Resource partitioning among five sympatric mammalian herbivores on Yanakie Isthmus, south- eastern

Naomi Ezra Davis

Submitted in total fulfilment of the requirements of the degree of Doctor of Philosophy

September 2010

Department of Zoology The University of Melbourne

i

Abstract

This thesis combines multiple approaches to improve our understanding of large herbivore ecology and organisation in a contemporary assemblage made up of with independent evolutionary histories on Yanakie Isthmus, Wilsons Promontory National Park, , Australia. In particular, this thesis compares niche parameters among populations of five sympatric native and introduced herbivore species by simultaneously assessing overlap in resource use along two dimensions (spatial and trophic) at multiple scales, thereby providing insight into resource partitioning and competition within this herbivore assemblage. Faecal pellet counts demonstrated that inter-specific overlap in herbivore habitat use on Yanakie Isthmus was low, suggesting that spatial partitioning of habitat resources had occured. However, resource partitioning appeared to be independent of coevolutionary history. Low overlap in habitat use implies low competition, and the lack of clear shifts in habitat use from preferred to suboptimal habitats suggested that inter-specific competition was not strong enough to cause competitive exclusion. However, low overlap in habitat use between the European rabbit Oryctolagus cuniculus and other species, and preferential use by rabbits (and avoidance by other species) of the habitat that appeared to have the highest carrying capacity, suggested that rabbits excluded other grazing herbivores from preferred habitat. High overlap in habitat use was apparent between some species, particularly grazers, indicating some potential for competition if resources are limiting. In particular, the eastern grey Macropus giganteus had a narrow niche, occurred at low densities and had low population metabolism relative to other species, consistent with competitive suppression. In contrast, the common wombat Vombatus ursinus appears to be the strongest competitor in this assemblage, being numerically dominant, utilising the greatest proportion of resources, and displaying a relatively broad habitat niche. Habitat modification by fire, including changes in vegetation composition and structure, altered fine-scale partitioning of habitat resources by sympatric herbivore species, and changed the composition of the herbivore community. Faecal pellet counts demonstrated a decrease in herbivore densities, particularly grazers, following the burn, probably associated with reduced ground layer

ii vegetation cover. In contrast, there was an increase in the density of browsers, particularly the swamp Wallabia bicolor, probably associated with increased shrub cover. Fine-scale partitioning of habitat resources was evident through inter-specific differences in abundance, population metabolism and use of fine-scale vegetation strata prior to and after the burn. Inter-specific differences in feeding strategies and thus resource requirements appear to facilitate coexistence within this assemblage. Further, overlap in fine-scale habitat use appears to be greater between native and introduced species than between native species, suggesting that species with independent evolutionary histories have inherently less resource partitioning than species with long coevolutionary histories. Microhistological diet analysis showed that the diets of the relatively large-bodied introduced hog deer Axis porcinus and relatively small native swamp wallaby consisted mainly of dicots. The diet of the small, introduced European rabbit contained similar proportions of monocots and dicots. The diets of the native eastern grey kangaroo, intermediate in size, and large native common wombats, consisted mainly of monocots, but also consumed moderate amounts of dicots. Overlap in food use by the five species was high, particularly between native and introduced species, but also between some native species. Despite a high potential for food resource competition, it appears that coexistence of herbivores on Yanakie Isthmus is facilitated by ecological separation. However, patterns of ecological separation, niche breadth and diet overlap in this guild did not conform well to body-size related predictions: the species with the narrowest and the broadest diet niches were intermediate in size, and the largest species consumed a greater proportion of dicots than did several smaller species. Interactions between intrinsic and extrinsic constraints on diet choice are likely to influence the diet of herbivores on Yanakie Isthmus. This study provides important preliminary insights into herbivore community niche dynamics on Yanakie Isthmus. High diet overlap and overlap in habitat use at some scales between some species, coupled with resource limitation is likely to result in inter-specific competition, particularly given indications of resource limitation through diet niche adjustments, broad niches and high diet overlap between the native and introduced herbivores in this community. Experimental manipulation is required to obtain a mechanistic understanding of species interactions and conclusively demonstrate competition.

iii

Declaration

This is to certify that i. the thesis comprises only my original work towards the PhD expect where indicated in the Preface; ii. due acknowledgement has been made in the text and all other material used; iii. the thesis is less than 100,000 words in length, exclusive of tables, maps, bibliographies and appendices.

______(Signature) (Date)

iv

v

Preface

The only data in this thesis that I did not collect myself were faecal pellet counts conducted by Parks Victoria staff after the ecological burn at Big Hummock between July 2004 and February 2005 (Chapter 4). Parks Victoria staff also contributed to this thesis by shooting animals used for the diet study (Chapter 5) and collecting stomach samples from animals on the sampling nights for which I was not present. Parks Victoria staff conducted ecological burn at Big Hummock that comprised the treatment for my study of the influence of fire on fine scale habitat use (Chapter 4). Ian Gordon, PhD Astat, Director of the Statistical Consulting Centre, University of Melbourne, provided advice on statistical analysis and in particular, undertook the Generalised Linear Modelling used to assess relationships between herbivore densities and vegetation attributes (Chapter 3) and made suggestions regarding analysis of pellet decay data (Chapter 2). As all chapters have been written as manuscripts for publication with Graeme Coulson and Dave Forsyth as co-authors, Graeme and Dave have provided editorial input on drafts and contributed to the ideas presented in each chapter. In addition, two anonymous referees provided constructive criticism that improved the methods and results sections of Chapter 5. These sections have been published as part of a paper that examines the diet data presented in this thesis from a management perspective:

Davis, N. E., Coulson, G. and Forsyth, D. M. 2008. Diets of native and introduced mammalian herbivores in shrub-encroached grassy woodland, south- eastern Australia. Wildlife Research, 7:684-694.

vi

vii

Acknowledgments

Thank you to my supervisors, Graeme Coulson and Dave Forsyth. Your patience, wisdom, time, and ongoing support and encouragement throughout a challenging few years have been greatly appreciated. This thesis has come together thanks to your guidance.

Many Parks Victoria staff offered support during this project, both practical, and more importantly, in the form of sharing of their extensive local knowledge of Yanakie Isthmus. Matt Hoskins, my project officer, supported me in all my field requests with a sense of humour (no matter how obscure he thought they were). During the early stages of this project, Elaine Thomas demonstrated a contagious enthusiasm for the restoration of Coastal Grassy Woodland on Yanakie Isthmus, and thus a passionate interest in my work – Elaine was always encouraging and never failed to deliver on requests of any nature, big or small! In the later stages of this project, Dan Jones, has been a major port of call within Parks Victoria, and like Elaine, has shown ongoing support and been keen to help out wherever possible. Jim Whelan’s great knowledge of Yanakie Isthmus has been invaluable. Thank you to John Wright, for making this project happen. Mick Keenan, Graeme Baxter, Scott Griggs, the late Stuart Judd and other Parks Victoria staff at all provided support in various ways.

Ian Gordon, PhD Astat, Director of the Statistical Consulting Centre, University of Melbourne, provided advice on statistical analysis, as did Jan Carey and Mick Keough, for which I am grateful. Ian Gordon also provided support and interest in my project above and beyond the call of duty: he was instrumental in convincing me that the depth and breadth of my project was worthy of conversion from MSc to PhD.

Ron Mayze (Para Park Co-Operative Game Reserve) and the late Geoff Moore (Australian Deer Research Foundation) generously shared their deer expertise. In particular, Ron volunteered many nights attempting to catch hog deer at the Prom with me, however, despite his extensive experience, we never succeeded! Nevertheless, Ron’s passion for hog deer and selfless giving of time and knowledge has taught me a lot. Thank you also to Para Park Game Cooperative for funding my trip to Sunday Island with Ron, and thanks to Para Park and Charles Franken (DSE) for lending deer traps to me.

Grant Norbury and Jac Cutter provided advice on diet analysis and David Meagher identified mosses. Sam Young, David Middleton and Michael Lynch provided veterinary expertise. Steve Elefteriadis, Institute of Land and Food Resources, made the greenhouse facilities available for my use. Graeme Baxter and the students of Foster High School constructed the herbivore exclosure plots.

This research was funded by Parks Victoria (Research Partners Program) and the Holsworth Wildlife Endowment. During my candidature, I also received financial support from the Alfred Nicholas and Drummond fellowships, the Australasian Wildlife Management Society and the Department of Zoology, The University of Melbourne.

viii

Although blissfully unaware, two friends have played pivotal roles in directing me towards this achievement. Emily Geraghty encouraged me to keep at it when I came to evaluate my progress at the one year mark and realised that nearly all aspects of my field work to date had failed! During a casual but decisive conversation at one of Graeme’s annual Australia Day BBQs, Julian Di Stefano encouraged me to convert my MSc to a PhD, which I subsequently did. I am grateful for the concern Julian showed. Office mates and members of the Vertecol Lab Group in the Department of Zoology, The University of Melbourne, have provided support, both academically and personally, as have many other friends. In particular, I thank Jenny Martin, Lisa Evans and Susan Campbell for taking such an interest in Leuwin – when I was frustrated at having to take time out from study, you guys reminded me that being a mum is a fantastic thing to be. Janine Hulston and Dawn Hayes have taken a genuine interest in my progress, for which I thank them. As I print this thesis, Leuwin plays happily with his friend Aiden. Thank you also to friends such as Amanda Honeybone and Shanti O’Brien for helping out in times of need! Thanks to all of you for your encouragement along the way.

Leuwin has spent many hours exploring with Popsi, and Nansi, Aunty Beth and Uncle Kev have also done more than their fair share of babysitting. This assistance has allowed me to keep chipping away at my thesis. Thank you, your continued willingness to help out has been vital in the completion of this thesis and greatly appreciated! On a more sentimental note, thank you to mum and dad for supporting me throughout my education. Thank you to dad for spending many hours patiently working through maths and science problems with me, and to mum, for telling me as a teenager that I could be whatever I wanted to be. Gillian Lueckenhausen imparted a love of learning and instilled in me a confidence in my ability to find the answers I sought. Jill Robbins has shown support from afar with amusing articles on feral deer, the gift of a laptop (which I never would have indulged in, but which has made all the difference), and ongoing faith in my ability.

This thesis is dedicated to my partner Jordan and to my son Leuwin. Their patience and understanding is appreciated more than words can say. I look forward to many joyful times together without the constant distraction of ‘the thesis’!

ix

Table of contents

Abstract…………………………………………………………………….. i Declaration…………………………………………………………………. iii Preface………………………………………………………………………v Acknowledgments…………………………………………………………..vii List of tables………………………………………………………………... xiii List of figures………………………………………………………………. xix

Chapter 1 General Introduction……………………………………………. 1 Inter-specific interactions…………………………………………...3 Competition, resource partitioning and niche differentiation……… 3 Habitat resource partitioning………………………………………..5 Food resource partitioning and its interplay with habitat use……… 6 Temporal resource partitioning…………………………………….. 6 Resource partitioning among native and introduced herbivore Communities……………………………………………………….. 7 Resource partitioning among herbivores in Australia………………8 The herbivore assemblage on Yanakie Isthmus…………………….9 The vegetation and management of Yanakie Isthmus……………... 13 Thesis aims………………………………………………………….15 Thesis outline………………………………………………………. 15

Chapter 2 Habitat-specific faecal pellet decay rates for five mammalian herbivores in south-eastern Australia……………………………………….21 Abstract…………………………………………………………………….. 23 Introduction………………………………………………………………… 25 Methods……………………………………………………………………. 26 Study area…………………………………………………………...26 Experimental design………………………………………………...28 Statistical analyses…………………………………………………. 29 Results……………………………………………………………………… 31 Site descriptions……………………………………………………. 31 Decay times………………………………………………………… 31 Influence of vegetation type and season on decay time and rate…... 32 Ageing faecal pellet groups………………………………………... 34 Discussion………………………………………………………………….. 35 Decay times………………………………………………………… 35 Influence of vegetation type and season on decay time and rate…... 37 Ageing faecal pellet groups………………………………………... 37 Applications………………………………………………………... 38 Conclusion…………………………………………………………. 40

Chapter 3 Broad scale habitat use by native and introduced mammalian herbivores on Yanakie Isthmus, south-eastern Australia…...... 53 Abstract...... 55 Introduction...... 57 Methods...... 62 Study site...... 63

x

Survey design...... 64 Faecal pellet counts...... 65 Habitat requirements…………………...... 66 Statistical analyses...... 66 Results...... 71 Habitat preference...... 71 Niche breadth and overlap...... 72 Herbivore density, population abundance and population metabolism…………………………………………………………..72 Relationships between herbivore faecal pellet density and vegetation type...... 73 Relationships between herbivore faecal pellet density and habitat requirements...... 74 Discussion...... 75 Broad scale habitat use: habitat preferences and requirements...... 76 Inter-specific overlap in habitat use………………………………... 82 Niche breadth………………………………………………………. 87 Herbivore density, population abundance and population metabolism…………………………………………………………. 88 Conclusion…………………………………………………………..90

Chapter 4 The influence of fire on fine scale habitat use by native and introduced mammalian herbivores in Coastal Grassy Woodland on Yanakie Isthmus, south- eastern Australia…………………...... 107 Abstract...... 109 Introduction...... 111 Methods...... 115 Study site...... 116 Study design...... 117 Ecological burning...... 118 Fine scale herbivore habitat use...... 118 Biomass accumulation...... 119 Floristic surveys...... 120 Statistical analyses...... 121 Results...... 125 Site comparability...... 125 Fine scale herbivore habitat use...... 126 Biomass accumulation...... 128 Floristic surveys...... 128 Discussion...... 131 Herbivore responses to ecological burning...... 131 Inter-specific differences in the use of vegetation strata...... 136 Conclusion…………………………………………………………..139

Chapter 5 Dietary niche relationships among sympatric native and introduced mammalian herbivores on Yanakie Isthmus, south-eastern Australia……... 163 Abstract…………………………………………………………………….. 165 Introduction………………………………………………………………… 167 Methods……………………………………………………………………. 172 Study area…………………………………………………………...172

xi

Collection and preparation of stomach samples…………………… 173 Microhistological analysis…………………………………………. 173 Statistical analyses…………………………………………………. 174 Results……………………………………………………………………… 176 Diets of the five herbivores………………………………………… 176 Interspecific variation in diet………………………………………. 177 Niche breadth and interspecific overlap in diet……………………. 178 Discussion………………………………………………………………….. 179 Herbivore diets……………………………………………………... 179 Competition, ecological separation and resource limitation ……... 179 Body size…………………………………………………………. 182 Inter-specific interactions: native versus introduced herbivore species……………………………………………………………… 183 Niche breadth: body size, resource limitation and competition…….186 Conclusion…………………………………………………………..188

Chapter 6 Conclusion...... 203 Food and habitat resource use……………………………………… 206 Niche adjustment…………………………………………………... 206 Resource use overlap………………………………………………..207 Species-specific implications of competition……………………….209 Resource partitioning………………………………………………. 211 The influence of body size on niche breadth and resource use overlap………………………………………………………………214 The influence of evolutionary history on niche breadth and resource use overlap…………………………………………………………..215 Future directions…………………………………………………….218

References...... 223

xii

xiii

List of tables

Table 2.1. Results of one-factor ANOVAs comparing % cover of each the ground, shrub and canopy layers, and litter depth between five vegetation types (n = 10 per type) on Greater Yanakie Isthmus…...... ….. 41 Table 2.2. Mean time to decay (days) and standard error for pellet groups of five herbivores (and for individual pellets for the three marsupial species) in five vegetation types (n = 60 pellet groups per vegetation type, except n = 55 pellet groups for the eastern grey kangaroo and the common wombat in Heath) on Greater Yanakie Isthmus……...... 42 Table 2.3. Results of two-factor repeated measures ANOVAs comparing the decay rate of pellet groups from each of five herbivores between trials in five vegetation types……………...... 43 Table 3.1. Mean density (number km-2) and population abundance estimates (with 95% confidence intervals) for five herbivores in five vegetation types on Greater Yanakie Isthmus (GYI): Coastal Grassy Woodland (CGW), Coastal Scrubs and Grasslands (CSG), Heath (H), Heathy Woodland (HW), and Moist Foothill Forest (MFF)...... 93 Table 3.2. Results of G-tests comparing the selection of vegetation types by each of five herbivore species (n = 5 vegetation types, except for the rabbit n = 4 and for the kangaroo n = 3) and results of pairwise chi-squared comparisons of standardised

ratio values for selection indices (Bi) for five vegetation types……. 95

Table 3.3. Standardised ratio selection index values (Bi) (with 95% confidence intervals, corrected for multiple comparisons using the Bonferroni correction) for herbivore selection of vegetation types………………………………………………………………... 96 Table 3.4. Values for Hulberts’s index of niche overlap (L) for proportional abundance of each of five herbivore species using five vegetation type resource states………………………………... 97

xiv

Table 3.5. Values for Smith’s measure of niche breadth (FT) (on a scale of 0-1) with upper and lower 95% confidence limits for the proportion of individuals of each of five herbivore species using five vegetation type resource states……………………… …. 97 Table 3.6. Results of a generalised linear model estimating the relationship between European rabbit faecal pellet group density and: (1) the vegetation type in which counts were conducted, (2) the season in which counts were conducted, (3) the distance of faecal pellet count plots to water, and (4) vegetation cover parameters measured within plots……………….. 98 Table 3.7. Results of a generalised linear model estimating the relationship between hog deer faecal pellet group density and: (1) the vegetation type in which counts were conducted, (2) the season in which counts were conducted, (3) the distance of faecal pellet count plots to water, and (4) vegetation cover parameters measured within plots…………………………………. 99 Table 3.8. Results of a generalised linear model estimating the relationship between eastern grey kangaroo faecal pellet density and: (1) the vegetation type in which counts were conducted, (2) the season in which counts were conducted, (3) the distance of faecal pellet count plots to water, and (4) vegetation cover parameters measured within plots……………….. 100 Table 3.9. Results of a generalised linear model estimating the relationship between swamp wallaby faecal pellet density and: (1) the vegetation type in which counts were conducted, (2) the season in which counts were conducted, (3) the distance of faecal pellet count plots to water, and (4) vegetation cover parameters measured within plots………………………………….. 101 Table 3.10. Results of a generalised linear model estimating the relationship between common wombat faecal pellet density and: (1) the vegetation type in which counts were conducted, (2) the season in which counts were conducted, (3) the distance of faecal pellet count plots to water, and (4) vegetation cover parameters measured within plots………………………………….. 102

xv

Table 4.1. Study design for herbivore faecal pellet counts, biomass sampling and floristic surveys at four sites on Greater Yanakie Isthmus (2003 – 2004). Units are the number of replicates per site per treatment…………………………………………………… 141 Table 4.2. Density of five herbivore species (with 95% confidence intervals) calculated from one pre-fire faecal pellet accumulation count (November 2003) at each of four sites and three post-fire counts (December 2004) at Big Hummock on Yanakie Isthmus………………………………...... …. 142 Table 4.3. Results of two-factor ANOVA comparing pre-fire faecal pellet (or pellet group) standing crop for five herbivore species, summed over two surveys (October and November 2003), between two vegetation strata (slashed swale vs. scrub/dune woodland) and among four sites on Yanakie Isthmus…………...... 143 Table 4.4. Results of two-factor ANOVA comparing the mean % cover for vegetation structure variables among sites and between vegetation strata (slashed swale vs. scrub/dune woodland) during the pre-fire survey in the Yanakie Isthmus…….. 144 Table 4.5. Mean % cover and standard error for vegetation structure variables in two vegetation strata (slashed swale and scrub/dune woodland) at four sites (n = 40 per site) during the pre-fire survey on Yanakie Isthmus………………………...... 145 Table 4.6. Results of three-factor ANOVA comparing pre-fire load of two states (live vs. dead) of fine fuel between vegetation strata (slashed swale vs. scrub/dune woodland) and among four sites on Yanakie Isthmus…………………………………...... …. 146 Table 4.7. Results of two-tailed t-tests comparing mean sapling and tree height, mean branch length, mean number of and mean number of leaders (all square-root transformed) between two sites on Yanakie Isthmus………………………...... 146 Table 4.8. Results of two-factor repeated measures ANOVA comparing daily pellet (or pellet group) accumulation rate for five herbivore species between pre-fire (October and November 2003) and post-fire (July, September, November 2003, and

xvi

February 2004) surveys in two vegetation strata (slashed swale vs. scrub/dune woodland) at Big Hummock on Yanakie Isthmus……………………………………………………………... 147 Table 4.9. Results of two-factor repeated measures ANOVA

comparing fine fuel load (log10-transformed) of two states (dead vs. live) in two vegetation strata (slashed swale vs. scrub/dune woodland) between the pre-fire period at four sites and the post-fire period at Big Hummock on Yanakie Isthmus…….148 Table 4.10. Results of three-factor ANOVA examining the effect of grazing (exclosure vs. control plots) and fire (burnt vs. unburnt patches) on fine fuel load in two vegetation strata (slashed swale vs. scrub/dune woodland) during the post-fire sampling period at Big Hummock on Yanakie Isthmus…………………...... 148 Table 4.11. Results of two-tailed t-tests comparing the mean % cover for vegetation structure variables between pre-fire (November 2003) and post-fire (December 2004) surveys at Big Hummock on Yanakie Isthmus……………………………………...... 149 Table 4.12. Results of two-factor ANOVA comparing the mean % cover of vegetation structure variables between burnt and unburnt patches in two vegetation strata (slashed swale vs. scrub/dune woodland) during the post-fire survey at Big Hummock on Yanakie Isthmus………………………………… …. 150 Table 5.1. Summary of gut samples (n = 93) collected from male and female herbivores on Yanakie Isthmus in five vegetation types over three seasons between 7 June 2004 and 11 February 2005…... 191 Table 5.2. Frequency of occurrence of species in gut samples from five herbivores……………………………...... 192 Table 5.3. Bootstrapped mean and upper and lower 95% confidence intervals for proportion of plant epidermal fragments identified within Structural, Broad Taxonomic, Functional Group and Plant Origin categories per gut sample for five herbivore species……………………………………………………………… 196 Table 5.4. Results of one-factor analysis of variance comparing the diets of five herbivore species……………………………………... 197

xvii

Table 5.5. Results of pair-wise comparisons using Tukey’s post-hoc analysis for one-factor analysis of variance (Table 2.1) comparing the diets of five herbivore species………………………198

