males maintain exclusive ranges, but overlap with sub-adults and females, or b) that the sample size was insufficient to identify overlapping males. Dominant male cats in southern Sweden held exclusive ranges outside the breeding season, presumably to maintain access to food resources, and overlapped subordinate males (Liberg et al. 2000). Further research would be required to determine the extent to which different sex-age classes overlapped at our site, and how overlap varies with season and resource availability.

Caveats and future research directions

Our study did not include all individuals occupying the study area, and was limited by small sample size, a frequent challenge of carnivore studies. We aimed to ensure that animals were sympatric by limiting trapping to a defined area, but it would be unrealistic to expect simultaneous tracking of all individuals. Dingo trap success declined rapidly following an initial naïve period, and their large ranges mean that individuals could easily enter or leave the study area during the tracking period. Unfortunately we could not collect data on more than one female cat, preventing any comment on female space use. Despite these issues, all individuals

126 Chapter 5: Spatial overlap overlapped with at least one other predator, and we collected data on dingoes from several packs.

Low overlap recorded between a dingo and a feral cat does not necessarily mean that that cat has low overlap with all sympatric dingoes. As we collected data on only a sample of the cat and dingo populations, we cannot rule out the potential for higher overlap with other, untracked dingoes in an area. Feral cats may be balancing overlap with different sympatric dingoes (see below), or there may not be any dingo-free areas available. However, if feral cats generally avoided dingoes in the landscape, I would still expect to observe an overall trend of avoidance between dingo-cat dyads. By also estimating the distance separating dingoes and feral cats, I can measure how feral cats respond to individual dingoes in time, which might not be influenced by uncollared dingoes in the vicinity to the same extent as spatial overlap.

My conclusions are also limited by methods that only measure overlap between two individuals (Minta 1992), hence the influence of overlap with one individual on overlap with another, could not be examined. If a greater proportion of the population could have been collared, it would have been interesting to explore whether feral cats avoided overlap in areas used by multiple dingoes with a compounded risk of encounter. Habitat complexity may also affect spatial overlap. Feral cats may come into proximity with dingoes, or enter high-use areas, if the habitat provides cover to avoid encounters. Alternatively, if an area contains valuable resources, the benefits of sharing the space may outweigh the risks of encounter.

In some cases it was difficult to determine how important the overlap zone was to the predators that occupied it. As overlap zones were generally on the outskirts of an animal's space use, it was difficult to confirm whether the area was actively avoided, whether it was just used less regardless of the other predator’s movements, or even whether it was used at all, or was actually an artefact of oversmoothing. Restrictions of the size and number of fixes within the overlap zone minimised identification of an overlap zone that did not actually exist. Information on the quality of the overlap zone, such as availability of prey, as well as long-term data, may improve the definition of shared space between the two predators. Further sampling in different

127 Chapter 5: Spatial overlap

seasons may also shed light on the influence of behavioural patterns on overlap, e.g. during the dingo breeding season when dingoes are more sociable (Thomson 1992d).

Implications for biodiversity conservation

As discussed in Chapter 4, the more even space use of cats may, over time, create more continuous overlap with prey populations, preventing recovery of prey populations susceptible to cat predation. This study provided evidence that male feral cats may be territorial, as they consistently had very little spatial overlap, and boundaries that appeared to abut nearby conspecifics (Figure 5.4). This potential for 'continuous cover' of male feral cats could further contribute to consistent levels of predation risk across the landscape.

It does appear that feral cats can coexist in space with dingoes to some extent. Although overlap in core areas was lower, they can coexist outside these areas. Dingoes concentrate much of their activity in core areas that form a smaller proportion of their space use compared to feral cats (Chapter 4). As a consequence, much of their occupied space could be used by feral cats with minimal risk of encounter, although increased dingo density may increase the spread of dingo core areas across the landscape, limiting this potential refuge. Further, while feral cats appeared to frequently avoid the overlap zone with adjacent dingoes, it was not rare for them to also be attracted to these shared landscapes. However, close encounters between feral cats and dingoes were rare, suggesting that the mesopredators may be largely excluded from the vicinity of high-use areas such as dens, where dingoes maintain a constant presence, and are excluded temporally from areas frequented by dingoes. Whether this provides benefits to native fauna in these high-use areas, that offsets the costs of dingo predation, requires further study.

Conclusion

This study suggests that feral cats do adjust their use of space through time and across the landscape in response to dingoes, as encounters between the two predators were rare.

However, feral cats were not restricted to areas unoccupied by the larger predator. As dingoes may cover large ranges outside their core areas, for feral cats to avoid these areas completely could represent an unreasonable trade-off in resources. Rather, feral cats apparently reduce the

128 Chapter 5: Spatial overlap risk of encounter by avoiding dingo core areas and maintaining a 'safe' distance from sympatric dingoes. Feral cats could thus avoid intraguild interactions by estimating immediate risk in terms of proximity and probability of use by dingoes, rather than circumventing all areas occupied by the apex predator.

Author contributions

LB conceived the study, LB and HM collected field data, LB analysed the data and wrote the paper, CJ and LS assisted with analysis, CJ, LS, SL and ER assisted with editing.

129 Chapter 5: Spatial overlap

130

Chapter 6 General discussion

Invasive species have caused extensive damage across the globe (Mack et al. 2000), often placing unsustainable predation pressure on naïve prey (Salo et al. 2007), and disrupting trophic interactions (McNatty et al. 2009), evolutionary processes (Mooney & Cleland 2001) and ecosystem services (Crooks 2002). Australia is no exception; introduced species have contributed to widespread ecological problems (Rolls 1969; McLeod 2004). For example, many

CWR mammals dig while foraging, creating pits that facilitate seed germination (James et al.

2009). Introduced mesopredators such as feral cats have contributed to declines and extinctions of digging marsupials (Johnson et al. 2007), potentially degrading this important function

(James & Eldridge 2007; Fleming et al. in press). Once invasive mesopredators are established on the mainland, they are virtually impossible to eradicate (Bomford & O'Brien 1995), and control efforts may need to be continuous and widespread to be effective (Sharp & Saunders

2005; Mosnier et al. 2008). The potential for dingoes to supply cost-effective, broad-scale suppression of mesopredators has hence garnered considerable interest in recent years (Glen et al. 2007), and encouraged research into how dingoes might provide conservation benefits and restore lost ecosystem services (Letnic et al. 2009b; Letnic et al. 2012; Ritchie et al. 2012).

However, dingoes stir conflicted opinions in science, management and the broader population. The positive concept of dingoes as a native apex predator contrasts with a negative view of the dingo as a pest that damages livestock and imposes social and economic costs on rural communities (Allen & Sparkes 2001; McLeod 2004; Gong et al. 2009). Therefore, it is important to understand the predatory, competitive and behavioural mechanisms by which dingoes may affect top-down trophic cascades, for example by suppressing mesopredators, and how they vary depending on landscape context. This information will help to predict the effectiveness of dingoes as trophic regulators, and to assess whether the benefits of apex predators outweigh the costs of retaining dingoes in the landscape.

131 Chapter 6: General Discussion

My thesis sought to enhance current understanding of the mechanisms that underpin ecological interactions between dingoes and feral cats in northern Australia. The findings highlight the need for far greater consideration of the impacts of dingoes on feral cat behaviour, in addition to more direct interactions. Risk effects not only affect the behaviour of individuals, but can also alter the dynamics of mesopredator populations (Heithaus et al. 2008). In this chapter, I synthesise the major findings of the research, discuss the potential for dingoes to suppress feral cats in northern Australia and make suggestions for future research.

Summary of major findings

Aim 1: Determine whether dingoes can numerically suppress feral cat populations.

Objective 1: Assess the effect of baiting on dingo abundance indices and consequent effects on feral cat indices

In Chapter 2, I assessed the effect of predator control on abundance indices of dingoes and feral cats. I found that indices of dingo abundance were lower on properties that practised predator control, although the effect size was variable. On properties where predator control strongly reduced dingo indices compared to the paired unbaited property, I recorded higher indices of feral cat abundance. In areas where predator control did not reduce the index of dingo abundance, feral cat indices tended to be higher on properties without predator control. This suggests that where dingo populations have been artificially reduced, feral cat populations can increase in size or activity.

Objective 2: Assess levels of dietary overlap between dingoes and feral cats and determine the extent of intraguild predation

In Appendix A, I report results from dietary analysis of dingo and feral cat samples collected from study sites in northern Australia. I could not collect sufficient feral cat samples to

132 Chapter 6: General Discussion analyse dietary overlap or comment on prey consumed by feral cats. I found that dingoes consumed a variety of mainly mammalian prey across the study areas. The most frequent or dominant prey appeared to differ by region, from cattle in the Kimberley, to medium and large native mammals in the Einasleigh Uplands. The large volume and percentage of small mammals on Cape York Peninsula and the Gulf Plains suggest that dingoes and feral cats may overlap in diet in these areas. I also recorded a case of intraguild predation on Cape York Peninsula, suggesting that territorial or predatory aggression occurred in at least one area of this study.

Aim 2: Further understanding of predator ecology

Objective 3: Compare space use patterns of dingoes and feral cats

In Chapter 4, I calculated seasonal space use estimates and compared patterns of space use intensity for sympatric dingoes and feral cats in the Kimberley, WA. Both dingoes and feral cats left their core areas in crepuscular periods, just after dawn or before dusk, suggesting that core areas are not the only locations in which these predators hunt. Dingoes used larger areas than feral cats, but they concentrated more of their activity in smaller, more intense core areas relative to overall space use. This result was consistent for general utilisation distributions, as well as areas that were used frequently or for long periods. Dingoes tended to shift the areas they used over time more than feral cats. Space use estimates of feral cats were more contiguous compared to dingoes and were split over fewer areas. These results suggest that predation pressure from feral cats may be spread relatively evenly across the landscape, while predation pressure from dingoes may vary from low-risk areas they use rarely, to high-risk core areas.

This variable gradient of predation risk may provide more 'valleys' of low predation risk across the landscape that could act as refuges for prey, provided they are not occupied by other dingo core areas.

In Chapter 5, I described spatial overlap between predators. Dingoes associated strongly when presumably belonging to the same pack, resulting in high levels of overlap, attraction to shared areas in space and time and higher than expected frequencies of close encounters.

133 Chapter 6: General Discussion

However, there appeared to be little contact between adjacent packs, indicated by lower levels of overlap between some individuals, avoidance of shared areas and lower than expected frequencies of close encounters. There was minimal overlap between the space use estimates of feral cats, most of which were large males. Space use appeared to abut adjacent conspecifics, with relatively small gaps between individuals. These results suggest that large male feral cats may hold exclusive territories, although other uncollared individuals were recorded within the study area, implying that overlap may occur with females or subordinate animals.

Aim 3: Explore how dingoes potentially influence the behaviour of feral cats

Objective 4: Determine whether feral cats alter their temporal activity to avoid dingoes

In Chapter 2, I assessed the effect of predator control on the temporal activity of dingoes, and whether feral cat activity patterns responded to potential differences in dingo activity. There was a negative relationship between the proportion of dingo and feral cat records in the early evening, between properties. Overall, dingoes exhibited a broad bimodal activity pattern in areas without predator control, with crepuscular peaks and more activity at night than during the day. In contrast, feral cat activity patterns peaked in the middle of the night, which corresponded to the nocturnal period with the lowest levels of dingo activity. In areas with predator control, dingo activity patterns lacked a peak in the early evening, replaced by a larger activity peak for feral cats during this period. This suggests that feral cats may be able to co- occur with dingoes by adjusting their activity to avoid interactions in time. In areas where predator control is used, differences in dingo activity may relax temporal restrictions on feral cat activity, providing them with extended access to prey.

Objective 5: Determine whether feral cats alter their use of habitat to avoid dingoes

In Chapter 3, I assessed whether feral cats use habitat features that would minimise interactions with dingoes, either by avoiding areas or habitats where or when dingoes are more

134 Chapter 6: General Discussion common, or by using protective habitats that reduce the risk of encounter or death. In the complex woodland matrix of Cape York Peninsula, feral cats appear to frequently use roads, despite these features being favoured by dingoes. Cats were also recorded in habitats with relatively less ground cover complexity, which could provide easier movement or improved visibility, either for locating prey, or for early detection of dingoes. In the more open grassland/woodland habitats of the Gulf Plains, I did not record strong habitat associations for feral cats, however I recorded them more frequently on the property without predator control, where dingo activity appeared to be low. Feral cat activity also tended to be higher in areas with relatively less tree cover within 3 km2. In open woodland habitats, early detection could allow feral cats to reach the safety of nearby trees upon encountering a dingo, reducing the likelihood of capture. Habitats with sufficient tree cover for cats to easily escape may therefore facilitate coexistence between dingoes and feral cats and reduce aggressive intraguild interactions

(Janssen et al. 2007). Although I would expect areas with little tree cover to be habitats in which cats are particularly vulnerable to aggression by dingoes, it may be that low dingo activity on the Gulf Plains reduced the risk of dingo interaction to an acceptable level, allowing feral cats to exploit this otherwise risky habitat. Increased dingo activity on the Gulf Plains could enhance the risk associated with open habitats, potentially leading to shifts in feral cat activity away from these areas.

Objective 6: Determine whether feral cats alter their use of space to avoid dingoes

In Chapter 2, I compared trap rates for feral cats and dingoes at individual camera stations across central and northern Australia. I found a negative triangular relationship between dingo and feral cat trap rates, similar to that observed between dingoes and foxes in eastern

Australia (Johnson & VanDerWal 2009). No cats were recorded at cameras that recorded more than one dingo per night, and feral cats reached their highest trap rates at cameras with little or no dingo activity. This suggests that dingoes may have a limiting effect on feral cats on a local scale, effectively excluding them from areas of frequent dingo activity.

135 Chapter 6: General Discussion

In Chapter 5, I investigated how feral cats use space when in sympatry with dingoes. I found that feral cats do use the same areas as dingoes, but that spatial overlap is much lower at the core compared to the 95% contour. Although the two predators share space in the landscape, feral cats were recorded less frequently than expected within 1,000 m of dingoes, and within

100 m on less than 0.01% of fixes. It appears that feral cats can maintain separation from dingoes, either by avoiding close encounters with the larger predator, or moving to a 'safe' distance if a dingo approaches. The variability in dynamic responses to shared space suggests that feral cats may adopt a flexible approach to minimising interactions with dingoes. They appear capable of using either a predictive response to risk, by avoiding dingo high-use areas, or a reactive response, by minimising close encounters (Broekhuis et al. 2013).

Can dingoes effectively suppress feral cats in northern

Australia?

This study has shown that feral cat behaviour can be influenced by dingoes, and that feral cat activity and potentially abundance can be higher when dingoes are removed. This suggests that mesopredator release can occur in northern Australia. It is unlikely that dingoes are capable of locally eradicating or permanently excluding feral cats, as feral cats are efficient hunters, which may facilitate coexistence (Polis & Holt 1992). Monitoring and control of mesopredators may still be required in some cases, despite dingo presence, to ensure that feral cats are not selecting vulnerable taxa. However, temporary exclusion, and limitations to feral cat behaviour, could provide tangible benefits to vulnerable prey populations that should be explored in further studies. Below, I describe how feral cats may avoid severe intraguild interactions, and outline factors that could mediate the strength of direct or indirect effects, and the potential for dingoes to limit the impacts of feral cats.

136 Chapter 6: General Discussion

Behavioural games between clever predators?

Srinivasan et al. (2010) modelled the behavioural game adopted by marine mammal predators and prey, noting that despite calculated antipredator strategies, clever prey can still be outwitted by clever predators. But who wins when the game is played between two clever predators? Although feral cats may shift behaviour to avoid intraguild encounters (Chapters 2 &

5), dingoes are more likely to be maximising overlap with prey rather than with feral cats.

Predators are more likely to influence prey or mesopredators in lower trophic guilds when they are focused on locating these species, i.e. when a behavioural response race is occurring (Sih

1984; Laundré 2010). As feral cats are not a main prey item for dingoes, they are unlikely to face targeted harassment from dingoes. Feral cats may use areas that dingoes use intensely or frequently, and as a mobile species, they can move when dingoes get too close, further minimising cases of close proximity (Chapter 5). Direct intraguild interactions may therefore only occur if feral cats fail to avoid encounters, by getting too close to dingoes, by using habitats that increase the risk of detection or capture, or when dingo density or the extent of competition is at a level where the first two options become unavoidable.

Habitat complexity

Habitat complexity is likely to play an important role in mediating intraguild interactions between dingoes and cats, as it should affect the probability of both detection and capture. Cats may be able to coexist with dingoes in areas where habitat provides an escape route (McGee et al. 2006; Heithaus et al. 2009). As in north Queensland, woodland habitats may facilitate co-occurrence of predators, allowing feral cats to avoid capture by climbing trees.

I would expect feral cats to be particularly vulnerable in open habitats, where dingoes may easily detect and outrun them before they can reach wooded areas. Although feral cats used areas with very little tree cover on the Gulf Plains, low levels of dingo activity may negate the risk associated with these habitats (Chapter 3).

137 Chapter 6: General Discussion

Strength of competition

Previous research has shown that intraguild interactions are stronger when dietary overlap is high (Polis et al. 1989), and when subordinate predators are not too close, or too different, in size to the dominant predators (Donadio & Buskirk 2006). Where this is true for feral cats and dingoes, one could expect intraguild aggression to occur, as dingoes attempt to remove the smaller competitors from their territories. However, the extent of dietary overlap, and hence competition, will vary spatially and temporally across northern Australia, depending on available prey. If resources are lacking, feral cats may accept greater levels of risk to acquire prey, and dingoes may be more motivated to pursue intraguild aggression. Hence the strength of mesopredator suppression is also likely to vary with landscape context.

Predator control

Predator control is used to reduce impacts of dingoes on livestock or wildlife, by reducing the size of dingo populations. A threshold population size may be required for dingoes to be ecologically effective as apex predators, with the number dependent on bottom-up processes and species composition (Soulé et al. 2003). If predator control reduces dingo populations, it may also have detrimental effects on dingo-entrained top-down cascades (Linnell

& Strand 2000; Choquenot & Forsyth 2013). As areas of low dingo activity appear to be available to feral cats (Chapter 5), the reduced frequency of core areas across the landscape may increase the area in which feral cats can occur with limited mesopredator suppression.

In addition to directly removing dingoes, artificial control can also effect behavioural changes in apex predators, by destabilising social structures (Wallach et al. 2009b) or by imposing risk effects, where predators adjust their activity, potentially to avoid interactions with humans (Lesmerises et al. 2012) (Chapter 2). Human-imposed risk effects may then affect how apex predators interact with mesopredators and prey. A dingo population sufficiently reduced in size, and with restricted behaviours, could be ecologically impotent, imposing negligible impacts on lower trophic guilds.

138 Chapter 6: General Discussion

Dynamic trophic interactions

Trophic interactions are dynamic, shifting in nature and magnitude, and in response to changes in productivity and human intervention (Terborgh et al. 2010). My results suggest that the landscape of fear imposed on feral cats by dingoes varies not only spatially, but through time. While feral cats in the Kimberley used space more consistently, dingoes focused their activity on smaller, temporally-variable core areas (Chapter 4). This gradient of dingo space use may facilitate coexistence and provide refugia for feral cats from dingo interference in adjacent areas of low use (Chapter 5). As dingo core areas shift over time, areas that feral cats are excluded from are also likely to shift both spatially and temporally. While this process may provide interim refuge for prey species from cat predation, it also implies that feral cats can co- occur with dingoes, as long as they avoid current areas of high use. Future research should explore whether increased densities of dingoes, and hence dingo core areas, excludes feral cats from a larger proportion of the landscape.

Future research directions

This research has highlighted several possible avenues for additional research, to further understand how both contextual and intrinsic factors influence intraguild interactions between dingoes and feral cats. Some suggestions would elaborate or contribute to outcomes of this thesis, while others represent ideal methodologies to best establish causal evidence and interactions through time.

Integrate prey and habitat information

Prey survey data would complement analyses of predator habitat use, potentially increasing the deviance explained in the modelling process. The distribution of resources in different habitats could be estimated from prey diversity and abundance indices, which could then clarify how feral cats trade off the benefits of resource-rich areas against the risk of dingo interactions.

139 Chapter 6: General Discussion

By combining spatial layers of prey diversity and abundance with predator space use data, one could explore how predation pressure from dingoes and feral cats influences prey communities, and whether the exclusion of cats from dingo high-use areas provides a net benefit to prey, despite potential increased predation from dingoes. Incorporating habitat layers with predator spatial data could also assess whether feral cats enter areas used frequently by dingoes, or accept closer proximity to dingoes, with the protection of higher habitat complexity.

Investigate prevalence of interspecific killing

The low frequency of intraguild predation in dingo diet studies suggests that dingoes do not target feral cats as a prey item, which has been interpreted as evidence that interference competition between the predators is weak. However, territorial aggression, which is more likely due to interference competition than hunger, may lead to interspecific killing without predation (Polis et al. 1989) and would not be recorded in diet studies. Long-term telemetry studies of feral cats could identify cause of death and estimate the prevalence of interspecific killing. This data would help to quantify the risk faced by feral cats, and justify their motivation to avoid dingoes.

Experimental studies

Despite the logistical difficulties and long-term commitments associated with a Before-

After/Control-Impact (BACI)-style study, a broad-scale experiment could provide the sought- after causal evidence that dingoes reduce feral cat abundance and behaviour and have beneficial effects on lower trophic guilds (Glen et al. 2007; Brashares et al. 2010). Ideally, an experiment would involve pairs of sites large enough to encompass a considerable dingo population, matched for habitat, climate and land use practices. The impact site would either relax predator control and allow dingoes to re-establish, or effectively remove dingoes with predator control, and then monitor the short- and long-term effects on predator and prey abundance and behaviour. Experiments could also provide an opportunity to quantify whether hunting behaviour of feral cats is compromised by risk effects that restrict behaviour, and how this may translate into reduced effects on prey populations. Novel methods to estimate feral cat kill rates

140 Chapter 6: General Discussion could clarify spatial and temporal variation in hunting success, and determine whether suppression by dingoes reduces the optimal hunting behaviour of feral cats. Predator removal experiments may need to consider the ethical consequences of removing dingoes, as increases in mesopredator populations could have long-lasting detrimental effects on wildlife.

Incorporate resource fluctuations

Bottom-up processes are also likely to influence the extent of dietary overlap and strength of interference competition between predators (Elmhagen & Rushton 2007). A long- term monitoring study of sympatric dingoes and feral cats, encompassing surveys before, during and following a resource boom period, could document changes in prey availability, dietary overlap, interspecific killing and feral cat behaviour as the strength of competition fluctuates.

Concluding remarks

Polis et al. (1989) stated that "interactions in natural communities are complex". This study has demonstrated that the ecological interactions between mobile, terrestrial predators, in this case dingoes and feral cats, are indeed complicated by context-dependent factors such as predator control and habitat complexity. Here, I have highlighted the importance of considering behavioural mechanisms when examining how apex predators might impose top-down control on mesopredators. Although direct intraguild interactions may not be evident, risk effects that change the activity patterns, and use of habitat and space by a mesopredator could influence their interactions with prey species. By assessing not only the abundance but the behaviour of intraguild predators, managers should derive a more detailed picture of how dingoes may limit feral cats and provide conservation benefits in northern Australia.

141 Chapter 6: General Discussion

142

References

Abbott, I. (2002) Origin and spread of the cat, Felis catus, on mainland Australia, with a discussion of the magnitude of its early impact on native fauna. Wildlife Research 29 (1): 51-74.

Abbott, I. (2008) The spread of the cat, Felis catus, in Australia: re-examination of the current conceptual model with additional information. Conservation Science Western Australia 7 (1): 1-17.

Agresti, A. (2007) An introduction to categorical data analysis. Second edn. John Wiley & Sons, Inc, Hoboken, New Jersey.

Algar, D., Angus, G.J., Williams, M.R. & Mellican, A.E. (2007) Influence of bait type, weather and prey abundance on bait uptake by feral cats (Felis catus) on Peron Peninsula, Western Australia. Conservation Science Western Australia 6 (1): 109-149.

Algar, D. & Burrows, N.D. (2004) Feral cat control research: Western Shield review - February 2003. Conservation Science Western Australia 5 (2): 131-163.

Algar, D.A., Burbidge, A.A. & Angus, G.J. (2002) Cat eradication on Hermite Island, Montebello Islands, Western Australia. Turning the tide: the eradication of invasive species: Proceedings of the International Conference on Eradication of Island Invasives (eds C.R. Veitch & M.N. Clout) pp. 14-18. IUCN SSC Invasive Species Specialist Group, Gland.

Allen, B.L. (2010) Skin and bone: observations of dingo scavenging during a chronic food shortage. Australian Mammalogy 32 (2): 207-208.

Allen, B.L. & Fleming, P.J.S. (2012) Reintroducing the dingo: the risk of dingo predation to threatened vertebrates of western New South Wales. Wildlife Research 39 (1): 35-50.

Allen, B.L. & Leung, L.K.P. (2012) Assessing predation risk to threatened fauna from their prevalence in predator scats: dingoes and rodents in arid Australia. Plos One 7 (5): e36426.

Allen, B.L. & Miller, H.A. (2009) The biodiversity benefits and the production costs of dingoes in the arid zone: Summary of research results for 2008. South Australian Arid Lands Natural Resources Management Board, Port Augusta.

Allen, L. & Byrne, D. (2008) Are we focussing wild dog control the wrong time of the year and going about it the wrong way? Proceedings of the 14th Australasian Vertebrate Pest Conference (eds G. Saunders & C. Lane). Darwin.

Allen, L. & Gonzalez, T. (1998) Baiting reduces dingo numbers, changes age structures yet often increases calf losses. Australian Vertebrate Pest Conference, pp. 421–428. Bunbury, WA.