Table 5.6. Values for Horn’s index of niche overlap (Ro) between five herbivore species…………………………………… ……………... 200

xviii

xix

List of figures

Figure 1.1. Location of the study area, Yanakie Isthmus, at Wilsons Promontory National Park, Victoria, Australia……………………. 17 Figure 1.2. Location of Wilsons Promontory National Park, Victoria, Australia……………………………………………………………. 19 Figure 2.1. Mean monthly rainfall, and mean monthly maximum and minimum temperatures averaged across months within each season between summer 2004-05 and summer 2006-07 on Yanakie Isthmus…………………………………………………….45 Figure 2.2. Pellet group decay trial design: one pellet group from each of the five species was deposited at equally-spaced compass bearings of 70o, in each of three concentric rings 30 cm apart (one ring per monthly trial) around a central peg………………….. 45 Figure 2.3. Comparison of bootstrapped and standard 95% upper and lower confidence intervals for the mean number of hog deer faecal pellets surviving over time (days) from pellet groups (n = 15) deposited in Coastal Grassy Woodland on Greater Yanakie Isthmus…………………………………………………………. …. 46 Figure 2.4. Mean ( standard error) % cover of the ground and shrub layers in five vegetation types (n = 10 per type) on Greater Yanakie Isthmus………………...... 46 Figure 2.5. Changes in the mean (with lower and upper 95% confidence intervals) number of pellets surviving over time (days) for hog deer pellet groups in five vegetation types on Greater Yanakie Isthmus……………………………………………47 Figure 2.6. Changes in the mean (with lower and upper 95% confidence intervals) number of pellets surviving over time (days) for European rabbit pellet groups in five vegetation types on Greater Yanakie Isthmus…………………………………. 48 Figure 2.7. Changes in the mean (with lower and upper 95% confidence intervals) number of pellets surviving over time (days) for eastern grey kangaroo pellet groups in five

xx

vegetation types on Greater Yanakie Isthmus…………….……….. 49 Figure 2.8. Changes in the mean (with lower and upper 95% confidence intervals) number of pellets surviving over time (days) for swamp wallaby pellet groups in five vegetation types on Greater Yanakie Isthmus……………………………………….. 50 Figure 2.9. Changes in the mean ( standard error) number of pellets surviving over time (days) for common wombat pellet groups in five vegetation types on Greater Yanakie Isthmus………………… 51 Figure 2.10. Mean % of pellet groups classified as fresh, medium or old at each time interval across monthly trials from February to August 2005 at five sites in the Greater Yanakie Isthmus…………. 52 Figure 3.1. Uniformity of habitat use (Smith’s measure of niche breadth: FT) versus body mass (M) for five herbivore species (hog deer, European rabbit, swamp wallaby, eastern grey kangaroo and common wombat) using five vegetation type resource states: Coastal Grassy Woodland, Coastal Shrubs and Grasslands, Heath, Heathy Woodland, and Moist Foothill Forest. Samples (n = 110) were collected from GYI, Wilsons Promontory National Park during autumn and winter 2004. The x-axis is logarithmic.....…………………………………………….. 103 Figure 3.2. Mean ( standard error) density of large mammalian herbivores per km2 calculated from faecal pellet surveys conducted in summer, autumn and winter 2004 in five vegetation types on Greater Yanakie Isthmus……………………... 104 Figure 3.3. Population abundance, biomass and basal metabolic rate for five herbivore species (i) over the area of Greater Yanakie Isthmus, and (ii) within Coastal Grassy Woodland on Greater Yanakie Isthmus…………………………………………………….105 Figure 4.1. Treatment (burnt) and control (unburnt) sites used for a trial of ecological burning at Big Hummock, Yanakie Isthmus…… 151 Figure 4.2. Mean (± standard error) pellet standing crop for five herbivore species over two pre-fire surveys (October and

xxi

November 2003) at four sites (n = 40 per site) on Yanakie Isthmus……………………………………………………………... 153 Figure 4.3. Non-metric multi-dimensional scaling two and three- dimensional configurations of ground layer plant species composition on Yanakie Isthmus………………………………...... 154 Figure 4.4. Non-metric multi-dimensional scaling three-dimensional configuration (each axis displayed in two dimensions) of shrub and canopy layer plant species composition on Yanakie Isthmus…………………………………………………………. …. 155 Figure 4.5. Mean (± standard error) sapling and tree height, mean branch length, mean number of leaves and mean number of leaders at two sites (n = 10 per site) on Yanakie Isthmus………….. 156 Figure 4.6. Population abundance, biomass and basal metabolic demand at the population level for five herbivore species (i) pre-fire at Big Hummock, and (ii) post-fire at Big Hummock on Greater Yanakie Isthmus. Population estimates are based on faecal pellet counts conducted between 2003-2005 and body mass values from the literature…………………………………….. 157 Figure 4.7. Mean (± standard error) pellet (or pellet group) accumulation rate for five herbivore species over four post-fire surveys (July, September, November 2003 and February 2004) at Big Hummock (n = 40 per survey) on Yanakie Isthmus………... 158 Figure 4.8. Mean (± standard error) fine fuel load (g m-2) of two vegetation states (live and dead) in two vegetation strata (slashed swale vs. scrub/dune woodland) during the pre-fire period (November 2003) at four sites (n = 40), and during the post-fire period (December 2004) at Big Hummock (n = 20) on Yanakie Isthmus…………………………………………………….159 Figure 4.9. Number of species recorded at Big Hummock on Yanakie Isthmus, Wilsons Promontory National Park during (i) the pre- fire (November 2003, n = 40) and post-fire (December 2004, n = 20) surveys, and (ii) in burnt (n = 10) and unburnt (n = 10) patches during the post-fire survey………………………………… 160 Figure 4.10. Mean (± standard error) % cover during pre-fire

xxii

(November 2003) and post-fire (December 2004) surveys at Big Hummock on Yanakie Isthmus…………………………………...... 161 Figure 4.11. Mean (± standard error) % cover of vegetation structure variables at Big Hummock on Yanakie Isthmus……………………162 Figure 5.1. Mean (with 95% confidence intervals) proportion of plant fragments identified in stomachs of individuals of five herbivore species (n = 20 for all species, except swamp wallaby n = 13): (i) broad taxonomic group; (ii) functional group; (iii) structural group; (iv) plant part; and (v) plant origin………………. 201 Figure 5.2. Non-metric multi-dimensional scaling three-dimensional configuration (each axis displayed in two dimensions) of individuals of five herbivore species (n = 20 for all species, except swamp wallaby n = 13) based on a Bray-Curtis matrix of dissimilarities between the number of fragments identified in stomach samples…………………………...... 202 Figure 6.1. Spatial and trophic resource use overlap among five herbivore species on GYI. High, medium and low overlap in broad-scale habitat use indicate Hulberts’ index of niche overlap values > 1.3, > 0.8 to < 1.3, and < 0.8, respectively (based on faecal pellet counts). High and low overlap in fine-scale habitat use indicate pairs of species for which faecal pellet counts are greater in the same vegetation strata, and pairs of species for which pellet counts are greater in contrasting vegetation strata, respectively, and medium overlap indicates pairs of species for which neither the high nor low overlap definition applied (i.e., faecal pellet counts were not greater in the same strata, nor were they greater in opposite strata). High, medium and low overlap in diet indicate Hulberts’ index of niche overlap values of > 0.8, > 0.3 to < 0.8, and < 0.3, respectively (based on microhistological diet analysis). * indicates introduced species…... 221

1

Chapter 1

General introduction ______

Chapter 1 Introduction 2

Chapter 1 Introduction 3

Chapter 1 General introduction

Inter-specific interactions Inter-specific interactions among sympatric species are much debated issues in theoretical ecology and practical wildlife management (Wegge et al. 2006). In particular, competitive interactions are pervasive in ecological systems (Schoener 1983) and are believed to be a central biotic factor structuring herbivore communities (Sinclair and Norton-Griffiths 1982, Schoener 1989). Investigation of competitive interactions, and the ways in which species coexist, is essential to improve our understanding of the ecological principles underlying ecosystem functioning (Duncan et al. 1998). Extensive work has been done to further our knowledge of inter-specific interactions in communities of native mammalian herbivores (e.g., Bell 1971). However, communities globally are increasingly composed of native and introduced mammalian herbivores (e.g., Voeten and Prins 1999, Baldi et al. 2004, Madhusudan 2004), adding a new dimension to this well-established area of study.

Competition, resource partitioning and niche differentiation A species’ niche is the limits, for all important environmental features, within which individuals can survive, grow and reproduce (Futuyma and Moreno 1988). A species ‘realised’ niche (i.e., resource use patterns) in a particular environment may be more restricted than its ‘fundamental’ niche (Hutchinson 1957) due to local biotic constraints, including predation, parasitism (Hutchinson 1957, Futuyma and Moreno 1988), resource availability (Levins 1968) and inter- specific competition (Connell 1983, Schoener 1983). Gause’s (1934) competitive exclusion principle states that when resource competition occurs between species with the same ecological niche, one species will out-compete the other, resulting in its exclusion: complete competitors cannot coexist (Hardin 1960). A resource is considered to be any environmental component to which an organism can gain exclusive access for some period of time, and that by its use increases the organisms fitness (MacNally 1995). Pioneering experiments by Gause (1934) and Park (1962) verified the potential

Chapter 1 Introduction 4 consequences of inter-specific competition for species persistence if resources are limiting, and demonstrated that for inter-specific resource competition to occur there must be niche overlap between species in their need for the resource (Hutchinson 1957). Hence, Gause (1934) suggested that sympatric species should avoid or reduce competition by selecting different resources (i.e., adjusting their niche widths in response to co-inhabitants; Chase and Leibold 2003). The use of multiple resources allows for resource partitioning among sympatric populations of ecologically similar species (Schoener 1974b, Whitfield 2002). Specialisation on resources along habitat, diet and temporal gradients results in niche differentiation, which is seen as the evolutionary outcome of competition: each species becomes adapted to exploit a unique niche, thus facilitating coexistence (Schoener 1974b, Whitfield 2002, Schmidt et al. 2010). Low levels of niche- differentiation relate to a higher competitive coefficient, with consequent effects on population dynamics of the interacting species (May 1973), for example, changes in population size, fecundity, age structure health and fitness (Petren and Case 1996). In a community, niche differentiation occurs along several dimensions (Schoener 1983). A species’ niche is conceptualised mathematically as an n- dimensional hypervolume, with each dimension representing a key environmental variable or resource (Hutchinson 1959). The basic tenet of competitive exclusion is that n number of species cannot coexist on fewer than n resources (Gause 1934, Hutchinson 1959), thus resource availability constrains species richness in ecological communities (MacArthur 1972, Tonn et al. 1990) and only those species that show trade-offs in niche utilisation in response to competition can coexist (Chase and Leibold 2003). Under the Hutchinsonian premise of n- dimensional hypervolume, niche differentiation is generally complementary; when species are similar on one niche dimension, they differ on another (Pianka 1976, Dunbar 1978, Fox 1989, Bagchi et al. 2003). In reality, competing species often differ in multiple niche dimensions (e.g., Pianka 1974, le Mar and McArthur 2005). Habitat is the most common dimension partitioned, followed by food resources (Schoener 1983). Temporal partitioning becomes important in environments where resources are rapidly renewed (Kotler et al. 1993). Sympatric herbivores tend to exploit their environments in different ways (Schwartz and Ellis 1981, Forsyth 2000) based on differences in feeding strategies

Chapter 1 Introduction 5

(Gwynne and Bell 1968, Schwartz and Ellis 1981) and/or differences in habitat preference (Batcheler 1960, Taylor 1983, Fox 1989). Understanding resource use overlap in terms of habitat and diet is critical to understanding competitive interactions (Krebs 1998, Wegge et al. 2006), and ultimately, the divergent exploitation patterns which determine community structure (Bell 1971, Hofmann and Stewart 1972, Jarman 1974).

Habitat resource partitioning Habitat partitioning is a major means of reducing inter-specific competition, by decreasing the spatial overlap of two populations (Schmidt et al. 2010). Habitat partitioning generally occurs before species evolve morphological adaptations to partition food (Pianka 1976). For example, although the large ungulate species in Kanha Park, India, occupy essentially the same environment, each species is adapted to certain habitat conditions which ecologically separate it, to some extent, from the other (Schaller 1967). Similarly, African studies have demonstrated habitat segregation between herbivore species (Lamprey 1963, Jarman 1972). Inter-specific competition tends to limit habitat use (Svarsden 1949) by forcing subordinate species into suboptimal habitat (Schoener 1983, Sih 1993). In turn, this has the potential to influence survival and reproduction of individuals of subordinate species (Sawyer et al. 2006). Resource utilisation can occur over multiple spatial hierarchies, each of which relate to the requirements of the individuals (Johnson 1980). Thus habitat selection at several scales can facilitate coexistence. Johnson (1980) describes three levels of habitat use: First order, which encompasses the broad distribution of the species; second order, which consists of the individuals home range; and third order, which refers to use of habitat components within the individuals home range. That is, while habitat partitioning can occur at a broad scale, broad habitat types are generally comprised of fine-scale mosaics within which individuals and species make diverse fine-scale choices between microhabitats over time (Clarke et al. 1989, Schmidt et al. 2010). In some cases fine-scale variation in habitat structure may be more important than broad habitat type (Schmidt et al. 2010). Microhabitat, aspects of habitat such as lateral and overhead cover that operate on a fine scale, can be a major dimension of habitat partitioning (Schmidt et al. 2010). Separation of habitat use at the broad habitat level may be more clearly

Chapter 1 Introduction 6 illustrated at the level of microhabitat (Schmidt et al. 2010). Moreover, species that do not separate at the home range scale may show distinct separation at the microhabitat scale (e.g., Scognamillo et al. 2003).

Food resource partitioning and its interplay with habitat use Food resource use is one of the most important components of the niche (Krebs 1998). Factors including mouth morphology, gut morphology and function, and body size result in differences in diet and foraging behaviour, and grouping patterns (Schwartz and Ellis 1981). Ecological separation may be achieved by feeding in different ways (Schwartz and Ellis 1981), but as noted by Child and Von Richter (1969) it is also commonly achieved by utilising different feeding habitats (e.g., ungulates in Tanzania; Voeten and Prins 1999). Further, in some cases ecological separation is achieved through a combination of these strategies (Krebs 1998). The strong interplay between habitat and food resource use means it can be difficult to distinguish between food and habitat partitioning: habitat preferences of large herbivores are often strongly associated with food resource abundance (Duncan 1983, Murden and Risenhoover 1993) and competition-driven divergence in habitat use can result in divergence in diet (niche compression hypothesis: MacArthur and Wilson 1967). Thus it is the sum of habitat patch and food choices which determine coexistence (Pyke et al. 1977). In particular, body-size related constraints on food intake rate commonly explain ecological phenomena such as grazing succession, facilitation and segregation of habitat use (Illius and Gordon 1993).

Temporal resource partitioning Habitat and food resources can be partitioned through time over the diel period, or seasonally (Kotler and Brown 1988). Hypotheses for temporal partitioning suggest species are active during the times when they are the most efficient foragers; predation and competition affect assessments of efficiency (Jones et al. 2001). Temporal partitioning throughout a diel period may allow competitors to avoid interference and resource competition, and has been demonstrated in a variety of taxa (e.g., Jones et al. 2001, Borgnia et al. 2007, Rouag et al. 2007). Competing species may also partition resources seasonally, as

Chapter 1 Introduction 7 has been demonstrated with respect to diet and/or habitat selection (e.g., Jones et al. 2001, Zapata et al. 2005, Schleuter and Eckmann 2007).

Resource partitioning among native and introduced herbivore communities Resource partitioning and niche differentiation are seen as evolutionary outcomes of competition, and have been well described among assemblages of co- evolved native herbivores (e.g., Bell 1970, Jarman and Sinclair 1979, McNaughton and Georgiadis 1986, Green 1987, Bodmer 1991). For example, native herbivore species in Nepal exhibit habitat partitioning associated with seasonal changes in food resource availability: a high degree of spatial and food resource overlap occurs during the monsoonal growing season, but in the resource-limited dry season, the majority of species move into lower quality habitats, associated with their abilities to utilise poor quality food resources (Wegge et al. 2006). More recently, there has been speculation on the effect of introducing exotic herbivore species into native herbivore assemblages (Bagchi et al. 2004, Madhusudan 2004, Mishra et al. 2004). Resource competition has been demonstrated between native and introduced herbivores (e.g., Fritz et al. 1996, Madhusudan 2004, Kuiters et al. 2005), in some cases leading to competitive exclusion (e.g., Prins 2000, Mishra et al. 2002). Studies such as that by Schwartz and Ellis (1981) have highlighted relationships between recent evolutionary history and the degree of resource use overlap between native and introduced species. Subsequently, it has been asserted that species with independent evolutionary histories have inherently less resource partitioning to facilitate coexistence than species with common evolutionary histories, as they have not evolved mechanisms of resource partitioning (Kirchhoff and Larsen 1998, Kelley et al. 2002, Madhusudan 2004). Therefore, it has been suggested that for an introduced species to become established, it must find a suitable niche that is unoccupied, or out-compete native species (Ovington 1978). However, the introduction of new species into an assemblage will not always result in competitive exclusion, even where niche overlap occurs. Namgail et al. (2009) discuss differences in the potential responses of species niche widths, with respect to food and habitat, to the introduction of species into an assemblage. Given that animals in multi-species assemblages may

Chapter 1 Introduction 8 use only a subset of available resources due to the presence of sympatric species sharing resources (Hutchinson 1957), species can be packed into assemblages as a result of either increasing the resource range, or narrowing the niche width of constituent species (MacArthur 1972). When resources are scarce, niche adjustment is expected to be the predominant way of accommodating additional species (Namgail et al. 2009). Thus, when a species becomes extinct from a community, niche space (habitat supplying resources for a species’ survival) is expected to become vacant and can be either occupied by an invading species or exploited by extant species, leading to readjustment (i.e., increase) in their niche widths (Diamond 1975, Ricklefs and Schluter 1993). However, such theoretical predictions are not always clear-cut or consistent. For example, in the case of diet, loss of a potentially competing sympatric species may have an opposite effect on niche width, as a species may narrow its niche by including fewer (but more nutritious) plant species in its diet in the absence of competitors (Namgail et al. 2009). Inversely, species may widen their diet spectrum when a new species invades the assemblage, as they may need to include less nutritious in their diet due to resource constraints imposed by the invading species (Namgail et al. 2009). Species responses to inter-specific competition are further complicated by resource availability. Pianka’s (1974) theory of maximal tolerable niche overlap attempts to predict the degree of niche overlap between competing species under situations of varying resource abundance. This theory predicts that there should be less niche overlap in resource limited situations as compared to ones with higher resource abundance (Pianka 1974). That is, if resources are not limiting, species can share them, leading to overlapping niches. In this situation, extensive overlap may indicate reduced competition. However, if resources are limiting, highly segregated niches may indicate a response to strong competition (Pianka 1974). Such diverse responses highlight the potential complexities of inter-specific relationships within communities of sympatric native and introduced species.

Resource partitioning among herbivores in Australia As in communities of sympatric herbivores globally, studies of Australian herbivores have demonstrated ecological separation based on habitat partitioning (e.g., between wallaroos Macropus robustus and eastern grey kangaroos

Chapter 1 Introduction 9

Macropus giganteus; Taylor 1983, and between eastern grey kangaroos and western grey kangaroos Macropus fuliginosus; Coulson 1990, McCullough and McCullough 2000), and diet partitioning (e.g., between northern hairy-nosed wombats Lasiorhinus krefftii and eastern grey kangaroos; Woolnough and Johnson 2000). Research has also demonstrated resource partitioning among native herbivores with respect to both habitat and diet (e.g., between eastern grey kangaroos and swamp Wallabia bicolor; de Munk 1999, and within other macropod communities; Jarman and Phillips 1989), and interactions between diet and habitat partitioning (e.g., between eastern grey kangaroos and wallaroos; Taylor 1983). Over the last two centuries, a wide variety of introduced mammalian herbivores have established populations in Australia (Forsyth et al. 2004) and many of these species are now sympatric with native herbivores (e.g., Dawson and Ellis 1979, Dierenfeld 1984, Duncan 1992, Dawson and Ellis 1996). Studies have begun to investigate inter-specific interactions between native and introduced herbivores in Australian communities. For example, Moriarty (2004) found that the introduced rusa deer Cervus timorensis competes with native herbivores for resources and is thought to be limiting recruitment in the swamp wallaby population in the Royal National Park. Similarly, work by Cooke (1998) implicates introduced rabbits Oryctolagus cuniculus in the decline of native common wombats Vombatus ursinus in . However, few studies have examined resource partitioning among multi-species assemblages, particularly those comprised of both native and introduced herbivore species.