143 Chapter 7: References

Allen, L., Goullet, M. & Palmer, R. (2012) The diet of the dingo (Canis lupus dingo and hybrids) in north-eastern Australia: a supplement to the paper of Brook and Kutt (2011). The Rangeland Journal 34 (2): 211-217.

Allen, L., Thomson, P.C. & Lisle, A. (2000) Pack size and prey behaviour affects prey selection and the predation of livestock by dingoes. Bureau of Rural Sciences, Australian Government.

Allen, L.R. (in press) Wild dog control impacts on calf wastage in extensive beef cattle enterprises. Animal Production Science.

Allen, L.R. & Sparkes, E.C. (2001) The effect of dingo control on sheep and beef cattle in Queensland. Journal of Applied Ecology 38 (1): 76-87.

Aplin, K.P. & Redhead, T.D. (2008) Canefield Rat. The mammals of Australia (eds S. Van Dyck & R. Strahan) pp. 697-698. Reed New Holland, Sydney.

Arias-Del Razo, I., Hernández, L., Laundré, J.W. & Velasco-Vázquez, L. (2012) The landscape of fear: habitat use by a predator (Canis latrans) and its main prey (Lepus californicus and Sylvilagus audubonii). Canadian Journal of Zoology 90 (6): 683-693.

Arjo, W.M. & Pletscher, D.H. (1999) Behavioral responses of coyotes to wolf recolonization in northwestern Montana. Canadian Journal of Zoology 77 (12): 1919-1927.

Austin, M.P. (2002) Spatial prediction of species distribution: an interface between ecological theory and statistical modelling. Ecological Modelling 157 (2-3): 101-118.

Banks, P.B. & Dickman, C.R. (2007) Alien predation and the effects of multiple levels of prey naivete. Trends in Ecology & Evolution 22 (5): 229-230.

Barclay, R.M.R. & Brigham, R.M. (1991) Prey detection, dietary niche breadth, and body size in bats: Why are aerial insectivorous bats so small? American Naturalist 137 (5): 693- 703.

Batschelet, E. (1981) Circular statistics in biology. Academic Press, London.

Bengsen, A., Butler, J. & Masters, P. (2011) Estimating and indexing feral cat population abundances using camera traps. Wildlife Research 38 (8): 732-739.

Bengsen, A.J., Butler, J.A. & Masters, P. (2012) Applying home-range and landscape-use data to design effective feral-cat control programs. Wildlife Research 39 (3): 258-265.

Benhamou, S. (2011) Dynamic approach to space and habitat use based on biased random bridges. Plos One 6 (1): e14592.

Benhamou, S. & Cornélis, D. (2010) Incorporating movement behavior and barriers to improve kernel home range space use estimates. Journal of Wildlife Management 74 (6): 1353- 1360.

144 Chapter 7: References

Benhamou, S. & Riotte-Lambert, L. (2012) Beyond the Utilization Distribution: Identifying home range areas that are intensively exploited or repeatedly visited. Ecological Modelling 227: 112-116.

Bennett, A.F. (1983) Ecological consequences of activity metabolism. Lizard ecology: Studies of a model organism (eds R.B. Huey, E.R. Pianka & T.W. Schoener) pp. 1-23. Harvard University Press, Cambridge, Massachusetts.

Berdoy, M., Webster, J.P. & Macdonald, D.W. (2000) Fatal attraction in rats infected with Toxoplasma gondii. Proceedings of the Royal Society B: Biological Sciences 267 (1452): 1591–1594.

Berger, J. (2010) Fear-mediated food webs. Trophic Cascades: Predators, prey, and the changing dynamics of nature (eds J. Terborgh & J.A. Estes) pp. 241-254. Island Press, Washington D.C.

Berger, J., Stacey, P.B., Bellis, L. & Johnson, M.P. (2001) A mammalian predator-prey imbalance: Grizzly bear and wolf extinction affect avian neotropical migrants. Ecological Applications 11 (4): 947-960.

Berger, K.M. & Gese, E.M. (2007) Does interference competition with wolves limit the distribution and abundance of coyotes? Journal of Animal Ecology 76: 1075-1085.

Berger, K.M., Gese, E.M. & Berger, J. (2008) Indirect effects and traditional trophic cascades: A test involving wolves, coyotes, and pronghorn. Ecology 89 (3): 818-828.

Beschta, R.L. & Ripple, W.J. (2009) Large predators and trophic cascades in terrestrial ecosystems of the western United States. Biological Conservation 142 (11): 2401-2414.

Bishop, K.L., Pai, A.K. & Schmitt, D. (2008) Whole body mechanics of stealthy walking in cats. Plos One 3 (11): e3808.

Bjørneraas, K., Van Moorter, B., Rolandsen, C.M. & Herfindal, I. (2010) Screening global positioning system location data for errors using animal movement characteristics. Journal of Wildlife Management 74 (6): 1361-1366.

Bolker, B.M., Brooks, M.E., Clark, C.J., Geange, S.W., Poulsen, J.R., Stevens, M.H.H. & White, J.S.S. (2009) Generalized linear mixed models: a practical guide for ecology and evolution. Trends in Ecology & Evolution 24 (3): 127-135.

Bomford, M. & O'Brien, P. (1995) Eradication or control for vertebrate pests? Wildlife Society Bulletin 23 (2): 249-255.

Borer, E.T., Seabloom, E.W., Shurin, J.B., Anderson, K.E., Blanchette, C.A., Broitman, B., Cooper, S.D. & Halpern, B.S. (2005) What determines the strength of a trophic cascade? Ecology 86 (2): 528-537.

145 Chapter 7: References

Börger, L., Dalziel, B.D. & Fryxell, J.M. (2008) Are there general mechanisms of animal home range behaviour? A review and prospects for future research. Ecology Letters 11 (6): 637-650.

Bradshaw, C.J.A., Sims, D.W. & Hays, G.C. (2007) Measurement error causes scale-dependent threshold erosion of biological signals in animal movement data. Ecological Applications 17 (2): 628-638.

Bradshaw, J.W.S. (2006) The evolutionary basis for the feeding behavior of domestic dogs (Canis familiaris) and cats (Felis catus). Journal of Nutrition 136 (7): 1927S-1931S.

Bradshaw, J.W.S., Goodwin, D., Legrand-Defrétin, V. & Nott, H.M.R. (1996) Food selection by the domestic cat, an obligate carnivore. Comparative Biochemistry and Physiology Part A: Physiology 114 (3): 205-209.

Brandl, R., Kristín, A. & Leisler, B. (1994) Dietary niche breadth in a local community of passerine birds: an analysis using phylogenetic contrasts. Oecologia 98 (1): 109-116.

Brashares, J.S., Prugh, L.R., Stoner, C.J. & Epps, C.W. (2010) Ecological and conservation implications of mesopredator release. Trophic Cascades: Predators, prey, and the changing dynamics of nature (eds J. Terborgh & J.A. Estes). Island Press, Washington D.C.

Brawata, R.L. & Neeman, T. (2011) Is water the key? Dingo management, intraguild interactions and predator distribution around water points in arid Australia. Wildlife Research 38 (5): 426-436.

Breed, B. & Ford, F. (2007) Native mice and rats. CSIRO Publishing, Collingwood.

Broekhuis, F., Cozzi, G., Valeix, M., McNutt, J.W., Macdonald, D.W. & Fryxell, J. (2013) Risk avoidance in sympatric large carnivores: reactive or predictive? Journal of Animal Ecology 82 (5): 1097-1105.

Brook, L.A. & Kutt, A.S. (2011) The diet of the dingo (Canis lupus dingo) in north-eastern Australia with comments on its conservation implications. The Rangeland Journal 33 (1): 79-85.

Brothers, N.P., Skira, I.J. & Copson, G.R. (1985) Biology of the feral cat, Felis catus (L), on Macquarie Island. Australian Wildlife Research 12 (3): 425-436.

Brown, J.S., Laundré, J.W. & Gurung, M. (1999) The ecology of fear: Optimal foraging, game theory, and trophic interactions. Journal of Mammalogy 80 (2): 385-399.

Brunner, H. & Coman, B.J. (1974) The Identification of Mammalian Hair. Inkata Press, Melbourne.

146 Chapter 7: References

Buckmaster, A.J. (2011) Ecology of the feral cat (Felis catus) in the tall forests of far east Gippsland. PhD thesis, University of Sydney, Sydney.

Burbidge, A.A., Johnson, K.A., Fuller, P.J. & Southgate, R.I. (1988) Aboriginal knowledge of the mammals of the central deserts of Australia. Wildlife Research 15 (1): 9-39.

Burbidge, A.A. & Manly, B.F.J. (2002) Mammal extinctions on Australian islands: causes and conservation implications. Journal of Biogeography 29: 465-473.

Burbidge, A.A. & McKenzie, N.L. (1989) Patterns in the modern decline of western Australia's vertebrate fauna: Causes and conservation implications. Biological Conservation 50 (1- 4): 143-198.

Burnham, K.P. & Anderson, D.R. (2002) Model selection and multimodel inference: a practical information-theoretic approach. 2nd edn. Springer-Verlag, New York.

Burrows, N.D., Algar, D., Robinson, A.D., Sinagra, J., Ward, B. & Liddelow, G. (2003) Controlling introduced predators in the Gibson Desert of Western Australia. Journal of Arid Environments 55: 691-713.

Burt, W.H. (1943) Territoriality and home range concepts as applied to mammals. Journal of Mammalogy 24 (3): 346-352.

Bustard, H.R. (1967) Activity cycle and thermoregulation in the Australian gecko Gehyra variegata. Copeia 1967 (4): 753-758.

Byrne, S. (2009) Relationships between dingoes and their prey in northern Australia (unpublished report). University of Queensland and Australian Wildlife Conservancy, Brisbane.

Cade, B.S., Terrell, J.W. & Schroeder, R.L. (1999) Estimating effects of limiting factors with regression quantiles. Ecology 80 (1): 311-323.

Calenge, C. (2006) The package "adehabitat" for the R software: A tool for the analysis of space and habitat use by animals. Ecological Modelling 197 (3-4): 516-519.

Carbone, C., Teacher, A. & Rowcliffe, J.M. (2007) The costs of carnivory. PLoS Biology 5 (2): 363-368.

Carothers, J.H. & Jaksić, F.M. (1984) Time as a niche difference - the role of interference competition. Oikos 42 (3): 403-406.

Carthey, A.J.R. & Banks, P.B. (2014) Naïveté in novel ecological interactions: lessons from theory and experimental evidence. Biological Reviews: doi: 10.1111/brv.12087.

Catling, P.C. (1988) Similarities and contrasts in the diets of foxes, Vulpes vulpes, and cats, Felis catus, relative to fluctuating prey populations and drought. Australian Wildlife Research 15 (3): 307-317.

147 Chapter 7: References

Catling, P.C. & Burt, R.J. (1995) Studies of the ground-dwelling mammals of eucalypt forests in south-eastern New South Wales - the effect of habitat variables on distribution and abundance. Wildlife Research 22 (3): 271-288.

Catling, P.C. & Burt, R.J. (1997) Studies of the ground-dwelling mammals of eucalypt forests in north-eastern New South Wales: the species, their abundance and distribution. Wildlife Research 24 (1): 1-19.

Catling, P.C., Corbett, L.K. & Newsome, A.E. (1992) Reproduction in captive and wild Dingoes (Canis familiaris dingo) in temperate and arid environments of Australia. Wildlife Research 19 (2): 195-209.

Caughley, G., Grigg, G.C., Caughley, J. & Hill, G.J.E. (1980) Does dingo predation control the densities of kangaroos and emus? Australian Wildlife Research 7: 1-12.

Choquenot, D. & Forsyth, D.M. (2013) Exploitation ecosystems and trophic cascades in non- equilibrium systems: pasture - red kangaroo - dingo interactions in arid Australia. Oikos 122 (9): 1292-1306.

Christensen, P. & Burrows, N. (1994) Project desert dreaming: experimental reintroduction of mammals to the Gibson Desert, Western Australia. Reintroduction biology of Australian and New Zealand fauna (ed. M. Serena) pp. 199-207. Surrey Beatty & Sons, Chipping Norton.

Ciucci, P., Boitani, L., Francisci, F. & Andreoli, G. (1997) Home range, activity and movements of a wolf pack in central Italy. Journal of Zoology 243: 803-819.

Claridge, A.W. & Hunt, R. (2008) Evaluating the role of the Dingo as a trophic regulator: Additional practical suggestions. Ecological Management and Restoration 9 (2): 116- 119.

Claridge, A.W., Mills, D.J. & Barry, S.C. (2010) Prevalence of threatened native species in canid scats from coastal and near-coastal landscapes in south-eastern Australia. Australian Mammalogy 32 (2): 117-126.

Claridge, A.W., Mills, D.J., Hunt, R., Jenkins, D.J. & Bean, J. (2009) Satellite tracking of wild dogs in south-eastern mainland Australian forests: Implications for management of a problematic top-order carnivore. Forest Ecology and Management 258 (5): 814-822.

Corbett, L. (1995a) The dingo in Australia and Asia. University of New South Wales Press Ltd, Sydney.

Corbett, L. (1995b) Does dingo predation or buffalo competition regulate feral pig populations in the Australian wet-dry tropics? An experimental study. Wildlife Research 22 (1): 65- 74.

Corbett, L. (2001) The dingo in Australia and Asia, 2nd edn. J.B. Books Pty Ltd, Marleston.

148 Chapter 7: References

Corbett, L. & Newsome, A.E. (1987) The feeding ecology of the dingo III. Dietary relationships with widely fluctuating prey populations in arid Australia: an hypothesis of alternation of predation. Oecologia 74: 215-227.

Corbett, L.K. (1979) Feeding ecology and social organisation of wildcats (Felis silvestris) and domestic cats (Felis catus) in Scotland. PhD, Aberdeen University, Aberdeen.

Corbett, L.K. (1988) Social dynamics of a captive dingo pack - population regulation by dominant female infanticide. Ethology 78 (3): 177-198.

Corbett, L.K. (1989) Assessing the diet of dingoes from feces: a comparison of three methods. Journal of Wildlife Management 53 (2): 343-346.

Corbett, L.K. (2008) Dingo. The mammals of Australia (eds S. Van Dyck & R. Strahan) pp. 737-739. Reed New Holland, Sydney.

Cornélis, D., Benhamou, S., Janeau, G., Morellet, N., Ouedraogo, M. & de Visscher, M.-N. (2011) Spatiotemporal dynamics of forage and water resources shape space use of West African savanna buffaloes. Journal of Mammalogy 92 (6): 1287-1297.

Costa, G.C. (2009) Predator size, prey size and dietary niche breadth relationships in marine predators. Ecology 90 (7): 2014-2019.

Côté, S.D., Rooney, T.P., Tremblay, J.-P., Dussault, C. & Waller, D.M. (2004) Ecological impacts of deer overabundance. Annual Review of Ecology, Evolution, and Systematics 35 (1): 113-147.

Coulson, G. (1996) Anti-predator behaviour in marsupials. Comparison of marsupial and placental behaviour (eds D.B. Croft & U. Ganslosser) pp. 158-186. Filander Verlag GmbH, Furth.

Cox, J.G. & Lima, S.L. (2006) Naiveté and an aquatic–terrestrial dichotomy in the effects of introduced predators. Trends in Ecology & Evolution 21 (12): 674-680.

Crabtree, R.L. & Sheldon, J.W. (1999) The ecological role of coyotes on Yellowstone's northern range. Yellowstone Science 7 (2): 15-23.

Creel, S. (2001) Four factors modifying the effect of competition on carnivore population dynamics as illustrated by African wild dogs. Conservation Biology 15 (1): 271-274.

Creel, S. & Christianson, D. (2008) Relationships between direct predation and risk effects. Trends in Ecology & Evolution 23 (4): 194-201.

Creel, S. & Christianson, D. (2009) Wolf presence and increased willow consumption by Yellowstone elk: implications for trophic cascades. Ecology 90 (9): 2454-2466.

Creel, S., Christianson, D., Liley, S. & Winnie, J.A. (2007) Predation risk affects reproductive physiology and demography of elk. Science 315 (5814): 960-960.

149 Chapter 7: References

Creel, S. & Creel, N.M. (1996) Limitation of African wild dogs by competition with larger carnivores. Conservation Biology 10 (2): 526-538.

Creel, S. & Creel, N.M. (2002) The African wild dog: behavior, ecology and conservation. Princeton University Press, Princeton.

Creel, S., Winnie Jr, J.A., Maxwell, B., Hamlin, K. & Creel, M. (2005) Elk alter habitat selection as an antipredator response to wolves. Ecology 86 (12): 3387-3397.

Crooks, J.A. (2002) Characterizing ecosystem-level consequences of biological invasions: the role of ecosystem engineers. Oikos 97 (2): 153-166.

Crooks, K.R. & Soulé, M.E. (1999) Mesopredator release and avifaunal extinctions in a fragmented system. Nature 400: 563-566.

Csurhes, S., Paroz, G. & Markula, A. (2009) Pest animal risk assessment: Feral horse Equus caballus. The State of Queensland, Department of Employment, Economic Development and Innovation.

Cumming, G.S. & Cornélis, D. (2012) Quantitative comparison and selection of home range metrics for telemetry data. Diversity and Distributions 18 (11): 1057-1065.

Cupples, J.B., Crowther, M.S., Story, G. & Letnic, M. (2011) Dietary overlap and prey selectivity among sympatric carnivores: could dingoes suppress foxes through competition for prey? Journal of Mammalogy 92 (3): 590-600.

D'Eon, R.G. & Delparte, D. (2005) Effects of radio-collar position and orientation on GPS radio-collar performance, and the implications of PDOP in data screening. Journal of Applied Ecology 42 (2): 383-388.

D'Eon, R.G., Serrouya, R., Smith, G. & Kochanny, C.O. (2002) GPS radiotelemetry error and bias in mountainous terrain. Wildlife Society Bulletin 30 (2): 430-439.

Davis, M.L., Kelly, M.J. & Stauffer, D.F. (2011) Carnivore co-existence and habitat use in the Mountain Pine Ridge Forest Reserve, Belize. Animal Conservation 14 (1): 56-65. de Solla, S.R., Bonduriansky, R. & Brooks, R.J. (1999) Eliminating autocorrelation reduces biological relevance of home range estimates. Journal of Animal Ecology 68 (2): 221- 234.

Deban, S.M., Schilling, N. & Carrier, D.R. (2012) Activity of extrinsic limb muscles in dogs at walk, trot and gallop. Journal of Experimental Biology 215 (2): 287-300.

DeCesare, N.J., Squires, J.R. & Kolbe, J.A. (2005) Effect of forest canopy on GPS-based movement data. Wildlife Society Bulletin 33 (3): 935-941.

Denny, E. (2008) Cat. The Mammals of Australia (eds S. Van Dyck & R. Strahan) pp. 742-744. Reed New Holland, Sydney.

150 Chapter 7: References

Denny, E., Yakovlevich, P., Eldridge, M.D.B. & Dickman, C. (2002) Social and genetic analysis of a population of free-living cats (Felis catus L.) exploiting a resource-rich habitat. Wildlife Research 29 (4): 405-413.

Denny, E.A. & Dickman, C.R. (2010) Review of cat ecology and management strategies in Australia. Invasive Animals Cooperative Research Centre, Canberra.

Department of the Environment (2009) EPBC Act List of Threatened Fauna. Australian Government. Retrieved 28 Oct 2013, from http://www.environment.gov.au/cgi- bin/sprat/public/publicthreatenedlist.pl?wanted=fauna.

DEWHA (2008) Background document for the threat abatement plan for predation by the European red fox. Department of the Environment, Water, Heritage and the Arts, Canberra.

Di Bitetti, M.S., De Angelo, C.D., Di Blanco, Y.E. & Paviolo, A. (2010) Niche partitioning and species coexistence in a Neotropical felid assemblage. Acta Oecologica 36 (4): 403- 412.

Dickman, C.R. (1996) Overview of the impacts of feral cats on Australian native fauna. Director of National Parks and Wildlife, Australian Nature Conservation Agency and Institute of Wildlife Research, University of Sydney, Canberra.

Dickman, C.R., Glen, A.S. & Letnic, M. (2009) Reintroducing the Dingo: Can Australia's conservation wastelands be restored? Reintroduction of Top-Order Predators (eds M.W. Hayward & M.J. Somers). Blackwell Publishing.

Dickman, C.R. & Watts, C.H.S. (2008) Black Rat. The mammals of Australia (eds S. Van Dyck & R. Strahan) pp. 707-709. Reed New Holland, Sydney.

Donadio, E. & Buskirk, S.W. (2006) Diet, morphology, and interspecific killing in carnivora. American Naturalist 167 (4): 524-536.

Doncaster, C.P. (1990) Non-parametric estimates of interaction from radio-tracking data. Journal of Theoretical Biology 143 (4): 431-443.

Doncaster, C.P. & Macdonald, D.W. (1991) Drifting territoriality in the red fox Vulpes vulpes. Journal of Animal Ecology 60 (2): 423-439.

Dray, S. & Dufour, A.B. (2007) The ade4 package: implementing the duality diagram for ecologists. Journal of Statistical Software 22 (4): 1-20.

Driscoll, C.A., Menotti-Raymond, M., Roca, A.L., Hupe, K., Johnson, W.E., Geffen, E., Harley, E.H., Delibes, M., Pontier, D., Kitchener, A.C., Yamaguchi, N., O'Brien, S.J. & Macdonald, D.W. (2007) The Near Eastern origin of cat domestication. Science 317 (5837): 519-523.

Duffy, D.C. & Capece, P. (2012) Biology and Impacts of Pacific Island Invasive Species. 7. The Domestic Cat (Felis catus). Pacific Science 66 (2): 173-212.

151 Chapter 7: References

Duong, T. (2012) ks: Kernel smoothing. R package version 1.8.10. http://CRAN.R- project.org/package=ks

Duong, T. & Hazelton, M. (2003) Plug-in bandwidth matrices for bivariate kernel density estimation. Journal of Nonparametric Statistics 15 (1): 17-30.

Durant, S.M. (1998) Competition refuges and coexistence: an example from Serengeti carnivores. Journal of Animal Ecology 67 (3): 370-386.

Durant, S.M. (2000) Living with the enemy: avoidance of hyenas and lions by cheetahs in the Serengeti. Behavioral Ecology 11 (6): 624-632.

Edwards, G.P., De Preu, N., Crealy, I.V. & Shakeshaft, B.J. (2002) Habitat selection by feral cats and dingoes in a semi-arid woodland environment in central Australia. Austral Ecology 27: 26-31.

Edwards, G.P., de Preu, N., Shakeshaft, B.J., Crealy, I.V. & Paltridge, R.M. (2001) Home range and movements of male feral cats (Felis catus) in a semiarid woodland environment in central Australia. Austral Ecology 26 (1): 93-101.

Elledge, A.E., Allen, L.R., Carlsson, B.L., Wilton, A.N. & Leung, L.K.P. (2008) An evaluation of genetic analyses, skull morphology and visual appearance for assessing dingo purity: implications for dingo conservation. Wildlife Research 35 (8): 812-820.

Elledge, A.E., Leung, L.K.P., Allen, L.R., Firestone, K. & Wilton, A.N. (2006) Assessing the taxonomic status of dingoes Canis familiaris dingo for conservation. Mammal Review 36 (2): 142-156.

Elmhagen, B., Ludwig, G., Rushton, S.P., Helle, P. & Lindén, H. (2010) Top predators, mesopredators and their prey: interference ecosystems along bioclimatic productivity gradients. Journal of Animal Ecology 79 (4): 785-794.

Elmhagen, B. & Rushton, S.P. (2007) Trophic control of mesopredators in terrestrial ecosystems: top-down or bottom-up? Ecology Letters 10 (3): 197-206.

ESRI (2010) ArcGIS Desktop: Release 10.0. Environmental Systems Research Institute, Redlands, CA, USA.

Estes, J.A. & Duggins, D.O. (1995) Sea otters and kelp forests in Alaska: generality and variation in a community ecological paradigm. Ecological Monographs 65 (1): 75-100.

Estes, J.A., Terborgh, J., Brashares, J.S., Power, M.E., Berger, J., Bond, W.J., Carpenter, S.R., Essington, T.E., Holt, R.D., Jackson, J.B.C., Marquis, R.J., Oksanen, L., Oksanen, T., Paine, R.T., Pikitch, E.K., Ripple, W.J., Sandin, S.A., Scheffer, M., Schoener, T.W., Shurin, J.B., Sinclair, A.R.E., Soulé, M.E., Virtanen, R. & Wardle, D.A. (2011) Trophic downgrading of Planet Earth. Science 333 (6040): 301-306.

Ewer, R.F. (1968) Ethology of Mammals. Paul Elek (Scientific Books) Ltd, London.

152 Chapter 7: References

Fanshawe, J.H. & Fitzgibbon, C.D. (1993) Factors influencing the hunting success of an African wild dog pack. Animal Behaviour 45 (3): 479-490.

Ferguson, J.W.H., Galpin, J.S. & de Wet, M.J. (1988) Factors affecting the activity patterns of black-backed jackals Canis mesomelas. Journal of Zoology 214 (1): 55-69.

Fieberg, J. (2007) Kernel density estimators of home range: Smoothing and the autocorrelation red herring. Ecology 88 (4): 1059-1066.

Fieberg, J. & Börger, L. (2012) Could you please phrase “home range” as a question? Journal of Mammalogy 93 (4): 890-902.

Fieberg, J. & Kochanny, C.O. (2005) Quantifying home-range overlap: The importance of the utilization distribution. Journal of Wildlife Management 69 (4): 1346-1359.

Fillios, M., Crowther, M.S. & Letnic, M. (2012) The impact of the dingo on the thylacine in Holocene Australia. World Archaeology 44 (1): 118-134.

Finke, D.L. & Denno, R.F. (2002) Intraguild predation diminished in complex-structured vegetation: Implications for prey suppression. Ecology 83 (3): 643-652.

Fisher, D.O., Blomberg, S.P. & Hoyle, S.D. (2001) Mechanisms of drought-induced population decline in an endangered wallaby. Biological Conservation 102 (1): 107-115.

Fisher, D.O., Johnson, C.N., Lawes, M.J., Fritz, S.A., McCallum, H., Blomberg, S.P., VanDerWal, J., Abbott, B., Frank, A., Legge, S., Letnic, M., Thomas, C.R., Fisher, A., Gordon, I.J. & Kutt, A. (in press) The current decline of tropical marsupials in Australia: is history repeating? Global Ecology and Biogeography.

Fitzgerald, B.M. & Turner, D.C. (2000) Hunting behaviour of domestic cats and their impact on prey populations. The Domestic Cat: the biology of its behaviour, second edition (eds D.C. Turner & P. Bateson) pp. 151-175. Cambridge University Press, Cambridge.

Fitzsimons, J.A., Legge, S., Traill, B. & Woinarski, J. (2010) Into oblivion? The disappearing native mammals of northern Australia. The Nature Conservancy, Melbourne.

Fleming, P.A., Anderson, H., Prendergast, A.S., Bretz, M.R., Valentine, L.E. & Hardy, G.E.S. (in press) Is the loss of Australian digging mammals contributing to a deterioration in ecosystem function? Mammal Review.

Fleming, P.J.S., Corbett, L., Harden, R.H. & Thomson, P.C. (2001) Managing the impacts of dingoes and other wild dogs. Bureau of Rural Sciences, Canberra.

Forsyth, D.M., Robley, A.J. & Reddiex, B. (2005) Review of methods used to estimate the abundance of feral cats. Arthur Rylah Institute for Environmental Research, Department of Sustainability and Environment, Melbourne.

Fox, J. & Weisberg, S. (2011) An {R} Companion to Applied Regression, Second Edition. Sage, Thousand Oaks CA.

153 Chapter 7: References

Frair, J.L., Fieberg, J., Hebblewhite, M., Cagnacci, F., DeCesare, N.J. & Pedrotti, L. (2010) Resolving issues of imprecise and habitat-biased locations in ecological analyses using GPS telemetry data. Philosophical Transactions of the Royal Society B-Biological Sciences 365 (1550): 2187-2200.

Garnett, S.T., Woinarski, J.C.Z., Crowley, G.M. & Kutt, A.S. (2010) Biodiversity conservation in Australian tropical rangelands. Wild Rangelands: Conserving wildlife while maintaining livestock in semi-arid ecosystems (eds J.T. du Toit, R. Kock & J.C. Deutsch) pp. 191-234. Wiley-Blackwell, London.

Gastwirth, J.L., Gel, Y.R., Hui, W.L.W., Lyubchich, V., Miao, W. & Noguchi, K. (2013) lawstat: An R package for biostatistics, public policy, and law. R package version 2.4. http://CRAN.R-project.org/package=lawstat

Gelman, A. (2008) Scaling regression inputs by dividing by two standard deviations. Statistics in Medicine 27 (15): 2865-2873.

Gelman, A. & Su, Y.S. (2013) arm: data analysis using regression and multilevel/hierarchical models. package version 1.6-07.01. http://CRAN.R-project.org/package=arm

Gese, E.M., Stotts, T.E. & Grothe, S. (1996) Interactions between coyotes and red foxes in Yellowstone National Park, Wyoming. Journal of Mammalogy 77 (2): 377-382.

Getz, W.M., Fortmann-Roe, S., Cross, P.C., Lyons, A.J., Ryan, S.J. & Wilmers, C.C. (2007) LoCoH: Nonparameteric kernel methods for constructing home ranges and utilization distributions. Plos One 2 (2): e207.

Gibson, D.F., Lundie-Jenkins, G., Langford, D.G., Cole, J.R., Clarke, D.E. & Johnson, K.A. (1994) Predation by feral cats, Felis catus, on the rufous hare-wallaby, Lagorchestes hirsutus, in the Tanami Desert. Australian Mammalogy 17: 103-107.

Gitzen, R.A., Millspaugh, J.J. & Kernohan, B.J. (2006) Bandwidth selection for fixed-kernel analysis of animal utilization distributions. Journal of Wildlife Management 70 (5): 1334-1344.

Glen, A.S. & Dickman, C.R. (2005) Complex interactions among mammalian carnivores in Australia, and their implications for wildlife management. Biological Reviews 80 (3): 387-401.

Glen, A.S. & Dickman, C.R. (2008) Niche overlap between marsupial and eutherian carnivores: does competition threaten the endangered spotted-tailed quoll? Journal of Applied Ecology 45 (2): 700-707.

Glen, A.S., Dickman, C.R., Soulé, M.E. & Mackey, B.G. (2007) Evaluating the role of the dingo as a trophic regulator in Australian ecosystems. Austral Ecology 32 (5): 492-501.

Glen, A.S., Pennay, M., Dickman, C.R., Wintle, B.A. & Firestone, K.B. (2011) Diets of sympatric native and introduced carnivores in the Barrington Tops, eastern Australia. Austral Ecology 36 (3): 290-296.

154 Chapter 7: References

Gong, W., Sinden, J., Braysher, M. & Jones, R. (2009) The economic impacts of vertebrate pests in Australia. Invasive Animals Cooperative Research Centre, Canberra.

Gorman, M.L., Mills, M.G., Raath, J.P. & Speakman, J.R. (1998) High hunting costs make African wild dogs vulnerable to kleptoparasitism by hyaenas. Nature 391 (6666): 479- 481.

Gosselink, T.E., Van Deelen, T.R., Warner, R.E. & Joselyn, M.G. (2003) Temporal habitat partitioning and spatial use of coyotes and red foxes in east-central Illinois. Journal of Wildlife Management 67 (1): 90-103.

Gotelli, N.J. (2001) Research frontiers in null model analysis. Global Ecology and Biogeography 10 (4): 337-343.

Grueber, C.E., Nakagawa, S., Laws, R.J. & Jamieson, I.G. (2011) Multimodel inference in ecology and evolution: challenges and solutions. Journal of Evolutionary Biology 24 (4): 699-711.

Gutman, R. & Dayan, T. (2005) Temporal partitioning: An experiment with two species of spiny mice. Ecology 86 (1): 164-173.

Hairston, N.G., Smith, F.E. & Slobodkin, L.B. (1960) Community structure, population control, and competition. American Naturalist 94 (879): 421-425.

Hamel, S., Killengreen, S.T., Henden, J.A., Yoccoz, N.G. & Ims, R.A. (2013) Disentangling the importance of interspecific competition, food availability, and habitat in species occupancy: Recolonization of the endangered Fennoscandian arctic fox. Biological Conservation 160 (0): 114-120.

Harden, R.H. (1985) The ecology of the dingo in northeastern New South Wales 1. Movements and home range. Australian Wildlife Research 12 (1): 25-37.

Harper, G.A. (2004) Feral cats on Stewart Island/Rakiura. Department of Conservation, Wellington, New Zealand.

Harrington, L.A., Harrington, A.L., Yamaguchi, N., Thom, M.D., Ferreras, P., Windham, T.R. & Macdonald, D.W. (2009) The impact of native competitors on an alien invasive: temporal niche shifts to avoid interspecific aggression? Ecology 90 (5): 1207-1216.

Harris, S., Cresswell, W.J., Forde, P.G., Trewhella, W.J., Woollard, T. & Wray, S. (1990) Home-range analysis using radio-tracking data - A review of problems and techniques particularly as applied to the study of mammals. Mammal Review 20 (2-3): 97-123.

Hayward, M.W. & Slotow, R. (2009) Temporal partitioning of activity in large African carnivores: tests of multiple hypotheses. South African Journal of Wildlife Research 39 (2): 109-125.

155 Chapter 7: References

Hebblewhite, M., White, C.A., Nietvelt, C.G., McKenzie, J.A., Hurd, T.E., Fryxell, J.M., Bayley, S.E. & Paquet, P.C. (2005) Human activity mediates a trophic cascade caused by wolves. Ecology 86 (8): 2135-2144.

Heithaus, M.R. (2001) Habitat selection by predators and prey in communities with asymmetrical intraguild predation. Oikos 92 (3): 542-554.

Heithaus, M.R. & Dill, L.M. (2006) Does tiger shark predation risk influence foraging habitat use by bottlenose dolphins at multiple spatial scales? Oikos 114 (2): 257-264.

Heithaus, M.R., Frid, A., Wirsing, A.J. & Worm, B. (2008) Predicting ecological consequences of marine top predator declines. Trends in Ecology & Evolution 23 (4): 202-210.

Heithaus, M.R., Wirsing, A.J., Burkholder, D., Thomson, J. & Dill, L.M. (2009) Towards a predictive framework for predator risk effects: the interaction of landscape features and prey escape tactics. Journal of Animal Ecology 78 (3): 556-562.

Henke, S.E. & Bryant, F.C. (1999) Effects of coyote removal on the faunal community in western Texas. Journal of Wildlife Management 63 (4): 1066-1081.

Heselton, R.R. (1998) Elevation effects on GPS positional accuracy. MSc, Virginia Polytechnic Institute and State University, Blacksburg, Virginia.

Hilmer, S.S. (2010) Ecophysiology of feral cats (Felis catus) in Australia. PhD Thesis, Johann Wolfgang Goeth University, Frankfurt.

Horsup, A. (2004) Recovery plan for the northern hairy-nosed wombat Lasiorhinus krefftii 2004-2008. Report to the Department of Environment and Heritage, Canberra. . Environmental Protection Agency/Queensland Parks and Wildlife Service, Brisbane.

Hothorn, T., Hornik, K., van de Wiel, M.A. & Zeileis, A. (2008) Implementing a class of permutation tests: The coin package. Journal of Statistical Software 28 (8): 1-23.

Hunter, J. & Caro, T. (2008) Interspecific competition and predation in American carnivore families. Ethology Ecology & Evolution 20 (4): 295-324.

Hurford, A. (2009) GPS measurement error gives rise to spurious 180 degrees turning angles and strong directional biases in animal movement data. Plos One 4 (5): e5632.

Husseman, J.S., Murray, D.L., Power, G., Mack, C., Wenger, C.R. & Quigley, H. (2003) Assessing differential prey selection patterns between two sympatric large carnivores. Oikos 101 (3): 591-601.

Hutchings, S. (2003) The diet of feral house cats (Felis catus) at a regional rubbish tip, Victoria. Wildlife Research 30 (1): 103-110.

Hyzer, G. (2012) Effects of GPS error on animal home range estimates MSc, University of South Florida, Florida.

156 Chapter 7: References

James, A.I. & Eldridge, D.J. (2007) Reintroduction of fossorial native mammals and potential impacts on ecosystem processes in an Australian desert landscape. Biological Conservation 138 (3-4): 351-359.

James, A.I., Eldridge, D.J. & Hill, B.M. (2009) Foraging animals create fertile patches in an Australian desert shrubland. Ecography 32 (5): 723-732.

Janssen, A., Sabelis, M.W., Magalhães, S., Montserrat, M. & van der Hammen, T. (2007) Habitat structure affects intraguild predation. Ecology 88 (11): 2713-2719.

Jerde, C.L. & Visscher, D.R. (2005) GPS measurement error influences on movement model parameterization. Ecological Applications 15 (3): 806-810.

Johnson, C.N. (2006) Australia's mammal extinctions: A 50,000 year history. Cambridge University Press, Port Melbourne.

Johnson, C.N. & Brook, B.W. (2011) Reconstructing the dynamics of ancient human populations from radiocarbon dates: 10 000 years of population growth in Australia. Proceedings of the Royal Society B-Biological Sciences 278 (1725): 3748-3754.

Johnson, C.N., Isaac, J.L. & Fisher, D.O. (2007) Rarity of a top predator triggers continent-wide collapse of mammal prey: dingoes and marsupials in Australia. Proceedings of the Royal Society B-Biological Sciences 274 (1608): 341-346.

Johnson, C.N. & VanDerWal, J. (2009) Evidence that dingoes limit abundance of a mesopredator in eastern Australian forests. Journal of Applied Ecology 46: 641-646.

Johnson, C.N. & Wroe, S. (2003) Causes of extinction of vertebrates during the Holocene of mainland Australia: arrival of the dingo, or human impact? Holocene 13 (6): 941-948.

Jones, E. (1977) Ecology of the feral cat, Felis catus (L.), (Carnivora: Felidae) on Macquarie Island. Australian Wildlife Research 4 (3): 249-262.

Jones, E. (1990) Physical characteristics and taxonomic status of wild canids, Canis familiaris, from the Eastern Highlands of Victoria. Australian Wildlife Research 17: 68-81.

Jones, E. & Coman, B.J. (1981) Ecology of the feral cat, Felis catus (L) in southeastern Australia I. Diet. Australian Wildlife Research 8 (3): 537-547.

Jones, E. & Coman, B.J. (1982) Ecology of the feral cat, Felis catus (L) in southeastern Australia III. Home ranges and population ecology in semi-arid northwest Victoria. Australian Wildlife Research 9 (3): 409-420.

Jones, M.C., Marron, J.S. & Sheather, S.J. (1996) A brief survey of bandwidth selection for density estimation. Journal of the American Statistical Association 91 (433): 401-407.

Kamler, J.F., Stenkewitz, U., Klare, U., Jacobsen, N.F. & Macdonald, D.W. (2012) Resource partitioning among cape foxes, bat-eared foxes, and black-backed jackals in South Africa. Journal of Wildlife Management 76 (6): 1241-1253.

157 Chapter 7: References

Kats, L.B. & Dill, L.M. (1998) The scent of death: Chemosensory assessment of predation risk by prey animals. Ecoscience 5 (3): 361-394.

Kauffman, M.J., Brodie, J.F. & Jules, E.S. (2010) Are wolves saving Yellowstone's aspen? A landscape-level test of a behaviorally mediated trophic cascade. Ecology 91 (9): 2742- 2755.

Kauffman, M.J., Brodie, J.F. & Jules, E.S. (2013) Are wolves saving Yellowstone’s aspen? A landscape-level test of a behaviorally mediated trophic cascade: reply. Ecology 94 (6): 1425-1431.

Kavanau, J.L. & Ramos, J. (1975) Influences of light on activity and phasing of carnivores. American Naturalist 109 (968): 391-418.

Kennedy, M., Phillips, B.L., Legge, S., Murphy, S.A. & Faulkner, R.A. (2012) Do dingoes suppress the activity of feral cats in northern Australia? Austral Ecology 37 (1): 134- 139.

Kerle, J.A., Foulkes, J.N., Kimber, R.G. & Papenfus, D. (1992) The decline of the Brushtail possum, Trichosurus vulpecula (Kerr 1798), in arid Australia. The Rangeland Journal 14 (2): 107-127.

Kie, J.G., Matthiopoulos, J., Fieberg, J., Powell, R.A., Cagnacci, F., Mitchell, M.S., Gaillard, J.M. & Moorcroft, P.R. (2010) The home-range concept: are traditional estimators still relevant with modern telemetry technology? Philosophical Transactions of the Royal Society B-Biological Sciences 365 (1550): 2221-2231.

Kimble, D.S., Tyers, D.B., Robison-Cox, J. & Sowell, B.F. (2011) Aspen recovery since wolf reintroduction on the Northern Yellowstone winter range. Rangeland Ecology & Management 64 (2): 119-130.

Kinnear, J.E., Sumner, N.R. & Onus, M.L. (2002) The red fox in Australia—an exotic predator turned biocontrol agent. Biological Conservation 108: 335–359.

Kitchen, A.M., Gese, E.M. & Schauster, E.R. (2000) Changes in coyote activity patterns due to reduced exposure to human persecution. Canadian Journal of Zoology 78 (5): 853-857.

Kitchener, A.C., Van Valkenburgh, B. & Yamaguchi, N. (2010) Felid form and function. Biology and Conservation of Wild Felids (eds D.W. MacDonald & A.J. Loveridge) pp. 83-106. Oxford University Press, Oxford.

Klare, U., Kamler, J.F. & Macdonald, D.W. (2011) A comparison and critique of different scat- analysis methods for determining carnivore diet. Mammal Review 41 (4): 294–312.

Koenker, R. (2011) quantreg: Quantile regression and related methods. R package version 4.76. Available at http://CRAN.R-project.org/package=quantreg [verified January 2012]. http://CRAN.R-project.org/package=quantreg

158 Chapter 7: References

Konecny, M.J. (1987) Home range and activity patterns of feral house cats in the Galapagos Islands. Oikos 50 (1): 17-23.

Kortello, A.D., Hurd, T.E. & Murray, D.L. (2007) Interactions between cougars (Puma concolor) and gray wolves (Canis lupus) in Banff National Park, Alberta. Ecoscience 14 (2): 214-222.

Kozlowski, A.J., Gese, E.M. & Arjo, W.M. (2008) Niche overlap and resource partitioning between sympatric kit foxes and coyotes in the Great Basin Desert of western Utah. American Midland Naturalist 160 (1): 191-208.

Kronfeld-Schor, N. & Dayan, T. (2003) Partitioning of time as an ecological resource. Annual Review of Ecology Evolution and Systematics 34: 153-181.

Kronfeld-Schor, N., Dayan, T., Elvert, R., Haim, A., Zisapel, N. & Heldmaier, G. (2001) On the use of the time axis for ecological separation: Diel rhythms as an evolutionary constraint. American Naturalist 158 (4): 451-457.

Kutt, A.S. (2011) The diet of the feral cat (Felis catus) in north-eastern Australia. Acta Theriologica 56 (2): 157-169.

Kutt, A.S. (2012) Feral cat (Felis catus) prey size and selectivity in north-eastern Australia: implications for mammal conservation. Journal of Zoology 287 (4): 292–300.

Kutt, A.S. & Gordon, I.J. (2012) Variation in terrestrial mammal abundance on pastoral and conservation land tenures in north-eastern Australian tropical savannas. Animal Conservation 15 (4): 416-425.

Laliberte, A.S. & Ripple, W.J. (2004) Range contractions of North American carnivores and ungulates. Bioscience 54 (2): 123-138.

Langham, N.P.E. (1992) Feral cats (Felis catus L.) on New Zealand farmland 2. Seasonal activity. Wildlife Research 19 (6): 707-720.

Larivière, S., Jolicoeur, H. & Crête, M. (2000) Status and conservation of the gray wolf (Canis lupus) in wildlife reserves of Québec. Biological Conservation 94 (2): 143-151.

Laundré, J.W. (2010) Behavioral response races, predator-prey shell games, ecology of fear, and patch use of pumas and their ungulate prey. Ecology 91 (10): 2995-3007.

Laundré, J.W., Calderas, J.M.M. & Hernández, L. (2009) Foraging in the landscape of fear, the predator’s dilemma: where should I hunt? Open Ecology Journal 2 (1): 1-6.

Laundré, J.W., Hernández, L. & Altendorf, K.B. (2001) Wolves, elk and bison: reestablishing the "landscape of fear" in Yellowstone National Park, U.S.A. Canadian Journal of Zoology 79: 1401-1409.

Laver, P.N. & Kelly, M.J. (2008) A Critical Review of Home Range Studies. The Journal of Wildlife Management 72 (1): 290-298.

159 Chapter 7: References

Legendre, P. (1993) Spatial autocorrelation: trouble or new paradigm? Ecology 74 (6): 1659- 1673.

Legge, S., Kennedy, M.S., Lloyd, R., Murphy, S.A. & Fisher, A. (2011) Rapid recovery of mammal fauna in the central Kimberley, northern Australia, following the removal of introduced herbivores. Austral Ecology 36 (7): 791-799.

Lesmerises, F., Dussault, C. & St-Laurent, M.H. (2012) Wolf habitat selection is shaped by human activities in a highly managed boreal forest. Forest Ecology and Management 276: 125-131.

Letnic, M., Bobby, T. & Dickman, C.R. (2005) The responses of mammals to La Niña (El Niño Southern Oscillation)-associated rainfall, predation, and wildfire in central Australia. Journal of Mammalogy 86 (4): 689-703.

Letnic, M., Crowther, M.S. & Koch, F. (2009a) Does a top-predator provide an endangered rodent with refuge from an invasive mesopredator? Animal Conservation 12 (4): 302- 312.

Letnic, M. & Dworjanyn, S.A. (2011) Does a top predator reduce the predatory impact of an invasive mesopredator on an endangered rodent? Ecography 34 (5): 827–835.

Letnic, M., Greenville, A., Denny, E., Dickman, C.R., Tischler, M., Gordon, C. & Koch, F. (2011) Does a top predator suppress the abundance of an invasive mesopredator at a continental scale? Global Ecology and Biogeography 20 (2): 343-353.

Letnic, M. & Koch, F. (2010) Are dingoes a trophic regulator in arid Australia? A comparison of mammal communities on either side of the dingo fence. Austral Ecology 35 (2): 167- 175.

Letnic, M., Koch, F., Gordon, C., Crowther, M.S. & Dickman, C.R. (2009b) Keystone effects of an alien top-predator stem extinctions of native mammals. Proceedings of the Royal Society B-Biological Sciences 276 (1671): 3249-3256.

Letnic, M., Ritchie, E.G. & Dickman, C.R. (2012) Top predators as biodiversity regulators: the dingo Canis lupus dingo as a case study. Biological Reviews 87: 390-413.

Lewis, J.S., Rachlow, J.L., Garton, E.O. & Vierling, L.A. (2007) Effects of habitat on GPS collar performance: using data screening to reduce location error. Journal of Applied Ecology 44 (3): 663-671.

Leyhausen, P. & Wolff, R. (1959) Das Revier einer Hauskatze. Zeitschrift für Tierpsychologie 16 (6): 666-670.

Liberg, O., Sandell, M., Pontier, D. & Natoli, E. (2000) Density, spatial organisation and reproductive tactics in the domestic cat and other felids. The Domestic cat: the biology of its behaviour, 2nd edition (eds D.C. Turner & P. Bateson) pp. 119-147. Cambridge University Press, Cambridge.

160 Chapter 7: References

Lichti, N.I. & Swihart, R.K. (2011) Estimating utilization distributions with kernel versus local convex hull methods. Journal of Wildlife Management 75 (2): 413-422.

Lima, S. (1998) Nonlethal effects in the ecology of predator-prey interactions. What are the ecological effects of anti-predator decision-making? Bioscience 48 (1): 25-34.

Lima, S. & Dill, L. (1990) Behavioural decisions made under the risk of predation: a review and prospectus. Canadian Journal of Zoology 68: 619-640.

Lima, S.L. (2002) Putting predators back into behavioral predator-prey interactions. Trends in Ecology & Evolution 17 (2): 70-75.

Linnell, J.D.C., Andersen, R., Kvam, T., Andren, H., Liberg, O., Odden, J. & Moa, P.F. (2001) Home range size and choice of management strategy for lynx in Scandinavia. Environmental Management 27 (6): 869-879.

Linnell, J.D.C. & Strand, O. (2000) Interference interactions, co-existence and conservation of mammalian carnivores. Diversity and Distributions 6: 169-176.

Loss, S.R., Will, T. & Marra, P.P. (2013) The impact of free-ranging domestic cats on wildlife of the United States. Nature Communications 4: 1396.

Lourenço, R., Penteriani, V., Rabaça, J.E. & Korpimäki, E. (2014) Lethal interactions among vertebrate top predators: a review of concepts, assumptions and terminology. Biological Reviews 89 (2): 270-283.

Lundie-Jenkins, G., Corbett, L.K. & Phillips, C.M. (1993) Ecology of the Rufous Hare-wallaby Lagorchestes hirsutus Gould (Marsupialia : Macropodidae), in the Tanami Desert, Northern Territory. III. Interactions with introduced mammal species. Wildlife Research 20: 495-511.

Lunney, D., Triggs, B., Eby, P. & Ashby, E. (1990) Analysis of scats of dogs Canis familiaris and Foxes Vulpes vulpes (Canidae, Carnivora) in coastal forests near Bega, New South Wales. Australian Wildlife Research 17 (1): 61-68.

MacDonald, D.W., Creel, S. & Mills, M.G.L. (2004) Society. Biology and Conservation of Wild Canids (eds D.W. MacDonald & C. Sillero-Zubiri) pp. 85-106. Oxford University Press, Oxford.

Macdonald, D.W., Mosser, A. & Gittleman, J.L. (2010) Felid Society. Biology and Conservation of Wild Felids (eds D.W. MacDonald & A.J. Loveridge) pp. 125-160. Oxford University Press, Oxford.

Macdonald, M.L., Rogers, Q.R. & Morris, J.G. (1984) Nutrition of the domestic cat, a mammalian carnivore. Annual Review of Nutrition 4: 521-562.

Mace, R.D. & Waller, J.S. (1997) Spatial and temporal interaction of male and female grizzly bears in northwestern Montana. Journal of Wildlife Management 61 (1): 39-52.

161 Chapter 7: References

MacFarland, D.M. & Van Deelen, T.R. (2011) Using simulation to explore the functional relationships of terrestrial carnivore population indices. Ecological Modelling 222 (15): 2761-2769.

Mack, R.N., Simberloff, D., Lonsdale, W.M., Evans, H., Clout, M. & Bazzaz, F.A. (2000) Biotic invasions: Causes, epidemiology, global consequences, and control. Ecological Applications 10 (3): 689-710.

Mahon, P.S., Banks, P.B. & Dickman, C.R. (1998) Population indices for wild carnivores: a critical study in sand-dune habitat, south-western Queensland. Wildlife Research 25 (1): 11-22.