The herbivore assemblage on Yanakie Isthmus On Yanakie Isthmus (Wilsons Promontory National Park, Victoria), introduced European rabbits Oryctolagus cuniculus and hog deer Axis porcinus occur in sympatry with three terrestrial native herbivores, the eastern grey kangaroo, swamp wallaby and common wombat. The ecology of these species has been studied in other parts of their Australian range (e.g., Taylor 1971, Coulson 1993, de Munk 1999, Moseby et al. 2005, Roger et al. 2007), as well as in the native range of the two introduced species (e.g., Mishra and Wemmer 1987, Fa et al. 1999, Calvete et al. 2004, Odden and Wegge 2007). On Yanakie Isthmus, Davis et al. (2008) used diet analysis to infer the potential impacts of these species

Chapter 1 Introduction 10 on Coastal Grassy Woodland vegetation. However, inter-specific relationships within this herbivore guild have not been studied. The hog deer Axis porcinus (c. 40 kg; Mayze and Moore 1990), native to south and south-east Asia (Biwas 2000), was introduced to Australia in 1861 primarily as a game animal (Mayze and Moore 1990). It is now established throughout much of the low-lying shrub-lands of Victoria’s plain (Menkhorst 1995a). In their native range, hog deer prefer floodplain grassland associations, which provide both food and shelter (Odden et al. 2005). In Victoria, hog deer are essentially limited to flat, swampy coastal plains (Taylor 1971). Hog deer are primarily grazers in their native range (Wegge et al. 2006) and in Victoria (Taylor 1971). Hog deer feeding patterns vary according to habitat, weather, season and the degree to which the population is disturbed (Mayze and Moore 1990). They may feed intermittently throughout the day if adequate shelter is available, and rest mainly during the middle of the night, but if shelter is limited they will feed mainly at night, with peaks just before dawn, just after dark and in the middle of the day (Mayze and Moore 1990). Hog deer are generally solitary (Roberts 1977), although they may congregate in family groups (Rowntree 1935). Hog deer are considered to be sedentary (Dhungel and O'Gara 1991): Taylor (1971) suggested that in Victoria they occupy a core home range of up to 20 ha, while Dhungel (1985b) suggested that in Nepal the mean hog deer home range for females is 60 ha and for males is 80 ha. European rabbits (c. 1.6 kg; Strahan 1995), native to Spain and southern France, are one of the most widely distributed and abundant mammals in Australia, occurring throughout the continent except in the northern most areas (Williams et al. 1995). European rabbits were first recorded at Yanakie in 1917 (Chesterfield et al. 1995). In their native range European rabbits favour open scrubland (Fa et al. 1999). Similarly, in sub-alpine regions of Australia rabbits occur in woodland associated with open grassy valleys (Williams and Myers 2008). Rabbits appear to prefer intermediate levels of cover (Carvalho and Gomes 2004) when predation pressure is low (Palomares and Delibes 1997) or when dense shrub cover reduces ground layer cover (e.g., Bennett 1994) or hinders movement (Fa et al. 1999), although interspersion of open and closed areas may be more important in habitat selection than the level of shrub cover (Carvalho and Gomes 2004, Fernández 2005). European rabbits graze selectively on forbs and

Chapter 1 Introduction 11 grasses (Leigh et al. 1991, Martin et al. 2007). Largely crepuscular, European rabbits are most active around dawn and dusk, but are also highly active throughout the night (Villafuerte et al. 1993). Rabbits are gregarious, social animals which occur in discrete breeding groups consisting of individuals who share access to underground refuges (Cowan 1987). In arid South Australia, Moseby et al. (2005) estimated home ranges for the European rabbit of 2.1 ha in summer and 4.2 ha in winter, while in the Mediterranean climatic zone of Australia Stott (2003) estimated rabbit home ranges of c.10 ha. The swamp wallaby Wallabia bicolor (c. 18 kg; Edwards 1969) has a broad distribution across eastern Australia from Cape York in Queensland to south-western Victoria (Merchant 2008). The swamp wallaby is capable of utilising a variety of environments (Hollis et al. 1986, Troy et al. 1992), but generally lives in thick undergrowth of forests, woodlands and heath (Merchant 2008). Habitat selection is heavily influenced by the availability of dense cover (Di Stefano et al. 2007, Merchant 2008, Schmidt et al. 2010) which provides both food and cover (e.g., Floyd 1980, Lunney and O'Connell 1988). However, habitat selection changes throughout the diel period (Di Stefano et al. 2009) and swamp wallabies commonly forage along ecotones between forest and grassland (Hume 1999), and will move into more open areas to feed during the night (Edwards and Ealey 1975). Swamp wallabies are solitary, but may aggregate for feeding (Croft 1989, Jarman and Coulson 1989) and individuals tend to occupy small, overlapping home ranges of c. 6 ha (Troy and Coulson 1993). Swamp wallabies have been classified as browsers on the basis of diet (Claridge 2001, Davis et al. 2008, Di Stefano and Newell 2008) and dentition (Sanson 1980), but will include a range of forage in their diets including grass (Norbury et al. 1989), herbs, woody perennials, ferns , tree seedlings, saplings and fungi (Edwards and Ealey 1975, Waters 1985, Hollis et al. 1986). Eastern grey kangaroos (c. 26 kg: ACT, n = 333, G. Coulson, The University of Melbourne, pers. comm.) are limited to the east of Australia, but are widespread, their range extending over Queensland, , Victoria and (Caughley et al. 1987). They occupy a variety of habitats including sclerophyll forest, shrubland and heathland (Coulson 2008), but are largely absent from wet forests (Hill 1981b). Eastern grey kangaroos require habitats with a mix of open grassy areas for foraging and more densely vegetated areas offering

Chapter 1 Introduction 12 lateral cover, for example, open woodland and forest/pasture ecotones (McCullough and McCullough 2000, Moore et al. 2002, Schmidt et al. 2010). The time eastern grey kangaroos spend foraging in a particular habitat is correlated to availability of their most preferred food category, high-protein grass (Taylor 1984). Eastern grey kangaroos are gregarious (Southwell 1984, Jarman and Coulson 1989). Home ranges often overlap (Coulson 1993), with a mean size of 317 ha (Norbury et al. 1989). Eastern grey kangaroos show peaks in foraging activity at dawn and dusk, when movement into open areas occurs, and they may also forage at low intensity throughout the night (Southwell 1987), generally resting in more sheltered habitats with dense vegetation during the day (Clarke et al. 1989). However, seasonal variations in these patterns occurs, with less time devoted to grazing in summer and more time resting in sheltered positions (Southwell 1987), while in winter they may forage periodically throughout the day (Clarke et al. 1989, Coulson 1993). Sanson (1978) classified eastern grey kangaroos as grazer on the basis of dentition and accordingly, their diet is dominated by high-protein, low-fibre grasses (Taylor 1985a, McCullough and McCullough 2000). The common wombat (c. 28 kg; Barboza et al. 1993) occurs in coastal regions of southeast Australia, including the southeast tip of Queensland, the eastern region of New South Wales, the eastern half and southern areas of Victoria, and the southeastern tip of South Australia (McIlroy 1973, Mallett and Cooke 1986). Common wombats utilise several vegetation types including sclerophyll forest, woodland, coastal scrub, heathland and temperate forest (McIlroy 2008), but they preferentially forage in habitats with abundant high quality grass (Evans et al. 2006) and habitat preferences may reflect suitability for burrowing (McIlroy 1995). Common wombats are nocturnal, generally emerging to feed at night, although in winter, they may emerge in the early evening before temperatures have dropped (Wells 1989). Common wombats have home ranges of c. 14 ha (McIlroy 1973). Home ranges of individuals often overlap and multiple burrow use is common, although semi-exclusive feeding areas occur within overlapping home ranges, suggesting some territoriality (McIlroy 1973). Common wombats are classified as grazers (Rishworth et al. 1995, Hume 1999). European rabbits compete with native herbivores for food (Dawson and Ellis 1979) and other resources (Martin and Sobey 1983) and have been

Chapter 1 Introduction 13 implicated in the decline of native mammal populations (e.g., Cooke 1998), although interactions between rabbits and other herbivores on Yanakie Isthmus have not been investigated. The occurrence of sympatric populations of hog deer and native herbivores has been used to infer a lack of competition (Frith 1979), and Taylor (1971) suggested that in Victoria, hog deer occupy a previously vacant niche. However, the existence of empty niches is unlikely (Williamson 1996). Taylor’s (1971) investigation of the ecological relationships between hog deer and native fauna suggested that although little competition occurred between hog deer, native eastern grey kangaroos and swamp wallabies, under conditions of increasing population densities, there is potential for competition between these species. Further, recent research suggests that like rabbits, introduced deer can compete with native species for food and other resources (NPWS 2001, Moriarty 2004). Studies on niche-related resource partitioning have been conducted on several taxa (e.g., Wheeler and Calver 1996, Toda et al. 1999, McDonald 2002), however, niche relationships are not as well understood in large herbivores, largely because of their low population densities and difficulties associated with manipulating their populations (Namgail et al. 2009). The herbivore guild on Yanakie Isthmus is complex and includes both introduced and native species, providing an opportunity to examine questions of ecological interest regarding community niche dynamics (Linklater et al. 2000).

The vegetation and management of Yanakie Isthmus Yanakie Isthmus (6874-ha) extends from the northern boundary of Wilsons Promontory National Park to and is bounded by coastline to the west and to the north-east, and by heathland to the south-east. Spatial heterogeneity is crucial for habitat selection to facilitate coexistence (Kotler and Brown 1988). Yanakie Isthmus has five distinct vegetation communities, and therefore provides a good basis for studies of broad scale habitat use. Davies and Oates (1999) describe the five broad vegetation types on Yanakie Isthmus as follows. Coastal Grassy Woodland generally consists of Eucalyptus pryoriana, , Allocasuarina littoralis and Allocasuarina verticillata, with occasional Acacia mearnsii shrubs in the understorey and a ground layer of Pteridium esculentum, grasses, sedges and herbs. Coastal Scrubs and Grasslands is dominated by grasses, sedges and salt-

Chapter 1 Introduction 14 tolerant herbs and shrubs. Heath has occasional trees (mainly Eucalyptus spp.) but is dominated by small ericoid-leaved shrubs and sedges. Heathy Woodland is dominated by scattered Eucalyptus nitida, heathy shrubs (predominantly Proteaceae and Epacridaceae) and restionaceous sedges. Moist Foothill Forest is a medium to tall forest of Eucalyptus species with a substantial shrub-layer, understorey trees and a herb-rich or ferny understorey. Given the potential for variation in habitat use at multiple levels (Johnson 1980, Clarke et al. 1989), it is essential to examine habitat use at broad scales as well as at a finer scale (Schmidt et al. 2010), as advocated by Southwell et al. (1999). The dominant vegetation community on Yanakie Isthmus, Coastal Grassy Woodland, has undergone dramatic changes in structure and composition (Bennett 1994). Originally consisting of open grassy woodland, it is now dominated by dense stands of the encroaching native shrubs Leptospermum laevigatum and Acacia longifolia var. sophorae, which suppress the regeneration of other native species (Bennett 1994, Costello et al. 2000). This process, and its subsequent management using manual vegetation slashing, provides a mosaic of microhabitats useful for studying fine-scale habitat use. In addition, Yanakie Isthmus is a focus area for management within Wilsons Promontory National Park due to its environmental values (Parks Victoria 2003b). Active involvement of land managers on Yanakie Isthmus provides access to important resources to aid in the study of herbivore niche partitioning in this system, for example, access to gut samples which enable accurate and comprehensive analysis of the diets of each of the terrestrial herbivores, and the ability to manipulate habitat using a natural ecological process (fire). Moreover, low predation rates on Yanakie Isthmus following the removal of native predators and control of red foxes Vulpes vulpes (Menkhorst and Seebeck 1998, Parks Victoria 2003a) mean that it is an ideal location for studies of competition: in the presence of predators, herbivore populations should be held down by predators and therefore, herbivores should not compete (Hairston et al. 1960), whereas in the absence of top-down control by predators, regulation should be bottom-up (Anderson et al. 2010, Hopcraft et al. 2010).

Chapter 1 Introduction 15

Thesis aims The comparative study of multiple species in a community can give insight into the ecology of individual species and ecological partitioning between them (Telfer et al. 2008). In this study I combine multiple approaches to improve our understanding of large herbivore ecology and organisation in a contemporary assemblage made up of species with disparate evolutionary histories on Yanakie Isthmus (Figure 1.1), Wilsons Promontory National Park, Victoria, Australia (Figure 1.2). In particular, I aim to compare niche parameters among populations of five sympatric native and introduced herbivore species by simultaneously assessing overlap in resource use along two dimensions (spatial and trophic) at multiple scales, thereby providing insight into resource partitioning and competition within this herbivore assemblage (Krebs 1998). I had four specific aims:  Improve the accuracy of density and abundance estimates from faecal pellet counts for the five terrestrial mammalian herbivore species on Yanakie Isthmus by quantifying species- and habitat-specific faecal pellet decay rates.  Quantify broad scale habitat use by five native and introduced herbivore species, identify habitat requirements and preferences of each species, and quantify inter-specific overlap (and thus partitioning) in habitat use, as well as variation in niche breadth among species on Yanakie Isthmus.  Quantify the effect of an ecological process (fire) on fine scale habitat use by (and thus resource partitioning among) five native and introduced herbivore species on Yanakie Isthmus.  Quantify the diets of five native and introduced herbivore species of varying body-size, and thus estimate inter-specific overlap in diet and examine resource partitioning among this community on Yanakie Isthmus.

Thesis outline The thesis is organised as a series of stand-alone chapters, written as manuscripts for publication. Therefore, some repetition of general background and site descriptions has been unavoidable. Chapter 2 describes habitat-specific faecal pellet decay rates for each herbivore species, which are used to estimate herbivore

Chapter 1 Introduction 16 densities in both Chapter 3 and Chapter 4. Chapter 3 describes broad scale habitat use by five native and introduced herbivore species, identifying habitat preferences of each species and quantifying inter-specific overlap in habitat use. Chapter 4 describes the effect of an ecological process (fire) on fine scale habitat use by (and thus resource partitioning among) five native and introduced herbivore species. Chapter 5 describes the diets of the five native and introduced herbivore species and estimates inter-specific overlap in diet. Chapters 3-5 provide discussion of niche partitioning within this community along particular dimensions and at particular scales, and these discussions are synthesised in Chapter 6, the concluding chapter.

Chapter 1 Introduction 17

Figure 1.1. Location of Yanakie Isthmus at Wilsons Promontory National Park, Victoria, Australia. Yanakie Isthmus extends from the northern boundary of Wilsons Promontory National Park to Darby River and is bounded by coastline to the west and to the north-east, and by heathland to the south-east. For the purposes of this study, the south-east boundary was extended to include heathland and Moist Foothill Forest, forming the Greater Yanakie Isthmus.

Chapter 1 Introduction 18

Chapter 1 Introduction 19

Figure 1.2. Location of Wilsons Promontory National Park, Victoria, Australia.

Chapter 1 Introduction 20

Chapter 1 Introduction 21

Chapter 2

Habitat-specific faecal pellet decay rates for five mammalian herbivores in south-eastern Australia ______

Chapter 2 Habitat-specific faecal pellet decay rates 22

Chapter 2 Habitat-specific faecal pellet decay rates 23

Chapter 2 Habitat-specific faecal pellet decay rates for five mammalian herbivores in south-eastern Australia

Abstract Faecal pellet counts are commonly used to estimate the abundance of mammalian herbivores, but the accuracy of such estimates is affected by pellet decay rates. I quantified habitat-specific and species-specific faecal pellet and pellet group decay times for the five mammalian herbivores present on Yanakie Isthmus, Victoria, Australia. I collected fresh pellet groups (n = 300 per species) over 12 months, deposited five pellet groups per species per month within each of five vegetation type, then monitored pellet group decay over 24 months. The mean (± s.e.) time for pellet groups to decay was 238 ± 16 days for hog deer Axis porcinus, 98 ± 12 days for European rabbits Oryctolagus cuniculus, 231 ± 17 days for eastern grey kangaroos Macropus giganteus, 194 ± 15 days for swamp wallabies Wallabia bicolor and 219 ± 16 days for common wombats Vombatus ursinus. Mean times for pellet groups to decay varied substantially within and among vegetation types, being longest in Coastal Grassy Woodland and Heathy Woodland, shortest in Heath, and intermediate in Coastal Scrubs and Grasslands and Moist Foothill forest. Pellet group decay rates varied idiosyncratically over time, influenced by interactions between the effects of vegetation type and season of trial commencement. My results highlight the need to incorporate habitat- specific decay times into estimates of herbivore abundance.

Chapter 2 Habitat-specific faecal pellet decay rates 24

Chapter 2 Habitat-specific faecal pellet decay rates 25

Introduction On Yanakie Isthmus, Wilsons Promontory National Park (WPNP), Victoria, five terrestrial mammalian herbivores occur in sympatry: hog deer Axis porcinus, European rabbits Oryctolagus cuniculus, eastern grey kangaroos Macropus giganteus, swamp wallabies Wallabia bicolor and common wombats Vombatus ursinus. Habitat is the most common resource partitioned in ecological communities (Schoener 1983) and mediation of competitive interactions through habitat selection often plays an important role in structuring herbivore assemblages (Mishra et al. 2002). Therefore, the study of habitat use by sympatric herbivores can provide insight into inter-specific interactions. To aid in the examination of habitat use by herbivores, information is required on their relative abundances. A method commonly used to estimate population abundance for mammalian herbivores is the faecal pellet count method (Bailey and Putman 1981). Pellet density is a function of the number of animals present, the rate at which these animals defecate, and the rate at which their pellets decay (Bailey and Putman 1981). Studies such as that by Forsyth et al. (2007) have demonstrated positive correlations between herbivore densities and pellet counts. Further, the faecal pellet count method is considered appropriate for simultaneously surveying species whose activity patterns and behaviour differ (e.g., Lunney and O'Connell 1988) and avoids many of the detection biases that occur for direct counts in dense vegetation with variable topography (Hill 1981b). In particular, pellet counts are useful when animals are nocturnal or shy (Ellis et al. 1998, Evans and Jarman 1999), or their foraging habitats necessitate ground surveys (e.g., swamp wallabies; de Munk 1999). Pellet decay rates have been shown to vary in time and space due to a variety of variables such as rainfall and temperature (Hone and Martin 1998, Massei et al. 1998, Vernes 1999). The ability to age pellets, using changes in characteristics such as colour or texture, could therefore improve the accuracy and efficiency of pellet counts (Prugh and Krebs 2004). Nonetheless, pellet decay should be measured in each habitat at sites where faecal pellet counts are used to estimate population abundance (Mayle et al. 1999). Faecal pellet counts have been used to estimate the abundance of hog deer (e.g., Dhungel and O'Gara 1991), European rabbits (e.g., Wood 1988, Fa et al.

Chapter 2 Habitat-specific faecal pellet decay rates 26

1999), eastern grey kangaroos (e.g., Hill 1981b), swamp wallabies (e.g., Floyd 1980) and common wombats (e.g., Lunney and O'Connell 1988) in other parts of the range of theses species. However, hog deer pellet group decay times have not been quantified; rather, previous researchers have used faecal accumulation rate counts and assumed that no pellet decay has occurred during the interval between clearance and re-surveying of plots (Dhungel and O'Gara 1991). Pellet group decay rate estimates have been made for the other four species, although trials have generally been conducted over short periods to determine appropriate intervals between faecal accumulation rate counts or to quantify decay between surveys (e.g., eastern grey kangaroos; Johnson and Jarman 1987) and in some cases are unpublished (e.g., for swamp wallabies; Bradley 1978). To improve the accuracy of density and abundance estimates from faecal pellet counts for the five terrestrial mammalian herbivore species on Yanakie Isthmus (Chapter 3 and Chapter 4), I aimed to quantify species- and habitat- specific faecal pellet decay times. I also aimed to determine whether decay rates varied among pellet groups deposited in five different vegetation types over four seasons, and to determine the accuracy of using characteristics such as colour and texture to classify pellets into three broad age classes.

Methods I determined decay times for faecal pellet groups of the hog deer, eastern grey kangaroo, swamp wallaby, common wombat and European rabbit by collecting fresh pellets of each species, depositing them in known field locations and monitoring decay for 24 months. It is possible that relocation of pellets to new microclimates may have altered decay rates, however, it was not logistically feasible to examine decay over long periods at the original sites of pellet deposition, and I ensured pellets were deposited in multiple representative microclimates within the five sites.

Study area I conducted this study on Yanakie Isthmus (38° 53' S; 146° 14' E), WPNP, Victoria, Australia, from 2 February 2005 to 5 February 2007. For the purposes of this study I extended the south-east boundary of Yanakie Isthmus to include 3700 ha of the dominant vegetation types at WPNP and excluded an uncommon forest

Chapter 2 Habitat-specific faecal pellet decay rates 27 community (22 ha), to create the 10524-ha ‘Greater Yanakie Isthmus’ (GYI; Figure 1.1). The GYI consists of five vegetation types. The climate on Yanakie Isthmus is mild and precipitation is reliable. During this study mean monthly maximum temperatures on Yanakie Isthmus ranged from 12.9°C (June) to 25.5°C (February) and mean minima from 6.1°C (June) to 17.1°C (February) (Figure 2.1; Parks Victoria, unpublished data). Mean annual rainfall during my study was 782.8 mm, the highest rainfall occurring during winter (June – August) 2005, spring (September – November) 2005 and autumn (March – May) 2006 (Figure 2.1; Parks Victoria, unpublished data). I measured pellet group decay at one site within each of the five broad vegetation types on GYI, described in Chapter 1. Large tracts of Coastal Grassy Woodland on Yanakie Isthmus are highly modified (Bennett 1994) and the Coastal Grassy Woodland site used (Old Burn Track) was dominated by a low canopy of Leptospermum laevigatum and occasional Acacia longifolia, with a sparse ground layer of herbs such as Dichondria repens and a thin litter layer. This site also included an open slashed area with a ground layer dominated by Hibbertia sericea, Acrotriche spp., Juncus spp. and mosses, with a low shrub layer of L. laevigatum. The Coastal Scrubs and Grasslands site (Yanakie Airstrip) was depleted, like much of this vegetation community on the isthmus (Parks Victoria, unpublished report). Most of this site consisted of a dense woodland formation with a low canopy of A. longifolia and a ground layer of occasional herbs and mosses, but the site also included an exposed, heavily grazed grassland with a patchy ground layer of grasses, Ficinia nodosa and mosses. Both the woodland and grassland areas were interspersed by patches of bare ground with a sparse litter layer. The Heath site (Darby River) was relatively moist, with a lush, heavily-grazed grass and herb-rich ground layer interspersed by patches of P. esculentum and rushes with a sparse litter layer. The shrub layer consisted of occasional Leucopogon parviflorus and Rhagodia candolleana, and the low overstorey consisted of scattered Melaleuca ericifolia, Solanum aviculare and Bursaria spinosa. The Heathy Woodland site (Five Mile Road) consisted of a woodland area with a coarse, dense litter layer, a dense ground layer of P. esculentum, Acrotriche spp. and non-graminoid monocots, a dense shrub layer of P. esculentum and scattered E. nitida, Allocasuarina spp. and Melaleuca spp. in the canopy. The Heathy Woodland site also incorporated a slashed area with

Chapter 2 Habitat-specific faecal pellet decay rates 28 similar but shorter ground layer vegetation, no shrub or canopy vegetation and sparse litter. The Moist Foothill Forest site (Meeniyan Promontory Road) had a thick, damp litter layer, occasional herbs (e.g., D. repens) in the ground layer, a discontinuous shrub layer of P. esculentum and Phragmites australis and a canopy dominated by Eucalyptus spp. and M. ericifolia. To quantify variability between vegetation types in characteristics that might influence pellet decay, I estimated percent cover (to the nearest 10%) of the ground, shrub and canopy layer and litter depth ( 0.1 cm) at each site in February 2005. I made the cover measurements in 3-m radius circles centred on ten random pegs per site (see Experimental design), and took litter measurements next to each peg.

Experimental design I collected fresh pellet groups of the five herbivore species at approximately monthly intervals between 2 February 2005 and 3 January 2006. I collected pellet groups opportunistically throughout GYI and during focused searches in areas with high herbivore densities on GYI and at the main camp ground, Tidal River, and froze them until the decay trials began. It is possible that freezing may affect microbial activity and thus alter pellet decay rates. However, it was not feasible to collect enough fresh pellets for each trial within a very short time prior to trial commencement and I considered the error potentially introduced by freezing to be less than that associated with storing pellets unfrozen. I identified fresh pellets by their soft and moist texture, intact and shiny outer layer, strong odour and lack of coprophagous insect activity (Mayle et al. 1999, Triggs 2003). I was able to identify natural pellet groups (an assemblage of one or more pellets judged to be separate from surrounding pellets on the basis of distance, pellet size, shape and age; Hill 1978) for all species except European rabbit, whose pellets are often concentrated in large latrines (Calvete et al. 2004). Therefore, I used rabbit pellet groups of varying size, ranging from 6-55 (mean 17.5 ± 0.04) pellets, based on the variation I observed in those natural rabbit pellet groups I could identify and similar to the pellet group size of 20 pellets used by Wood (1988) for this species is south-eastern Australia.

Chapter 2 Habitat-specific faecal pellet decay rates 29

I began one pellet decay trial at each of the five sites each month for 12 consecutive months, the first in February 2005 and the last in January 2006. Each monthly trial at each site consisted of 25 pellet groups: one pellet group from each species deposited at equally-spaced compass bearings in a ring around each of five pegs (i.e., one deer pellet group at 0°, kangaroo at 70°, rabbit at 140°, wombat at 210° and wallaby at 280°). I used each set of five pegs for three consecutive monthly trials, laid in concentric rings c. 30 cm apart (Figure 2.2). I allocated pegs to representative microhabitats in approximate proportion to their area (Mayle et al. 1999). This design gave a sample size of 60 pellet groups per species per vegetation type, exceeding the minimum of 4-6 groups per habitat type recommended by Mayle et al. (1999). To mimic natural deposition, I dropped pellet groups from a height of c. 30-40 cm and exposed them to natural conditions (Hemami and Dolman 2005). To quantify the decay of pellets and pellet groups, I counted the number of pellets remaining at set time intervals. After counting pellets weekly during the first month and fortnightly for the second and third months, I observed that decay was slow enough to detect changes at longer intervals. I subsequently counted pellets monthly until the first trial (begun in February 2005) had been running for 12 months, and bimonthly thereafter. I monitored pellet groups until they had decayed (i.e., < 6 intact pellets remaining for deer (Mayle et al. 1999) and rabbits and < 1 intact pellet remaining for kangaroos, wallabies and wombats). I defined intact pellets as retaining a complete outer coating (Hickling 1986). Consistent with pellet count methodology, I pushed vegetation aside to search for pellets, but did not disturb litter unless one pellet was visible above the litter (Hickling 1986). To assess the accuracy of ageing pellets, I classified pellet groups from the first seven monthly trials as: ‘fresh’ – pellets soft with an intact, dark, slimy outer layer, no evidence of decay and a strong odour, ‘medium’ – pellets firm with an intact outer layer that had dried black or grey, little evidence of decay, or ‘old’ – pellets grey and showing evidence of decay (Mayze and Moore 1990) at each count.

Statistical analyses To quantify variability among vegetation communities in characteristics thought to influence pellet decay, I compared litter depth and cover of the ground,

Chapter 2 Habitat-specific faecal pellet decay rates 30 shrub and canopy layers between the five vegetation types using one-factor ANOVA with post-hoc Tukey’s tests. I plotted residuals for these and other linear models against the corresponding fitted values to check for distributional problems. If these plots were wedge shaped rather than data points being evenly spread accros plots, I applied arcsine or square-root transformations to improve residual distributions. I performed all analyses using SYSTAT Version 10.2, and for all analyses I used an  = 0.05% level of significance. I based analyses of pellet decay on two reciprocal measures (Laing et al. 2003): (1) mean time to decay per pellet group, and (2) decay rate (the proportion of pellets persisting per unit of time). I considered the time to decay as the mid- point between the counts before and after which decay occurred. To calculate mean time (days) to decay for pellet groups of each species (and for individual pellets of each of the three marsupial species) within each vegetation type, I took the average of the decay times for the five pellet groups (and pellets for marsupials) deposited in each vegetation type per month. I calculated mean time (days) to decay for pellet groups from trials commenced in each season by averaging over vegetation types the time to decay for pellet groups (and pellets for marsupials) deposited in each season. I excluded from analyses pellet groups that became flooded at the Heath site, as well as two hog deer pellet groups in Coastal Grassy Woodland that had not decayed by the end of monitoring. For the hog deer pellet group decay trial commenced in summer in Coastal Grassy Woodland, I calculated 95% confidence intervals for the mean proportion of pellets persisting at each time point in two ways. The first approach essentially considered the 15 observations at each time point as a random sample from this vegetation type, independent of the samples at other time points. This simple analysis gave means and standard errors which I used to obtain confidence intervals in the usual way, assuming normality. However, the samples were not spatially independent over time, because I returned to the same 15 pegs on each occasion. To account for possible serial correlation, I used a bootstrap approach (Davison and Hinkley 2006), treating the 15 pegs as the essential sample, and using each peg as a replicate. I obtained 1000 bootstrap samples, and from these, I obtained bootstrap confidence intervals for each time point. However, the bootstrapped confidence intervals I obtained using this procedure were very

Chapter 2 Habitat-specific faecal pellet decay rates 31 similar to those found using the simple approach (Figure 2.3), suggesting that serial correlation over time was negligible. Therefore, I estimated 95% confidence intervals for pellet group decay curves for each species within each vegetation type using the standard approach, assuming independence. I used two-factor repeated measures ANOVA (Sokal and Rohlf 1995, Underwood 1997) to assess variation in the decay rate of pellet groups from each species among the five vegetation types and among the seasons in which I commenced trials. I based these analyses on the proportion of pellets persisting at four time intervals: 4 months, 8 months, 12 months and 16 months. I did not model decay rates because each model would have had too much evidence against it, given large variation in pellet group decay rates between replicates within vegetation types. I also used data for pellet groups deposited in the first month of each season to calculate the mean pellet group decay rates per week during the first three months after deposition within each season (averaged over vegetation types) for each species. I determined the age of pellet groups of each species in each of the three age classes to assess the accuracy of pellet ageing. I did this by plotting the mean percentage of pellet groups classified as fresh, medium or old at each time interval across the first seven monthly trials.

Results Site descriptions There were significant differences among vegetation types in ground cover: cover was greater in Heathy Woodland than in Coastal Scrubs and Grasslands or Coastal Grassy Woodland, and greater in Heath than in Coastal Grassy Woodland (Table 2.1; Figure 2.4). There were also significant differences among vegetation types in shrub cover: cover was greater in Moist Foothill Forest than in Coastal Scrubs and Grasslands. Canopy cover and litter depth did not vary significantly among vegetation types.

Decay times The mean number of pellets per natural pellet group (± standard error) was 75.1 ± 1.8 for hog deer, 6.1 ± 0.1 for the eastern grey kangaroo, 5.0 ± 0.1 for the swamp wallaby and 8.0 ± 0.1 for the common wombat. The mean time to decay of

Chapter 2 Habitat-specific faecal pellet decay rates 32 faecal pellet groups was generally long, but varied greatly (Table 2.2): 14 – 728 days for deer, 7 – 406 days for rabbits, 14 – 574 days for kangaroos, 14 – 518 days for both wallabies and wombats. Mean time to decay was longest for deer pellet groups and shortest for rabbit pellet groups: on average, rabbit pellet groups decayed in less than half the time in which deer pellet groups decayed (Table 2.2). Mean times to decay for pellet groups of kangaroo, wallaby and wombat were intermediate and similar. Mean times to decay for individual pellets were also similar among the three marsupial species.

Influence of vegetation type and season on decay time and rate The decay rates of pellet groups varied significantly between vegetation types for all five species: the most rapid decay rates occurred in Heath and, except for the rabbit, the slowest decay rates occurred in Heathy Woodland (Table 2.3; Figures 2.5-2.9). The decay rate of rabbit pellet groups was also slow in Heathy Woodland, but it was slowest in Coastal Grassy Woodland (Table 2.3; Figure 2.6). These patterns for pellet group decay rates are consistent with mean times to decay, which were longest in Coastal Grassy Woodland and Heathy Woodland, shortest in Heath and intermediate in Coastal Scrubs and Grasslands and Moist Foothill forest (Table 2.2). Although pellet group decay rates varied between vegetation types, there was large variation in decay rates within each vegetation type. The season of trial commencement had little effect on mean time to decay (Table 2.2). Overall, variation in mean time to decay between trials commenced in different seasons was minor, with < 20 days difference between the longest overall mean time to decay (for trials commenced in spring) and the shortest overall mean time to decay (for trials commenced in autumn) (Table 2.2). Season of trial commencement had no significant effect on the decay rate of hog deer, kangaroo, wombat or rabbit pellet groups (Table 2.3; Figures 2.5-2.9). In contrast, the decay rate of swamp wallaby pellet groups was significantly affected by season of trial commencement: the most rapid decay occurred for trials commenced in summer and the slowest for trials commenced in spring (Tables 2.2 and 2.3; Figure 2.8). Mean weekly pellet group decay rates during the first three months after deposition varied between species and seasons. During summer, autumn, winter

Chapter 2 Habitat-specific faecal pellet decay rates 33 and spring respectively, mean (± s.e.) weekly decay rates were 5.5 ± 0.4, 2.9 ± 0.5, 3.2 ± 0.4, 3.5 ± 0.6 deer pellets, 0.8 ± 0.1, 1.0 ± 0.1, 0.7 ± 0.1, 1.0 ± 0.1 rabbit pellets, 0.2 ± 0.0, 0.2 ± 0.0, 0.2 ± 0.0, 0.1 ± 0.0 kangaroo pellets, 0.2 ± 0.0, 0.1 ± 0.0, 0.2 ± 0.0, 0.1 ± 0.0 swamp wallaby pellets, and 0.2 ± 0.0, 0.2 ± 0.0, 0.3 ± 0.1, 0.3 ± 0.0 wombat pellets. There were significant interactions between the effects of time, season of trial commencement and vegetation type on pellet group decay rates (Table 2.3; Figures 2.5-2.9). In Moist Foothill Forest, the rate of decay for hog deer pellet groups was rapid for trials commenced in autumn, and slow for winter trials, relative to other seasons (Table 2.3; Figure 2.5). Deer pellet group decay rates in Coastal Scrubs and Grasslands were relatively rapid for winter trials and relatively slow for spring trials, while decay rates in Coastal Grassy Woodland were relatively rapid for spring trials. In Heathy Woodland and Coastal Grassy Woodland, deer pellet group decay rates were relatively slow for autumn trials. The decay rate of pellet groups for summer trials in Heathy Woodland varied from the decay curves typical of other seasons in that there was a considerable period where little decay occurred. The rate of decay for European rabbit pellet groups in Heath was slowest for trials commenced in spring relative to other seasons, while in Moist Foothill Forest decay was slowest for winter trials (Table 2.3; Figure 2.6). In Heathy Woodland, rabbit pellet groups from autumn and spring trials exhibited relatively rapid initial decay rates. In Coastal Grassy Woodland, trials commenced in winter exhibited a period of rapid decay at ~100 days; more even decay rates were observed in trials commenced in other seasons. In Coastal Scrubs and Grasslands, initial rabbit pellet group decay rates were similar among seasons, although pellet groups from summer and spring trials persisted longest. The rate of decay for eastern grey kangaroo pellet groups in Heath varied between seasonal trials (Table 2.3; Figure 2.7): pellet groups from trials commenced in spring exhibited an initial period of rapid decay, decay of pellet groups from summer trials was relatively even and rapid, pellet groups from autumn trials decayed little between ~30 and 50 days, and pellet groups from winter trials persisted for a relatively long time. Kangaroo pellet groups from winter trials in Moist Foothill Forest decayed relatively slowly. Decay rates for winter trials in Coastal Scrubs and Grasslands were relatively rapid, while those

Chapter 2 Habitat-specific faecal pellet decay rates 34 from spring trials were relatively slow. Kangaroo pellet group decay rates in Coastal Grassy Woodland and Heathy Woodland were slowest for autumn trials, and the most rapid decay rates in Coastal Grassy Woodland and Heathy Woodland were for spring and summer trials respectively. The rates of decay of swamp wallaby pellet groups in Heath were slow for trials commenced in winter relative to trials commenced in other seasons (Table 2.3; Figure 2.8). Although there was variation in pellet group decay curves among seasonal trials in Coastal Grassy Woodland, overall decay rates were similar. Decay curves varied among seasonal trials in Coastal Scrubs and Grasslands, with overall decay rates slowest for autumn trials, as was the case in Heathy Woodland. In Moist Foothill Forest and Heathy Woodland there was an increase in the mean number of pellets persisting at ~90 days for summer and spring trials, an anomaly also observed for some other species (e.g., wombat pellets). The decay rates of common wombat pellet groups in Coastal Grassy Woodland and Heath were relatively rapid for trials commenced in summer (Table 2.3; Figure 2.9). In Heath, wombat pellet group decay rates were slowest for spring trials and pellet groups from winter and spring trials persisted for relatively long periods. The decay rate of pellet groups from winter and spring trials in Moist Foothill Forest were also relatively slow. Pellet groups from autumn trials in Coastal Scrubs and Grasslands decayed slowly relative to pellet groups from other seasons. Although pellet group decay times in Heathy Woodland were similar for trials commenced in different seasons, decay rates varied temporally.

Ageing faecal pellet groups For all five species, pellet groups classified as fresh were < 1 week old with the exception of two rabbit pellet groups that were two weeks old (Figure 2.10). For deer, pellet groups classified as medium were ~1-2 weeks old and old pellet groups  3 weeks old (i.e., > 50% of pellet groups fell into those categories). For the kangaroo, wallaby and wombat, medium pellet groups were ~1-6 weeks old and old pellet groups > 6 weeks old. For the rabbit, medium pellet groups were ~1-8 weeks old and old pellets > 8 weeks old. Pellet groups classified

Chapter 2 Habitat-specific faecal pellet decay rates 35 as old were sometimes later classified as medium, and two rabbit pellet groups classified as medium were classified as fresh at a later time.

Discussion To my knowledge, this is the first study to quantify habitat-specific and season-specific faecal pellet decay times for an entire assemblage of sympatric mammalian herbivores. Specifically, this is the first study to quantify pellet decay times for hog deer, and although estimates of pellet group decay have been made for the swamp wallaby, common wombat, eastern grey kangaroo, and European rabbit, trials have generally been conducted over short periods and in most cases are unpublished. My habitat-specific estimates of pellet decay can be used to improve the accuracy of population abundance estimates from faecal pellet counts, a commonly used technique for mammalian herbivores. More broadly, my study provides guidelines regarding spatial and temporal generalisation of pellet decay rates, following demonstration of great variability in the decay times of mammalian herbivore pellet groups within and among species, vegetation types and seasons.

Decay times The mean decay time for hog deer pellet groups on GYI was 238 days, although decay times ranged from as little as two weeks to over two years. These results support the findings of studies in other locations showing that most deer pellet groups decay within 12 months, as reported for muntjac (Muntiacus reevesi) and roe deer (Capreolus capreolus) in England (Hemami and Dolman 2005), but pellet group decay can take several years, as shown in Columbian black-tailed deer (Odocoileus hemionus columbianus) on northern Vancouver Island (Harestad and Bunnell 1987) or as little as a week, as occurs in muntjac in England (Chapman 2004). The mean decay time for rabbit pellet groups on GYI was 98 days, slower than recorded by Taylor and Williams (1956) in New Zealand, but faster than recorded by Wood (1988) in Australian subalpine habitat or by Cabrera- Rodriquez (2006) in Spain. However, decay times on GYI ranged from one week to over a year, supporting observations by Taylor and Williams (1956) that rabbit pellet group decay rates are highly variable.

Chapter 2 Habitat-specific faecal pellet decay rates 36

The mean decay time for kangaroo pellet groups on GYI was 231 days, although decay times ranged from two weeks to over 18 months. Great variation in eastern grey kangaroo pellet group decay times has been demonstrated by other studies that have recorded decay of pellets within 24 hours (Johnson and Jarman 1987) and persistence of pellets after 240 days (Stewart 1982). Mean pellet group decay times for wallaby and wombat pellet groups on GYI were 194 days and 219 days respectively, and for both species ranged from 14 days to almost 18 months. My estimates of decay time for wallaby pellet groups converts to a decay rate at the upper limit of the range of pellet group decay rates (5-15% per month) recorded by Bradley (1978) for this species in wet and dry sclerophyll forest in NSW. Work by McKenzie (1976) and also by Lunney and O’Connell (1988) has also assessed decay rates for swamp wallabies, and in the latter case for wombats also, but these values are not published. Mean time to decay was longer for deer pellet groups than for pellet groups of the other four species. Conversely, the shortest mean time to decay was recorded for rabbit pellet groups. Inter-specific differences in pellet group decay times may be related to pellet and pellet group size (Massei et al. 1998), differential activity of invertebrates on pellets or inter-specific differences in diet and thus faecal material characteristics (Hemami and Dolman 2005). For example, pellets with high fibre content may be more resistant to decomposition than those with low fibre content (Lehmkuhl et al. 1994). On GYI, pellet group decay rates for the browsing swamp wallaby were slightly higher than those of the grazing wombat and kangaroo, supporting Southwell’s (1989) findings of faster decay rates for browsing than grazing macropods. However, the relatively slow decay of deer pellet groups and fast decay of rabbit pellet groups is unlikely to be explained by diet, as these deer consume more browse on Yanakie Isthmus than do rabbits (Chapter 5; Davis et al. 2008). The anomaly observed for some pellet group decay trials commenced in spring, in which there was an increase in the mean number of pellets at about 90 days, could indicate either the deposition of new pellets on top of the groups being monitored (Taylor and Williams 1956) and/or temporal changes in pellet visibility (Harestad and Bunnell 1987). Given that the increase occurred simultaneously at several sites, I suspect that pellet visibility changed with time.

Chapter 2 Habitat-specific faecal pellet decay rates 37

Influence of vegetation type and season on decay time and rate Pellet decay rates vary over time and space in relation to biotic factors (Chapman 2004), and uncertainty in pellet decay estimates can be reduced if these predictors can be identified (e.g., Barnes and Dunn 2002). However, variation in pellet group decay rates on GYI could not be simply explained by habitat or season. Inter-habitat differences in litter depth and vegetation cover, factors that commonly influence pellet decay (e.g., Kuehl et al. 2007), were inconsistent with variation in pellet decay. Similarly, there were no clear correlations between seasonal variation in pellet decay and weather conditions. For example, rapid decay of swamp wallaby pellet groups from trials commenced in summer could have been associated with high temperatures (e.g., Hone and Martin 1998, Hemami and Dolman 2005), yet wallaby pellet decay was relatively slow for trials commenced in spring, when temperatures were also warm. Likewise, relatively slow decay for pellets of some species during winter and autumn could have been associated with cool temperatures (e.g., Massei et al. 1998), but again this pattern was inconsistent across vegetation types and species. Highly variable spatial and temporal patterns of pellet decay suggest that complex interactions between habitat, season and pellet composition control decay. For example, the influence that exposure to weather conditions has on pellet decay (e.g., Johnson and Jarman 1987, Hemami and Dolman 2005) may vary depending on vegetation structure. More detailed information on processes effecting decay and greater replication at the site level are required to explain species- and habitat-specific idiosyncrasies in pellet group decay rates.

Ageing faecal pellet groups Accurate ageing of pellet groups would enable collection of data during the establishment of pellet accumulation rate plots and would reduce error associated with the persistence of old pellets when using standing crop counts (Prugh and Krebs 2004). Except for two rabbit pellet groups, I was able to reliably classify pellet groups as ‘fresh’ (i.e., < 1 week old). ‘Medium’ pellet groups were generally one week to three months old, although the age of ‘medium’ pellet groups varied among species (e.g., ~1-2 weeks for hog deer and ~1-8 weeks for rabbits). The classification of ‘old’ pellets was most problematic; after about three months, most pellet groups were reliably classified as old, but in some cases,

Chapter 2 Habitat-specific faecal pellet decay rates 38 pellet groups were classified as old after one week. Furthermore, it was not uncommon for pellets classified as old to be classified as medium at the next survey. My results highlight the difficulties associated with determining pellet age due to the influence of factors such as weather, vegetation and diet on the colour and texture of pellets (Taylor and Williams 1956, Brodie 2006) and indicate that subjective classification of pellets according to broad age classes is inaccurate. I suggest that pellet features can be used to identify fresh pellets and give an approximate indication of their age, but cannot be accurately applied to older pellets. Therefore, ageing pellets is unlikely to improve the accuracy of pellet count surveys for mammalian herbivores on GYI.

Applications If pellet group decay rates vary among habitats, as occurred on GYI, the assumption of constant pellet decay among habitats can lead to errors in estimates of herbivore densities from pellet counts (Hemami and Dolman 2005). Therefore, my habitat-specific estimates of pellet decay rates may be valuable in reducing this type of bias (Brodie 2006). However, application of my pellet decay estimates to future surveys requires caution, as pellet decay is likely to vary over time (Hone and Martin 1998), particularly as my study was conducted during drought, which can influence pellet group size (Perry and Braysher 1986) and dung beetle abundance (Tyndale-Biscoe 1994). Similarly, my pellet decay rate estimates should not be extrapolated beyond GYI, as this could bias density estimates (Kuehl et al. 2007). Rather, it is preferable to estimate local pellet decay rates immediately prior to pellet counts (Laing et al. 2003). Faecal pellet accumulation rate surveys commonly assume that pellet groups persist for the duration of the accumulation period (Harestad and Bunnell 1987) and therefore that corrections for pellet loss are not needed (Johnson and Jarman 1987). This assumption should not be made on GYI because in some cases, pellet decay times were rapid. Clearance of plots at intervals frequent enough to avoid pellet decay (i.e., < 1 week) is likely to be logistically unfeasible and therefore corrections for pellet group decay will be required (Murray et al. 2005). Following Mayle et al. (1999), my results suggest that for mammalian herbivores on GYI appropriate intervals between clearing and resurveying pellet accumulation plots (i.e., just before the expected pellet decay time) range from

Chapter 2 Habitat-specific faecal pellet decay rates 39 monthly (e.g., for rabbits in Heath) up to yearly (e.g., for hog deer in Heathy Woodland). Variation in pellet decay times between species and vegetation types means that survey intervals need to be considered carefully in the context of study objectives. During some studies of pellet decay, workers have been able to identify the best time of year to conduct pellet accumulation counts to minimise biases associated with pellet decay (Vernes 1999), for example, during cold and dry periods (Massei et al. 1998). On GYI there was variation in pellet group decay rates during the initial period after deposition within each season, but this variation was not consistent between species. For example, mean hog deer pellet group decay rates in the first three months after deposition were fastest in summer and slowest autumn, while European rabbit pellet group decay rates in the first three months were fastest in autumn and spring and slowest in winter. Inconsistencies between species in seasonal trends in decay rate suggest that the best time to conduct accumulation trials will vary depending on the species being surveyed. The temporal variation in decay rates observed on GYI was probably associated with changes in environmental conditions. This variation is likely to result in transient periods during which the number of pellets decaying is not equal to the number being built (Kuehl et al. 2007). Such transient periods can result in a failure of the steady-state assumption and bias pellet counts (Kuehl et al. 2007). For example, population estimates may be biased if decomposition is measured during only one season (Plumptre and Harris 1995). However, such bias is likely to be minimal on GYI because most pellet groups persist for at least two seasons. Thus, my estimates of decay times averaged over trials commencing in all seasons are unlikely to be severely biased. However, caution must be exercised in inferring recent changes in population density from changes in pellet density, or recent presence of herbivores from the presence of intact pellets because the long persistence of herbivore pellets on GYI is likely to result in a substantial lag time between changes in population density and changes in pellet density (Hone and Martin 1998).

Chapter 2 Habitat-specific faecal pellet decay rates 40

Conclusion The mean time to decay for herbivore faecal pellet groups on GYI ranged from 98 ± 12 days for European rabbits to 238 ± 16 days for hog deer, although decay times were highly variable within and among species. Ageing pellet groups was inaccurate and unlikely to improve the efficiency of pellet counts. However, the application of habitat-specific pellet decay rates should reduce the bias of population estimates derived from pellet counts, because pellet group decay rates varied significantly between vegetation types. The causes of variation in pellet group decay were unclear; pellet group decay rates varied idiosyncratically over time, due to complex interactions between the effects of habitat and season. The large variability in decay rate suggests that caution should be exercised if estimates of pellet decay are to be generalised over long time-frames or areas with multiple habitats, particularly if these elements are combined during long-term studies conducted at broad spatial scales.

Chapter 2 Habitat-specific faecal pellet decay rates 41

Table 2.1. Results of one-factor ANOVAs comparing % cover of each of the ground, shrub and canopy layers, and litter depth between five vegetation types (n = 10 per type) on Greater Yanakie Isthmus: Coastal Grassy Woodland (CGW), Coastal Scrubs and Grasslands (CSG), Heath (H), Heathy Woodland (HW), and Moist Foothill Forest (MFF) 2005. P values in bold are significant effects. Where applicable, results for pair-wise comparisons using Tukey’s post-hoc analysis are included.