Major, J.T. & Sherburne, J.A. (1987) Interspecific relationships of coyotes, bobcats, and red foxes in western Maine. The Journal of Wildlife Management 51 (3): 606-616.

Mannocci, L., Laran, S., Monestiez, P., Dorémus, G., Van Canneyt, O., Watremez, P. & Ridoux, V. (in press) Predicting top predator habitats in the Southwest Indian Ocean. Ecography.

Marsack, P. & Campbell, G. (1990) Feeding behaviour and diet of dingoes in the Nullabor region, Western Australia. Australian Wildlife Research 17 (4): 349-357.

Marshall, K.N., Hobbs, N.T. & Cooper, D.J. (2013) Stream hydrology limits recovery of riparian ecosystems after wolf reintroduction. Proceedings of the Royal Society B: Biological Sciences 280 (1756): 1471-2954.

Martin, J., Basille, M., Van Moorter, B., Kindberg, J., Allaine, D. & Swenson, J.E. (2010) Coping with human disturbance: spatial and temporal tactics of the brown bear (Ursus arctos). Canadian Journal of Zoology 88 (9): 875-883.

Martin, J., Rey, B., Pons, J.-B., Natoli, E. & Pontier, D. (2013a) Movements and space use of feral cats in Kerguelen archipelago: a pilot study with GPS data. Polar Biology 36 (10): 1531-1536.

Martin, J., van Moorter, B., Revilla, E., Blanchard, P., Dray, S., Quenette, P.Y., Allaine, D. & Swenson, J.E. (2013b) Reciprocal modulation of internal and external factors determines individual movements. Journal of Animal Ecology 82 (2): 290-300.

May, R., Van Dijk, J., Wabakken, P., Swenson, J.E., Linnell, J.D.C., Zimmermann, B., Odden, J., Pedersen, H.C., Andersen, R. & Landa, A. (2008) Habitat differentiation within the large-carnivore community of Norway's multiple-use landscapes. Journal of Applied Ecology 45 (5): 1382-1391.

McDonald, J.E., Jr. & Fuller, T.K. (2005) Effects of spring acorn availability on black bear diet, milk composition, and cub survival. Journal of Mammalogy 86 (5): 1022-1028.

McGee, B.K., Ballard, W.B., Nicholson, K.L., Cypher, B.L., Lemons II, P.R. & Kamler, J.F. (2006) Effects of artificial escape dens on swift fox populations in northwest Texas. Wildlife Society Bulletin 34 (3): 821-827.

162 Chapter 7: References

McLaren, B.E. & Peterson, R.O. (1994) Wolves, moose and tree-rings on Isle Royale. Science 266 (5190): 1555-1558.

McLeod, R. (2004) Counting the cost: Impact of Invasive Animals in Australia 2004. Cooperative Research Centre for Pest Animal Control, Canberra.

McNatty, A., Abbott, K.L. & Lester, P.J. (2009) Invasive ants compete with and modify the trophic ecology of hermit crabs on tropical islands. Oecologia 160 (1): 187-194.

Mech, L.D. (2012) Is science in danger of sanctifying the wolf? Biological Conservation 150 (1): 143-149.

Medway, D.G. (2004) The land bird fauna of Stephens Island, New Zealand in the early 1890s, and the cause of its demise. Notornis 51 (4): 201-211.

Melis, C., Jędrzejewska, B., Apollonio, M., Bartoń, K.A., Jędrzejewski, W., Linnell, J.D.C., Kojola, I., Kusak, J., Adamic, M., Ciuti, S., Delehan, I., Dykyy, I., Krapinec, K., Mattioli, L., Sagaydak, A., Samchuk, N., Schmidt, K., Shkvyrya, M., Sidorovich, V.E., Zawadzka, B. & Zhyla, S. (2009) Predation has a greater impact in less productive environments: variation in roe deer, Capreolus capreolus, population density across Europe. Global Ecology and Biogeography 18 (6): 724-734.

Melis, C., Teurlings, I., Linnell, J.D.C., Andersen, R. & Bordoni, A. (2004) Influence of a deer carcass on Coleopteran diversity in a Scandinavian boreal forest: a preliminary study. European Journal of Wildlife Research 50 (3): 146-149.

MELP (2001) British Columbia Standards, Specifications and Guidelines for Resource Surveyors Using Global Positioning System (GPS) Technology., Victoria, Canada.

Menzel, U. (2013) EMT: Exact Multinomial Test: Goodness-of-Fit Test for Discrete Multivariate data. R package version 1.1. http://CRAN.R-project.org/package=EMT

Miao, W., Gel, Y.R. & Gastwirth, J.L. (2006) A new test of symmetry about an unknown median. Random Walk, Sequential Analysis and Related Topics - A Festschrift in Honor of Yuan-Shih Chow (eds A. Hsiung, C.-H. Zhang & Z. Ying). World Scientific Publisher, Singapore.

Middleton, A.D., Kauffman, M.J., McWhirter, D.E., Jimenez, M.D., Cook, R.C., Cook, J.G., Albeke, S.E., Sawyer, H. & White, P.J. (2013) Linking anti-predator behaviour to prey demography reveals limited risk effects of an actively hunting large carnivore. Ecology Letters 16 (8): 1023–1030.

Minta, S.C. (1992) Tests of spatial and temporal interaction among animals. Ecological Applications 2 (2): 178-188.

Minta, S.C. (1993) Sexual differences in spatiotemporal interaction among badgers. Oecologia 96 (3): 402-409.

163 Chapter 7: References

Mitchell, B.D. & Banks, P.B. (2005) Do wild dogs exclude foxes? Evidence for competition from dietary and spatial overlaps. Austral Ecology 30 (5): 581-591.

Moehrenschlager, A., List, R. & Macdonald, D.W. (2007) Escaping intraguild predation: Mexican kit foxes survive while coyotes and golden eagles kill Canadian swift foxes. Journal of Mammalogy 88 (4): 1029-1039.

Mohr, C.O. (1947) Table of equivalent populations of north American small mammals. American Midland Naturalist 37 (1): 223-249.

Moller, H. & Alterio, N. (1999) Home range and spatial organisation of stoats (Mustela erminea), ferrets (Mustela furo) and feral house cats (Felis catus) on coastal grasslands, Otago Peninsula, New Zealand: implications for yellow-eyed penguin (Megadyptes antipodes) conservation. New Zealand Journal of Zoology 26 (3): 165-174.

Molsher, R., Dickman, C., Newsome, A. & Muller, W. (2005) Home ranges of feral cats (Felis catus) in central-western New South Wales, Australia. Wildlife Research 32 (7): 587- 595.

Molsher, R., Newsome, A. & Dickman, C. (1999) Feeding ecology and population dynamics of the feral cat (Felis catus) in relation to the availability of prey in central-eastern New South Wales. Wildlife Research 26 (5): 593-607.

Moodie, E. (1995) The potential for biological control of feral cats in Australia. Australian Nature Conservation Agency, Canberra.

Mooney, H.A. & Cleland, E.E. (2001) The evolutionary impact of invasive species. Proceedings of the National Academy of Sciences of the United States of America 98 (10): 5446-5451.

Moore, J. (2002) Parasites and the Behavior of Animals. Oxford University Press, New York.

Moseby, K.E., Hill, B.M. & Read, J.L. (2009a) Arid Recovery - A comparison of reptile and small mammal populations inside and outside a large rabbit, cat and fox-proof exclosure in arid South Australia. Austral Ecology 34 (2): 156-169.

Moseby, K.E., Neilly, H., Read, J.L. & Crisp, H.A. (2012) Interactions between a top order predator and exotic mesopredators in the Australian rangelands. International Journal of Ecology vol. 2012 (Article ID 250352).

Moseby, K.E., Read, J.L., Paton, D.C., Copley, P., Hill, B.M. & Crisp, H.A. (2011) Predation determines the outcome of 10 reintroduction attempts in arid South Australia. Biological Conservation 144 (12): 2863-2872.

Moseby, K.E., Stott, J. & Crisp, H. (2009b) Movement patterns of feral predators in an arid environment - implications for control through poison baiting. Wildlife Research 36 (5): 422-435.

164 Chapter 7: References

Moser, B.W. & Garton, E.O. (2007) Effects of telemetry location error on space-use estimates using a fixed-kernel density estimator. Journal of Wildlife Management 71 (7): 2421- 2426.

Mosnier, A., Boisjoly, D., Courtois, R. & Ouellet, J.-P. (2008) Extensive predator space use can limit the efficacy of a control program. The Journal of Wildlife Management 72 (2): 483-491.

Nagelkerke, N.J.D. (1991) A note on a general definition of the coefficient of determination. Biometrika 78 (3): 691-692.

National Land and Water Resource Audit (2008) Assessing invasive animals in Australia 2008. NLWRA and Invasive Animals Cooperative Research Centre, Canberra.

Newsome, A.E., Catling, P.C., Cooke, B.D. & Smyth, R. (2001) Two ecological universes separated by the dingo barrier fence in semi-arid Australia: interactions between landscapes, herbivory and carnivory, with and without dingoes. The Rangeland Journal 23 (1): 71-98.

Newsome, A.E., Catling, P.C. & Corbett, L.K. (1983a) The feeding ecology of the dingo 2. Dietary and numerical relationships with fluctuating prey populations in southeastern Australia. Australian Journal of Ecology 8 (4): 345-366.

Newsome, A.E. & Corbett, L. (1982) The Identity of the Dingo 2. Hybridisation with domestic dogs in captivity and in the wild. Australian Journal of Zoology 30 (2): 365-374.

Newsome, A.E. & Corbett, L.K. (1985) The Identity of the Dingo 3. The incidence of Dingoes, Dogs and Hybrids and their coat colours in remote and settled regions of Australia. Australian Journal of Zoology 33 (3): 363-373.

Newsome, A.E., Corbett, L.K., Catling, P.C. & Burt, R.J. (1983b) The Feeding Ecology of the Dingo 1. Stomach contents from trapping in southeastern Australia, and the non-target wildlife also caught in dingo traps. Australian Wildlife Research 10 (3): 477-486.

Newsome, T.M., Ballard, G.-A., Dickman, C.R., Fleming, P.J.S. & Howden, C. (2013a) Anthropogenic resource subsidies determine space use by australian arid zone dingoes: An improved resource selection modelling approach. Plos One 8 (5): e63931.

Newsome, T.M., Ballard, G.-A., Dickman, C.R., Fleming, P.J.S. & van de Ven, R. (2013b) Home range, activity and sociality of a top predator, the dingo: a test of the Resource Dispersion Hypothesis. Ecography 36 (8): 914–925.

Newsome, T.M., Stephens, D., Ballard, G.-A., Dickman, C.R. & Fleming, P.J.S. (2013c) Genetic profile of dingoes (Canis lupus dingo) and free-roaming domestic dogs (C. l. familiaris) in the Tanami Desert, Australia. Wildlife Research 40 (3): 196-206.

Noble, G.K. (1939) The role of dominance in the life of birds. Auk 56: 263-273.

165 Chapter 7: References

Nogales, M., Vidal, E., Medina, F.M., Bonnaud, E., Tershy, B.R., Campbell, K.J. & Zavaleta, E.S. (2013) Feral cats and biodiversity conservation: The urgent prioritization of island management. Bioscience 63 (10): 804-810.

Norbury, G., Byrom, A., Pech, R., Smith, J., Clarke, D., Anderson, D. & Forrester, G. (2013) Invasive mammals and habitat modification interact to generate unforeseen outcomes for indigenous fauna. Ecological Applications 23 (7): 1707-1721.

O'Connell Jr, A.F., Talancy, N.W., Bailey, L.L., Sauer, J.R., Cook, R. & Gilbert, A.T. (2006) Estimating site occupancy and detection probability parameters for meso- and large mammals in a coastal ecosystem. The Journal of Wildlife Management 70 (6): 1625- 1633.

Oksanen, L., Fretwell, S.D., Arruda, J. & Niemela, P. (1981) Exploitation ecosystems in gradients of primary productivity. American Naturalist 118 (2): 240-261.

Oksanen, L. & Oksanen, T. (2000) The logic and realism of the hypothesis of exploitation ecosystems. American Naturalist 155 (6): 703-723.

Ordiz, A., Støen, O.G., Sæbø, S., Kindberg, J., Delibes, M. & Swenson, J.E. (2012) Do bears know they are being hunted? Biological Conservation 152: 21-28.

Oskarsson, M.C.R., Klütsch, C.F.C., Boonyaprakob, U., Wilton, A., Tanabe, Y. & Savolainen, P. (2012 ) Mitochondrial DNA data indicate an introduction through Mainland Southeast Asia for Australian dingoes and Polynesian domestic dogs. Proceedings of the Royal Society B: Biological Sciences 279 (1730): 967-974.

Otis, D.L. & White, G.C. (1999) Autocorrelation of location estimates and the analysis of radiotracking data. Journal of Wildlife Management 63 (3): 1039-1044.

Pace, M.L., Cole, J.J., Carpenter, S.R. & Kitchell, J.F. (1999) Trophic cascades revealed in diverse ecosystems. Trends in Ecology & Evolution 14 (12): 483-488.

Packer, C., Holt, R.D., Hudson, P.J., Lafferty, K.D. & Dobson, A.P. (2003) Keeping the herds healthy and alert: implications of predator control for infectious disease. Ecology Letters 6 (9): 797-802.

Paine, R.T. (1969) A note on trophic complexity and community stability. American Naturalist 103 (929): 91-93.

Palomares, F. & Caro, T.M. (1999) Interspecific killing among mammalian carnivores. American Naturalist 153 (5): 492-508.

Palomares, F., Gaona, P., Ferreras, P. & Delibes, M. (1995) Positive effects on game species of top predators by controlling smaller predator populations - an example with lynx, mongooses and rabbits Conservation Biology 9 (2): 295-305.

Paltridge, R. (2002) The diets of cats, foxes and dingoes in relation to prey availability in the Tanami Desert, Northern Territory. Wildlife Research 29 (4): 389-403.

166 Chapter 7: References

Paltridge, R., Gibson, D. & Edwards, G. (1997) Diet of the feral cat (Felis catus) in central Australia. Wildlife Research 24 (1): 67-76.

Pascoe, J.H., Mulley, R.C., Spencer, R. & Chapple, R. (2012) Diet analysis of mammals, raptors and reptiles in a complex predator assemblage in the Blue Mountains, eastern Australia. Australian Journal of Zoology 59 (5): 295-301.

Paton, D.C. (1993) Impacts of domestic and feral cats on wildlife. Cat Management Workshop proceedings (eds G. Siepen & C. Owens). Australian Nature Conservation Agency and the Queensland Department of Environment and Heritage, Brisbane.

Pavey, C.R., Eldridge, S.R. & Heywood, M. (2008) Population dynamics and prey selection of native and introduced predators during a rodent outbreak in arid Australia. Journal of Mammalogy 89 (3): 674-683.

Petren, K. & Case, T.J. (1996) An experimental demonstration of exploitation competition in an ongoing invasion. Ecology 77 (1): 118-132.

Pettigrew, J.D. (1993) A burst of feral cats in the Diamantina - a lesson for the management of pest species? Cat Management Workshop (eds G. Siepen & C. Owens). Australian Nature Conservation Agency and the Queensland Department of Environment and Heritage, Brisbane.

Pimentel, D., McNair, S., Janecka, J., Wightman, J., Simmonds, C., O'Connell, C., Wong, E., Russel, L., Zern, J., Aquino, T. & Tsomondo, T. (2001) Economic and environmental threats of alien plant, animal, and microbe invasions. Agriculture Ecosystems & Environment 84 (1): 1-20.

Polis, G.A. (1981) The evolution and dynamics of intraspecific predation. Annual Review of Ecology and Systematics 12: 225-251.

Polis, G.A. & Holt, R.D. (1992) Intraguild predation: the dynamics of complex trophic interactions. Trends in Ecology & Evolution 7 (5): 151-154.

Polis, G.A., Myers, C.A. & Holt, R.D. (1989) The ecology and evolution of intraguild predation: potential competitors that eat each other. Annual Review of Ecology and Systematics 20 (1): 297-330.

Polis, G.A. & Strong, D.R. (1996) Food web complexity and community dynamics. American Naturalist 147 (5): 813-846.

Pople, A.R., Grigg, G.C., Cairns, S.C., Beard, L.A. & Alexander, P. (2000) Trends in the numbers of red kangaroos and emus on either side of the South Australian dingo fence: evidence for predator regulation? Wildlife Research 27: 269-276.

Portas, T.J. (2010) Toxoplasmosis in macropodids: A review. Journal of Zoo and Wildlife Medicine 41 (1): 1-6.

167 Chapter 7: References

Powell, R.A. (2012) Movements, home ranges, activity, and dispersal. Carnivore ecology and conservation: A handbook of techniques (eds L. Boitani & R.A. Powell) pp. 188-217. Oxford University Press, Oxford.

Powell, R.A. & Mitchell, M.S. (2012) What is a home range? Journal of Mammalogy 93 (4): 948-958.

Preisser, E.L. & Bolnick, D.I. (2008) The many faces of fear: Comparing the pathways and impacts of nonconsumptive predator effects on prey populations. Plos One 3 (6): e2465.

Preisser, E.L., Bolnick, D.I. & Benard, M.F. (2005) Scared to death? The effects of intimidation and consumption in predator-prey interactions. Ecology 86 (2): 501-509.

Preisser, E.L., Orrock, J.L. & Schmitz, O.J. (2007) Predator hunting mode and habitat domain alter nonconsumptive effects in predator-prey interactions. Ecology 88 (11): 2744-2751.

Prentiss, P.G., Wolfe, A.V. & Eddy, H.A. (1959) Hydropenia in cat and dog. Ability of the cat to meet its water requirements solely from a diet of fish or meat. American Journal of Physiology 196 (3): 625-632.

Priddel, D. & Wheeler, R. (2004) An experimental translocation of brush-tailed bettongs (Bettongia penicillata) to western New South Wales. Wildlife Research 31: 421-432.

Prugh, L.R., Stoner, C.J., Epps, C.W., Bean, W.T., Ripple, W.J., Laliberte, A.S. & Brashares, J.S. (2009) The rise of the mesopredator. Bioscience 59 (9): 779-791.

Purcell, B.V. (2010) Order in the Pack: Ecology of Canis lupus dingo in the Southern Greater Blue Mountains World Heritage Area. PhD thesis, University of Western Sydney, Hawkesbury.

R Development Core Team (2011) R: A language and environment for statistica l computing. version 2.14.1. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3- 900051-07-0, http://www.R-project.org/

R Development Core Team (2013) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, http://www.R-project.org/

Radloff, F.G.T. & du Toit, J.T. (2004) Large predators and their prey in a southern African savanna: a predator’s size determines its prey size range. Journal of Animal Ecology 73: 410-423.

Randall, W., Cunningham, J.T., Randall, S., Liittschwager, J. & Johnson, R.F. (1987) A 2-peak circadian system in body temperature and activity in the domestic cat Felis catus L. Journal of Thermal Biology 12 (1): 27-37.

Rasmussen, G.S.A. & Macdonald, D.W. (2011) Masking of the zeitgeber: African wild dogs mitigate persecution by balancing time. Journal of Zoology 286 (3): 232-242.

168 Chapter 7: References

Read, J. & Bowen, Z. (2001) Population dynamics, diet and aspects of the biology of feral cats and foxes in arid South Australia. Wildlife Research 28 (2): 195-203.

Rempel, R.S. & Rodgers, A.R. (1997) Effects of differential correction on accuracy of a GPS animal location system. The Journal of Wildlife Management 61 (2): 525-530.

Richard, E., Calenge, C., Saïd, S., Hamann, J.-L. & Gaillard, J.-M. (2013) Studying spatial interactions between sympatric populations of large herbivores: a null model approach. Ecography 36 (2): 157–165.

Rigby, R.A. & Stasinopoulos, D.M. (2005) Generalized additive models for location, scale and shape, (with discussion). Applied Statistics 54 (3): 507-554.

Ripple, W.J. & Beschta, R.L. (2004) Wolves and the ecology of fear: Can predation risk structure ecosystems? Bioscience 54 (8): 755-766.

Ripple, W.J. & Beschta, R.L. (2006) Linking a cougar decline, trophic cascade, and catastrophic regime shift in Zion National Park. Biological Conservation 133 (4): 397-408.

Ripple, W.J. & Beschta, R.L. (2012) Trophic cascades in Yellowstone: The first 15 years after wolf reintroduction. Biological Conservation 145 (1): 205-213.

Ripple, W.J. & Larsen, E.J. (2000) Historic aspen recruitment, elk, and wolves in northern Yellowstone National Park, USA. Biological Conservation 95 (3): 361-370.

Risbey, D.A., Calver, M. & Short, J. (1997) Control of feral cats for nature conservation .1. Field tests of four baiting methods. Wildlife Research 24 (3): 319-326.

Ritchie, E.G., Elmhagen, B., Glen, A.S., Letnic, M., Ludwig, G. & McDonald, R.A. (2012) Ecosystem restoration with teeth: what role for predators? Trends in Ecology & Evolution 27 (5): 265-271.

Ritchie, E.G. & Johnson, C.N. (2009) Predator interactions, mesopredator release and biodiversity conservation. Ecology Letters 12 (9): 982-998.

Ritchie, E.G., Martin, J.K., Johnson, C.N. & Fox, B.J. (2009) Separating the influences of environment and species interactions on patterns of distribution and abundance: competition between large herbivores. Journal of Animal Ecology 78 (4): 724-731.

Robert, K., Garant, D. & Pelletier, F. (2012) Keep in touch: Does spatial overlap correlate with contact rate frequency? The Journal of Wildlife Management 76 (8): 1670-1675.

Robley, A., Gormley, A., Forsyth, D.M., Wilton, A.N. & Stephens, D. (2010a) Movements and habitat selection by wild dogs in eastern Victoria. Australian Mammalogy 32 (1): 23-32.

Robley, A., Gormley, A., Woodford, L., Lindeman, M., Whitehead, B., Albert, R., Bowd, M. & Smith, A. (2010b) Evaluation of camera trap sampling designs used to determine change in occupancy rate and abundance of feral cats. Department of Sustainability and Environment, Heidelberg, Victoria.

169 Chapter 7: References

Roemer, G.W., Gompper, M.E. & Van Valkenburgh, B. (2009) The ecological role of the mammalian mesocarnivore. Bioscience 59 (2): 165-173.

Rolls, E.C. (1969) They all ran wild. Angus and Robertson, Sydney.

Romero-Muñoz, A., Maffei, L., Cuéllar, E. & Noss, A.J. (2010) Temporal separation between jaguar and puma in the dry forests of southern Bolivia. Journal of Tropical Ecology 26: 303-311.

Ron, M., Pastor, J. & Cohen, Y. (1997) Accuracy of GPS telemetry collar locations with differential correction. The Journal of Wildlife Management 61 (2): 530-539.

Rosalino, L.M., Macdonald, D.W. & Santos-Reis, M. (2005) Resource dispersion and badger population density in Mediterranean woodlands: is food, water or geology the limiting factor? Oikos 110 (3): 441-452.

Rosenzweig, M.L. (1966) Community structure in sympatric carnivora. Journal of Mammalogy 47 (4): 602-612.

Rovero, F. & Marshall, A.R. (2009) Camera trapping photographic rate as an index of density in forest ungulates. Journal of Applied Ecology 46: 1011-1017.

Rowcliffe, M.J., Carbone, C., Jansen, P.A., Kays, R. & Kranstauber, B. (2011) Quantifying the sensitivity of camera traps: an adapted distance sampling approach. Methods in Ecology and Evolution 2 (5): 464-476.

Ruxton, G.D. & Lima, S.L. (1997) Predator-induced breeding suppression and its consequences for predator-prey population dynamics. Proceedings of the Royal Society B-Biological Sciences 264 (1380): 409-415.

Salo, P., Korpimäki, E., Banks, P.B., Nordström, M. & Dickman, C.R. (2007) Alien predators are more dangerous than native predators to prey populations. Proceedings of the Royal Society B-Biological Sciences 274 (1615): 1237-1243.

Salo, P., Nordström, M., Thomson, R.L. & Korpimäki, E. (2008) Risk induced by a native top predator reduces alien mink movements. Journal of Animal Ecology 77 (6): 1092-1098.

Samson, C. & Huot, J. (2001) Spatial and temporal interactions between female American black bears in mixed forests of eastern Canada. Canadian Journal of Zoology 79 (4): 633-641.

Samuel, M.D., Pierce, D.J. & Garton, E.O. (1985) Identifying areas of concentrated use within the home range. Journal of Animal Ecology 54 (3): 711-719.

Sargeant, G.A., Johnson, D.H. & Berg, W.E. (1998) Interpreting carnivore scent-station surveys. Journal of Wildlife Management 62 (4): 1235-1245.

Savolainen, P., Leitner, T., Wilton, A.N., Matisoo-Smith, E. & Lundeberg, J. (2004) A detailed picture of the origin of the Australian dingo, obtained from the study of mitochondrial DNA. Proceedings of the National Academy of Sciences 101 (33): 12387-12390.

170 Chapter 7: References

Scheffer, M. (2010) Alternative states in ecosystems. Trophic cascades : predators, prey, and the changing dynamics of nature (eds J. Terborgh & J.A. Estes) pp. 287-298. Island Press, Washington D.C.

Scheffer, M., Carpenter, S., Foley, J.A., Folke, C. & Walker, B. (2001) Catastrophic shifts in ecosystems. Nature 413: 591-596.

Schlexer, F.V. (2008) Attracting animals to detection devices. Noninvasive survey methods for carnivores (eds R.A. Long, P. Mackay, W.J. Zielinski & J.C. Ray) pp. 263-292. Island Press, Washington.