Significant P (significant Comparison m.s. d.f. F P pair-wise pair-wise comparisons comparisons) % ground cover Site 5233.000 4 5.669 0.001 HW vs CSG 0.027 Error 923.111 45 HW vs CGW 0.003 H vs CGW 0.012 % shrub cover Site 23.738 4 4.847 0.002 CSG vs MFF 0.001 Error 4.898 45 % canopy cover Site 1843.000 4 1.408 0.247 Error 1308.889 45 Litter depth Site 1.979 4 1.807 0.144 Error 1.095 45

Chapter 2 Habitat-specific faecal pellet decay rates 42

Table 2.2. Mean time to decay (days) and standard error for pellet groups of five herbivores (and for individual pellets for the three marsupial species) in five vegetation types (n = 60 pellet groups per vegetation type, except n = 55 pellet groups for the eastern grey kangaroo and the common wombat in Heath) on Greater Yanakie Isthmus, commenced over four seasons (n = 75 pellet groups per season) in 2005.

Hog deer European Swamp wallaby Eastern grey kangaroo Common wombat Mean for all rabbit species Pellet group Pellet group Pellet group Pellet Pellet group Pellet Pellet group Pellet Pellet group

Mean SE Mean SE Mean SE Mean SE Mean SE Mean SE Mean SE Mean SE Mean SE Vegetation type Coastal Grassy Woodland 351 12 167 9 257 22 134 7 296 20 173 7 274 26 189 6 Heath 71 8 41 5 88 7 59 3 89 14 59 3 108 9 71 2 Moist Foothill Forest 221 15 68 9 215 15 106 3 239 15 140 5 242 20 144 5 Coastal Scrubs and Grasslands 167 19 52 11 130 10 72 4 178 16 89 4 146 10 73 3 Heathy Woodland 381 26 161 24 279 22 174 7 353 21 219 8 326 16 193 6

Mean for all vegetation types 238 16 98 12 194 15 109 5 231 17 136 5 219 16 134 5

Season Summer 251 35 113 20 178 19 102 5 220 25 130 5 195 25 124 5 191 25 Autumn 222 40 78 13 189 23 118 6 249 25 151 7 209 22 128 5 189 25 Winter 228 32 111 23 198 25 125 5 231 31 147 6 239 31 147 5 201 28 Spring 252 30 89 17 216 28 124 6 244 30 142 6 240 25 130 5 208 26

Chapter 2 Habitat-specific faecal pellet decay rates 43

Table 2.3. Results of two-factor repeated measures ANOVAs comparing the decay rate of pellet groups from each of five herbivores between trials in five vegetation types commenced over four seasons (n = 5 per vegetation type per season) on Greater Yanakie Isthmus, 2005-2007. Analyses were based on the proportion of pellets (arcsine-transformed) persisting at four time intervals: 4 months, 8 months, 12 months and 16 months. P values in bold are significant effects.

Comparison m.s. d.f. F P Hog deer Between subjects Season 0.147 3 1.969 0.119 Vegetation type 4.164 4 55.678 <0.001 Season*Vegetation type 0.063 12 0.841 0.608 Error 0.075 278 Within subjects Time 5.803 3 388.213 <0.001 Time*Season 0.064 9 4.302 <0.001 Time*Vegetation type 0.483 12 32.302 <0.001 Time*Season*Vegetation type 0.023 36 1.513 0.028 Error 0.015 834 European rabbit Between subjects Season 0.066 3 1.293 0.277 Vegetation type 1.741 4 34.056 <0.001 Season*Vegetation type 0.075 12 1.468 0.136 Error 0.051 275 Within subjects Time 3.647 3 194.280 <0.001 Time*Season 0.023 9 1.245 0.264 Time*Vegetation type 0.567 12 30.209 <0.001 Time*Season*Vegetation type 0.028 36 1.478 0.036 Error 0.019 825 Eastern grey kangaroo Between subjects Season 0.129 3 1.170 0.321 Vegetation type 6.211 4 56.180 <0.001 Season*Vegetation type 0.192 12 1.737 0.059 Error 0.111 280 Within subjects Time 12.245 3 357.017 <0.001 Time*Season 0.052 9 1.517 0.137 Time*Vegetation type 0.720 12 21.004 <0.001 Time*Season*Vegetation type 0.056 36 1.635 0.011 Error 0.034 840

Chapter 2 Habitat-specific faecal pellet decay rates 44

Table 2.3 (cont.) Common wombat Between subjects Season 0.325 3 2.221 0.086 Vegetation type 5.603 4 38.226 <0.001 Season*Vegetation type 0.379 12 2.582 0.003 Error 0.147 280 Within subjects Time 18.329 3 358.971 <0.001 Time*Season 0.095 9 1.854 0.056 Time*Vegetation type 1.142 12 22.372 <0.001 Time*Season*Vegetation type 0.130 36 2.544 <0.001 Error 0.051 840 Swamp wallaby Between subjects Season 0.302 3 2.752 0.043 Vegetation type 3.959 4 36.122 <0.001 Season*Vegetation type 0.185 12 1.691 0.069 Error 0.110 275 Within subjects Time 17.533 3 338.520 <0.001 Time*Season 0.071 9 1.380 0.193 Time*Vegetation type 1.058 12 20.431 <0.001 Time*Season*Vegetation type 0.147 36 2.837 <0.001 Error 0.052 825

Chapter 2 Habitat-specific faecal pellet decay rates 45

Figure 2.1. Mean monthly rainfall, and mean monthly maximum and minimum temperatures averaged across months within each season between summer 2004- 05 and summer 2006-07 on Yanakie Isthmus, Wilson Promontory National Park.

Figure 2.2. Pellet group decay trial design: one pellet group from each of the five species was deposited at equally-spaced compass bearings of 70o, in each of three concentric rings 30 cm apart (one ring per monthly trial) around a central peg.

Chapter 2 Habitat-specific faecal pellet decay rates 46

120 Variable Bootstrapped estimate

p 100 Boostrapped lower CI u

o Boostrapped upper CI

r g

Mean r

e 80 Lower CI p

Upper CI

s

t

e

l l

e 60

p

f

o

r

e 40

b

m

u

n

n 20

a

e M 0

0 10 20 30 40 50 60 70 80 90 Time (days)

Figure 2.3. Comparison of bootstrapped and standard 95% upper and lower confidence intervals for the mean number of hog deer faecal pellets persisting over time (days) from pellet groups (n = 15) deposited in Coastal Grassy Woodland on Greater Yanakie Isthmus, Wilson Promontory National Park, during summer 2005.

Ground cover Shrub cover 90 80 70 60 50 40

% cover 30 20 10 0 Moist Foothill Heath Coastal Heathy Coastal Forest Scrubs & Woodland Grasssy Grasslands Woodland Vegetation type

Figure 2.4. Mean ( standard error) % cover of the ground and shrub layers in five vegetation types (n = 10 per type) on Greater Yanakie Isthmus, Wilson Promontory National Park, 2005.

Chapter 2 Habitat-specific faecal pellet decay rates 47

(i) Summer

120 Coastal Grassy Woodland 100 Coastal Scrubs & Grasslands 80 Heath 60 Heathy Woodland 40 Moist Foothill Forest 20 0 0 56 112 168 224 280 336 392 448 504 560 616 Time (days)

(ii) Autumn 100

80

60 40 20 0 0 56 112 168 224 280 336 392 448 504 560 616 700 812 924 Time (days)

(iii) Winter 100

80

60

Mean number of pellets per group per Meanof pellets number 40 20 0 0 56 112 168 224 280 336 392 448 504 560 616 Time (days)

(iv) Spring 100

80

60 40

20 0 0 56 112 168 224 280 336 392 448 504 560 Time (days)

Figure 2.5. Changes in the mean (with lower and upper 95% confidence intervals) number of pellets persisting over time (days) for hog deer pellet groups in five vegetation types on Greater Yanakie Isthmus, Wilson Promontory National Park 2005-2007, deposited in four seasons: (i) summer; (ii) autumn; (iii) winter; and (iv) spring (n = 15 per vegetation type for most time intervals, but varied between 5-15 due to flooding at some sites).

Chapter 2 Habitat-specific faecal pellet decay rates 48

(i) Summer

30 Coastal Grassy Woodland 25 Coastal Scrubs & Grasslands 20 Heath 15 Heathy Woodland 10 Moist Foothill Forest 5 0 0 28 56 84 112 140 168 196 224 252 280 308 336 364 392 448 504 Time (days)

(ii) Autumn 20

15

10

5

0 0 28 56 84 112 140 168 196 224 252 280 308 336 364 392 420 Time (days)

(iii) Winter 25 20

15 Mean number of pellets per group per Meanof pellets number 10 5 0 0 28 56 84 112 140 168 196 224 252 280 308 336 364 392 420 448 476 504 Time (days)

(iv) Spring 20

15

10

5

0 0 28 56 84 112 140 168 196 224 252 280 308 336 364 392 420 448 504 Time (days)

Figure 2.6. Changes in the mean (with lower and upper 95% confidence intervals) number of pellets persisting over time (days) for European rabbit pellet groups in five vegetation types on Greater Yanakie Isthmus, Wilson Promontory National Park 2005-2007, deposited in four seasons: (i) summer; (ii) autumn; (iii) winter; and (iv) spring (n = 15 per vegetation type for most time intervals, but varied between 5-15 due to flooding at some sites).

Chapter 2 Habitat-specific faecal pellet decay rates 49

(i) Summer

8 Coastal Grassy Woodland Coastal Scrubs & Grasslands 6 Heath 4 Heathy Woodland Moist Foothill Forest 2

0 0 28 56 84 112 140 168 196 224 252 280 308 336 364 392 420 448 Time (days)

(ii) Autumn 8

6

4

2

0 0 56 112 168 224 280 336 392 448 504 560 Time (days)

(iii) Winter 8

6

4 Mean number of pellets per group per Meanof pellets number 2

0 0 28 56 84 112 140 168 196 224 252 280 308 336 364 392 420 448 476 Time (days)

(iv) Spring 8

6

4

2

0 0 56 112 168 224 280 336 392 448 504 Time (days)

Figure 2.7. Changes in the mean (with lower and upper 95% confidence intervals) number of pellets persisting over time (days) for eastern grey kangaroo pellet groups in five vegetation types on Greater Yanakie Isthmus, Wilson Promontory National Park 2005-2007, deposited in four seasons: (i) summer; (ii) autumn; (iii) winter; and (iv) spring (n = 15 per vegetation type for most time intervals, but varied between 5-15 due to flooding at some sites).

Chapter 2 Habitat-specific faecal pellet decay rates 50

(i) Summer

6 Coastal Grassy Woodland 5 Coastal Scrubs & Grasslands 4 Heath 3 Heathy Woodland Moist Foothill Forest 2 1 0 0 28 56 84 112 140 168 196 224 252 280 308 336 364 392 420 448 476 Time (days)

(ii) Autumn 7 6 5

4 3 2 1 0 0 56 112 168 224 280 336 392 448 504 Time (days)

(iii) Winter 7 6 5 4 Mean number of pellets per group per Meanof pellets number 3 2 1 0 0 28 56 84 112 140 168 196 224 252 280 308 336 364 392 420 448 Time (days)

(iv) Spring 7 6 5 4 3 2 1 0 0 28 56 84 112 140 168 196 224 252 280 308 336 364 392 420 Time (days)

Figure 2.8. Changes in the mean (with lower and upper 95% confidence intervals) number of pellets persisting over time (days) for swamp wallaby pellet groups in five vegetation types on Greater Yanakie Isthmus, Wilson Promontory National Park 2005-2007, deposited in four seasons: (i) summer; (ii) autumn; (iii) winter; and (iv) spring (n = 15 per vegetation type for most time intervals, but varied between 5-15 due to flooding at some sites).

Chapter 2 Habitat-specific faecal pellet decay rates 51

(i) Summer

10 Coastal Grassy Woodland Coastal Scrubs & Grasslands 8 Heath 6 Heathy Woodland 4 Moist Foothill Forest 2 0 0 28 56 84 112 140 168 196 224 252 280 308 336 364 392 420 448 476 504 Time (days)

(ii) Autumn 10

8

6

4 2

0 0 56 112 168 224 280 336 392 448 504 Time (days)

(iii) Winter 12 10 8 6

Mean number of pellets per group per Meanof pellets number 4 2 0 0 56 112 168 224 280 336 392 448 504 560 Time (days)

(iv) Spring 10 8 6 4 2 0 0 28 56 84 112 140 168 196 224 252 280 308 336 364 392 420 448 476 504 Time (days)

Figure 2.9. Changes in the mean ( standard error) number of pellets persisting over time (days) for common wombat pellet groups in five vegetation types on Greater Yanakie Isthmus, Wilson Promontory National Park 2005-2007, deposited in four seasons: (i) summer; (ii) autumn; (iii) winter; and (iv) spring (n = 15 per vegetation type for most time intervals, but varied between 5-15 due to flooding at some sites).

Chapter 2 Habitat-specific faecal pellet decay rates 52

(i) Hog deer

Fresh Medium Old 40 5 15 5 40 10 40 5 40 40 40 40 40 100% 80% 60% 40% 20% 0%

(ii) European rabbit 100% 80% 60% 40% 20% 0%

(iii) Eastern grey kangaroo

100% 80% 60% 40% 20% 0%

% pellet groups pellet % (iv) Swamp wallaby 100% 80% 60% 40% 20% 0%

(v) Common wombat 100% 80% 60% 40% 20% 0% 0 7 14 21 28 42 56 70 84 112 140 168 196 Time (days)

Figure 2.10. Mean % of pellet groups classified as fresh, medium or old at each time interval across monthly trials from February to August 2005 at five sites in the Greater Yanakie Isthmus, Wilson Promontory National Park, for five species: (i) hog deer; (ii) European rabbit; (iii) eastern grey kangaroo; (iv) swamp wallaby; (v) common wombat. Numbers above bars on (i) are sample sizes for (i-v).

Chapter 2 Habitat-specific faecal pellet decay rates 53

Chapter 3

Broad scale habitat use by native and introduced mammalian herbivores on Yanakie Isthmus, south- eastern Australia ______

Chapter 3 Broad scale habitat use 54

Chapter 3 Broad scale habitat use 55

Chapter 3 Broad scale habitat use by native and introduced mammalian herbivores in shrub-encroached grassy woodland, south-eastern Australia

Abstract Sympatric species should reduce competition by partitioning resources along one or more dimensions, habitat being the most common dimension partitioned. At Wilsons Promontory National Park, Victoria, native and introduced mammalian herbivores occur in sympatry. I estimated habitat use by five mammalian herbivore species using faecal pellet counts. Inter-specific overlap in habitat use was generally low, suggesting spatial partitioning of habitat resources, possibly aided by temporal partitioning; resource partitioning appeared to be independent of coevolutionary history. Low overlap in habitat use implies low competition. The lack of clear shifts in habitat use from preferred to suboptimal habitats supports the suggestion that inter-specific competition is not strong enough to cause competitive exclusion. However, low overlap in habitat use between the European rabbit and other species, both native and introduced, and preferential use by rabbits (and avoidance by other species) of the habitat which appears to have the highest carrying capacity, suggests that rabbits exclude other grazing herbivores from preferred habitat. High overlap was apparent between some species, particularly grazers, indicating some potential for competition if resources are limiting. In particular, eastern grey kangaroo had a narrow niche, occurred at low densities and had low metabolic demand at the population level relative to other species, consistent with competitive suppression. In contrast, common wombats appear to be the strongest competitors in this assemblage, being numerically dominant, utilising the greatest proportion of resources, and displaying a relatively broad habitat niche. Population metabolism scaled positively with body mass, suggesting that larger herbivore species use a disproportionately greater share of local resources than smaller species on GYI, although there was no clear relationship between body size and the diversity of habitats used. Large-scale adaptive experimental management is required to provide a mechanistic understanding of patterns of habitat selection observed.

Chapter 3 Broad scale habitat use 56

Chapter 3 Broad scale habitat use 57

Introduction Investigation of competitive interactions, and the ways in which species coexist, is essential to improve our understanding of the ecological principles underlying ecosystem functioning (Duncan et al. 1998). Gause (1934) first suggested that sympatric species should reduce competition by selecting different resources. Specialisation on resources along habitat, diet and temporal gradients results in niche differentiation, facilitating coexistence (Schoener 1974b, Whitfield 2002, Schmidt et al. 2010). Sympatric herbivores tend to exploit their environments in different ways (Schwartz and Ellis 1981, Forsyth 2000) based on differences in feeding strategies (Gwynne and Bell 1968, Schwartz and Ellis 1981) which lead to differences in habitat preference (Batcheler 1960, Taylor 1983, Fox 1989), ultimately determining community structure (Bell 1971, Hofmann and Stewart 1972, Jarman 1974). Habitat is the most common resource partitioned (Schoener 1983), as habitat partitioning generally occurs before species evolve morphological adaptations to partition food (Pianka 1976). Thus mediation of competitive interactions through habitat selection often plays an important role in structuring herbivore assemblages (Mishra et al. 2002). Body size has a particularly strong influence on habitat use (duToit and Owen-Smith 1989) through its influence on dietary tolerance (e.g., larger animals have a greater ability to utilise low-quality foods; May and MacArthur 1972, Schoener 1974b) and resource domination (e.g., large animals generally have advantages in inter-specific aggression and predator evasion; Brown and Maurer 1986). The balance between population density and energy use per individual determines how evenly community resources are shared by species of different size (duToit and Owen-Smith 1989). Although large species generally occur at lower population densities than small species, being larger, they require more energy per individual (duToit and Owen-Smith 1989). In many mammalian communities, population metabolism (i.e., the product of population density and energy use per individual) scales positively with body mass (1984), thus larger species use a disproportionately greater share of local resources (duToit and Owen-Smith 1989). As highlighted by Schoener (1986), many studies suggest that foragers respond to inter-specific competition by reducing niche overlap and narrowing

Chapter 3 Broad scale habitat use 58 patch use (i.e., competition tends to limit habitat use; Svarsden 1949). By decreasing spatial overlap between populations, habitat partitioning provides a major means of reducing inter-specific competition (Schmidt et al. 2010), yet the forced movement of some species into suboptimal habitat (Schoener 1983, Sih 1993) can have a negative effect on population dynamics of these subordinate species (Sawyer et al. 2006). This type of competitive interaction can be detected in the use of similar habitats by species when they occur separately, but use of different habitats when they occur together (Werner 1977). For example, impala Aepyceros melampus in competition with cattle for habitat and food resources in their preferred habitat (Jarman and Jarman 1974, Dunham 1981) may adapt their foraging behaviour by switching to 'refuge' habitat (Bell 1970, Illius and Gordon 1987). Thus the presence of competitors can be seen to decrease habitat quality, for example, as food resources are depleted (Abramsky et al. 1991). Resource partitioning and niche differentiation are seen as evolutionary outcomes of competition, and have been well described among assemblages of co- evolved native herbivores (e.g., Bell 1970, Jarman and Sinclair 1979, McNaughton and Georgiadis 1986, Green 1987, Bodmer 1991). In particular, many early studies showed habitat segregation among native herbivore species (e.g., Lamprey 1963, Jarman 1972). However, communities globally are increasingly composed of native and introduced mammalian herbivores (e.g., Voeten and Prins 1999, Baldi et al. 2004, Madhusudan 2004) and there has been speculation on the effect of introducing exotic herbivore species into native herbivore assemblages (Bagchi et al. 2004, Madhusudan 2004, Mishra et al. 2004). Studies such as that by Schwartz and Ellis (1981) have found that inter- specific overlap in resource use can be linked to recent evolutionary history. For example, Schwartz and Ellis (1981) demonstrated low overlap in food resource use between native ungulate species relative to overlap between native and introduced species. Subsequently, it has been asserted that assemblages of species with independent evolutionary histories have inherently less resource partitioning to facilitate coexistence than species with common evolutionary histories, as they have not had the opportunity to evolve mechanisms of resource partitioning (Kirchhoff and Larsen 1998, Kelley et al. 2002, Madhusudan 2004). Therefore, it is unsurprising that competition for habitat resources has been demonstrated between native and introduced herbivores globally (e.g., Fritz et al. 1996,

Chapter 3 Broad scale habitat use 59

Madhusudan 2004, Kuiters et al. 2005), in some cases leading to competitive exclusion (Prins 2000, Kelley et al. 2002). As in communities of sympatric herbivores globally, studies of Australian herbivores have demonstrated ecological separation based on habitat partitioning (e.g., between wallaroos Macropus robustus and eastern grey kangaroos Macropus giganteus; Taylor 1983, and between eastern grey kangaroos and western grey kangaroos Macropus fuliginosus; Coulson 1990, McCullough and McCullough 2000). Over the last two centuries, a wide variety of introduced mammalian herbivore species have become established in Australia (Forsyth et al. 2004) and all of these introduced herbivores are now sympatric with native herbivores (e.g., Dawson and Ellis 1979, Dierenfeld 1984, Duncan 1992, Dawson and Ellis 1996). However, few studies have examined partitioning of habitat resources among multi-species assemblages comprised of native and introduced herbivore species. On Yanakie Isthmus (Wilsons Promontory National Park, Victoria), introduced European rabbits Oryctolagus cuniculus and hog deer Axis porcinus occur in sympatry with three native herbivores, the eastern grey kangaroo, swamp wallaby Wallabia bicolor and common wombat Vombatus ursinus. Habitat use by these species has been studied in other parts of their Australian range (e.g., Taylor 1971, Coulson 1993, de Munk 1999, Moseby et al. 2005, Roger et al. 2007), as well as in the native range of the two introduced species (e.g., Fa et al. 1999, Calvete et al. 2004, Odden et al. 2005, Odden and Wegge 2007). However, habitat use by the herbivore assemblage on Yanakie Isthmus has not been studied and provides an opportunity to examine community niche dynamics in a complex herbivore assemblage comprised of introduced and native species. The introduced hog deer (c. 40 kg; Mayze and Moore 1990) is established throughout much of the low-lying shrub-lands of Victoria’s Gippsland plain (Menkhorst 1995a). In their native range, hog deer prefer floodplain grassland associations (Odden et al. 2005) and in Victoria, they are essentially limited to flat, swampy coastal plains (Taylor 1971). In their native range, European rabbits (c. 1.6 kg; Strahan 1995) favour open scrubland (Fa et al. 1999). Similarly, in sub- alpine regions of Australia rabbits occur in woodland associated with open grassy valleys (Williams and Myers 2008). The swamp wallaby (c. 18 kg; Edwards 1969) has a broad distribution across eastern Australia from Cape York in

Chapter 3 Broad scale habitat use 60

Queensland to south-western Victoria and generally lives in thick undergrowth of forests, woodlands and heath (Merchant 2008), although it is capable of utilising a variety of environments (Hollis et al. 1986, Troy et al. 1992). Eastern grey kangaroos (c. 26 kg: ACT, n = 333, G. Coulson, The University of Melbourne, pers. comm.) are limited to the east of Australia, their range extending over Queensland, New South Wales, Victoria and Tasmania (Caughley et al. 1987). Eastern grey kangaroos occupy a variety of habitats including sclerophyll forest, shrubland and heathland (Coulson 2008), but are largely absent from wet forests (Hill 1981b). The common wombat (c. 28 kg; Barboza et al. 1993) occurs in coastal regions of southeast Australia (McIlroy 1973, Mallett and Cooke 1986) and utilises several vegetation types including sclerophyll forest, woodland, coastal scrub, heathland and temperate forest (McIlroy 2008). Understanding herbivore resource partitioning requires knowledge of spatial patterns of habitat use and the environmental factors influencing these patterns. Previous research suggests that habitat use by hog deer, European rabbits, swamp wallabies, eastern grey kangaroos and common wombats is influenced by the availability of food (e.g., Taylor 1984, Lunney and O'Connell 1988, Evans et al. 2006, Odden and Wegge 2007) and shelter (e.g., Mayze and Moore 1990, McIlroy 1995, Carvalho and Gomes 2004, Di Stefano et al. 2007, Schmidt et al. 2010). Further, studies have shown that the food and/or habitat requirements of at least some of these species may overlap, although a degree of ecological partitioning has been recognised (Taylor 1971, de Munk 1999, Schmidt et al. 2010). Yanakie Isthmus offers a good basis for studies of broad scale habitat use among introduced and native species in sympatry, as its five distinct vegetation communities provide spatial heterogeneity crucial for habitat selection (Kotler and Brown 1988). Moreover, low predation rates on Yanakie Isthmus (Menkhorst and Seebeck 1998, Parks Victoria 2003a) mean that it is an ideal location for studies of competition: in the presence of predators, herbivore populations should be held down by predators and therefore, herbivores should not compete (Hairston et al. 1960), whereas in the absence of top-down control by predators, regulation should be bottom-up (Anderson et al. 2010, Hopcraft et al. 2010). I aimed to improve our understanding of large herbivore ecology and organisation in a contemporary assemblage made up of species with disparate

Chapter 3 Broad scale habitat use 61 evolutionary histories. I aimed to do this by quantifying broad scale habitat use within a guild of sympatric native and introduced species, identifying habitat preferences of each species, identifying habitat characteristics with which each species were associated to provide insight into their habitat requirements, and quantifying inter-specific overlap (and thus partitioning) in habitat use, as well as variation in niche breadth, among species. Five general hypotheses emerge:

1. Overlap in habitat use among the five herbivore species should be low. Sympatric species should reduce competition by selecting different resources (Gause 1934).