Schmidt, K., Jędrzejewski, W., Okarma, H. & Kowalczyk, R. (2009) Spatial interactions between grey wolves and Eurasian lynx in Białowieża Primeval Forest, Poland. Ecological Research 24 (1): 207-214.

Schmitz, O.J. (2008) Effects of predator hunting mode on grassland ecosystem function. Science 319 (5865): 952-954.

Schmitz, O.J., Beckerman, A.P. & O'Brien, K.M. (1997) Behaviorally mediated trophic cascades: Effects of predation risk on food web interactions. Ecology 78 (5): 1388- 1399.

Schmitz, O.J., Hamback, P.A. & Beckerman, A.P. (2000) Trophic cascades in terrestrial systems: A review of the effects of carnivore removals on plants. American Naturalist 155 (2): 141-153.

Schoener, T.W. (1983) Field experiments on interspecific competition. American Naturalist 122 (2): 240-285.

Seaman, D.E., Millspaugh, J.J., Kernohan, B.J., Brundige, G.C., Raedeke, K.J. & Gitzen, R.A. (1999) Effects of sample size on kernel home range estimates. Journal of Wildlife Management 63 (2): 739-747.

Seaman, D.E. & Powell, R.A. (1990) Identifying patterns and intensity of home range use. Bears: their Biology and Management. Vol. 8. A selection of papers from the Eighth International Conference on Bear Research and Management (eds L.M. Darling & W.R. Archibald) pp. 243-249. International Association of Bear Research and Management, Victoria, British Columbia, Canada.

Seaman, D.E. & Powell, R.A. (1996) An evaluation of the accuracy of kernel density estimators for home range analysis. Ecology 77 (7): 2075-2085.

Selva, N., Jedrzejewska, B., Jedrzejewski, W. & Wajrak, A. (2005) Factors affecting carcass use by a guild of scavengers in European temperate woodland. Canadian Journal of Zoology 83 (12): 1590-1601.

Sergio, F., Caro, T., Brown, D., Clucas, B., Hunter, J., Ketchum, J., McHugh, K. & Hiraldo, F. (2008) Top predators as conservation tools: ecological rationale, assumptions, and efficacy. Annual Review of Ecology Evolution and Systematics 39: 1-19.

171 Chapter 7: References

Sergio, F. & Hiraldo, F. (2008) Intraguild predation in raptor assemblages: a review. Ibis 150: 132-145.

Sergio, F., Marchesi, L., Pedrini, P. & Penteriani, V. (2007) Coexistence of a generalist owl with its intraguild predator: distance-sensitive or habitat-mediated avoidance? Animal Behaviour 74: 1607-1616.

Sergio, F., Newton, I., Marchesi, L. & Pedrini, P. (2006) Ecologically justified charisma: preservation of top predators delivers biodiversity conservation. Journal of Applied Ecology 43 (6): 1049-1055.

Serpell, J.A. (2000) Domestication and history of the cat. The domestic cat: the biology of its behaviour, 2nd edition (eds D.C. Turner & P. Bateson) pp. 179-192. Cambridge University Press, Cambridge.

Sharp, T. & Saunders, G. (2005) Model code of practice for the humane control of feral cats. NSW Department of Primary Industries and Natural Heritage Trust, Sydney.

Shimada, T., Jones, R., Limpus, C. & Hamann, M. (2012) Improving data retention and home range estimates by data-driven screening. Marine Ecology Progress Series 457: 171- 180.

Short, J. (1998) The extinction of rat-kangaroos (Marsupialia : Potoroidae) in New South Wales, Australia. Biological Conservation 86 (3): 365-377.

Short, J. & Calaby, J.H. (2001) The status of Australian mammals in 1922 –collections and field notes of museum collector Charles Hoy. Australian Zoologist 31: 533-562.

Sih, A. (1980) Optimal behaviour: Can foragers balance two conflicting demands? Science 210 (4473): 1041-1043.

Sih, A. (1984) The behavioral response race between predator and prey. American Naturalist 123 (1): 143-150.

Sillero-Zubiri, C. & Macdonald, D.W. (1998) Scent-marking and territorial behaviour of Ethiopian wolves Canis simensis. Journal of Zoology 245: 351-361.

Silva-Rodriguez, E.A., Ortega-Solis, G.R. & Jimenez, J.E. (2010) Conservation and ecological implications of the use of space by chilla foxes and free-ranging dogs in a human- dominated landscape in southern Chile. Austral Ecology 35: 765-777.

Smith, A.P. & Quin, D.G. (1996) Patterns and causes of extinction and decline in Australian conilurine rodents. Biological Conservation 77 (2-3): 243-267.

Smith, I.D. & Munday, B.L. (1965) Observations on the incidence of Toxoplasma gondii in native and introduced feral fauna in eastern Australia. Australian Veterinary Journal 41 (9): 285-286.

172 Chapter 7: References

Smucker, T.D., Lindsey, G.D. & Mosher, S.M. (2000) Home range and diet of feral cats in Hawaii forests Pacific Conservation Biology 6 (3): 229-237.

Soulé, M.E., Bolger, D.T., Alberts, A.C., Wright, J., Sorice, M. & Hill, S. (1988) Reconstructed dynamics of rapid extinctions of chaparral-requiring birds in urban habitat islands. Conservation Biology 2 (1): 75-92.

Soulé, M.E., Estes, J.A., Berger, J. & Del Rio, C.M. (2003) Ecological effectiveness: Conservation goals for interactive species. Conservation Biology 17 (5): 1238-1250.

Soulé, M.E., Estes, J.A., Miller, B. & Honnold, D.L. (2005) Strongly interacting species. conservation policy, management, and ethics. Bioscience 55 (2): 168-176.

Spencer, P.B.S. (1991) Evidence of predation by a feral cat, Felis catus (Carnivora: Felidae) on an isolated rock-wallaby colony in tropical Queensland. Australian Mammalogy 14 (2): 143-144.

Srinivasan, M., Grant, W.E., Swannack, T.M. & Rajan, J. (2010) Behavioral games involving a clever prey avoiding a clever predator: An individual-based model of dusky dolphins and killer whales. Ecological Modelling 221 (22): 2687-2698.

Stasinopoulos, D.M., Rigby, R.A. & Akantziliotou, C. (2008) Instructions on how to use the gamlss package in R, second edition.

Sunde, P., Overskaug, K. & Kvam, T. (1999) Intraguild predation of lynxes on foxes: evidence of interference competition? Ecography 22 (5): 521-523.

Sunquist, M. & Sunquist, F. (2002) Wild cats of the world. University of Chicago Press, Chicago.

Swihart, R.K. & Slade, N.A. (1985) Testing for independence of observations in animal movements. Ecology 66 (4): 1176-1184.

Symonds, M. & Moussalli, A. (2011) A brief guide to model selection, multimodel inference and model averaging in behavioural ecology using Akaike’s information criterion. Behavioral Ecology and Sociobiology 65 (1): 13-21.

Taylor, R.H. (1979) How the Macquarie Island parakeet became extinct. New Zealand Journal of Ecology 2: 42-45.

Terborgh, J., Holt, R.D. & Estes, J.A. (2010) Trophic cascades: What they are, how they work, and why they matter. Trophic Cascades: Predators, prey, and the changing dynamics of nature (eds J. Terborgh & J.A. Estes) pp. 1-18. Island Press, Washington D.C.

Terborgh, J., Lopez, L., Nuñez, P., Rao, M., Shahabuddin, G., Orihuela, G., Riveros, M., Ascanio, R., Adler, G.H., Lambert, T.D. & Balbas, L. (2001) Ecological meltdown in predator-free forest fragments. Science 294 (5548): 1923-1926.

173 Chapter 7: References

Thaker, M., Vanak, A.T., Owen, C.R., Ogden, M.B., Niemann, S.M. & Slotow, R. (2011) Minimizing predation risk in a landscape of multiple predators: effects on the spatial distribution of African ungulates. Ecology 92 (2): 398-407.

Theuerkauf, J. (2009) What drives wolves: fear or hunger? Humans, diet, climate and wolf activity patterns. Ethology 115 (7): 649-657.

Thomson, P.C. (1992a) The behavioural ecology of dingoes in north-western Australia 1. The Fortescue River study area and details of captured dingoes. Wildlife Research 19 (5): 509-518.

Thomson, P.C. (1992b) The behavioural ecology of dingoes in north-western Australia 2. Activity patterns, breeding season and pup rearing. Wildlife Research 19 (5): 519-530.

Thomson, P.C. (1992c) The behavioural ecology of dingoes in north-western Australia 3. Hunting and feeding behaviour, and diet. Wildlife Research 19 (5): 531-541.

Thomson, P.C. (1992d) The behavioural ecology of dingoes in north-western Australia 4. Social and spatial organisation and movements. Wildlife Research 19 (5): 543-563.

Thomson, P.C., Rose, K. & Kok, N.E. (1992a) The behavioural ecology of dingoes in north- western Australia 5. Population dynamics and variation in the social system. Wildlife Research 19 (5): 565-584.

Thomson, P.C., Rose, K. & Kok, N.E. (1992b) The behavioural ecology of dingoes in north- western Australia 6. Temporary extraterritorial movements and dispersal. Wildlife Research 19 (5): 585-595.

Thornton, D.H., Sunquist, M.E. & Main, M.B. (2004) Ecological separation within newly sympatric populations of coyotes and bobcats in south-central Florida. Journal of Mammalogy 85 (5): 973-982.

Thornton, I.W.B., Cook, S., Edwards, J.S., Harrison, R.D., Schipper, C., Shanahan, M., Singadan, R. & Yamuna, R. (2001) Colonization of an island volcano, Long Island, Papua New Guinea, and an emergent island, Motmot, in its caldera lake. VII. Overview and discussion. Journal of Biogeography 28 (11-12): 1389-1408.

Towne, E.G. (2000) Prairie vegetation and soil nutrient responses to ungulate carcasses. Oecologia 122 (2): 232-239.

Treves, A. & Karanth, K.U. (2003) Human-carnivore conflict and perspectives on carnivore management worldwide. Conservation Biology 17 (6): 1491-1499.

Triggs, B. & Brunner, H. (2002) Hair ID: An interactive tool for identifying Australian mammalian hair. Ecobyte Pty Ltd, CSIRO Publishing, Victoria, Australia.

Triggs, B., Brunner, H. & Cullen, J.M. (1984) The food of fox, dog and cat in Croajingalong National Park, southeastern Victoria. Australian Wildlife Research 11 (3): 491-499.

174 Chapter 7: References

Trites, A.W. & Joy, R. (2005) Dietary analysis from fecal samples: how many scats are enough? Journal of Mammalogy 86 (4): 704-712.

Van Dyck, S. & Strahan, R. (2008) The mammals of Australia. Third edn. Reed New Holland, Sydney. van Winkle, W. (1975) Comparison of several probabilistic home-range models. The Journal of Wildlife Management 39 (1): 118-123.

Vanak, A.T., Fortin, D., Thaker, M., Ogden, M.B., Owen, C.R., Greatwood, S. & Slotow, R. (2013) Moving to stay in place: behavioral mechanisms for coexistence of African large carnivores. Ecology 94 (11): 2619-2631.

Vander Wal, E. & Rodgers, A.R. (2012) An individual-based quantitative approach for delineating core areas of animal space use. Ecological Modelling 224 (1): 48-53.

Vázquez-Domínguez, E., Ceballos, G. & Cruzado, J. (2004) Extirpation of an insular subspecies by a single introduced cat: the case of the endemic deer mouse Peromyscus guardia on Estanque Island, Mexico. Oryx 38 (3): 347-350.

Vernes, K. (2000) Immediate effects of fire on survivorship of the northern bettong (Bettongia tropica): an endangered Australian marsupial. Biological Conservation 96 (3): 305-309.

Vernes, K., Dennis, A. & Winter, J. (2001) Mammalian diet and broad hunting strategy of the dingo (Canis familiaris dingo) in the wet tropical rain forests of northeastern Australia. Biotropica 33 (2): 339-345.

Vigne, J.D., Guilaine, J., Debue, K., Haye, L. & Gerard, P. (2004) Early taming of the cat in Cyprus. Science 304 (5668): 259-259.

Visser, R.L., Watson, J.E.M., Dickman, C.R., Southgate, R., Jenkins, D.J. & Johnson, C.N. (2009) Developing a national framework for Dingo trophic regulation research in Australia: Outcomes of a national workshop. Ecological Management & Restoration 10 (2): 168-170.

Wallach, A.D., Johnson, C.N., Ritchie, E.G. & O'Neill, A.J. (2010) Predator control promotes invasive dominated ecological states. Ecology Letters 13 (8): 1008-1018.

Wallach, A.D., Murray, B.R. & O'Neill, A.J. (2009a) Can threatened species survive where the top predator is absent? Biological Conservation 142 (1): 43-52.

Wallach, A.D. & O'Neill, A.J. (2009) Artificial water points: Hotspots of extinction or biodiversity? Biological Conservation 142 (7): 1253-1254.

Wallach, A.D., Ritchie, E.G., Read, J. & O'Neill, A.J. (2009b) More than mere numbers: the impact of lethal control on the social stability of a top-order predator. Plos One 4 (9): 1- 8.

175 Chapter 7: References

Walters, J.R., Doerr, P.D. & Carter, J.H. (1988) The cooperative breeding system of the red- cockaded woodpecker. Ethology 78 (4): 275-305.

Wand, M.P. & Jones, M.C. (1995) Kernel Smoothing. Chapman & Hall, London.

Wang, Y. & Fisher, D.O. (2012) Dingoes affect activity of feral cats, but do not exclude them from the habitat of an endangered macropod. Wildlife Research 39 (7): 611-620.

Watts, C.H.S. & Aslin, H.J. (1981) The rodents of Australia. Angus and Robertson, Sydney.

Welsh, A.H., Lindenmayer, D.B. & Donnelly, C.F. (2013) Fitting and interpreting occupancy models. Plos One 8 (1): e52015.

Whitehouse, S.J.O. (1977) The diet of the dingo in Western Australia. Australian Wildlife Research 4: 145-150.

Wikenros, C., Liberg, O., Sand, H. & Andren, H. (2010) Competition between recolonizing wolves and resident lynx in Sweden. Canadian Journal of Zoology 88 (3): 271-279.

Wikenros, C., Sand, H., Ahlqvist, P. & Liberg, O. (2013) Biomass flow and scavengers use of carcasses after re-colonization of an apex predator. Plos One 8 (10): e77373.

Williams, C.K., Parer, I., Coman, B.J., Burley, J. & Braysher, M.L. (1995) Managing Vertebrate Pests: Rabbits. Bureau of Resource Sciences/CSIRO Division of Wildlife and Ecology. Australian Government Publishing Service, Canberra.

Williams, R.J., Carter, J., Duff, G.A., Woinarski, J.C.Z., Cook, G.D. & Farrer, S.L. (2005) Carbon accounting, land management, science and policy uncertainty in Australian savanna landscapes: introduction and overview. Australian Journal of Botany 53 (7): 583-588.

Wilson, R.R., Blankenship, T.L., Hooten, M.B. & Shivik, J.A. (2010) Prey-mediated avoidance of an intraguild predator by its intraguild prey. Oecologia 164 (4): 921-929.

Winnie Jr, J. (2012) Predation risk, elk, and aspen: tests of a behaviorally mediated trophic cascade in the Greater Yellowstone Ecosystem. Ecology 93 (12): 2600-2614.

Wirsing, A.J. & Heithaus, M.R. (2009) Olive-headed sea snakes Disteria major shift seagrass microhabitats to avoid shark predation. Marine Ecology-Progress Series 387: 287-293.

Wirsing, A.J., Heithaus, M.R., Frid, A. & Dill, L.M. (2008) Seascapes of fear: evaluating sublethal predator effects experienced and generated by marine mammals. Marine Mammal Science 24 (1): 1-15.

Woinarski, J., Mackey, B., Nix, H. & Traill, B. (2007) The nature of Northern Australia: its natural values, ecological processes and future prospects. ANU E Press, Canberra.

Woinarski, J.C.Z., Armstrong, M., Brennan, K., Fisher, A., Griffiths, A.D., Hill, B., Milne, D.J., Palmer, C., Ward, S., Watson, M., Winderlich, S. & Young, S. (2010) Monitoring

176 Chapter 7: References

indicates rapid and severe decline of native small mammals in Kakadu National Park, northern Australia. Wildlife Research 37 (2): 116-126.

Woinarski, J.C.Z., Legge, S., Fitzsimons, J.A., Traill, B.J., Burbidge, A.A., Fisher, A., Firth, R.S.C., Gordon, I.J., Griffiths, A.D., Johnson, C.N., McKenzie, N.L., Palmer, C., Radford, I., Rankmore, B., Ritchie, E.G., Ward, S. & Ziembicki, M. (2011a) The disappearing mammal fauna of northern Australia: context, cause, and response. Conservation Letters 4 (3): 192-201.

Woinarski, J.C.Z., Milne, D.J. & Wanganeen, G. (2001) Changes in mammal populations in relatively intact landscapes of Kakadu National Park, Northern Territory, Australia. Austral Ecology 26 (4): 360-370.

Woinarski, J.C.Z., Ward, S., Mahney, T., Bradley, J., Brennan, K., Ziembicki, M. & Fisher, A. (2011b) The mammal fauna of the Sir Edward Pellew island group, Northern Territory, Australia: refuge and death-trap. Wildlife Research 38 (4): 307-322.

Woinarski, J.C.Z., Williams, R.J., Price, O. & Rankmore, B. (2005) Landscapes without boundaries: wildlife and their environments in northern Australia. Wildlife Research 32: 377-388.

Wolf, E.C., Cooper, D.J. & Hobbs, N.T. (2007) Hydrologic regime and herbivory stabilize an alternative state in Yellowstone National Park. Ecological Applications 17 (6): 1572- 1587.

Worton, B.J. (1989) Kernel methods for estimating the utilization distribution in home range studies. Ecology 70 (1): 164-168.

Zar, J.H. (2010) Biostatistical Analysis. Fifth edn. Prentice Hall Inc., Upper Saddle River, New Jersey.

Ziembicki, M.R., Woinarski, J.C.Z. & Mackey, B. (2013) Evaluating the status of species using Indigenous knowledge: Novel evidence for major native mammal declines in northern Australia. Biological Conservation 157 (0): 78-92.

177 Chapter 7: References

178

Appendix A Notes on prey consumed by dingoes and feral cats in northern Australia

Introduction

Dietary analysis is a useful tool to examine relationships between predators and prey, by detailing the diversity and abundance of prey consumed. Diet studies can also provide evidence for direct and indirect interactions between sympatric predator populations, by uncovering the extent of dietary overlap and intraguild predation. Where dietary overlap is high, indicating predators are competing for the same resources, exploitative competition can result in the dominant competitor depriving the other of access to resources (Schoener 1983). Intraguild predation occurs when aggression imposed by the larger predator, known as interference competition, leads to interspecific killing, followed by consumption of the smaller predator

(Polis & Holt 1992). Dietary studies can therefore highlight the probability and potential that an apex predator is suppressing a sympatric mesopredator either through exploitative or interference competition.

Diet and potential for dietary overlap

Extent of dietary overlap may be influenced by the relative size and dietary breadth of each predator. While larger predators may be expected to have broader dietary niches given the larger range of available prey, several studies have found no relationship between body size and niche breadth (Brandl et al. 1994; Radloff & du Toit 2004; Costa 2009). Small prey may not provide sufficient benefit to justify the energy expenditure associated with hunting (Costa

2009). It can be more cost-effective for large predators (>14.5 kg) to hunt larger-bodied prey

(Carbone et al. 2007), while smaller predators are generally restricted to smaller prey species

179 Appendix A: Dietary analysis

(Barclay & Brigham 1991). However, other traits or specialisations may allow some predators to hunt prey outside their predicted size range (Rosenzweig 1966).

Feral cats, although a small predator, are effective hunters, and can kill relatively large species such as the Rufous Hare-wallaby Lagorchestes hirsutus (Paltridge et al. 1997) and

Allied Rock-wallaby Petrogale assimilis (Spencer 1991). However, generally they consume a broad range of prey smaller than themselves, in multiple meals per day (Bradshaw 2006), with different proportions of mammals, reptiles, birds and other items constituting the diet across

Australia (Paltridge et al. 1997; Glen et al. 2011; Kutt 2011). Rabbits Oryctolagus cuniculus form a large part of the cat diet where they occur in abundance (Catling 1988; Molsher et al.

1999). Carrion consumption appears rare (Jones & Coman 1981; Paltridge et al. 1997; but see

Molsher et al. 1999), as cats are generally thought to prefer live prey (Bradshaw et al. 1996).

Feral cats can readily switch to different prey when current prey become unavailable, reducing crashes in their population when prey numbers decline (Read & Bowen 2001).

Being within the 'transition range' of Carbone et al. (2007), dingoes can hunt both small and large prey, which is reflected in the broad dietary range in some areas (e.g. Triggs et al.

1984; Paltridge 2002; Brook & Kutt 2011). In open areas, dingoes appear to focus on larger prey such as red kangaroos Macropus rufus, aided by their ability to hunt cooperatively in suitable habitats (Whitehouse 1977; Thomson 1992c). However they appear to be opportunistic hunters (Newsome et al. 1983a; Vernes et al. 2001), consuming a variety of mainly mammalian prey from small to large species (Newsome et al. 1983b; Corbett & Newsome 1987; Glen &

Dickman 2008; Cupples et al. 2011), but also preying upon other taxa such as reptiles (Paltridge

2002) and crustaceans (Allen et al. 2012), eating human-provided waste (Newsome et al.

2013a) and scavenging carrion when resources are limited (Allen 2010).

Dietary overlap may also vary with spatial and temporal fluctuations in resources, particularly in Australia where stochastic climatic variation can lead to irruptions in potential prey species (Breed & Ford 2007). Both dingoes and feral cats can target these explosions in prey, leading to particularly high levels of dietary overlap when they occur (Pavey et al. 2008).

180 Appendix A: Dietary analysis

However, particularly abundant resources may enable coexistence with high levels of overlap without enhancing competition (Lundie-Jenkins et al. 1993; Letnic et al. 2012).

Dietary overlap has also been recorded between dingoes and feral cats outside irruption periods. In the Tanami Desert , dingoes and cats overlapped considerably, but dingoes ate more prey larger than 100 g (Paltridge 2002). In south-eastern Australia, dingoes and cats consumed similar small and medium-sized mammals, but again, dingoes in addition hunted larger macropods (Triggs et al. 1984).

Intraguild predation

There is considerable evidence for intraguild predation of feral cats by dingoes across

Australia. Cat remains have been recorded in dingo scats or stomachs in semi-arid (Paltridge

2002) and arid areas (Marsack & Campbell 1990; Thomson 1992c; Lundie-Jenkins et al. 1993;

Pavey et al. 2008; Cupples et al. 2011; Allen & Leung 2012), in tropical north Australia

(Corbett 1995b; Allen et al. 2012) and in forests in eastern Australia (Newsome et al. 1983b;

Lunney et al. 1990; Glen & Dickman 2008; Pascoe et al. 2012). Although feral cat records in the dingo diet are usually rare, some studies have found cat remains in almost 10% of samples

(Lundie-Jenkins et al. 1993; Paltridge 2002). There is also evidence that dingoes will kill feral cats and not consume them (Moseby et al. 2012), which suggests that interference competition, rather than food acquisition, is the primary motive (Polis et al. 1989; Sunde et al. 1999). If dingoes frequently kill cats without consuming them, dietary analysis may potentially underestimate the frequency of interspecific killing.

In this analysis, I aimed to compare dietary samples from dingoes and feral cats across my study sites, to provide evidence for a) dietary overlap that may indicate competition for resources and/or b) intraguild predation of feral cats by dingoes, signifying that interference competition occurs between the two predators.

181 Appendix A: Dietary analysis

Methods

Scat collection and processing

Scats were collected either opportunistically, or by searching areas of conspicuous scat deposition such as roadsides, track intersections and river crossings, during fieldwork contributing to this thesis in north QLD and the Kimberley, WA from 2009-2011 and in previous surveys on one of the same sites (Mt Zero) in 2007-2008. Samples were collected from surveys on the Coane and Hervey's Ranges west of Townsville in the Einasleigh Uplands (EU), on properties north of Julia Creek on the Gulf Plain (GP), on a pair of properties on Cape York

Peninsula (CYP) and at Marion Downs and Mornington Wildlife Sanctuaries in the Kimberley, western Australia (KM). All properties experienced a tropical climate, with a dry winter and most rain falling in the summer wet season. Mean annual rainfall in the survey areas ranged from 479 mm in the Gulf Plain, to over 1000 mm on CYP. Habitat was characterised by

Eucalyptus/Corymbia open forest and woodland, Melaleuca vindifolia open woodland and dry rainforest thickets on Cape York Peninsula (Figure A.1a), Eucalyptus/Corymbia woodland or open woodland in the Einasleigh Uplands (Figure A.1b), and Astrebla tussock grasslands and

Eucalyptus/Acacia woodlands in the Gulf Plain (Figure A.1c). In the Kimberley, Eucalyptus woodlands with understoreys of native grasses such as Themeda triandra and Heteropogon contortus and shrubs such as Melaleuca minutifolia thinned to hummock and tussock grasslands including Triodia spp, interspersed by rocky outcrops (Figure A.1d). Further details of the properties are provided in Chapters 2-4.

Scats were identified to species by their size and smell (Triggs et al. 1984). Samples were oven-dried at 100 °C for 12 h to remove parasites, then placed in individual bags and washed in a washing machine. Prey items were identified to the lowest taxonomic level possible from remains such as bones, hair or scales using established techniques (Brunner &

Coman 1974) and compared to confirmed reference material and the literature (Watts & Aslin

1981; Triggs & Brunner 2002). Reptiles were identified to family and birds were grouped in one

182 Appendix A: Dietary analysis

category. Insects were grouped by order. Small amounts of dingo hair (<10%) were not

included as prey as they likely represent grooming hairs.

a. b.

c. d.