2. Overlap in habitat use should be greater between native and introduced herbivore species than among native herbivore species. To facilitate coexistence, mechanisms of resource partitioning have evolved in species with common evolutionary histories, while species with independent evolutionary histories have inherently less resource partitioning (Kirchhoff and Larsen 1998, Kelley et al. 2002, Madhusudan 2004).

3. If inter-specific overlap in habitat use is high, temporal (i.e., seasonal) partitioning may be evident. Specialisation on resources along habitat, diet and temporal gradients results in niche differentiation, facilitating coexistence (Schoener 1974b, Whitfield 2002, Schmidt et al. 2010). Niche differentiation is generally complementary; when species are similar on one niche dimension, they differ on another (Pianka 1976, Dunbar 1978, Fox 1989, Bagchi et al. 2003).

4. Habitat use will be influenced by body size: smaller herbivores will be more selective (i.e., have narrower habitat niche breadths) than larger herbivores. Due to increased dietary tolerance among larger species for a wider range of food items, either in terms of nutritional quality or food-item size (May and MacArthur 1972, Schoener 1974b), a wider range of habitat patches and, hence, more even use of environmental resources is expected (duToit and Owen-Smith 1989).

5. Population metabolism will scale positively with body mass.

Chapter 3 Broad scale habitat use 62

The balance between population density and energy use per individual determines how evenly community resources are shared by species of different size (duToit and Owen-Smith 1989). A comprehensive study of the relationships between mammalian population density and body mass by Peters and Raelson (1984) indicates that population metabolism generally scales positively with body mass, reflecting advantages of large body size which may result in resource domination (Brown and Maurer 1986) and greater energy use per individual (duToit and Owen-Smith 1989).

Methods I used the faecal pellet count method (Bailey and Putman 1981) to examine use of the vegetation matrix on Yanakie Isthmus by eastern grey kangaroos, swamp wallabies, common wombats, hog deer and European rabbits. Faecal pellet density is a function of: (i) the number of animals present, (ii) the rate at which these animals defecate, and (iii) the rate at which their pellets decay (Bailey and Putman 1981). Faecal pellet counts can be used to assess habitat use because increased pellet density is expected to be proportional to the time spent in that habitat (Hannan and Whelan 1989). There are known problems with this technique (Taylor and Williams 1956, Robinette et al. 1958, Wallmo et al. 1962, Van Etten and Bennet 1965), for example the assumption of a proportional relationship between pellet density and time spent in a habitat can be violated by non-regular defaecation (e.g., related to activity and diet), and differential abilities to locate pellet groups in different vegetation types (Bailey and Putman 1981). However, a pilot study showed that direct counts are impractical on Yanakie Isthmus due to dense vegetation and low contact rates. The faecal pellet count method is considered appropriate for simultaneously surveying species whose activity patterns and behaviour differ (e.g., Lunney and O'Connell 1988) and avoids many of the detection biases that occur for direct counts in dense vegetation with variable topography (Hill 1981b). In particular, pellet counts are useful when animals are nocturnal or shy (Ellis et al. 1998, Evans and Jarman 1999), or their foraging habitats necessitate ground surveys (e.g., swamp wallabies; de Munk 1999). For these reasons, many studies have used faecal pellet counts to estimate habitat resource use (e.g., Hill 1978, Floyd 1980, Taylor 1985b, Lunney and O'Connell 1988, Ramsey and Engeman 1994, de Munk 1999), and

Chapter 3 Broad scale habitat use 63 studies such as that by Forsyth et al. (2007) have demonstrated positive correlations between herbivore densities and pellet counts. I used faecal standing crop (whereby all pellets on the ground are counted at a given time in plots) because it is less time and resource intensive than faecal accumulation rate counts (Mayle et al. 1999). To provide insight into the habitat requirements of each species, I examined the influence of habitat characteristics on herbivore densities. I did this by examining relationships between herbivore faecal pellet densities and the availability of three resources recognised as major contributors to herbivore habitat choice: forage (e.g., Caughley 1964, Hill 1981c, Duncan 1983, Murden and Risenhoover 1993, Fritz et al. 1996), surface water (e.g., Chamaille-Jammes et al. 2007), and shelter (e.g., Caughley 1964, Hill 1981c, Lunney and O'Connell 1988).

Study site I conducted this study on Yanakie Isthmus (38º 53' S; 146º 14' E), a 6874- ha section of Wilsons Promontory National Park (WPNP), Victoria, Australia. For the purposes of this study I extended the south-east boundary of Yanakie Isthmus to include 3700 ha of Heathy Woodland, Heath and Moist Foothill Forest (dominant vegetation types at WPNP) and excluded Lowland Forest (which comprises only 22 ha of WPNP) to form the 10524-ha ‘Greater Yanakie Isthmus’ (GYI; Figure 1.1). The GYI consists of five vegetation types, Coastal Grassy Woodland (5591 ha), Coastal Scrubs and Grasslands (1425 ha), Heath (720 ha), Heathy Woodland (2551 ha) and Moist Foothill Forest (237 ha), each of which are described in Chapter 1. The climate is mild and precipitation is reliable; between 1985-2005, mean monthly maximum temperatures ranged from 11.9 - 26.4°C and mean minima from 4.4 - 16.7°C, and the mean annual rainfall was 958.5 mm (Parks Victoria, unpublished data). Historically, the two top-level predators at Wilsons Promontory were indigenous people and dingoes Canis lupus dingo (Parks Victoria 2011). Predation rates on Yanakie Isthmus are likely to be low following the removal of indigenous people and native predators, including dingoes and spotted-tailed quolls Dasyurus maculatus (Menkhorst and Seebeck 1998, Parks Victoria 2003a), although eastern quoll Dasyurus viverrinus may be present in low numbers and lace monitors Varanus varius are present (J. Whelan, Parks Victoria, pers.

Chapter 3 Broad scale habitat use 64 comm.). The last dingo at Wilsons Promontory was shot in 1948 and the last confirmed sighting of a spotted-tailed quoll was on Yanakie Isthmus is 1961 (J. Whelan, Parks Victoria, pers. comm.). The Vulpes vulpes is the largest, and together with the cat Felis catus the only, terrestrial mammalian predator present in Wilsons Promontory National Park (Parks Victoria 2003a). The red fox is a predator of small to medium-sized animals (critical weight range 35–5500 grams) (May 1997). The European rabbit has been identified as the preferred prey species of the red fox in Australia (Lunney et al. 1990) and throughout Yanakie Isthmus, rabbits are also controlled by managers (Parks Victoria 2003a) through baiting with pindone (J. Whelan, Parks Victoria, pers. comm.). However, there is a lack of predators of mature wombats, hog deer and eastern grey kangaroos, although foxes may predate young of these species and mortality also occurs from road kill and for hog deer, from illegal hunting (Parks Victoria 2003a).

Survey design To enable efficient sampling in the dense vegetation on GYI, I used restricted random sampling based on the track system: tracks were distributed throughout the study area, and disturbance along tracks was low. Using a 1:50,000 Geocentric Datum of Australia map of Wilsons Promontory National Park (Natural Resources and Environment 2000), I placed potential start points for sampling at 200-m intervals along the length of both sides of each track on GYI. I randomly selected 55 of these potential start points, stratified by broad vegetation type (Department of Natural Resources and Environment 1998) in approximate proportion to area, and adjusted sample allocation to ensure a minimum of six samples per vegetation type (Mayle et al. 1999). To enable stratification of sampling by vegetation type, the broad vegetation types were overlaid from Figure 1.1, a map created by Parks Victoria, onto the 1:50,000 Geocentric Datum of Australia map of Wilsons Promontory National Park. I repeated sampling using this design four times, once each in spring 2003, summer 2003, autumn 2004 and winter 2004, resampling 55 new start points each season. Sampling was conducted seasonally because decisions about food and shelter resource selection vary according to the time of year (Clarke et al. 1989).

Chapter 3 Broad scale habitat use 65

Faecal pellet counts Using the odometer on a 4WD vehicle, I located the start point for each sample, from which I ran a transect on a random compass bearing. If a transect intersected a body of water I added or subtracted 90º from the bearing so that the transect continued in the direction of the largest angle (Forsyth 2005). In each of 10 plots along each transect, I counted the faecal pellet standing crop for each of the five herbivore species. I counted individual faecal pellets for each of the three marsupial species, given their smaller pellet group sizes and difficulties that can arise in distinguishing groups (Southwell 1989). Although Forsyth et al. (2007) demonstrated a positive relationship between deer density and both individual pellets and pellet groups, I counted pellet groups for hog deer and European rabbits, because Smith (1964) showed that the number of pellet groups produced daily is a more reliable index of deer density than the number of individual pellets produced. Following work on red deer Cervus elaphus by Hickling (1986), I defined pellet groups as  6 pellets of the same defaecation, with at least one visible above the ground litter. I applied this definition to both hog deer and rabbits for simplicity, as no rabbit-specific definition was available in the literature. I identified pellets based on size, shape and colour according to Triggs (2003). I systematically searched plots, pushing the vegetation aside (Hickling 1986), but did not disturb the litter except to look for additional deer or rabbit pellets when one was visible above the litter (Hickling 1986). When pellet groups were on the perimeter of plots, I counted them if > 50% of pellets fell inside the plot (Mayle et al. 1999). My sampling units during the first survey in spring 2003 were 200-m transects, each with 10 circular plots (5.64-m radius) placed at 20-m intervals (the first placed 20 m from the track to reduce the potential for bias associated with tracks). During spring I counted deer pellet groups in the 5.64 m radius plot and noted their approximate age as: ‘fresh’ – pellets soft with an intact, dark, slimy outer layer, no evidence of decay and a strong odour, ‘medium’ – pellets firm with an intact outer layer that had dried black or grey, little evidence of decay, or ‘old’ – pellets grey and showing evidence of decay (Mayze and Moore 1990). Initially, my primary interest was in hog deer, so I invested greater effort counting deer pellets in the larger 5.64-m radius plots to increase the probability of detecting

Chapter 3 Broad scale habitat use 66 pellets. In contrast, I counted fresh kangaroo, wallaby and wombat pellets, and rabbit pellet groups within the central 1-m radius only to minimise search time spent on these species of secondary interest. For the summer 2003-04 survey, I improved the efficiency of the survey design by using 100-m transects, each with 10 plots (3-m radius) placed at 10-m intervals (the first 10 m from the track): I counted deer pellet groups of all ages in the 3-m radius plots and counted marsupial pellets and rabbit pellet groups of all ages in the central 1-m radius only. In autumn and winter 2004, I broadened the focus of the project to the entire herbivore assemblage and counted pellets (or pellet groups) of all ages for all species in the 3-m radius, to reduce the frequency of pellet counts that were zero.

Habitat requirements To examine relationships between herbivore densities and habitat requirements, I examined associations between faecal pellet counts and a number of habitat characteristics. I recorded the cover of ground, shrub and canopy layer (to the nearest 10%), as estimates of forage and shelter availability. I also measured the distance to water from each transect using GPS latitude and longitude coordinates, a GIS hydrology layer that included streams, rivers and wetlands, and local knowledge of permanent free water bodies (J. Whelan, Parks Victoria, pers. comm.). To further identify potential habitat requirements, I also recorded cover (to the nearest 10%) of a number of focal plant species thought to be potentially important as food or shelter resources, either due to their relative dominance or scarcity within the vegetation community: (i) native Poa spp., (ii) the overstorey trees B. integrifolia and A. verticillata, (iii) the dominant native shrubs Acacia longifolia and L. laevigatum, and (iv) exotic weeds (Conyza bonariensis and Senecio jacobaea). These measures were made in conjunction with pellet counts. During the spring survey I recorded all vegetation variables within the 5.64 m radius plots. For subsequent surveys, I recorded the cover of ground, shrub and canopy layer (to the nearest 10%) only.

Statistical analyses My analysis of faecal pellet data was divided into two parts. Firstly, I examined habitat preference, niche breadth, inter-specific overlap in habitat use, and population metabolism, based on the population abundance of each herbivore

Chapter 3 Broad scale habitat use 67 species within each vegetation type. Secondly, I examined relationships between herbivore pellet densities and both vegetation type and habitat characteristics. To measure the preference of each herbivore species for each vegetation type available to them, I used the selection index suggested by Savage (1931). I then used the standardisation procedure from Manly et al. (2002) to convert selection index values to standardised ratios, where values equal to 1/ K (where K is the number of resources; 0.2 in this case) indicate no preference, values below this indicate relative avoidance, and values above this indicate relative preference (Krebs 1998). I calculated confidence intervals for these selection ratios following Krebs (1998), corrected using the Bonferroni correction for multiple comparisons. To test whether each species was selecting between the five resource states (vegetation types) at random, I used G-tests as recommended by Manly et al. (2002). I then used chi-squared tests to compare the preference for pairs of vegetation types for which upper and lower 95% confidence intervals for selection ratios did not overlap (Manly et al. 2002). I selected Hulbert’s index for niche overlap to measure overlap in resource use between pairs of species in this community because this index allows resource availability to vary (Krebs 1998). This index gives values of 1 when both species utilise each resource state in proportion to its abundance, 0 when the two species share no resources, and > 1 when the two species both use certain resource states more intensively than others, and their preferences tend to coincide (Krebs 1998). Smith’s measure quantifies the uniformity of distribution of individuals among resource states (Krebs 1998). I chose this standardised measure of niche breadth, varying from 0 to 1 (Krebs 1998), because it takes resource availability into account and is more sensitive to the selectivity of common than rare resources (Smith 1982). I then used niche breadth to examine the relationship between habitat diversity and body size graphically, following du Toit and Owen-Smith (1989). To calculate each of these three measures of habitat use I used the proportional abundance of each herbivore species per vegetation type, estimated by weighting herbivore density within each vegetation type according to its area ˆ (Mayle et al. 1999). To estimate herbivore densities Da , I used mean pellet (or

Chapter 3 Broad scale habitat use 68 pellet group) densities summed across plots within each transect per vegetation type following Laing et al. (2003):

Dˆ s Dˆ a  pˆ tˆ

ˆ ˆ where Ds is the estimated density of pellets or pellet groups, t is the estimated mean time to decay of the pellets or pellet groups and pˆ is the estimated rate of production of pellets or pellet groups per animal. I used species- and habitat- specific pellet or pellet group decay rates from a previous study on Yanakie Isthmus (Chapter 2; Davis et al. in review) and species-specific mean daily defaecation rates from published literature (Table 3.1): 493 eastern grey kangaroo pellets (Johnson et al. 1987); 150 swamp wallaby pellets (Floyd 1980); 90 wombat pellets, the mid-point of the range provided by Triggs (1996); 20.3 hog deer pellet groups (Dhungel 1985a); and 18.5 European rabbit pellet groups, converted from 325 pellets (Wood 1988) based on an estimate of 17.5 ± 0.04 pellets per group on GYI (Chapter 2; Davis et al. in review). I selected defaecation rates measured in habitats most similar to GYI, and sought studies of free-ranging animals in preference to those using captive animals. However, for hog deer, the only data available was for captive animals and it was unclear whether data for swamp wallaby, common wombat and European rabbit defaecation rates were from captive or free-ranging animals. For the common wombat, the only estimate of defaecation available was a range, so I used the mid-point of this range. Where estimates of the error associated with defaecation rates were not provided, I assumed standard errors of ± 20% to approximate the error observed in other studies (e.g., Hill 1978). I estimated the precision of density estimates following Laing et al. (2003): ˆ 2 ˆ 2 ˆ 2 2 [cv(Da )]  [cv(Ds )] [cv(t)] [cv(pˆ)] where cvtˆis the coefficient of variation of , and similarly for the other terms. I then used coefficient of variation values to calculate 95% confidence intervals for density estimates. I estimated population abundance for each herbivore species in each broad vegetation type and for the entire area of GYI by weighting herbivore density estimates within each vegetation type according to its area (Mayle et al. 1999). For estimates of habitat preference, niche breadth and inter-specific overlap in

Chapter 3 Broad scale habitat use 69 habitat use, I based estimates of density on pellet count data collected during autumn and winter, because these were the two seasons when pellet count methodology was constant for all five species, thus providing good comparative data. However, to ensure overall estimates density, abundance and population metabolic demand (see below) were as accurate as possible (i.e., represented as many seasons as possible), I based estimates for hog deer on pellet count data collected during all four seasons, and for the other four species I used data collected during all seasons except spring data (because only fresh pellets were counted during spring, due to the initial focus of the study on hog deer). I also used estimates of herbivore abundance to estimate the percentage of herbivores in Coastal Grassy Woodland that have predominantly browser diets on Yanakie Isthmus (swamp wallabies and hog deer, Chapter 5; Davis et al. 2008) and the percentage that have predominantly grazer diets (eastern grey kangaroos, common wombats and European rabbits, Chapter 5; Davis et al. 2008). Metabolic rate can be used to predict energy and nutrient requirements (Hume 1999). Following Schaller (1967), I converted estimates of density for each herbivore species within Coastal Grassy Woodland and over the entire area of GYI, to estimates of metabolic demand, as a more meaningful indicator of their relative resource use than relative abundance. Du Toit and Owen-Smith (1989) define population metabolism as the product of population density and energy use per individual. I calculated population metabolism for the entire GYI as the product of mean population abundance for GYI and basal metabolic rate per individual. There are several measures of metabolic rate (Hume 1999). Field metabolic rate is the energy cost of free existence (Hume and Stevens 1995) and provides the most realistic estimate of the actual energetic needs of an animal (Hume 1999). However, field metabolic rates include energetic costs associated with factors such as growth and reproduction and can be highly variable between sexes and seasons (Hume 1999). Therefore, inter-specific comparisons are commonly made using Basal Metabolic Rate (BMR) (Hume 1999), which is the minimum rate of metabolism compatible with endothermy (McNab 1988) and correlates strongly with the energy required for longer term energy balance (Hume and Stevens 1995). I calculated BMR following Hume (1999): BMR = aMb

Chapter 3 Broad scale habitat use 70 where a is a proportionality coefficient that differs among species, and b is an empirically determined exponent that expresses the rate of change of R with changes in body mass (M). The estimates of a and b that I used were from Dawson and Hulbert’s (1970) ‘marsupial line’ and Kleiber’s (1961) ‘Kleiber line’, which yield BMRs for the average marsupial of 201 kJ kg-0.75 day-1 and for the average eutherian of 293 kJ kg-0.75 day-1, respectively. These traditional values are based on small sample sizes, but are widely used for comparative purposes (Hume 1999). Dawson and Hulbert’s (1970) marsupial line may overestimate metabolic demand for wombats, as species such as the southern hairy-nosed wombat Lasiorhinus latifrons have low metabolic rates relative to other marsupials (Wells and Green 1998). However, work by Gowland (1973) suggests that the standard metabolic rate of common wombats is near the marsupial average. I obtained weights for each species from the literature (Table 3.1). For the rabbit, I took the mid-point between the weights given for males and females by Myers (1983). After du Toit and Owen-Smith (1989), I then assessed the relationship between population metabolism and body mass according to the least-squares-regression equation: log E = a + b(log M) where E is population metabolism and M is body mass. I used generalised linear modelling (Hilborn and Mangel 1997) to examine relationships between herbivore faecal pellet densities and both vegetation type and habitat characteristics. The response variable I modelled was the number of pellets (for the three marsupial species) or pellet groups (for hog deer and rabbits) per plot: all pellet and pellet group counts included pellets of all ages, except the non-deer counts in spring, which were fresh pellets only. For inclusion in the model, I selected seven explanatory variables that I considered likely to have an influence on pellet counts, based on published findings on habitat requirements: (1) broad vegetation type, (2) % ground cover, (3) % shrub cover, (4) % canopy cover, (5) distance to water, (6) cover of L. laevigatum in the ground, shrub and canopy layers, and (7) cover of A. longifolia, A. verticillata and B. integrifolia in the canopy layer. In addition, because decisions about food and shelter resource selection vary according to the time of year (Clarke et al. 1989), I included season in the model.

Chapter 3 Broad scale habitat use 71

I fitted all of the variables in a separate model for each species: I made no attempt to reduce the model on grounds of parsimony. Because individual pellets are unlikely to arise from independent events, a model with ‘pure’ Poisson variation was unlikely to be realistic for these data (e.g., if a single animal deposits several pellets in one defaecation event, this would cause clustering, and more variation than would be expected from a Poisson model). A common way to deal with this ‘extra-Poisson’ variation is to model these data as having a Poisson- like structure, but with more variation than that implied by the Poisson distribution. This can be achieved by using a generalised linear model for the mean and modelling the variance as this mean multiplied by an over-dispersion factor estimated from the data. I used a log link function to model the data. To account for differing plot sizes I used an ‘offset’ term equal to the area, with the implication that a count from a circle is proportional to the area of the circle. The estimates given by the model were rate ratios: the ratio of the pellet density at one level of the explanatory variable, relative to another specified level. For categorical explanatory variables, this was simply a comparison between variables. For numerical explanatory variables, it was the rate ratio for a change (increase) in one unit (or a more meaningful unit of increment, based on examination of the distributions of the variables).