Figure A.1. Examples of habitat from a) Cape York Peninsula, b) Einasleigh Uplands, c) Gulf Plain and d) the Kimberley.

Analysis

I used two techniques to analyse the composition of scats: percent frequency of

occurrence and percent volume. Percent frequency of occurrence identifies the number of

occurrences of the food item divided by the total number of scats, including trace items (Corbett

1989; Klare et al. 2011). Frequency of occurrence tends to overestimate smaller prey, but is

useful for identifying rarer components in the diet (Klare et al. 2011).

183 Appendix A: Dietary analysis

Percent volume estimates the relative proportions of food items within scats by volume

(McDonald & Fuller 2005), providing an estimate of ingested biomass, and hence the importance of prey items in the diet, when a correction factor is not available (Klare et al.

2011). Percent volume was estimated visually using a grid system.

Mammal species weights were estimated as the mean adult weight from Van Dyck &

Strahan (2008), except for the feral horse estimate (Csurhes et al. 2009). When grouping species into prey sizes, I used the size categories of Glen and Dickman (2008) for small (<500 g), medium (500-6999 g) and large mammals (≥7000 g). I also categorised species as native or introduced. Any probable identifications were categorised in more general categories, e.g. unknown Macropus.

Results

Dingo diet

I recorded 288 prey items from 165 dingo scats collected in the four survey areas (CYP:

N = 100, GP: N = 8, KM: N = 20, EU: N = 37; Table A.1). Some scats recorded multiple individuals of each prey item, e.g. multiple Rattus individuals, but within each scat, each prey type was counted as one item.

Results from % frequency occurrence and % volume differed in identifying the main prey items in the samples (Figures A.2 & A.3). The differences were most evident in the non- mammal prey such as birds and invertebrates. Although these prey were recorded relatively frequently, for example invertebrates were recorded in 12.7% of all scats, they contributed very little to the overall volume of prey (0.6%).

184 Appendix A: Dietary analysis

Table A.1. Summary of dietary items recorded in dingo scats in each area and in the total sample.

No. scats with each prey item % frequency occurrence % volume

Prey item size (g) CYP GP EU KM Total CYP GP EU KM Total CYP GP EU KM Total

Delicate mouse, Psuedomys delicatulus 10.5 0 0 1 0 1 2.7 0.6 0.5 0.1

Common rock rat, Zyzomys argurus 36 0 0 0 3 3 15 1.8 9 1.1

Canefield rat, Rattus sordidus 118 43 0 0 0 43 43 26.1 33.6 20.4

Cape York rat, Rattus leucopus 132 9 0 0 0 9 9 5.5 6.1 3.7

Long-haired rat, Rattus villosissimus 134 0 2 0 0 2 25 1.2 24.9 1.2

Black rat, Rattus rattus 280 0 3 0 0 3 37.5 1.8 33.6 1.6

Unknown rodent, Rattus sp. 5 1 0 0 6 5 12.5 3.6 1.1 12.5 1.2

Unknown rodent 1 0 0 0 1 1 0.6 0.1 0.1

Small native prey 52 2 1 3 58 52 25 2.7 15 35.2 39.6 24.9 0.5 9 26.4

Small introduced prey 0 3 0 0 3 37.5 1.8 33.6 1.6

Small prey 58 6 1 3 68 58 75 2.7 15 41.2 40.8 71 0.5 9 29.4

Flying fox, Pteropus sp. 620 2 0 0 0 2 2 1.2 1 0.6

Long-nosed bandicoot, Perameles nasuta 925 0 0 3 0 3 8.1 1.8 6.6 1.5

Northern brown bandicoot, Isoodon macrourus 1,600 7 0 10 0 17 7 27 10.3 4.4 23 7.8

Rufous bettong, Aepyprymnus rufescens 2,750 0 0 1 0 1 2.7 0.6 2.7 0.6

Short-beaked echidna, Tachyglossus aculeatus 4,500 2 0 0 0 2 2 1.2 0.2 0.1

Red-legged pademelon, Thylogale stigmatica 4,600 0 0 2 0 2 5.4 1.2 5.4 1.2

Unknown medium-sized mammal 1 0 1 0 2 1 2.7 1.2 1 1.1 0.8

European rabbit, Oryctolagus cuniculus 1,580 0 0 3 0 3 8.1 1.8 8.1 1.8

Feral cat, Felis catus 4,550 1 0 0 0 1 1 0.6 0.4 0.2

Medium native prey 12 0 17 0 29 12 45.9 17.6 6.6 38.8 12.7

Medium introduced prey 1 0 3 0 4 1 8.1 2.4 0.4 8.1 2.1

Medium prey 13 0 20 0 33 13 54.1 20 7 46.9 14.7

185 Appendix A: Dietary analysis

Table A.1. continued.

No. scats with each prey item % frequency occurrence % volume

Prey item size (g) CYP GP EU KM Total CYP GP EU KM Total CYP GP EU KM Total

Swamp wallaby, Wallabia bicolor 15,000 0 0 2 0 2 5.4 1.2 4.3 1.0

Agile wallaby, Macropus agilis 15,000 24 0 5 1 30 24 13.5 5 18.2 19.8 12.2 5.0 15.3

Common wallaroo, Macropus robustus 25,375 0 0 0 1 1 5 0.6 5.0 0.6

Antilopine wallaroo, Macropus antilopinus 27,025 1 0 0 1 2 1 5 1.2 1.0 5.0 1.2

Eastern grey kangaroo, Macropus giganteus 40,750 0 1 3 0 4 12.5 8.1 2.4 7.5 6.8 1.9

Unknown macropod, Macropus sp. 6 0 7 0 13 6 18.9 7.9 2.3 11.1 3.9

Dingo, Canis familiaris 14,000 0 1 0 1 2 12.5 5 1.2 12.5 2.0 0.8

Feral pig, Sus scrofa 86,250 13 1 1 1 16 13 12.5 2.7 5 9.7 4.5 4.4 0.8 1.0 3.3

Feral horse, Equus caballus 400,000 0 0 0 1 1 5 0.6 5.0 0.6

Cattle, Bos sp. 700,000 28 0 4 14 46 28 10.8 70 27.9 16.5 10.3 50.5 18.4

Large native prey 31 2 17 4 54 31 25 45.9 20 32.7 23.1 20.0 34.4 17.0 24.7

Large introduced prey 41 1 5 16 63 41 12.5 13.5 80 38.2 21.0 4.4 11.1 56.5 22.3

Large prey 72 3 22 20 117 72 37.5 59.5 100 70.9 44.1 24.4 45.4 73.5 47

Skink, Family Scincidae 6 2 0 2 10 6 25 10 6.1 0.7 1.4 1.3 0.6

Goanna, Family Varanidae 1 0 0 0 1 1 0.6 0.2 0.1

Gecko, Family Gekkonidae 0 1 0 0 1 12.5 0.6 0.1 0.0

Snake, Multiple families 1 0 0 0 1 1 0.6 0.3 0.2

Reptiles 8 3 0 2 13 8 37.5 10 7.9 1.2 1.5 1.3 0.9

186 Appendix A: Dietary analysis

Table A.1. continued.

No. scats with each prey item % frequency occurrence % volume

Prey item CYP GP EU KM Total CYP GP EU KM Total CYP GP EU KM Total

Birds 12 1 0 3 16 12 12.5 15 9.7 2.0 0.6 1.8 1.5

Beetles or Cockroaches, Coleoptera/Blattodea 8 0 0 1 9 8 5 5.5 0.2 0.3 0.1

Ants or termites Hymenoptera/Blattodea 2 0 1 1 4 2 2.7 5 2.4 0.0 0.5 0.1 0.1

Grasshoppers or Crickets, Orthoptera 2 1 0 1 4 2 12.5 5 2.4 0.1 2.5 0.5 0.2

Maggots, Diptera 4 0 0 0 4 4 2.4 0.1 0.1

Invertebrates 16 1 1 3 21 16 12.5 2.7 15 12.7 0.4 2.5 0.5 0.8 0.6

Vegetation 9 0 3 5 17 9 8.1 25 10.3 2.5 6.6 13.8 4.6

Bone 3 0 0 0 3 3 1.8 2.1 1.3

No. items 191 14 47 36 288

No. scats 100 8 37 20 165

187 Appendix A: Dietary analysis

Small prey 90% Medium prey Large native prey 80%

Large introduced prey

70% Reptiles Bird 60% Invertebrates

50%

40%

30%

Frequencyoccurrenceof scats in 20%

10%

0% CYP GP EU KM Total (n=100) (n=8) (n=37) (n=20) (n=165)

Figure A.2. Percent frequency of occurrence of different prey categories for each survey area and total sample.

Recorded prey categories were not consistent between survey areas (Figures A.2 &

A.3). In the Einasleigh Uplands medium-sized mammals such as Isoodon macrourus and

Oryctolagus cuniculus constituted about half the sample, with macropods and cattle also dominant prey items. Small prey such as native and introduced Rattus spp. were more prevalent in the more open habitats of Cape York Peninsula and Gulf Plain, but dingoes frequently consumed large prey such as macropods and pigs on Cape York Peninsula, while reptiles were found in over a third of samples on the Gulf Plain. Diet samples from the Kimberley were dominated by large prey, in particular cattle, found in 70% of scats and comprising 50.5% of the area's scat volume (Table A.1). Levels of predation on birds were similar (~9.7% across all samples) in all areas except in the Einasleigh Uplands.

188 Appendix A: Dietary analysis

Intraguild predation was rarely recorded, with only one scat containing cat remains on

Cape York Peninsula. However, two cases of cannibalism were recorded, one on the Gulf Plain and one in the Kimberley.

Small prey 90% Medium prey Large native prey 80% Large introduced prey Reptiles 70%

Bird

60% Invertebrates

50%

40% Volume in total Volume in total diet 30%

20%

10%

0% CYP GP EU KM Total (n=100) (n=8) (n=37) (n=20) (n=165)

Figure A.3. Volume of different prey categories in the dietary samples for each area and the total sample.

Cat diet

Only five cat scats were collected: two from the Kimberley and one from each other area. Dietary items found ranged from larger species such as the long-nosed bandicoot

Perameles nasuta and the European rabbit, to agamids and grasshoppers (Table A.2). The small sample prevents any inference about cat diet in the study areas.

189 Appendix A: Dietary analysis

Table A.2. Summary of dietary items recorded in feral cat scats in the total sample.

Prey item No. scats with each prey item % volume Common rock rat, Zyzomys argurus 1 0.14 Long-nosed bandicoot, Perameles nasuta 1 0.20 European rabbit, Oryctolagus cuniculus 1 0.20 Dragon, Agamidae 1 0.01 Bird, Aves 3 0.42 Ants/Termites, Hymenoptera/Blattodea 1 0.01 Grasshoppers/crickets , Orthoptera 1 0.02

Discussion

Due to the small sample sizes and sampling windows, I cannot extrapolate too widely outside the geographic areas in which samples were collected, nor compare general diet between survey areas (Trites & Joy 2005). Further, I could not examine levels of dietary overlap between dingoes and feral cats due to the low return of cat samples.

Similarly to previous studies, mammals formed the largest volume of prey in the dietary samples. There did appear to be variation in prey consumed by dingoes across the different sites, which may reflect the different compositions of prey communities in each landscape.

Almost half the scats from Cape York Peninsula contained Rattus sordidus remains, a native rodent that inhabits a range of habitats (Aplin & Redhead 2008), suggesting this species may have been particularly prevalent in the area. Similarly, on the Gulf Plain, more than half the scats contained Rattus species, although the sample size was small. Rattus rattus had not previously been recorded in this area (Dickman & Watts 2008), highlighting the utility of dietary analysis in describing prey communities. In the Einasleigh Uplands, medium and large native mammals formed the largest portion of frequency and volume. Medium mammals such as Isoodon macrourus were frequently recorded on remote camera in woodland habitats, while larger species such as Macropus giganteus and M. agilis were observed in a range of habitats.

The consumption of large species in all areas suggests that dingoes may be occasionally hunting

190 Appendix A: Dietary analysis in groups, which may be necessary to bring down adults in this size range (Newsome et al.

1983a).

Dingoes consumed large introduced species across all areas. Feral pigs were taken by dingoes in all areas, particularly on Cape York Peninsula where they are abundant (NLWRA

2008). Cattle were found in low numbers of scats in both the Einasleigh Uplands and Gulf

Plain, despite both properties in the latter region being active cattle stations. In the Kimberley, cattle dominated both the volume and frequency of prey consumed, and they were also recorded in almost a third of scats from Cape York Peninsula. In both these areas, cattle are killed during control operations, hence cattle consumption may represent direct predation or scavenging of readily-available carrion (Allen 2010).

Dingoes consumed conspecifics both on the Gulf Plain and in the Kimberley, however I could not determine whether this was due to cannibalism or scavenging. Cannibalism of subordinate animals and juveniles has previously been recorded in captive animals (Corbett

1988) and dingo remains are occasionally recorded in dietary studies (Marsack & Campbell

1990; Thomson 1992c; Brook & Kutt 2011).

Potential for competition

The high volume and frequency of small mammals in the samples from Cape York

Peninsula and Gulf Plain suggest that dingoes may overlap in diet with feral cats in these areas, potentially increasing competition for resources. The prevalence of small prey species may be due to scarcity of larger prey such as macropods in these areas, which is unlikely (L. Brook, pers. obs.), or particularly high densities of rodents, which can irrupt under optimal conditions

(Breed & Ford 2007). In the first situation, we might expect higher resource competition with feral cats, but in the latter, the abundance of prey may negate competition.

I only recorded one case of intraguild predation between dingoes and feral cats, from

Cape York Peninsula, suggesting cat consumption was rare. However, most previous studies have also recorded low rates of predation; this does not necessarily confirm that interspecific killing is not occurring. Moseby et al. (2012) noted dingoes killed feral cats without consuming

191 Appendix A: Dietary analysis them, and similar deaths of mesopredators without predation have been recorded for coyotes on swift foxes (Moehrenschlager et al. 2007) and lynx on red foxes (Sunde et al. 1999). In these cases, 'territorial aggression' may be the more likely motive for interspecific killing rather than acquiring food (Polis et al. 1989).

Due to the low number of feral cat diet samples, I could not examine the extent of dietary overlap between dingoes and feral cats across my study sites. However, I have shown that dingoes do consume a range of predominantly mammalian prey in these areas, and that dietary overlap may be particularly high on Cape York Peninsula and the Gulf Plain, where dingoes appear to frequently consume small mammals. Further, I found evidence of intraguild predation on feral cats in one area, confirming dingoes exert direct lethal effects on the mesopredator in this study site.

192

Appendix B Sun-time calibration of activity data

Animal activity is likely to be influenced by the timing of sunrise and sunset, as well as other physical and biological parameters. The timing of sunrise and sunset change through the year, and vary with latitude and longitude. To ensure that temporal data are calibrated around these important astronomical events, I calibrated activity data to reflect the time relative to sunrise and sunset. Activity records are calibrated to a circular proportion with equal distance between sunset (0.5) and sunrise (0 and 1). Hence each time record reflects its position relative to sunrise and sunset.

193 Appendix B: Sun-time calibration

STEP ONE sunset Convert the time record, sunrise and sunset to minutes, by multiplying decimal time by 1440 (no. midnight minutes in day) sunrise STEP TWO Calculate the length of day and night in minutes

STEP THREE Calculate day and night ratios, as equinox length (720 min) divided by the actual day or night length, to convert actual time points to an equinox scale

STEP FOUR If the time record is between midnight and sunrise, add 1440 min, so values continuously increase between sunset and sunrise

STEP FIVE Give the time record in terms of minutes from sunset, then multiply by the relevant day or night ratio, to convert to the equinox scale

STEP SIX If the time record is at night, calculate as 720 (equinox sunset) plus the time record. If the record is during the day, calculate as 720 minus the time record

RESULT Suntime is given first as minutes after sunrise. Decimal suntime has sunrise at 0 and 1 and sunset at 0.5.

194

Appendix C The effect of screening methods on bias and data loss in GPS fixes

Introduction

Data screening is an important precursor to analysis of GPS location data, as GPS components can frequently record erroneous fixes at a considerable distance from the animal's true position. Kernel density methods may be relatively resilient to average levels of GPS error (Moser & Garton 2007); however, polygon-based methods (Hyzer 2012), and analysis of movement behaviour, habitat selection or spatial overlap, which require precise locations, can be considerably biased by inaccurate datasets (Bradshaw et al. 2007; Frair et al. 2010).

Erroneous fixes can be caused by inherent GPS error in the collars, however error rates may be exacerbated by physical barriers that reduce contact with satellites (D'Eon et al. 2002). Habitat cover (DeCesare et al. 2005), complex terrains such as riparian areas (D'Eon et al. 2002) and the collar position on the animal (D'Eon & Delparte 2005) may affect fix precision. GPS collars can record the dilution of precision (DOP), a measure of accuracy related to satellite orientation, for each fix (MELP

2001), with a low DOP value indicating a fix closer to the true position. Other indicators of imprecision include the number of satellites used to calculate the fix (Shimada et al. 2012), fixes calculated with only three satellites (known as two-dimensional fixes) (Lewis et al. 2007) and incorrect altitude values (Ron et al. 1997; Heselton 1998). Data are frequently screened by removing fixes with limited satellite numbers or DOP values above a certain threshold (Moseby et al. 2009b;

Martin et al. 2013b), however, conservative blanket screening methods can remove a large amount of data, which may negate the benefits of increased precision (D'Eon & Delparte 2005; Bjørneraas et al.

2010).

195 Appendix C: GPS screening methods

The non-movement method, recently proposed by Bjørneraas et al. (2010), identifies fixes which are incongruous with movement behaviour of the study animal. It identifies step lengths that exceed species-specific maximum speeds, to identify fixes that would be unlikely or impossible for an animal to travel to within the fix interval. It then finds fixes which are separated from the general trajectory by a specified distance, with a small inner angle. These 'spikes' are also likely to represent errors. The non-movement method is limited in its ability to identify spikes associated with step lengths shorter than these specified distances. These errors may be common when data are recorded at high frequency, and can contribute considerable bias to analyses of animal movement, as they suggest an animal is moving when it is in fact stationary (Jerde & Visscher 2005; Hurford 2009). Hence a more sensitive technique, that can remove GPS errors without excessive data loss, may be a valuable addition to the screening of datasets recorded during this study.

In this appendix, I outline the results of a post-hoc analysis on stationary test data, to characterise measurement error of the deployed GPS collars as a function of various indicators. I then determine an optimal filtering rule, to be applied to data following the non-movement method, that balances increased precision with limited data loss.

Methods

I collected data from sixteen (13 dingo and 3 cat) collars in stationary positions. Eleven of the dingo collars and one cat collar were located randomly, where they had fallen from the animal once the drop-off was activated (dingoes) or where the animal had died (cat). Two additional dingo collars were deployed in tests, one in open habitat and the other in more complex habitat. Two of the cat collars were similarly deployed in test positions. Collars fell in a range of habitats and orientations, which should encompass some of the inherent variability in GPS performance. Fixes were collected at

15-min or 5-min intervals. Other variables were generally consistent with collars deployed on animals.

One feral cat test collar collected on a varying "add-time" schedule, which had a significant effect on mean error (two-sample t test: t = 2.643, df = 51.103, P = 0.011), hence data were restricted to 5-sec

196 Appendix C: GPS screening methods add-time, consistent with the other collars. One collar recorded three fixes that were in excess of

200,000 km from the collar position. These fixes were removed prior to analysis.

I used the median from each sample's coordinates as the barycentre to represent actual collar positions. The median was more appropriate than the mean, which could be influenced by outlying errors. I then calculated the 'error distance' of each fix as the Euclidean distance from the median point, then graphically compared the pooled error distance against a range of variables that may be correlated with fix position:

 Horizontal dilution of precision (hdop)

 No. satellites used to calculate the position (sat)

 Maximum satellite signal strength (maxsnr)

 Fix type: 2D or 3D (fix)

 Difference (alt_diff) between the elevation estimated by the GPS collar and a 1-second

Digital Elevation Model (DEM) (Source: Geoscience Australia) (approximately 30 m at

16-17 ° latitude) at the position of the recorded fix. The altitude of the recorded fix was

used rather than that of the barycentre, as the true fix position would not be known with

actual data.

I then applied different filtering rules to the individual data sets to examine their effect on accuracy (mean error), precision (standard deviation of the error) and number of errors removed. I used a combination of rules to remove 2D fixes, fixes at high hdop values, and fixes with large differences in altitude (Table C.1).

I calculated the weighted mean to compare mean error distance across different stationary tests and used formulas for pooled standard deviation and standard error. To compare levels of precision and data loss among tests, I took the arithmetic mean of the number of errors and standard deviation values for each dataset and calculated standard deviation and standard error using general rules. I used R (R Development Core Team 2013) and Microsoft Excel for all analysis.

197 Appendix C: GPS screening methods

Table C.1. Filtering rules applied to the test data.

Abbreviation Filtering rule No screening No screening applied HDOP<9 Fixes with hdop ≥ 9 removed HDOP<7 Fixes with hdop ≥ 7 removed HDOP<5 Fixes with hdop ≥ 5 removed 3D All 2D fixes removed 3D, HDOP<9 All 2D fixes, and those with hdop ≥ 9 removed 3D, HDOP<7 All 2D fixes, and those with hdop ≥ 7 removed 3D, HDOP<5 All 2D fixes, and those with hdop ≥ 5 removed AD<100 Fixes with alt_diff ≥ 100 m removed AD<100, HDOP<9 Fixes with alt_diff ≥ 100 m, and hdop ≥ 9 removed AD<100, HDOP<7 Fixes with alt_diff ≥ 100 m, and hdop ≥ 7 removed AD<100, HDOP<5 Fixes with alt_diff ≥ 100 m, and hdop ≥ 5 removed 3D,AD<100 All 2D fixes and those with alt_diff ≥ 100 m removed 3D,AD<100,HDOP<9 All 2D fixes, those with alt_diff ≥ 100 m and hdop ≥ 9 removed 3D,AD<100,HDOP<7 All 2D fixes, those with alt_diff ≥ 100 m and hdop ≥ 7 removed 3D,AD<100,HDOP<5 All 2D fixes, those with alt_diff ≥ 100 m and hdop ≥ 5 removed

Results

I noted many obvious errors during exploratory analysis of the GPS data that were not identified by the indicator variables. Furthermore, fixes with extreme indicator values were not necessarily suspicious or outside the general trajectory. The four indicators produced by the GPS collars did not appear to be strongly correlated with the pooled error distance (Figure 0.1), hence they could not provide a clear rule to identify large errors. Although the mean error distance for 2D fixes

( 24.27) was higher than for 3D fixes ( 6.45), there were more large outlying 3D errors

(Figure 0.1a,b). High hdop values tend to have larger error distances ( 5.10 and 38.79 for hdop = 1 and 9.9 respectively), but some of the largest errors have hdop values that would be considered within a safe range (Figure 0.1c,d). A very large number of satellites do tend to result in smaller errors, but at lower values the relationship is unclear. Maximum signal strength does not appear to have a linear relationship to error distance. I did not use regression to further explore these relationships, as error distance was highly overdispersed and could not be transformed to normality.

198 Appendix C: GPS screening methods

a. b.

c. d.

e. f.

g. h.

199 Appendix C: GPS screening methods

The discrepancy between the GPS-recorded and DEM altitude values appeared to have a stronger positive relationship with error (Figure C.2). Error was considerably lower when alt_diff

<100 m ( 5.19) compared fixes with an alt_diff value between 100 and 200 m ( 104.93)

(Figure C.2c). Reducing the alt_diff cut-off to 50 m reduced the error further, but with a greater increase in data loss (Figure C.2d).

a. b.

c. d.

Figure C.2. Relationship between the difference in altitude reading and a) all errors, b) errors < 1 km and c) errors < 100 m. Fixes in c) are also restricted to those with < 200 m difference in altitude reading, and show the cut-off used in screening for altitude reading, and show the cut-off used in screening for altitude difference at 100 m (red dashed line). The trade-off between mean error (thick line) and data loss (thin line) with altitude value cut -off is shown in 2d.

There were minimal declines in mean error and mean standard deviation as screening rules removed fixes with progressively lower cut-off points for hdop values (Figure C.3). The weighted mean error when only fixes with hdop values <5 were retained was 6.48 compared to 6.62 without screening. Removing all 2D fixes slightly lowered the mean standard deviation from 23.82 prior to

200 Appendix C: GPS screening methods screening to 22.30, but resulted in higher rates of data loss ( 16.81 fixes). However, removing fixes with large differences in altitude reading considerably lowered the mean standard deviation to

11.67 and produced a small but noticeable reduction in mean error to 5.17. Further, this technique removed fewer errors ( 10.25 fixes) than the 3D or HDOP < 5 ( 15.75 fixes) filtering rules.

Further removal of 3D fixes or those with high hdop values following screening for altitude, had little effect on the mean or standard deviation of errors, but led to increased data loss (Figure C.4).

Of the five fixes in the pooled data that had errors > 1 km, all were 3D fixes and had hdop values under 2.5. However, all but one had differences in altitude reading above 100 m. Similarly, of the 14 fixes with errors > 500 m, all were 3D with low hdop values, and all but two were removed by screening by altitude, making this the only filter rule that removed most of the extreme outliers.

201 Appendix C: GPS screening methods

202 Appendix C: GPS screening methods

Figure C.4. The mean (weighted) and standard deviation of error in fixes following screening methods that removed fixes with differences in altitude ≥ 100 m.