Results Habitat preference G-tests, based on estimates of herbivore abundance during autumn and winter, indicate that selection between the five vegetation types was not random for any species (Table 3.2). Habitat selection indices suggested that deer preferred Coastal Scrubs and Grasslands and Heath, but avoided Coastal Grassy Woodland, Heathy Woodland and Moist Foothill Forest (Table 3.3). Rabbits preferred Coastal Grassy Woodland and Coastal Scrubs and Grasslands, but avoided Heath, Heathy Woodland and Moist Foothill Forest. Kangaroos preferred Coastal Scrubs and Grasslands, but avoided the other four vegetation types. Wallabies and wombats preferred Heath and Heathy Woodland; wallabies avoided Coastal Grassy Woodland, Coastal Scrubs and Grasslands and Moist Foothill Forest, while wombats avoided Coastal Grassy Woodland and Moist Foothill Forest, and had no preference for or against Coastal Scrubs and Grasslands.

Chapter 3 Broad scale habitat use 72

In particular, Chi-squared tests showed that habitat selection indices for Coastal Grassy Woodland and Heathy Woodland differed significantly for all species except kangaroos and hog deer (Table 3.2). The only species for which selection indices differed significantly between Coastal Grassy Woodland and Moist Foothill Forest was the European rabbit. Selection indices differed significantly between Coastal Scrubs and Grassland and both Coastal Grassy Woodland and Heathy Woodland for kangaroos and rabbits. Selection indices for wombats differed significantly between Heath and both Coastal Grassy Woodland and Moist Foothill Forest, and between Heathy Woodland and Moist Foothill Forest.

Niche breadth and overlap Values for Hulberts’ index of niche overlap between rabbits and each of the other species (deer, kangaroo, wallaby and wombat) were relatively small, as was the value for niche overlap between the kangaroo and wallaby, indicating that these pairs of species shared relatively few habitats (Table 3.4). The hog deer and kangaroo, and the wallaby and wombat had niche overlap values greater than one, indicating that these pairs of species used certain resource states more intensively than others, and their preferences for resources tended to coincide. Niche overlap values between the deer and both the wallaby and wombat, and also between the kangaroo and wombat, were close to one. This indicates that these pairs of species used each habitat resource state in proportion to its abundance, although in each case, values were just over one, indicating some overlap in vegetation type preferences. Smith’s measure of niche breadth, based on autumn and winter pellet counts in five vegetation types, was narrowest for the kangaroo, intermediate for the deer and rabbit, and broadest for the wallaby and wombat (Table 3.5). There was no apparent relationship between habitat diversity and body size (Figure 3.3).

Herbivore density, population abundance and population metabolism The total abundance of all five species of mammalian herbivore in combination on GYI, estimated during spring 2003 to winter 2004 for hog deer, and during summer 2003-4 and autumn and winter 2004 for the other four species, was 3,692 animals (95% CI, 1,409-5,974), equating to a density of 35 animals per

Chapter 3 Broad scale habitat use 73 km2 (95% CI, 13-57). The greatest density of herbivores occurred in Coastal Scrubs and Grasslands (Figure 3.1). The rank order for the abundance of herbivore species on GYI was: common wombat > European rabbit > swamp wallaby > hog deer > eastern grey kangaroo (Table 3.1; Figure 3.2). Within the dominant vegetation type, Coastal Grassy Woodland, the rank order for the abundance of herbivore species was: rabbit > wombat > wallaby > deer > kangaroo. Herbivores with predominantly browser diets on Yanakie Isthmus (wallabies and deer, Chapter 5; Davis et al. 2008) comprised 15% of the total herbivore population in Coastal Grassy Woodland, while those with predominantly grazer diets (kangaroos, wombats and rabbits, Chapter 5; Davis et al. 2008) comprised 85% of the herbivore population. Daily basal metabolic rates varied substantially between herbivore species, being lowest for rabbits, highest for deer, and intermediate and similar for the three marsupial species (Table 3.1). The relative metabolic demand by each population of herbivores on GYI varied considerably from relative abundances (Figure 3.2). Although the relative abundance of rabbits was high, metabolic demand at the population level was relatively low. Although small to intermediate in size and BMR, metabolic demand of wallabies and kangaroos was also relatively low due to their low abundance. Metabolic demand at the population level was highest for wombats, reflecting their greater biomass. Metabolic demand at the population level was intermediate for deer, despite their relatively low abundance, due to relatively high biomass and BMR. Similarly, the relative metabolic demand by each population of herbivores within Coastal Grassy Woodland on GYI varied considerably from relative abundances (Figure 3.2). Within Coastal Grassy Woodland, the rank order for metabolic demand at the population level was: wombat > deer > rabbit> wallaby > kangaroo. Population metabolism increases approximately in relation to M0.77 (log E = -3.42 + 0.77(log M); n = 5; r = 0.50). Thus, a 10-fold increase in species body mass would be associated with an approximately six-fold increase in resource use.

Relationships between herbivore faecal pellet density and vegetation type Generalised linear modelling (GLM) showed that herbivore pellet (or pellet group) densities varied significantly among vegetation types and seasons for all species except the rabbit (Tables 3.6-3.10). However, no rabbit pellet groups

Chapter 3 Broad scale habitat use 74 were recorded in Heath, making pairwise comparisons of rabbit pellet densities between Heath and other vegetation types meaningless (Table 3.6). Hog deer pellet group densities in Coastal Grassy Woodland, Coastal Scrubs and Grasslands, and Heath were estimated to be 18, 20 and 17 times greater respectively, than in Heathy Woodland (Table 3.7). Deer pellet group density in Heathy Woodland was also estimated to be 7% of the pellet group density in Moist Foothill Forest. Eastern grey kangaroo pellet density in Coastal Grassy Woodland was estimated to be ~2% of that in Coastal Scrubs and Grasslands, where pellet density was ~19 times greater than in Heathy Woodland (Table 3.8). No kangaroo pellets were recorded in Heath or Moist Foothill Forest. Swamp wallaby pellet density in Coastal Scrubs and Grasslands was estimated to be ~35% of that in Heathy Woodland (Table 3.9). Wallaby pellet density in Heathy Woodland was estimated to be ~3 times greater than in Moist Foothill Forest. Common wombat pellet densities in both Coastal Grassy Woodland and Heathy Woodland were estimated to be ~5 times greater than wombat pellet densities in Moist Foothill Forest, and pellet densities in Heath were estimated to be 9 times greater than in Moist Foothill Forest (Table 3.10). Wombat pellet density in Coastal Scrubs and Grasslands was estimated to be 30% of the pellet density in Heath.

Relationships between herbivore faecal pellet density and habitat requirements GLM showed significant relationships between herbivore pellet (or pellet group) densities and vegetation cover in particular structural layers, as well as the cover of encroaching native shrubs, but no relationships were detected between pellet densities and cover of overstorey tree species (Tables 3.6-3.10). Hog deer pellet group density was estimated to increase by ~20% with 10% increases in total ground cover, and Acacia longifolia canopy cover, but to decrease by 11% with a 10% increase in shrub cover of Leptospermum laevigatum (Table 3.7). European rabbit pellet group density was estimated to decrease by ~30% with 10% increases in shrub cover, but to increase by ~20% with a 10% increase in L. laevigatum shrub cover (Table 3.6). Eastern grey kangaroo pellet density was estimated to decrease by ~30% with a 10% increase in canopy cover of L. laevigatum (Table 3.8). Swamp wallaby pellet density was estimated to decrease by ~15% with 10% increases in canopy cover of A. longifolia (Table 3.9).

Chapter 3 Broad scale habitat use 75

Common wombat pellet density was estimated to increase by 16% with a 10% increase in shrub cover, and more specifically, to increase by 10% with a 10% increase in shrub cover of L. laevigatum (Table 3.10). Significant relationships between pellet (or pellet group) densities and distance from water were detected for hog deer and eastern grey kangaroos, but not for common wombats, swamp wallabies or European rabbit (Table 3.6-3.10). With each 500 m increase in distance from water, deer pellet group density was estimated to decrease by 32% (Table 3.7), and kangaroo pellet density to increase by ~2 times (Table 3.8).

Discussion This study utilises faecal pellet counts to assess habitat use by five herbivore species in sympatry on Greater Yanakie Isthmus. Faecal pellet counts can be used to assess habitat use if animals defaecate randomly and indiscriminately with respect to space and time (Bailey and Putman 1981), because increased pellet density is expected to be proportional to the time spent in that habitat (Hannan and Whelan 1989). However, the assumption of a proportional relationship between pellet density and time spent in a habitat can be violated by factors such as non-regular defaecation (e.g., related to activity and diet), and differential abilities to locate pellet groups in different vegetation types (Bailey and Putman 1981). Both wombats (Triggs 1996) and rabbits (Fa et al. 1999) tend to deposit faecal pellets throughout their home range. However, macropods defaecate mainly while feeding and defaecate little while resting (Hill 1978), thus swamp wallaby (Ramsey and Engeman 1994) and eastern grey kangaroo faecal pellets are most concentrated at feeding sites, despite defaecation throughout the day (Johnson et al. 1987). For this reason faecal pellet counts may overestimate the utilisation and importance of feeding habitats, although pellet distributions are considered to provide a reliable indicator of preferred feeding zones for macropods (Hill 1981c). Therefore, although the results of this study should be interpreted with caution, they provide valuable preliminary insights into large herbivore ecology and organisation in a contemporary assemblage made up of species with disparate evolutionary histories.

Chapter 3 Broad scale habitat use 76

Broad scale habitat use: habitat preferences and requirements On Greater Yanakie Isthmus, hog deer, eastern grey kangaroos, swamp wallabies, common wombats and European rabbits used habitats selectively (e.g., Senft et al. 1985, Prins and Olff 1998). Habitat use on GYI broadly reflected patterns of habitat use described elsewhere for these native and introduced herbivore species, although in some cases, relationships between herbivore density and the availability of water, food and shelter resources diverged from those previously presented in the literature. In their native range, hog deer prefer floodplain grassland associations, which provide both food and shelter (Odden et al. 2005). In Victoria, hog deer are essentially limited to flat, swampy coastal plains (Taylor 1971). Hog deer occurred in all five vegetation types on GYI, but showed a preference for Heath and Coastal Scrubs and Grasslands. On GYI, Heath flanks Darby River, and preferential use of this habitat supports the suggestion by Mayze and Moore (1990) that wetland adjacent to scrubland provides the mix of food and cover required by hog deer. The other vegetation type used preferentially by hog deer on GYI, Coastal Scrubs and Grasslands, is also likely to provide adjacent food-cover associations as it typically consists of scrubland interspersed by patches of grassland. When predation pressure is low (Odden et al. 2005), as it is likely to be on GYI (Menkhorst and Seebeck 1998, Parks Victoria 2003a), habitat selection by hog deer is likely to be influenced more by forage availability than by the availability of shelter. In line with this, hog deer showed a positive association with ground cover. Therefore, despite the potential of Coastal Grassy Woodland and Heathy Woodland to provide shelter from predation and weather (Mayze and Moore 1990, Menkhorst 1995a), these vegetation types are probably avoided due to low ground cover of grasses and herbs (Davies and Oates 1999, University of Ballarat 1999). Moist Foothill Forest may be avoided by hog deer because it occurs at higher elevations (>100 m; Parks Victoria 2003b) than used by hog deer (Mayze and Moore 1990). The availability of surface water limits the distribution of hog deer in Nepal (Dinerstein 1979). Similarly, hog deer pellet group density on GYI decreased as the distance from water increased, supporting the assertion by Mayze and Moore (1990) that the availability of surface water limits hog deer populations in Victoria.

Chapter 3 Broad scale habitat use 77

European rabbits on GYI showed a particularly strong preference for Coastal Grassy Woodland and also used Coastal Scrubs and Grasslands preferentially. In contrast, rabbits avoided Heathy Woodland and Moist Foothill Forest, and did not occur in Heath. These observations mirror habitat use by European rabbits in their native range, where they favour open scrubland (Fa et al. 1999). Similarly, in sub-alpine regions of Australia rabbits occur in woodland associated with open grassy valleys (Williams and Myers 2008). Habitat selection by European rabbits is a compromise between food availability and protection from predators (Moreno et al. 1996). Although the cover of favoured grasses and forbs (Martin et al. 2007) is low in Coastal Grassy Woodland and Coastal Scrubs and Grasslands on Yanakie Isthmus (University of Ballarat 1999), it is likely to be greater in these habitats than in Heathy Woodland, where the ground layer is dominated by sedges, or in Moist Foothill Forest, where the ground layer is commonly ferny (Davies and Oates 1999). However, this does not explain the avoidance of Heath by rabbits and it is likely that factors not addressed during this study influence use of this habitat type. On Yanakie Isthmus, high shrub densities enable rabbits to shelter above ground instead of using burrows (Parks Victoria 2003a), yet rabbit pellet group density decreased as shrub cover increased. Rabbits may prefer intermediate levels of cover (Carvalho and Gomes 2004) when predation pressure is low (Palomares and Delibes 1997). Although rabbits are the favoured prey of red foxes Vulpes vulpes (Lunney et al. 1990), foxes are controlled on GYI and predation rates are probably low (Parks Victoria 2003a). Further, park managers use baiting rather than shooting to control rabbits on GYI. Therefore, the availability of shelter from predation is unlikely to be a strong driver of habitat use by rabbits on GYI. However, the influence of predation on habitat use is unknown and may vary spatially and temporally if predation is not uniformly distributed. Rabbits may also prefer intermediate levels of cover when dense shrub cover hinders movement (Fa et al. 1999) or reduces ground layer cover, as occurs on GYI (Bennett 1994). Moreover, the interspersion of open and closed areas may be more important for habitat selection by rabbits than the level of shrub cover (Carvalho and Gomes 2004, Fernández 2005). There was no relationship between distance to water and rabbit pellet group density, probably because they rarely need to visit water to drink if the moisture content of food is high (Williams and Myers 2008).

Chapter 3 Broad scale habitat use 78

Eastern grey kangaroos occupy a variety of habitats including sclerophyll forest, shrubland and heathland (Coulson 2008). They show a preference for grassland and woodland habitats (McCullough and McCullough 2000), but are largely absent from wet forests (Hill 1981b). Eastern grey kangaroos require habitats with a mix of open grassy areas for foraging and more densely vegetated areas offering lateral cover for shelter while resting, for example, open woodland and forest/pasture ecotones (McCullough and McCullough 2000, Moore et al. 2002, Schmidt et al. 2010). On GYI, eastern grey kangaroos preferred Coastal Scrubs and Grasslands, but avoided the other four vegetation types. Information on the relative availability of food resources in different habitat types is not available. However, selective use of Coastal Scrubs and Grasslands by kangaroos suggests that grass availability and quality, strong predictors of habitat use by kangaroos (Hill 1982, Taylor 1984), are likely to be relatively high in this vegetation type compared to other vegetation types on GYI, particularly Heath and Moist Foothill Forest, where no kangaroos were recorded. Furthermore, eastern grey kangaroos can utilise selected forage in woodland if it is present (de Munk 1999). The scrubland formation of Coastal Scrubs and Grasslands is relatively heterogeneous, and thus likely to be better suited to supporting eastern grey kangaroo populations than more homogeneous habitats (Coulson 1993), for example contiguous woodland (Beale 1973). In particular, shrub cover in Coastal Scrubs and Grasslands is interspersed with grassy patches, thus providing shelter, another key element of favoured eastern grey kangaroo habitat (Hill 1981b). In contrast, vegetation types such as Heathy Woodland were probably avoided by kangaroos due to the scarcity of adequate open grassy patches. Eastern grey kangaroos avoided Coastal Grassy Woodland, possibly due to inhibition of movement by dense shrubby vegetation (Taylor 1980, Bennett 1994), and low grass cover (University of Ballarat 1999). Similarly, Schmidt et al. (2010) found that eastern grey kangaroos avoided grassy woodland. Eastern grey kangaroo pellet density increased with increasing distance from water. Native marsupials have low water turnover rates (Denny and Dawson 1975), but this does not explain the observed trend, as it would result in eastern grey kangaroo faecal pellet densities independent of distance from water. It is possible that predation risk is greater close to water sources, resulting in avoidance of these areas by kangaroos.

Chapter 3 Broad scale habitat use 79

Swamp wallabies are able to utilise diverse community structures and plant associations (de Munk 1999), but generally live in thick undergrowth of forests, woodlands and heath (Merchant 2008). Indeed, local distribution of swamp wallabies appears to be determined by the availability of dense vegetation (Merchant 2008), which provides both food and cover (e.g., Floyd 1980, Lunney and O'Connell 1988), although they commonly forage along ecotones between forest and grassland (Hume 1999) and will utilise open pasture (Edwards 1969). On GYI, swamp wallabies preferred Heath and Heathy Woodland, but avoided Coastal Grassy Woodland, Coastal Scrubs and Grasslands and Moist Foothill Forest. No relationship was detected between swamp wallaby pellet density and shrub cover, suggesting that swamp wallabies are responding to more complex factors such as shrub palatability (e.g., as observed for browsers in an African savanna woodland; Fritz et al. 1996), or shrub diversity. Their use of Heath and Heathy Woodland is consistent with their ability to obtain forage from contiguous woodland (de Munk 1999). The common wombat occurs in several vegetation types including sclerophyll forest, woodland, coastal scrub, heathland and temperate forest (McIlroy 2008), preferentially foraging in habitats with abundant high quality grass (Evans et al. 2006). On GYI, common wombats preferred Heath and Heathy Woodland where grasses are uncommon, although both vegetation types contain sedges (Davies and Oates 1999), which wombats will consume if grasses are scarce (Rishworth et al. 1995). Wombats showed strong avoidance of Moist Foothill Forest, probably due to reduced growth of grasses under conditions of low light (e.g., McIlroy 1995). Similarly, avoidance of Coastal Grassy Woodland by wombats, and their lack of preference for or against Coastal Scrubs and Grassland, may be explained by low grass cover in these habitats (University of Ballarat 1999), although the preference of rabbits for these habitats suggests otherwise, as do vegetation descriptions (Davies and Oates 1999); forage availability needs to be quantified to clarify such possibilities. Alternatively, wombat habitat preferences may reflect the suitability of habitats for burrowing (McIlroy 1995). The use of riparian zones by wombats for burrowing (Skerratt et al. 2004, McIlroy 2008) has been used to explain positive associations between common wombats and watercourses (Roger et al. 2007, Roger and Ramp 2009). However, there was no relationship between wombat pellet density and distance

Chapter 3 Broad scale habitat use 80 from water on GYI. It may be that characteristics of the riparian zones on GYI render them unsuitable for burrowing. In addition to broad differences in the vegetation types used by each herbivore species, examination of similarities and differences in habitat characteristics with which species were associated provides insight into the potential for resource partitioning and competition within the herbivore assemblage on GYI. In particular, relationships between herbivore densities and cover of the dominant shrub, L. laevigatum, illustrate inter-specific difference in the use of habitats on GYI. Hog deer can utilise dense vegetation (Odden et al. 2005), including Leptospermum scrub (Mayze and Moore 1990). However, deer pellet group density on GYI was estimated to decrease with increases in L. laevigatum shrub cover, possibly due to suppression of other plant species by L. laevigatum (Bennett 1994) which may be favoured food species, and conversely, the likely low palatability of L. laevigatum (Clarke 2002). Similarly, eastern grey kangaroo pellet density decreased as L. laevigatum canopy cover increased. Herbage yields are inversely proportional to tree density (Beale 1973), yet there was no relationship between total canopy cover and kangaroo pellet density, indicating that kangaroos are responding to a reduction in ground cover specific to L. laevigatum scrub (Bennett 1994). These results are not surprising, given that eastern grey kangaroos favour open habitats with a ground layer of grass and herbs, but few shrubs (1978). Kangaroos defecate little while resting (Johnson et al. 1987), making it difficult to comment on their potential use of L. laevigatum for shelter, although this is unlikely as dense L. laevigatum can impede herbivore movement (Bennett 1994). Similarities in the responses of deer and kangaroos to L. laevigatum cover in the shrub/canopy stratum indicates similar food and/or shelter requirements of these two species, therefore suggesting the potential for inter-specific competition for habitat resources. A similar scenario could be true of wombats and rabbits, as both species responded positively to an increase in shrub cover of L. laevigatum, indicating preferential use of dense L. laevigatum scrub for shelter by both species. However, rabbits and wombats displayed opposite relationships to total shrub cover: wombat pellet density was positively associated with shrub cover and rabbit pellet density was negatively associated with shrub cover. This difference is likely to be due to differences in diet between

Chapter 3 Broad scale habitat use 81 these species, for example, palatable grasses preferred by rabbits (Leigh 1989) may be less abundant under shrub cover. Swamp wallabies and deer showed opposite responses to canopy cover of another dominant shrub species, A. longifolia. Wallaby pellet density decreased with increases in A. longifolia canopy cover, which was surprising, given that swamp wallaby habitat selection is positively correlated with lateral cover (Di Stefano et al. 2009) and wallabies consume A. longifolia on Yanakie Isthmus (Chapter 5; Davis et al. 2008). However, swamp wallaby habitat selection is also influenced by forage quality (Di Stefano et al. 2009) and the negative relationship with A. longifolia could be explained by its suppression of other more nutritious forage plants (Costello et al. 2000). In contrast, hog deer pellet group density increased with increasing A. longifolia canopy cover, which may reflect a positive association between hog deer and one of their food sources on Yanakie Isthmus (Chapter 5; Davis et al. 2008), as well as shelter. The differential associations of hog deer and swamp wallabies with dominant habitat features suggest differential food and/or shelter requirements and thus a low likelihood of competition between these species. The relationships detected between herbivore densities and habitat characteristics (and thus presumably habitat requirements) may in reality be more pronounced than suggested here, because biomass and food quality may be more meaningful measures of food supply than vegetation cover (Carvalho and Gomes 2004), and there may be relationships with vegetation attributes such as habitat interspersion (Fernández 2005) and grass cover (Catling and Burt 1995), which could not be modelled during this study due to few records. For example, increased grass cover and interspersion of open and closed habitat would likely favour rabbits (Carvalho and Gomes 2004, Fernández 2005, Cabrera-Rodriguez 2006), wombats (Evans et al. 2006, Roger et al. 2007), kangaroos (Calaby 1966) and deer (Mayze and Moore 1990, Odden et al. 2005), despite the lack of relationship between ground cover and pellet density for kangaroos, wombats or rabbits. Similarly, no relationships were detected between pellet densities and canopy cover, contradicting Catling and Burt (1995), who found that wombat density was related to canopy cover. Such contradictions indicate that more research is needed to clarify habitat requirements of the herbivore species on GYI.

Chapter 3 Broad scale habitat use 82

The above explanations for habitat use focus on the suitability of each vegetation type for providing the food and shelter requirements of each species. There are also other ecological requirements not considered during this study, such as for reproduction, which can contribute to habitat use choices (de Munk 1999). However, habitat use is not necessarily correlated with resource quality (Garshelis 2000): competing species have an interacting effect on habitat availability and hence selection (Ovadia and Abramsky 1995). Species may use similar habitats when they occur separately but show complete segregation when they occur in sympatry (Lawlor and Maynard Smith 1976, Werner 1977). One species may competitively exclude others from mutually preferred habitat (e.g., Fritz et al. 1996, Wegge et al. 2006), while the other behaves opportunistically, giving the illusion of ‘preference’ for suboptimal habitat (Garshelis 2000). On GYI, no clear shifts in habitat use from preferred habitats to suboptimal habitats were apparent, given general similarities between habitat use by herbivore species on GYI and other locations in their Australian ranges, and in the native ranges of the two introduced species. However, examining niche parameters provides further insight into potential inter-specific interactions within this community.