Discussion

This analysis demonstrated that the traditional bias indicators of dilution of precision and fix type do not consistently detect erroneous fixes. Relationships between the indicators and the error distance were neither clear nor linear, and large errors could be recorded with indicator values that would suggest good precision. There appeared to be no relationship between satellite signal strength and fix error, and the number of satellites only indicated precision when there were ≥10 satellites deployed.

Although removing fixes with high hdop values has improved fix accuracy in other studies (e.g. hdop > 5, Rempel & Rodgers 1997), I did not see any considerable improvement in mean error or mean standard deviation by removing fixes at stricter hdop cut-off values. Some dilution of precision metrics, such as geometric DOP and positional DOP, incorporate elevation into the indicator (MELP 2001), and have been found to more reliably predict erroneous fixes

(Lewis et al. 2007). As our collars only recorded horizontal DOP, which does not incorporate

203 Appendix C: GPS screening methods discrepancy in altitude, this indicator was less useful. I would not recommend relying on hdop as a filter rule without incorporating some measure of altitude discrepancy.

Screening out two-dimensional fixes only resulted in slight improvements in mean error and standard deviation. Previous studies have found removal of 2D fixes to increase precision of datasets (Lewis et al. 2007), but this technique may also incur large data losses (D'Eon &

Delparte 2005). The stationary tests in this analysis did not record large numbers of 2D fixes, which would explain why fewer errors were removed using 3D filtering rules.

The difference in altitude between the GPS-recorded value and the DEM was the most reliable indicator to detect erroneous fixes. Heselton (1998) found that error in elevation reading was related to horizontal error for 2D fixes, but this analysis found similar discrepancies with

3D fixes. Further, screening data by difference in altitude reading was the only method to identify the majority of extreme outlying errors.

This analysis estimated the error in GPS datasets specific to the study site and collars used in the research of this thesis. It provides justification for screening rules used in the subsequent analysis of dingo and feral cat movement data. The most appropriate filter rule derived from this analysis, which reduced the mean and standard deviation of error distance, while maximising data retention, is to screen out all fixes with a difference in altitude of ≥ 100 m. Further filtering rules that remove 2D fixes or those with high hdop values, did not provide considerable benefits to the precision of the data.

204

Appendix D Additional tables from Chapter 4

Table D.1. Summary of data collection and screening outcomes for all collared individuals.

No. No. 15- No. % fix % 3D Date Species ID Year screened min Fixes start Fixes end days rate rate caught fixes fixes fixes Dingoes DS_M1 2011 1,134 561 98.44% 98.94% 22/06/2011 20/07/2011 30/07/2011 6 DS_M2 2011 7,327 3,728 98.03% 98.89% 23/06/2011 20/07/2011 06/10/2011 40 DS_M3 2011 7,584 3,979 98.75% 98.54% 23/06/2011 20/07/2011 10/10/2011 42 DS_M4 2011 7,202 3,654 98.33% 98.74% 24/06/2011 20/07/2011 05/10/2011 39 DS_F1 2011 7,759 4,133 98.75% 99.08% 23/06/2011 20/07/2011 14/10/2011 44 DS_F2 2011 7,372 3,774 98.45% 98.14% 24/06/2011 20/07/2011 06/10/2011 40 DM_M1a 2011 7,381 3,773 98.57% 98.97% 14/07/2011 20/07/2011 06/10/2011 40 DM_M1b 2012 4,402 2,140 97.54% 99.36% 09/06/2012 12/06/2012 23/07/2012 23 DM_M2 2011 6,646 3,418 98.90% 98.77% 26/07/2011 28/07/2011 06/10/2011 36 DM_F1 2011 1,435 694 77.03% 92.82% 14/07/2011 20/07/2011 06/08/2011 9 DM_F2 2011 3,078 1,530 98.56% 98.96% 27/07/2011 28/07/2011 29/08/2011 16 DM_F3 2012 5,930 2,932 98.33% 99.44% 08/06/2012 12/06/2012 06/08/2012 31 DM_F4 2012 8,202 4,090 97.89% 98.12% 07/06/2012 11/06/2012 26/08/2012 44 DM_F5 2012 8,532 4,173 97.22% 99.74% 15/06/2012 17/06/2012 04/09/2012 46 DM_F6 2012 4,807 2,363 98.52% 99.94% 19/06/2012 23/06/2012 06/08/2012 25

Feral CM_M1 2011 4,748 4,748 97.02% 98.15% 12/07/2011 20/07/2011 29/10/2011 51 cats CM_M2 2011 4,611 4,611 98.67% 99.11% 16/07/2011 20/07/2011 25/10/2011 49 CM_M3 2011 4,929 4,929 98.74% 99.37% 18/07/2011 20/07/2011 30/10/2011 52 CM_M4a 2011 1,566 1,566 97.75% 97.64% 27/07/2011 28/07/2011 30/08/2011 17 CM_M4b 2012 5,075 5,075 89.33% 97.26% 06/06/2012 10/06/2012 02/10/2012 59 CM_M6 2012 2,094 2,094 96.90% 85.63% 17/01/2012 04/05/2012 10/08/2012 23 CM_M7 2012 5,773 5,773 93.96% 80.69% 02/05/2012 03/05/2012 28/08/2012 64 CM_M8 2012 7,702 7,702 94.98% 89.28% 02/05/2012 06/05/2012 13/10/2012 85 CM_F1 2012 5,077 5,077 94.07% 89.30% 16/05/2012 20/05/2012 03/09/2012 56

Animal ID codes: D = dingo, C = cat; S = Siddins Valley, M = Mornington; F = female, M = male; number indicates order of capture. a 2011 data. b 2012 data.

205 Appendix D: Space use tables

Table D.2. Utilisation distribution estimates (KDE) fo r all collared individuals.

KDE Utilisation Distribution

UD95 Core Core as % Intensity ID year Core (ha) Asymptote (ha) isopleth UD95 index

DS_F1 2011 1,725.44 71.10% 510.24 29.57% 2.40 Yes DS_F2 2011 644.48 76.30% 46.24 7.17% 10.63 No DS_M1 2011 1,476.8 72.90% 288 19.50% 3.74 No DS_M2 2011 2,018.72 73.80% 347.04 17.19% 4.29 Yes DS_M3 2011 71,863.04 67.00% 22,042.24 30.67% 2.18 No DS_M4 2011 2,363.04 70.80% 646.72 27.37% 2.59 Yes DM_F1 2011 1,949.76 66.80% 293.12 15.03% 4.44 No DM_F2 2011 1,278.4 75.80% 222.08 17.37% 4.36 Yes DM_M1a 2011 1,437.6 75.50% 341.76 23.77% 3.18 No DM_M2 2011 9,405.28 67.90% 2,723.36 28.96% 2.34 Yes DM_F3 2012 1,430.08 67.40% 428.48 29.96% 2.25 Yes DM_F4 2012 715.52 73.00% 86.56 12.10% 6.03 Yes DM_F5 2012 1,976.48 67.80% 477.92 24.18% 2.80 Yes DM_F6 2012 3,342.72 70.40% 819.84 24.53% 2.87 Yes DM_M1b 2012 2,728.48 65.40% 892.48 32.71% 2.00 Yes DM_M1c both 2,444.48 70.40% 646.88 26.46% 2.66 No

CM_M1 2011 872 69.00% 292.48 33.54% 2.06 Yes CM_M2 2011 415.84 69.20% 91.68 22.05% 3.14 Yes CM_M3 2011 576 67.20% 149.44 25.94% 2.59 Yes CM_M4a 2011 740.16 62.10% 245.6 33.18% 1.87 Yes CM_F1 2012 127.52 69.40% 41.92 32.87% 2.11 Yes CM_M4b 2012 1,094.4 66.60% 335.04 30.61% 2.18 Yes CM_M6 2012 1,321.6 65.40% 479.04 36.25% 1.80 Yes CM_M7 2012 416.96 66.80% 130.08 31.20% 2.14 Yes CM_M8 2012 988.64 63.60% 326.56 33.03% 1.93 Yes CM_M4c both 1176 64.90% 346.72 29.48% 2.20 Yes

Animal ID codes: D = dingo, C = cat; S = Siddins Valley, M = Mornington; F = female, M = ma le; number indicates order of capture. a 2011 data. b 2012 data. c Combined data from both years

206 Appendix D: Space use tables

Table D.3. Utilisation distribution estimates (MKDE) for all collared individuals.

MKDE Utilisation Distribution

UD95 Core Core Core as % Intensity ID Asymptote (ha) isopleth (ha) UD95 index

DS_F1 1,615.04 71.40% 438.56 27.15% 2.63 Yes DS_F2 1,389.6 74.30% 123.36 8.88% 8.37 No DS_M1 734.24 76.10% 121.92 16.60% 4.58 No DS_M2 1,687.84 74.90% 264.96 15.70% 4.77 Yes DS_M3 13,748.96 72.50% 2,349.4 17.09% 4.24 No DS_M4 2,096.8 72.80% 490.24 23.38% 3.11 Yes DM_F1 1,222.56 69.20% 173.76 14.21% 4.87 No DM_F2 1,244 76.70% 206.88 16.63% 4.61 Yes DM_M1a 1,551.04 76.30% 366.56 23.63% 3.23 No DM_M2 2,490.24 74.40% 632.96 25.42% 2.93 No DM_F3 1,278.88 68.60% 392.64 30.70% 2.23 Yes DM_F4 1,083.36 70.20% 132.8 12.26% 5.73 Yes DM_F5 1,932.96 68.00% 456.16 23.60% 2.88 Yes DM_F6 2,314.08 74.00% 475.36 20.54% 3.60 No DM_M1b 1,804.96 71.10% 529.6 29.34% 2.42 Yes DM_M1c 2,254.88 72.10% 605.28 26.84% 2.69 Yes

CM_M1 589.6 67.50% 178.88 30.34% 2.22 Yes CM_M2 386.88 68.80% 88.48 22.87% 3.01 Yes CM_M3 513.76 67.00% 130.88 25.47% 2.63 Yes CM_M4a 516.96 66.60% 139.2 26.93% 2.47 Yes CM_F1 132.96 66.60% 45.44 34.18% 1.95 Yes CM_M4b 857.12 67.10% 226.08 26.38% 2.54 Yes CM_M6 630.88 68.40% 174.88 27.72% 2.47 Yes CM_M7 375.68 65.10% 112.64 29.98% 2.17 Yes CM_M8 793.44 64.90% 236.32 29.78% 2.18 Yes CM_M4c 960.64 65.50% 245.76 25.58% 2.56 Yes

Animal ID codes: D = dingo, C = cat; S = Siddins Valley, M = Mornington; F = female, M = male; number indicates order of capture. a 2011 data. b 2012 data. c Combined data from both years

207 Appendix D: Space use tables

Table D.4. Intensity and recursion distribution estimates (MKDE) for all collared individuals.

MKDE Intensity Distribution MKDE Recursion Distribution

ID95 Core Core Core as Intensity Core Core Core as Intensity ID RD95 (ha) (ha) isopleth (ha) % ID95 index isopleth (ha) % RD95 index

DS_F1 2,424.48 70.20% 463.2 19.11% 3.67 2,488.48 65.50% 955.36 38.39% 1.71 DS_F2 3,912 57.50% 629.92 16.10% 3.57 3,711.52 64.80% 1,042.72 28.09% 2.31 DS_M1 1,059.68 66.70% 179.52 16.94% 3.94 1,228 56.60% 397.28 32.35% 1.75 DS_M2 3,744.64 71.50% 599.84 16.02% 4.46 4,135.2 62.50% 1,002.24 24.24% 2.58 DS_M3 16,402.4 67.60% 2,662.08 16.23% 4.17 20,471.36 57.40% 7,297.6 35.65% 1.61 DS_M4 3,807.84 70.70% 674.88 17.72% 3.99 4,063.2 63.00% 1,251.52 30.80% 2.05 DM_F1 1,986.24 62.40% 437.44 22.02% 2.83 2,170.56 57.80% 657.76 30.30% 1.91 DM_F2 2,483.52 69.00% 376.48 15.16% 4.55 2,967.04 59.00% 700.96 23.62% 2.50 DM_M1a 3,378.72 65.40% 718.72 21.27% 3.07 3,339.2 63.60% 899.84 26.95% 2.36 DM_M2 3,807.36 70.10% 851.68 22.37% 3.13 4,614.56 60.00% 1,309.12 28.37% 2.11 DM_F3 1,924.32 66.40% 510.56 26.53% 2.50 1,897.92 63.90% 600.64 31.65% 2.02 DM_F4 1,897.92 67.70% 408.64 21.53% 3.14 1,955.2 62.70% 631.52 32.30% 1.94 DM_F5 2,899.84 68.00% 649.44 22.40% 3.04 2,835.04 63.70% 802.08 28.29% 2.25 DM_F6 3,334.72 68.10% 632.48 18.97% 3.59 3,776.64 61.90% 1,190.72 31.53% 1.96 DM_M1b 2,132 69.80% 452.8 21.24% 3.29 2,644.8 60.60% 868.96 32.86% 1.84 DM_M1c 3,664.8 66.80% 744.32 20.31% 3.29 3,867.36 62.70% 1,053.76 27.25% 2.30

CM_M1 946.4 64.10% 268.64 28.39% 2.2 6 937.12 64.40% 315.36 33.65% 1.91 CM_M2 600.96 63.10% 138.56 23.06% 2.74 593.44 62.40% 232.96 39.26% 1.59 CM_M3 691.2 63.00% 189.76 27.45% 2.29 697.92 60.00% 248.64 35.63% 1.68 CM_M4a 601.76 60.50% 141.6 23.53% 2.57 629.76 59.10% 228.64 36.31% 1.63 CM_F1 193.92 62.00% 58.88 30.36% 2.04 169.28 68.40% 64.48 38.09% 1.80 CM_M4b 1,162.4 60.20% 317.76 27.34% 2.20 1,125.12 62.10% 391.2 34.77% 1.79 CM_M6 710.56 66.60% 183.84 25.87% 2.57 826.56 59.60% 299.52 36.24% 1.64 CM_M7 543.2 59.10% 180 33.14% 1.78 459.68 64.70% 168.48 36.65% 1.77 CM_M8 995.04 61.60% 298.56 30.00% 2.05 956.48 59.60% 314.4 32.87% 1.81 CM_M4c 1,304.64 60.00% 346.4 26.55% 2.26 1,235.04 62.30% 424.16 34.34% 1.81

Animal ID codes: D = dingo, C = cat; S = Siddins Valley, M = Mornington; F = female, M = male; number indicates order of capture. a 2011 data. b 2012 data. c Combined data from both years

208 Appendix D: Space use tables

Table D.5. Minimum Convex Polygon (MCP) estimates for all collared individuals using different proportions of the data.

Minimum Convex Polygon

ID MCP100 (ha) MCP95 (ha) MCP90 (ha) MCP50 (ha)

DS_F1 4,816.88 2,670.62 2,510.17 837.14 DS_F2 6,974.02 1,964.40 884.69 0.41 DS_M1 3,431.27 3,026.91 2,186.01 121.62 DS_M2 16,181.30 11,053.52 5,106.01 436.39 DS_M3 179,025.23 149,691.47 144,550.13 52,308.87 DS_M4 7,261.99 4,764.40 3,394.59 678.71 DM_F1 7,365.35 2,043.72 1,512.14 178.98 DM_F2 3,816.17 2,959.32 2,261.64 162.88 DM_M1a 5,411.71 2,377.19 1,799.03 196.76 DM_M2 29,343.83 28,220.44 27,219.17 17,157.10 DM_F3 3,533.64 2,204.72 1,408.75 394.55 DM_F4 2,082.29 1,957.76 1,799.65 42.24 DM_F5 4,849.15 2,993.62 2,413.56 464.73 DM_F6 5,169.00 5,076.54 4,431.05 1,138.50 DM_M1b 4,167.26 3,527.08 2,878.40 768.60 DM_M1c 6,650.67 3,764.50 2,786.56 392.77

CM_M1 4,649.76 841.39 603.28 181.48 CM_M2 785.48 546.18 482.75 186.14 CM_M3 887.06 696.16 628.41 255.02 CM_M4a 791.67 655.07 586.82 301.95 CM_F1 214.65 166.06 104.52 33.47 CM_M4b 1,541.09 1,126.40 988.04 574.94 CM_M6 1,599.24 1,452.45 1,336.81 697.61 CM_M7 790.58 636.10 459.59 116.19 CM_M8 1,322.48 1,169.43 1,006.24 444.82 CM_M4c 1,720.64 1,274.45 1,060.51 499.26

Animal ID codes: D = dingo, C = cat; S = Siddins Valley, M = Mornington; F = female, M = male; number indicates order o f capture. a 2011 data. b 2012 data. c Combined data from both years

209 Appendix D: Space use tables

210

Appendix E Summary of Incremental Area Analysis

Table E.1. Summary of animals for who the density estimates for each distribution type reached sufficient asymptote.

Dingoes Feral cats KDE MKDE KDE MKDE DM_F3 DM_F3 CM_M1 CM_M1

DM_M1 (2012) DM_M1 (2012) CM_F1 CM_F1

DM_M1 (both years) CM_M8 CM_M8

DS_F1 DS_F1 CM_M4 (2011) CM_M4 (2011)

DM_F2 DM_F2 CM_M4 (2012) CM_M4 (2012)

DS_M2 DS_M2 CM_M4 (both years) CM_M4 (both years)

DM_F4 DM_F4 CM_M6 CM_M6

DM_F5 DM_F5 CM_M3 CM_M3

DS_M4 DS_M4 CM_M7 CM_M7

DM_M2 CM_M2 CM_M2

DM_F6

211 Appendix E: Space use asymptotes

Dingoes a.

Figure E.1. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug -in method) and b) MKDE utilisation distributions for DS_M1, a male dingo collared in 2011 in Siddins Valley. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line).

212 Appendix E: Space use asymptotes a.

Figure E.2. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug -in method) and b) MKDE utilisation distributions for DS_M2, a male dingo collared in 2011 in Siddins Valley. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line).

213 Appendix E: Space use asymptotes a.

Figure E.3. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug-in method) and b) MKDE utilisation distributions for DS_M3, a male dingo collared in 2011 in Siddins Valley. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line).

214 Appendix E: Space use asymptotes a.

Figure E.4. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug-in method) and b) MKDE utilisation distributions for DS_M4, a male dingo collared in 2011 in Siddins Valley. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line).

215 Appendix E: Space use asymptotes a.

Figure E.5. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug-in method) and b) MKDE ut ilisation distributions for DS_F1, a female dingo collared in 2011 in Siddins Valley. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line).

216 Appendix E: Space use asymptotes a.

Figure E.6. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug-in method) and b) MKDE utilisation distributions for DS_F2, a female dingo collared in 2011 in Siddins Valley. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line).

217 Appendix E: Space use asymptotes a.

Figure E.7. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug-in method) and b) MKDE ut ilisation distributions using data from 2011 for DM_M1, a male dingo collared in both years at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line).

218 Appendix E: Space use asymptotes a.

Figure E.8. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug-in method) and b) MKDE utilisation distributions using data from 2012 for DM_M1, a male dingo collared in both years at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line).

219 Appendix E: Space use asymptotes a.

Figure E.9. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug-in method) and b) MKDE utilisation distributions using data from both years for DM_M1, a male dingo collared in both years at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line).

220 Appendix E: Space use asymptotes a.

Figure E.10. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug-in method) and b) MKDE utilisation distributions for DM_M2, a male dingo collared in 2011 at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line).

221 Appendix E: Space use asymptotes a.

Figure E.11. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug-in method) and b) MKDE utilisation distributions for DM_F1, a female dingo collared in 2011 at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line).

222 Appendix E: Space use asymptotes a.

Figure E.12. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug-in method) and b) MKDE utilisation distributions for DM_ F2, a female dingo collared in 2011 at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line).

223 Appendix E: Space use asymptotes a.

Figure E.13. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug-in method) and b) MKDE utilisation distributions for DM_F3, a female dingo collared in 2012 at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line).

224 Appendix E: Space use asymptotes a.

Figure E.14. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug-in method) and b) MKDE utilisation distributions for DM_F4, a female dingo collared in 2012 at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line).

225 Appendix E: Space use asymptotes a.

Figure E.15. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug-in method) and b) MKDE utilisation distributions for DM_F5, a female dingo collared in 2012 at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively , around the 95% KDE calculated with all data (red line).

226 Appendix E: Space use asymptotes a.

Figure E.16. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug-in method) and b) MKDE utilisation distributions for DM_F6, a female dingo collared in 2012 at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line).

227 Appendix E: Space use asymptotes

Feral cats a.

Figure E.17. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug-in method) and b) MKDE utilisation distributions for CM_M1, a male cat collared in 2011 at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line).

228 Appendix E: Space use asymptotes a.

Figure E.18. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug-in method) and b) MKDE utilisation distributions for CM_M2, a male cat collared in 2011 at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line).

229 Appendix E: Space use asymptotes a.

Figure E.19. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug-in method) and b) MKDE utilisation distributions for CM_M3, a male cat collared in 2011 at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line).

230 Appendix E: Space use asymptotes a.

Figure E.20. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug-in method) and b) MKDE utilisation distributions using data from 2011 for CM_M4, a male cat collared in both years at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line).

231 Appendix E: Space use asymptotes

Figure E.21. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug-in method) and b) MKDE utilisation distributions using data from 2012 for CM_M4, a male cat collared in both years at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line).

232 Appendix E: Space use asymptotes a.

Figure E.22. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug-in method) and b) MKDE utilisation distributions using data from both years for CM_M4, a male cat collared in both years at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line).

233 Appendix E: Space use asymptotes a.

Figure E.23. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug-in method) and b) MKDE utilisation distributions for CM_M 6, a male cat collared in 2012 at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line).

234 Appendix E: Space use asymptotes a.

Figure E.24. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug-in method) and b) MKDE utilisation distributions for CM_M 7, a male cat collared in 2012 at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line).

235 Appendix E: Space use asymptotes a.

Figure E.25. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug-in method) and b) MKDE utilisation distributions for CM_M 8, a male cat collared in 2012 at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line).

236 Appendix E: Space use asymptotes a.

Figure E.26. Seasonal space use area (ha) against proportion of data included, for a) KDE (plug-in method) and b) MKDE utilisation distributions for CM_F1, a female cat collared in 2012 at Mornington. Blue and purple lines represent 5% and 10% boundaries respectively, around the 95% KDE calculated with all data (red line).

237 Appendix E: Space use asymptotes

238

Appendix F Figures of density estimates for dingoes and feral cats

a. d.

b. e.

c. f.

239 Appendix F: Space use figures

a. b. c.

d. e. f.

Figure F.2. Ninety-five percent and core isopleths for the a) KDE and b) MKDE utilisation distribut ions and c) fixes and trajectory of DS_M 3, a male dingo, and the d) KDE and e) MKDE utilisation distributions and f) fixes and trajectory of DS_ F1, a female dingo, both collared in 2011 in Siddins Valley.

240 Appendix F: Space use figures

a. d.

b. e.

c. f.

Figure F.3. Ninety-five percent and core isopleths for the a) KDE and b) MKDE utilisation distributions and c) fixes and trajectory of DS_M 4, a male dingo, and the d) KDE and e) MKDE utilisation distributions and f) fixes and trajectory of DS_F2, a female dingo, both collared in 2011 in Siddins Valley.

241 Appendix F: Space use figures

a. b. c.

d. e. f.

242 Appendix F: Space use figures

a. b. c.

d. e. f.

Figure F.5. Ninety-five percent and core isopleths for the a) KDE and b) MKDE utilisation distributions and c) fixes and trajectory of DM_ F1, and the d) KDE and e) MKDE utilisation distributions and f) fixes and trajectory of DM_ F2, both male dingoes collared in 2011 at Mornington.

243 Appendix F: Space use figures

a. b. c.

d. e. f.

Figure F.6. Ninety-five percent and core isopleths for the a) KDE and b) MKDE utilisation distributions and c) fixes and trajectory of DM_F 3, and the d) KDE and e) MKDE utilisation distributions and f) fixes and trajectory of DM_F4, both female dingoes collared in 2012 at Mornington.

244 Appendix F: Space use figures

a. b. c.

d. e. f.

Figure F.7. Ninety-five percent and core isopleths for the a) KDE and b) MKDE utilisation distributions and c) fixes and trajectory of DM_F 5, and the d) KDE and e) MKDE utilisation distribut ions and f) fixes and trajectory of DM_F6, both female dingoes collared in 2012 at Mornington.

245 Appendix F: Space use figures

a. b. c.

d. e. f.

246 Appendix F: Space use figures

a. d.

b. e.

c. f.

Figure F.9. Ninety-five percent and core isopleths for the a) KDE and b) MKDE utilisation distributions and c) fixes and trajectory of CM_M 3, a male cat collared in 2011, and the d) KDE and e) M KDE utilisation distributions and f) fixes and trajectory of CM_M4, a male cat collared in both years at Mornington.

247 Appendix F: Space use figures

a. d.

b. e.

c. f.

248 Appendix F: Space use figures

a. b. c.

d. e. f.

Figure F.11. Ninety-five percent and core isopleths for the a) KDE and b) MKDE utilisation distributions and c) fixes and trajectory of CM_M 8, a male cat collared in 2012, and the d) KDE and e) MKDE utilisation distributions and f) fixes and trajectory of CM_ F1, a female cat collared in 2012 at Mornington.

249 Appendix F: Space use figures

250

Appendix G Additional plots and tables from Chapter 5

251 Appendix G: Spatial overlap supporting information

Figure G.1. Boxplots comparing a) UDOI and b) PHR between dyad types at the 95% and core contours for the MKDE utilisation distribution.

252 Appendix G: Spatial overlap supporting information

Dyad type

Distribution contour Dingo-dingo Dingo-cat Cat-dingo Cat-cat Utilisation 95% 0.22 ± 0.12 (14) 0.03 ± 0.02 (22) 0.002 ± 0.001 (3)

0.19 ± 0.05 (28) 0.06 ± 0.02 (22) 0.21 ± 0.04 (22) 0.05 ± 0.01 (6)

Utilisation core 0.04 ± 0.02 (6) 0.003 ± 0.002 (13) 0.0001 ± 0.0001 (3)

0.08 ± 0.03 (12) 0.02 ± 0.01 (13) 0.04 ± 0.01 (13) 0.006 ± 0.003 (6)

Intensity core 0.01 ± 0.01 (12) 0.002 ± 0.001 (20) 0.0004 ± 0.0002 (3)

0.05 ± 0.02 (24) 0.01 ± 0.00 (20) 0.05 ± 0.01 (20) 0.02 ± 0.01 (6)

Recursion core 0.05 ± 0.03 (11) 0.004 ± 0.001 (17) 0.00003 ± 0.0000 1 (3)

0.11 ± 0.04 (22) 0.02 ± 0.01 (17) 0.09 ± 0.02 (17) 0.005 ± 0.002 (6)

253 Appendix G: Spatial overlap supporting information

Table G.2. Wilcoxon signed -rank test statistics and exact P-values for comparisons between overlap at different distributions for each dyad type, estimated using UDOI. No comparisons were made between distribution types for cat -cat dyads, as there was no significant difference between groups (Asymptotic Friedman test:

= 4.2, P = 0.24). Data sets that could not be log -transformed to ensure the distribution of differences was symmetrical are marked NA.

Dingo-dingo Dingo-feral cat Overall Friedman test: = 31.19, P < 0.001 Overall Friedman test: = 29.51, P < 0.001

UD 95% UD core ID core UD 95% UD core ID core UD 95% - - - UD 95% - - - Z = -3.30 Z = -4.11 UD core - - UD core - - P < 0.001 P < 0.001 Z = -3.30 Z = -1.26 Z = 3.46 Z = -0.91 ID core - ID core - P < 0.001 P = 0.23 P < 0.001 P = 0.38 Z = -3.30 Z = -2.89 Z = 0.94 Z = 3.52 Z = 0.75 RD core RD core NA P < 0.001 P = 0.002 P = 0.37 P < 0.001 P = 0.47

254 Appendix G: Spatial overlap supporting information

Table G.3. Wilcoxon signed -rank test statistics and exact P-values for comparisons between overlap at different distributions for each dyad type, estimated using PHR. Data sets that could not be log -transformed to ensure the distribution of differences was symmetrical are marked NA.

Pr(dingo overlapping with dingo) Pr(dingo overlapping with feral cat) Overall Friedman test: = 57.90, P < 0.001 Overall Friedman test: = 28.77, P < 0.001

UD 95% UD core ID core UD 95% UD core ID core UD 95% - - - UD 95% - - - Z = -4.62 Z = -4.11 UD core - - UD core - - P < 0.001 P < 0.001 Z = 4.53 Z = -1.94 Z = 3.13 Z = -1.04 ID core - ID core - P < 0.001 P = 0.05 P = 0.001 P = 0.31 Z = 4.37 Z = -4.02 Z = 1.40 Z = 3.30 Z = -1.97 Z = 0.32 RD core RD core P < 0.001 P < 0.001 P = 0.17 P < 0.001 P = 0.05 P = 0.76

Pr(feral cat overlapping with dingo) Pr(feral cat overlapping with feral cat) Overall Friedman test: = 38.89, P < 0.001 Overall Friedman test: = 9.80, P = 0.02

UD 95% UD core ID core UD 95% UD core ID core UD 95% - - - UD 95% - - - Z = -4.11 Z = -2.20 UD core - - UD core - - P < 0.001 P = 0.03 Z = 3.75 Z = -1.33 Z = 1.78 Z = -1.57 ID core - ID core - P < 0.001 P = 0.19 P = 0.09 P = 0.16 Z = 3.85 Z = 0.19 Z = 2.20 Z = 0.73 Z = -1.57 RD core NA RD core P < 0.001 P = 0.86 P = 0.03 P = 0.56 P = 0.16

255 Appendix G: Spatial overlap supporting information

Mean Fifth quantile Under 100 m Observed Null Observed Null Observed Null Null Null mean P Null mean P P Dyad type Dyad metric variance metric variance metric mean variance DD DM_F3-DM_M1 2285.00 2865.02 327.96 0.001 6.85 474.88 893.44 0.001 322 13.86 13.27 0.001 DM_F3-DM_F4 3273.10 3426.60 100.98 0.001 1591.24 1723.13 503.65 0.001 1 0.65 0.54 1 DM_F3-DM_F5 2044.84 2659.96 297.69 0.001 8.85 451.42 508.05 0.001 363 16.41 13.62 0.001 DM_F3-DM_F6 6844.34 6856.65 292.00 0.46 3499.90 3500.23 14988.68 1 0 0.02 0.02 1 DM_M1-DM_F2 7117.91 7195.23 228.58 0.001 3063.10 3037.30 8033.28 0.84 0 0.01 0.01 1 DM_M1-DM_M2 13330.95 13434.11 236.14 0.001 1344.32 2045.63 1923.71 0.001 1 0.88 0.87 1 DM_M1-DM_F1 9030.48 9004.55 105.68 0.01 4695.75 4603.53 16506.99 0.46 0 0.00 0.00 1 DM_M1-DM_F4 4868.83 4828.86 102.47 0.001 2220.14 2141.35 967.28 0.03 0 0.09 0.09 1 DM_M1-DM_F5 1726.57 3039.26 776.67 0.001 3.08 570.39 1520.38 0.001 585 12.98 11.98 0.001 DM_M1-DM_F6 6079.26 6348.76 923.00 0.001 2454.32 2853.57 13696.76 0.001 0 0.01 0.01 1 DM_F2-DM_F1 5210.37 5530.20 455.35 0.001 2541.25 2530.51 3655.57 0.86 0 0.14 0.15 1 DM_F4-DM_F5 4046.45 4107.92 38.17 0.001 2001.71 2048.15 292.99 0.002 0 0.10 0.10 1 DM_F5-DM_F6 6270.77 6481.88 411.33 0.001 2658.26 3074.98 7994.49 0.001 0 0.01 0.01 1 DM_M2-DM_F1 7729.84 7759.55 291.18 0.09 3364.65 3827.10 75227.19 0.05 1 0.31 0.27 0.29

CC CM_M1- CM_M3 3580.59 3629.87 14.88 0.001 1410.40 1426.03 694.84 0.57 0 0.65 0.71 0.68 CM_M1-CM_M5 2737.34 2726.49 9.93 0.001 1380.39 1347.96 3976.49 0.61 0 0 0 NA CM_M8-CM_M6 4650.86 4675.91 27.44 0.001 1449.62 1560.85 1977.64 0.011 0 0.60 0.60 0.66 CM_M4-CM_M7 2911.58 2926.38 44.60 0.03 1198.06 1180.93 516.24 0.44 0 0.88 0.87 0.63

256 Appendix G: Spatial overlap supporting information

Table G.4 continued.

Under 500 m Under 1000 m Under 1500 m Under 2000 m Observed Null Null Observed Null Null Observed Null Null Observed Null Null Dyad P P P P type Dyad metric mean variance metric mean variance metric mean variance metric mean variance DD DM_F3-DM_M1 524 114 89.69 0.001 760 306 199.99 0.001 1016 547 250.79 0.001 1186 790 258.56 0.001 DM_F3-DM_F4 5 5 3.95 1 14 21 15.37 0.10 106 93 50.65 0.09 410 306 132.45 0.001 DM_F3-DM_F5 598 153 126.05 0.001 914 413 264.23 0.001 1208 738 344.24 0.001 1508 1052 349.26 0.001 DM_F3-DM_F6 0 0 0.31 1 0 2 1.94 0.17 0 12 9.03 0.001 1 21 15.07 0.001 DM_M1-DM_F2 0 0 0.10 1 0 1 1.26 0.42 0 9 6.74 0.001 0 19 11.84 0.001 DM_M1-DM_M2 11 9 7.99 0.48 58 37 29.06 0.001 192 83 54.20 0.001 318 158 93.59 0.001 DM_M1-DM_F1 0 1 0.83 0.61 0 3 2.74 0.13 0 7 6.54 0.007 0 9 7.79 0.004 DM_M1-DM_F4 0 2 1.60 0.25 0 7 5.66 0.007 40 29 18.08 0.01 75 77 48.62 0.80 DM_M1-DM_F5 778 77 64.48 0.001 957 216 169.47 0.001 1174 401 253.62 0.001 1314 603 302.46 0.001 DM_M1-DM_F6 0 2 1.95 0.17 0 15 11.73 0.001 2 20 15.84 0.001 9 33 24.49 0.001 DM_F2-DM_F1 1 0 0.39 0.34 2 1 1.05 0.62 6 2 1.92 0.02 12 9 3.91 0.20 DM_F4-DM_F5 0 1 1.30 0.40 4 6 5.05 0.37 54 35 22.48 0.001 180 154 76.89 0.003 DM_F5-DM_F6 0 0 0.38 1 0 5 4.65 0.02 11 10 8.37 0.74 33 22 18.69 0.016 DM_M2-DM_F1 3 1 0.83 0.11 5 4 1.81 0.44 11 7 2.81 0.03 14 11 2.82 0.08

CC CM_M1- CM_M3 18 14 13.47 0.34 87 84 64.04 0.69 279 274 146.05 0.68 565 613 254.34 0.003 CM_M1-CM_M5 1 1 0.75 1 11 8 4.93 0.19 60 33 16.05 0.001 138 82 22.37 0.001 CM_M8-CM_M6 3 11 9.62 0.009 40 45 34.54 0.43 138 116 72.83 0.01 339 219 119.94 0.001 CM_M4-CM_M7 13 30 25.34 0.001 106 118 82.20 0.20 452 380 198.97 0.001 918 850 319.52 0.001

257 Appendix G: Spatial overlap supporting information

Table G.5. Distance analysis results for dingo -cat dyads. P-values are from randomisation tests comparing the observed metric with the null distribution. P-values in italics should be treated with caution, as the null distribution was not considered symmetrical by the MGG test (Miao et al. 2006).

Mean 5th Quantile Under 100 m Observed Null Observed Null Observed Null Null Null mean P Null mean P P Dyad metric variance metric variance metric mean variance DM_F3-CM_M8 3099.23 3155.49 100.86 0.001 943.74 798.63 682.59 0.001 0 3.15 3.34 0.09 DM_F3-CM_M4 6225.82 6213.76 21.33 0.01 3433.54 3151.71 3699.27 0.001 0 0.07 0.07 1 DM_F3-CM_M6 2431.53 2362.25 360.48 0.001 687.34 665.79 2134.85 0.67 0 3.49 3.42 0.10 DM_M1-CM_M1 7225.99 7200.64 29.26 0.001 4307.46 4485.01 2808.48 0.001 0 0.01 0.01 1 DM_M1-CM_M3 4124.38 4102.71 27.74 0.002 2087.00 2094.05 1087.39 0.84 0 0.15 0.14 1 DM_M1-CM_M8 3535.33 3582.65 175.29 0.001 1209.77 918.73 1031.42 0.001 1 1.39 1.29 1 DM_M1-CM_M6 3244.26 3024.45 702.28 0.001 834.24 672.11 10056.29 0.10 0 1.14 1.04 0.40 DM_F2-CM_M1 3846.05 3880.78 278.81 0.03 499.72 825.57 1167.12 0.001 0 0.87 0.77 0.60 DM_F2-CM_M4 4134.83 4107.50 158.52 0.02 1972.10 1960.23 1329.55 0.78 0 0.03 0.03 1 DM_F2-CM_M3 4631.36 4588.44 119.54 0.001 3067.37 2691.56 6350.48 0.001 0 0.96 0.87 0.63 DM_F2-CM_M2 3601.23 3557.48 53.94 0.001 1670.52 1628.38 1520.84 0.24 0 0.13 0.14 1 DM_F4-CM_M8 4631.37 4683.64 45.34 0.001 2259.68 2616.78 710.42 0.001 0 0.01 0.01 1 DM_F4-CM_M6 3567.80 3578.28 18.53 0.009 1827.06 1849.21 53.86 0.003 0 0.00 0.001 1 DM_F5-CM_M8 3243.54 3333.47 113.00 0.001 1235.47 1012.08 659.02 0.001 0 2.58 2.32 0.11 DM_F5-CM_M4 6132.84 6117.13 20.37 0.002 3441.69 3380.96 2311.68 0.21 0 0.14 0.13 1 DM_F5-CM_M6 2665.39 2500.28 377.35 0.001 724.84 590.41 1514.07 0.001 1 1.82 1.83 0.75 DM_M2-CM_M1 16265.80 16025.98 287.96 0.001 3403.41 3967.08 2496.48 0.001 0 0.00 0.00 1 DM_M2-CM_M3 13760.15 13685.20 49.69 0.001 2295.61 2143.84 685.11 0.001 0 0.05 0.05 1 DM_F6-CM_F1 4120.69 4078.27 21.65 0.001 975.46 1074.91 756.28 0.001 0 0.77 0.76 0.66 DM_F6-CM_M4 4448.80 4288.11 224.97 0.001 1087.61 707.94 4010.91 0.001 1 10.52 9.66 0.00 DM_F6-CM_M6 5831.03 5782.68 143.35 0.002 2618.20 2470.24 2480.91 0.004 0 0.21 0.19 1 DM_F6-CM_M7 5689.20 5679.39 111.92 0.36 1591.66 1393.86 2994.24 0.001 0 1.37 1.36 0.42 DM_F1-CM_M1 4162.29 4348.49 703.82 0.001 1830.43 1871.86 7680.72 0.64 0 0.00 0.00 1 DM_F1-CM_M3 6351.55 6206.51 347.86 0.001 1883.43 2008.43 7531.37 0.153 0 0.06 0.07 1

258 Appendix G: Spatial overlap supporting information

Table G.5 continued.

Under 500 m Under 1000 m Under 1500 m Under 2000 m Observed Null Null Observed Null Null Observed Null Null Observed Null Null P P P P Dyad metric mean variance metric mean variance metric mean variance metric mean variance DM_F3-CM_M8 25 52 45.42 0.001 161 234 165.57 0.001 511 541 276.89 0.07 800 876 294.57 0.001 DM_F3-CM_M4 0 2 1.93 0.17 0 7 5.05 0.002 2 19 7.71 0.001 11 33 10.81 0.001 DM_F3-CM_M6 30 40 33.55 0.10 136 137 103.31 0.95 304 318 169.00 0.30 531 539 171.75 0.54 DM_M1-CM_M1 0 1 0.55 0.66 1 3 2.16 0.33 10 7 4.80 0.17 13 13 7.79 1 DM_M1-CM_M3 4 4 3.46 1 14 29 20.32 0.004 63 73 38.42 0.11 168 162 70.82 0.48 DM_M1-CM_M8 31 28 25.83 0.64 71 128 92.50 0.001 230 289 158.05 0.001 371 476 202.22 0.001 DM_M1-CM_M6 16 22 15.38 0.16 38 49 30.68 0.06 102 89 45.83 0.05 202 164 62.68 0.001 DM_F2-CM_M1 76 31 25.87 0.001 219 125 70.72 0.001 297 220 99.96 0.001 409 378 148.74 0.013 DM_F2-CM_M4 1 2 1.94 0.52 20 10 8.43 0.001 27 28 18.87 0.81 82 83 49.85 0.87 DM_F2-CM_M3 0 6 4.91 0.008 0 14 9.45 0.001 1 26 14.45 0.001 5 43 14.81 0.001 DM_F2-CM_M2 0 8 7.32 0.006 7 29 16.98 0.001 54 60 31.63 0.33 131 133 61.76 0.80 DM_F4-CM_M8 0 0 0.24 1 6 2 2.42 0.03 47 14 12.45 0.001 145 59 43.32 0.001 DM_F4-CM_M6 0 0 0.11 1 3 1 0.76 0.05 15 6 4.48 0.001 188 148 46.42 0.001 DM_F5-CM_M8 3 30 29.20 0.001 47 171 140.71 0.001 422 485 273.49 0.001 762 855 341.81 0.001 DM_F5-CM_M4 0 2 1.57 0.24 0 8 5.87 0.001 0 19 10.56 0.001 3 36 21.24 0.001 DM_F5-CM_M6 37 57 43.64 0.005 136 167 112.11 0.005 381 354 193.58 0.05 606 604 226.06 0.89 DM_M2-CM_M1 2 1 1.04 0.62 6 3 2.88 0.14 13 6 4.70 0.001 16 8 5.94 0.004 DM_M2-CM_M3 0 1 0.92 0.39 0 15 12.88 0.001 10 51 37.60 0.001 70 129 60.71 0.001 DM_F6-CM_F1 28 34 16.13 0.13 111 95 22.30 0.001 193 183 35.70 0.10 331 310 51.78 0.002 DM_F6-CM_M4 8 68 56.79 0.001 69 143 99.09 0.001 272 272 156.70 1 448 439 174.12 0.49 DM_F6-CM_M6 0 2 1.91 0.15 0 5 4.57 0.04 1 13 9.79 0.001 2 30 21.57 0.001 DM_F6-CM_M7 4 19 15.97 0.001 18 53 38.33 0.001 101 124 59.04 0.003 181 199 59.51 0.02 DM_F1-CM_M1 0 1 0.77 0.63 4 6 4.19 0.46 27 17 10.95 0.003 40 41 21.27 0.92 DM_F1-CM_M3 2 2 2.38 1 5 8 5.98 0.30 13 16 11.23 0.38 43 34 18.21 0.05

259 Appendix G: Spatial overlap supporting information

Table G.6. Extended results of dynamic overlap analysis.

Animal Area (ha) Likelihood and P-values for total and partitions Odds ratios Interpretation

Dyad

Spatial Temporal type Dingo- DM_F3 DM_M1 186 2114 4385 1840 <0.001 1.34 <0.001 1.04 <0.001 0.68 0.026 1.78 0.40 0.62 0.21 Symmetrical attraction Simultaneous attraction dingo DM_F3 DM_F4 245 2135 879 62 0.042 -0.18 0.665 -1.11 0.005 -0.18 0.889 0 0.36 0.90 1.05 Avoidance by B

DM_F3 DM_F5 226 2186 3494 1719 <0.001 1.03 <0.001 0.79 <0.001 0.55 0.009 1.64 0.46 0.70 0.37 Symmetrical attraction Simultaneous attraction

DM_M1 DM_M2 275 2286 20500 1834 <0.001 0.95 <0.001 0.79 <0.001 1.08 <0.001 2.48 0 1.01 0.48 Symmetrical attraction Simultaneous attraction

DM_M1 DM_F5 161 4784 3770 3442 <0.001 0.95 <0.001 0.90 0.025 0.44 0.378 1.28 0.49 0.50 0.25 Symmetrical attraction

DM_M1 DM_F6 134 4620 4672 619 0.064 -0.65 0.043 0.25 0.280 -0.72 0.157 0 1.43 0.64 1.01 Avoidance by A

DM_F5 DM_F6 198 3392 5191 350 0.035 -0.49 0.083 -1.03 0.018 -0.01 0.988 0 0.42 0.68 1.09 Symmetrical avoidance

Dingo - DM_F3 CM_M8 243 2172 1495 894 0.090 -0.25 0.066 0.04 0.735 -0.24 0.083 0.77 1.19 1.00 0.96 Random/NS cat Solitary use by A, some DM_F3 CM_M6 112 2227 1555 941 <0.001 1.06 <0.001 -0.14 0.467 -0.45 0.012 1.33 0.66 2.03 0.39 Attraction by A simultaneous use DM_M1 CM_M3 305 2252 682 81 0.001 -1.32 0.015 -0.82 0.001 0.32 0.756 0 0.49 0.31 1.10 Symmetrical avoidance

DM_M1 CM_M8 182 4469 1406 934 0.125 0.13 0.471 -0.28 0.062 -0.23 0.188 0.87 0.91 1.57 1.10 Avoidance by B

DM_M1 CM_M6 54 2555 1582 765 0.008 0.78 0.004 -0.47 0.095 0.19 0.455 1.53 0.44 1.68 1.02 A is attracted, B avoids

DM_F2 CM_M1 127 2159 2837 876 0 0.27 0.127 1.58 0 0.12 0.418 2.51 2.02 0.56 0.38 Attraction by B

DM_F2 CM_M4 125 2250 1017 192 0.165 -1.04 0.033 -0.14 0.554 -0.29 0.662 0 0.97 0.46 1.08 Avoidance by A

DM_F2 CM_M3 128 2161 564 114 0.351 -0.84 0.140 0.15 0.475 -0.57 0.444 0 1.19 0.56 0.99 Random/NS

260 Appendix G: Spatial overlap supporting information

Table G.6 continued.

Animal Area (ha) Likelihood and P-values for total and partitions Odds ratios Interpretation

Dyad

Spatial Temporal type Dingo- DM_F5 CM_M8 306 3138 1391 819 <0.001 -0.52 <0.001 -0.31 0.007 -0.04 0.758 0.51 1.00 0.88 1.29 Symmetrical avoidance cat DM_F5 CM_M4 301 3125 1351 72 0.501 -0.84 0.134 -0.06 0.810 -0.34 0.813 0 0.96 0.46 1.02 Random/NS

DM_F5 CM_M6 126 2412 1823 1250 <0.001 0.80 <0.001 1.01 <0.001 0.81 0.001 1.81 0.65 0.39 0.52 Symmetrical attraction Simultaneous attraction

DM_M2 CM_M1 275 20534 1423 783 0.002 NA NA -0.22 0.063 0.22 0.711 0 0.93 0 1.17 Random

DM_M2 CM_M3 282 20215 672 634 0.003 NA NA 0.74 0.036 -0.74 0.706 0 1.06 0 0.51 Attraction by B

DM_F6 CM_F1 189 5034 126 126 NA -0.47 0.417 NA NA NA NA 0.63 1.00 NA NA NS

DM_F6 CM_M4 171 5608 1083 945 <0.001 1.06 <0.001 -0.38 0.060 -0.07 0.718 2.11 0.71 2.72 1.10 A is attracted, B avoids

Solitary use by A, some DM_F6 CM_M7 188 5317 734 267 0.166 0.34 0.236 0.01 0.930 -0.62 0.055 0.58 1.03 1.83 0.95 Random/NS simultaneous avoidance

Cat- cat CM_M1 CM_M3 388 1334 706 123 <0.001 -1.07 <0.001 -0.26 0.072 0.45 0.168 0.64 0.82 0.31 1.12 Symmetrical avoidance

CM_M1 CM_M5 192 727 456 30 0.300 -0.99 0.077 -0.24 0.465 -0.18 0.890 0 0.83 0.41 1.04 Avoidance by A

CM_M8 CM_M6 224 1527 2069 239 0.248 0.10 0.583 -0.34 0.152 -0.48 0.183 0.25 0.83 1.19 1.00 Random/NS

CM_M4 CM_M7 310 1338 691 83 0.283 0.28 0.176 -0.07 0.689 -0.53 0.177 0.43 0.97 1.42 0.98 Random/NS

261

Appendix H Overlap in MCP as a measure of territoriality

Liberg et al. (2000) consider ranges to be exclusive when neighbouring collared individuals in a population exhibit on average less than 10% overlap in Minimum Convex

Polygon (MCP). Exclusive ranges may indicate that individuals are territorial, defending the resources within an area against conspecifics, usually of the same sex (Liberg et al. 2000). In this appendix, I determine whether the animals collared during this study would be considered to occupy exclusive ranges or territories, given these assumptions.

Methods

To determine whether the individuals collared in this study could be considered to maintain exclusive ranges, I calculated the Minimum Convex Polygon for a) all data (MCP100) and b) all but the furthest 5% of data points (MCP95) for each animal in dyads of dingoes and feral cats. I only considered fixes collected simultaneously at 3-hour intervals.

To consider overlap in space use, I used the % overlap in HR, a simple indicator that calculates , the overlap area of animal with a neighbouring individual as a percentage of animal 's home range (Fieberg & Kochanny 2005).

I only considered individuals that had positive overlap for each metric, when calculating the mean overlap.

263 Appendix H: MCP overlap

Results

Eleven dingo-dingo dyads, eighteen dingo-cat dyads and three cat-cat dyads had positive overlap in space use, as estimated using MCP95, while including the furthest 5% of fixes increased this number to twelve dingo-dingo dyads and twenty-three dingo-cat dyads

(Table H.1).

Table H.1. Percentage of dyads for each dyad type that had less than 10% overlap for MCP95 and MCP100, estimated using HR.

MCP95 MCP100 Dyads with <10% Dyads with <10% Dyad type N N overlap overlap Dingo-dingo 22 28.35 ± 6.76% 24 28.85 ± 5.42% Dingo-cat 18 14.19 ± 4.13% 23 11.45 ± 2.12% Cat-dingo 18 37.77 ± 7.38% 23 42.64 ± 6.79% Cat-cat 6 7.22 ± 1.89% 6 11.68 ± 4.33%

From Table H.1, it can be seen that dingoes overlap with conspecifics on average in just over 25% of their space use. Feral cats have considerable overlap with adjacent dingoes, with around 40% overlap with each adjacent dingo. When overlap occurs between adjacent cats, overlap is lower: under 10% when excluding the furthest 5% of fixes, and just over 10% when all fixes are included.

The data suggest that male feral cats may maintain home ranges exclusive to other large males; however, the extent of overlap with sub-adults or females is unknown.

264