Inter-specific overlap in habitat use Inter-specific interactions can have a strong influence on habitat use in mixed-herbivore communities (Fritz et al. 1996). Theory suggests that sympatric species should reduce competition by selecting different resources (Gause 1934). As predicted (hypothesis 1), inter-specific overlap in habitat use by the herbivore assemblage on GYI was generally low, suggesting that partitioning of habitat resources occurs within this community (e.g., Jarman 1972, Madhusudan 2004). This suggestion is supported by generally varied habitat preferences among the five species, despite some coinciding preferences. These results reflect patterns of habitat use observed in many other multi-species herbivore assemblages globally, for example, large ungulates in Kanha Park, India, occupy essentially the same environment, but each is adapted to certain habitat conditions which ecologically separate it to some extent from other species (Schaller 1967) and habitat segregation between wild herbivore species has been widely demonstrated in Africa (Lamprey 1963, Jarman 1972).

Chapter 3 Broad scale habitat use 83

Low overlap in habitat use between some species on GYI probably reflects differences in habitat requirements and therefore low probabilities of competition (Schoener 1983). For example, contrasting dietary requirements of eastern grey kangaroos and swamp wallabies (Jarman and Phillips 1989) are likely to explain their differential use of habitats (Taylor 1971). However, low overlap may result from aggressive exclusion, among other things (Schmidt et al. 2010). It is possible that low overlap in habitat use between the European rabbit and sympatric herbivore species indicates competitive interactions. This possibility was particularly evident in Coastal Grassy Woodland, which was used preferentially by rabbits, but avoided by all other herbivore species. Avoidance of Coastal Grassy Woodland by browsers such as the swamp wallaby is not surprising, as browsers have been observed to avoid vegetation communities dominated by shrubs of low palatability (e.g., Fritz et al. 1996). However, it is likely that grass availability was high in Coastal Grassy Woodland relative to other habitats on GYI (Davies and Oates 1999). Therefore, preferential use of this habitat by only one of the grazing species, the rabbit, and avoidance of this habitat by the other three grazers, as well as low overlap in habitat use between these species in general, may represent competitive exclusion (MacArthur and Levins 1967, Abrams 1983). In particular, the two native grazers may be being excluded by rabbits, as they commonly use similar habitats (e.g., Dierenfeld 1984, Cooke 1998) and food (Chapter 5; Davis et al. 2008) to rabbits. Therefore, this may be an example of adaptation of foraging behaviour by subordinate species forced to switch from ‘preferred’ to 'refuge' habitats due to competition for food and/or habitat resources, possibly due to their inability to use the grazing resource in the preferred habitat as efficiently as the dominant species (Bell 1970, Illius and Gordon 1987). High overlap in habitat use was evident between some species, indicating a high potential for competition between these species if resources are limiting. However, if species do not exclude one another from common habitats, it is often assumed that competition between them is minimal due to the evolution of mechanisms of coexistence along other dimensions to reduce niche overlap in response to high spatial overlap (Schoener 1983). For example, swamp wallabies and eastern grey kangaroos are able to coexist in woodland environments because they occupy different niches: habitat used for shelter by kangaroos will be used

Chapter 3 Broad scale habitat use 84 for browsing by wallabies, and although both may forage in the same open areas (Taylor 1985a, Troy and Coulson 1993), their diets differ considerable (Sanson 1978). Niche overlap can also be reduced by temporal variation in the use of habitats by different species, which would not have been detected using pellet counts. Alternatively, high overlap in habitat use may indicate low levels of competition between ecologically similar species due to resource abundance (e.g., if herbivore densities are below the carrying capacity of the habitat). Habitat use was high between swamp wallabies and wombats, which probably indicates an ability of these species to coexist due to strong differences in food use: swamp wallabies are browsers (Claridge 2001, Di Stefano and Newell 2008), whereas common wombats are grazers (Rishworth et al. 1995, Hume 1999). This mechanism is consistent with findings in other herbivore communities consisting of species with low dietary niche overlap (e.g., Fritz et al. 1996). Overlap in habitat use was also high between hog deer and kangaroos, and in this case competition is more likely than between wombats and wallabies. Hog deer and kangaroos both demonstrated a preference for Coastal Scrubs and Grasslands, as did rabbits, and this habitat type supported the greatest density of herbivores, suggesting that herbivores may compete for this habitat resource. Taylor (1971) suggested overlapping habitat preferences of kangaroos and hog deer were distinct enough to minimize food competition, but predicted competition under conditions of resource limitation. High spatial overlap on GYI between grazing kangaroos (Sanson 1978) and hog deer (Taylor 1971, Wegge et al. 2006) indicates a high potential for inter-specific competition between these two species if resources are limited on GYI, contrasting the clear ecological separation between hog deer and sympatric herbivores in their native range due to inter-specific variation in spatial and/or temporal patterns of habitat use (Dinerstein 1979). As discussed, mechanisms of resource partitioning are likely to have evolved to facilitate coexistence among species with common evolutionary histories. However, species with independent evolutionary histories have inherently less resource partitioning (Kirchhoff and Larsen 1998, Kelley et al. 2002, Madhusudan 2004). Therefore, it was predicted (hypothesis 2) that overlap in habitat use would be greater between introduced and native herbivore species, than between native herbivore species (Kelley et al. 2002). However as outlined

Chapter 3 Broad scale habitat use 85 above, there was high overlap in habitat use between one pair of native species (the swamp wallaby and wombat), as well as between an introduced and a native species (the hog deer and kangaroo), suggesting that evolutionary history does not govern resource partitioning within this community. It is possible that in the case of ecologically similar native and introduced species, adaptation resulting in niche adjustment by one or both species may have occurred in response to competition within the evolutionary timeframe for which this assemblage has coexisted, as it is possible for invaders to evolve rapidly in response to novel abiotic and biotic conditions, and for native species to evolve in response to invasions (Sakai et al. 2001). For example, sympatric native and introduced cervids in North America have evolved patterns of habitat use that minimise inter-specific competition (Hanley and Hanley 1982, Hanley 1984). It is likely that food resource limitation (Madhusudan 2004) has a strong influence on the spatial patterns of resource partitioning observed on GYI, however, it was beyond the scope of this study to quantify these effects. Moreover, some introduced species become established not because they competitively exclude native species from shared resources, but because they are able to occupy a different niche to native species. Thus native and introduced species may not always show high levels of overlap in resource use. For example, in Australia spatial overlap between native red kangaroos Macropus rufus and introduced cattle Bos taurus is low except during drought (Dudzinski et al. 1982), and in Tanzania, introduced cattle Bos indicus select different feeding sites to native ungulates (Voeten and Prins 1999). The approach taken during this study could not detect trends in competition between native and introduced species with certainty. For example, introduced rabbits may compete strongly with native herbivores (as well as with hog deer, with which they also have no shared evolutionary history), but competition results in competitive displacement, resulting in low spatial overlap, and thus apparently low or ambiguous levels of competition between these native and introduced species. Indeed, this suggestion is supported by the extinction of common wombats from the north-western part of their range in South Australia after the arrival of rabbits, except on one pastoral lease free of rabbits, a trend which has been attributed to the loss of native perennial grasses following their selection by rabbits, and subsequent successional changes in vegetation

Chapter 3 Broad scale habitat use 86 composition (Cooke 1998). Further, high spatial overlap and thus potentially high competition between native species such as swamp wallabies and wombats may be a follow-on effect of forced movement by wombats into suboptimal habitat preferred by swamp wallabies, due to competition with rabbits. In addition, there is uncertainty about whether all three native herbivores present on GYI are indigenous to the area (Whelan 2008), and native herbivore species that once occured on GYI, such as the Tasmanian pademelon Thylogale billardierii, are no longer present in this system. The potentially complex inter-specific interactions within this system, and their effects on community structure, cannot be fully understood without experimental manipulation of populations (Schoener 1983, Sih 1993). Our understanding of inter-specific interactions on GYI would also be improved through the use of seasonal comparisons of resource availability and associated changes in niche parameters. During this study, estimates of habitat preference, niche breadth and inter-specific overlap in habitat use were based on pellet count data collected during autumn and winter only. If resource availability is greater in spring and summer than in autumn and winter (e.g., Mayze and Moore 1990), it could be predicted that habitat niche breadths would be broader in spring and summer than in winter and autumn due to a reduction in inter-specific competition for food resources. This shift would likely be associated with an increase in the range of habitats used by each species and an increase in inter-specific overlap in habitat use. As well as niche differentiation with respect to habitat use, coexistence can be facilitated by specialisation along diet and temporal gradients (Schoener 1974b, Whitfield 2002, Schmidt et al. 2010). Niche differentiation is generally complementary; when species are similar on one niche dimension, they differ on another (Pianka 1976, Dunbar 1978, Fox 1989, Bagchi et al. 2003). This suggests that if inter-specific overlap in habitat use were high, temporal (i.e., seasonal) partitioning may be evident (hypothesis 3). For example, large herbivores in Nepal show a high degree of spatial and food resource overlap in the monsoonal growing season, but habitat use changes in the resource-limited dry season, with most species moving into sub-optimal habitats (Wegge et al. 2006). At GYI, herbivore pellet densities did vary significantly among seasons for the hog deer, common wombat, eastern grey kangaroo and swamp wallaby. If

Chapter 3 Broad scale habitat use 87 these results reflect seasonal changes in habitat use, they indicate the likelihood of temporal partitioning due to competition within this assemblage (Rosenzweig 1981). However, it is unclear whether these seasonal differences represent differences in habitat use over time; rather, they could be an artefact of differential pellet decay rates between seasons (Chapter 2). High variability in faecal pellet decay rates among habitats (Chapter 2) is important to consider when using pellet counts to examine herbivore habitat use. When measures of pellet decay are taken into account, for example, in estimation of herbivore abundances, potential biases resulting from variable pellet decay rates among habitats are generally accounted for. Therefore, estimates based on herbivore abundances, such as habitat selection, should not be biased by differential pellet decay among habitats. However, caution must be used in interpreting analyses of habitat use where pellet count data is not adjusted to account for differential pellet decay rates among habitats, such as the generalised linear modelling used in this study.

Niche breadth Inter-specific competition tends to limit habitat use (Svarsden 1949), that is, foragers respond to inter-specific competition by narrowing patch use (Schoener 1986). In a heterogeneous environment, differential habitat selection can allow ecologically similar, sympatric populations to coexist (MacArthur and Levins 1964, Rosenzweig 1981). Thus as predicted by niche compression theory, populations that occur in sympatry tend to occupy a narrower range of habitats than in allopatry (MacArthur and Levins 1967). In particular, as discussed previously, competition can force subordinate species from preferred habitat into suboptimal habitat (Schoener 1983, Sih 1993). As outlined, there was no strong evidence for competitive displacement from optimal to sub-optimal habitats within this community. However, it was also predicted that habitat use by species in this multi-species assemblage would be more limited (i.e., niche breadths would be narrower) relative to when these species occur in assemblages with fewer species and therefore presumably, lower levels of inter-specific competition. Habitat niche breadth varied among species on GYI. The eastern grey kangaroo displayed a particularly narrow niche with respect to habitat use. This

Chapter 3 Broad scale habitat use 88 contrasts the suggestion by Hill (1981a) that this species has wide ecological tolerances with respect to habitat, and as predicted, may by due to limitation of habitat use as a result of inter-specific competition in this multi-species assemblage. Alternatively, narrow niche breadth may simply reflect limited availability of suitable habitat types for kangaroos to occupy on GYI. In contrast, wallabies and wombats displayed particularly broad habitat use niches, indicating that these species are capable of using a wide variety of habitats. This finding conforms with the suggestion by Roger and Ramp (2009) that the common wombat is highly adaptable, but contradicts the prediction of narrow niche breadths with respect to habitat use in this assemblage. In contrast to inter-specific competition, intra-specific competition tends to broaden habitat use (Svarsden 1949). It may be that high densities, particularly for wombats, have resulted in broad habitat use by these species. Further, it was expected that habitat use would be influenced by body size (hypothesis 4), with smaller herbivore species being more selective (i.e., having narrower habitat niche breadths) than larger herbivore species (duToit and Owen- Smith 1989). This is because increased dietary tolerance among larger species (i.e., tolerance of a wider range of food items in terms of nutritional quality or food-item size; May and MacArthur 1972, Schoener 1974b) generally allows use of a wider range of habitat patches and, hence, more even use of environmental resources (duToit and Owen-Smith 1989). For example, habitat use by ungulates in the Serengeti is strongly correlated with body size and diet selectivity (Bell 1970). However, there was no clear relationship between body size and the diversity of habitats used by herbivores on GYI. Diversity in habitat use by large species is limited by the diversity of available habitats (duToit and Owen-Smith 1989). Being a temperate ecosystem, GYI may not support great enough spatial heterogeneity to detect a relationship between body mass and habitat diversity (i.e., relative to tropical ecosystems where relationships have been detected; Peters and Raelson 1984, duToit and Owen-Smith 1989).

Herbivore density, population abundance and population metabolism The balance between population density and energy use per individual determines how evenly community resources are shared by species of different size (duToit and Owen-Smith 1989). This can be particularly important when

Chapter 3 Broad scale habitat use 89 patterns of resource use by sympatric herbivore species are similar, as resource limitation can suppress population densities (Madhusudan 2004). Moreover, density dependent effects are central to models of habitat selection (Lawlor and Maynard Smith 1976): competition is most intense at high population densities, whereas under certain density combinations, no competition may be apparent (Manor and Saltz 2008). The species that occurred in the lowest density on GYI was the eastern grey kangaroo. Eastern grey kangaroos also had a relatively low metabolic demand at the population level, closely followed by swamp wallabies. This could indicate competitive suppression (Madhusudan 2004) of macropod populations, particular grazing kangaroos, although these results may reflect in part the low metabolic rate of marsupials (Hume 1982). The density of hog deer was also relatively low; however hog deer biomass and metabolic demand at the population level were relatively high, given its relatively great body mass. In contrast, the most abundant herbivore species on GYI, and within Coastal Grassy Woodland, the dominant habitat type, were the native wombat and the introduced rabbit, indicating that these two species may be competitively superior to the other herbivore species present. In support of this assertion, the wombat population had a high metabolic demand at the population level relative to other species. In contrast, metabolic demand for rabbits at the population level was relatively low, indicating relatively low use of resources compared to the other species. Rather, hog deer ranked second with respect to metabolic demand at the population level, suggesting that wombats and deer may be the most efficient competitors in this assemblage, dominating food resource use. This finding appears to contradict the suggestion that wombats may have been displaced from preferred habitat by rabbits, based on habitat overlap and preferences within this system. However, it is possible that even following displacement from their preferred habitat, wombats are able to numerically dominate the system as a whole and utilise the greatest proportion of resources. The wombat has adaptations that allow it to save energy and reach high population densities in low productivity habitats (Johnson 1998), and an ability to burrow and utilise low quality forage may give the wombat a competitive advantage over other herbivore species (Hume 1999). Overall, metabolic demand from native herbivores on GYI was c. 40% greater than that of introduced herbivores, suggesting that introduced eutherian

Chapter 3 Broad scale habitat use 90 species are not superior competitors relative to native marsupials, as has been suggested (McNab 2005). Regardless of their origins, the abundance of herbivores with grazing diets on GYI (i.e., rabbits, wombats and kangaroos) was almost four times greater than the abundance of herbivores with browsing diets (i.e., wallabies and deer), and the metabolic demand at the population level of grazing species combined was almost twice that of browsing species. These trends are even more pronounced if hog deer are considered grazers as they are in their native range (Wegge et al. 2006) (i.e., over seven times as many grazers as browsers, and metabolic demand for grazers almost eight times that of browsers). This supports the assertion that competition among grazers is great and has forced some species into suboptimal habitat, and in the case of the hog deer and eastern grey kangaroo, has resulted in a shift towards browsing (Chapter 5). Depending on the herbivore community being examined, studies have found that population metabolism can scale either neutrally (Damuth 1981, 1987) or negatively (Peters 1983) with body mass. However, a comprehensive study of the relationships between mammalian population density and body mass by Peters and Raelson (1984) indicated that population metabolism generally scales positively with body mass, reflecting advantages of large body size which may result in greater energy use per individual (duToit and Owen-Smith 1989) and subsequent resource domination (Brown and Maurer 1986). In line with the suggestion of Peters and Raelson (1984), population metabolism on GYI scaled positively with body mass (hypothesis 5). However, on GYI, population metabolism increases approximately in relation to M0.77, which is much higher than other herbivore assemblages which have been studied, for example M0.45 for large tropical herbivores (Peters and Raelson 1984) and M0.43 for large African herbivores (unpublished data cited in duToit and Owen-Smith 1989). These results suggest that on GYI, larger species use a disproportionately larger share of local resources.

Conclusion Faecal pellet counts on Greater Yanakie Isthmus suggest that hog deer, eastern grey kangaroos, swamp wallabies, common wombats and European rabbits used habitats selectively in ways which broadly reflect patterns of habitat use described in other parts of their ranges. No clear shifts in habitat use from

Chapter 3 Broad scale habitat use 91 preferred habitats to suboptimal habitats were apparent on GYI, suggesting that inter-specific competition is not strong enough to cause competitive exclusion, although patterns of use of Coastal Scrubs and Grasslands suggest the possibility that rabbits exclude other grazing herbivores, both native and introduced, from this habitat. Overall though, levels of inter-specific overlap in habitat use were generally low, suggesting that inter-specific partitioning of habitat resources was operating, possibly aided by temporal partitioning. Low overlap in habitat use implies low competition (Schoener 1983), but is possibly the ghost of competition past (Connell 1980). Despite generally low overlap in habitat use, high overlap was apparent between some species, indicating the potential for competition, particularly between grazers, if resources are limiting. In particular, eastern grey kangaroos displayed relatively narrow habitat niche breadth, occurred at relatively low densities and their metabolic demand at the population level was relatively low, suggesting that the kangaroo population may be competitively suppressed. In contrast, common wombats appear to be the most efficient competitors in this assemblage, being numerically dominant, utilising the greatest proportion of resources, and displaying a relatively broad habitat niche. Population metabolism scaled positively with body mass, suggesting that larger herbivore species use a disproportionately larger share of local resources on GYI. In contrast, there was no clear relationship between body size and the diversity of habitats used, nor did evolutionary history appear to govern resource partitioning within this community. However, my ability to detect such trends may have been limited by inadequate spatial heterogeneity, or discrepancies regarding the evolutionary histories of the five study species (e.g., distinct evolutionary histories of the two introduced species). The results of this study are correlative, and it is difficult to distinguish between competitive interactions and habitat selection without experimental manipulations (Manor and Saltz 2008). Further, although competition is believed to be a central biotic factor structuring herbivore communities (Sinclair and Norton-Griffiths 1982, Schoener 1989), factors such as food quality and availability (Fritz et al. 1996), weather conditions, and predator and parasite avoidance are also potentially important in influencing habitat use (Duncan 1983), and foragers may trade-off certain components of patch quality for others. Large- scale experimental manipulation (Walters and Holling 1990), involving the

Chapter 3 Broad scale habitat use 92 measurement of species’ responses to altered abundance of potential competitors (Schoener 1983, Sih 1993), and habitat manipulation (e.g., by fire) with respect to multiple factors potentially effecting habitat use (Catling et al. 2000), is required to provide a mechanistic understanding of habitat selection (e.g., Parrish 1995) in this herbivore assemblage.

Chapter 3 Broad scale habitat use 93

Table 3.1. Mean density (number km-2) and population abundance estimates (with 95% confidence intervals) for five herbivores in five broad vegetation types (BVT) on Greater Yanakie Isthmus (GYI): Coastal Grassy Woodland (CGW), Coastal Scrubs and Grasslands (CSG), Heath (H), Heathy Woodland (HW), and Moist Foothill Forest (MFF). Estimates are based on faecal pellet counts conducted during summer 2003-4 and autumn and winter 2004 for all species and also during spring 2003 for hog deer. Species specific faecal pellet decay a rates and defaecation rates were taken from the literature. Also given is the basal metabolic rate (BMR) per day and mass for individuals of each species.

Species BVT BVT area N Pellet or pellet SE Pellet or pellet SE Pellet or SE Density (km-2) 95% CI Abundance 95% CI Mass kg BMR (kJ (ha) group density group decay rate pellet group day-1)

(ha-1) (day-1) defaecation rate (day-1) b g Hog deer CGW 5591 106 122.5 21.7 351 12 20.3 4.1 1.7 0.9 96.1 51.4 40.3 181 CSG 1425 33 264.0 57.4 167 19 7.8 5.0 111.0 71.5 H 720 9 199.5 109.2 71 8 13.8 18.9 99.6 136.4 HW 2551 56 6.2 3.2 381 26 0.1 0.1 2.1 2.3 MFF 237 16 161.2 51.9 221 15 3.6 2.9 8.5 7.0 GYI 10524 220 317.3 268.6 4687 c h European rabbit CGW 5591 80 503.5 82.7 167 9 18.5 1.0 16.3 5.8 909.3 326.9 1.58 6 CSG 1425 26 221.7 65.8 52 11 23.0 17.4 327.7 248.1 H 720 6 0.0 0.0 41 5 0.0 0.0 0.0 0.0 HW 2551 42 20.6 20.6 161 24 0.7 1.4 17.6 35.9 MFF 237 11 74.0 49.7 68 9 5.9 7.2 13.9 17.1 GYI 10524 165 1268.5 628.0 413 Eastern grey CGW 5591 80 220.4 78.9 173 7 d 98.6 0.3 0.2 14.4 11.8 i 9 kangaroo 493.0 26.3 CSG 1425 26 5456.1 1988.4 89 4 12.4 10.7 177.2 152.3 H 720 6 0.0 0.0 59 3 0.0 0.0 0.0 0.0 HW 2551 42 333.4 107.8 219 8 0.3 0.2 7.9 6.1 MFF 237 11 0.0 0.0 140 5 0.0 0.0 0.0 0.0 GYI 10524 165 199.5 170.2 2334

Chapter 3 Broad scale habitat use 94

Table 3.1 (cont.) e j Swamp wallaby CGW 5591 80 616.3 85.4 134 7 150 30 3.1 1.5 171.4 84.9 18 54 CSG 1425 26 331.9 107.9 72 4 3.1 2.4 43.8 34.7 H 720 6 542.3 278.7 59 3 6.1 8.3 44.1 59.8 HW 2551 42 1895.9 219.8 174 7 7.3 3.4 185.3 87.7 MFF 237 11 569.1 184.1 106 3 3.6 2.5 8.5 5.8 GYI 10524 165 453.1 272.9 1757 f k Common wombat CGW 5591 80 1722.8 235.0 189 6 90 18 10.1 4.9 566.3 275.1 27.9 386 CSG 1425 26 1375.3 251.1 73 3 20.9 11.8 298.3 168.0 H 720 6 1414.7 777.6 71 2 22.1 31.7 159.4 228.4 HW 2551 42 2849.2 678.5 193 6 16.4 10.3 418.4 263.7 MFF 237 11 588.4 184.0 144 5 4.5 3.0 10.8 7.2 GYI 10524 165 1453.2 942.5 2440 a Davis et al. (Chapter 2; in review) b Dhungel (1985a) c Wood (1988) d Johnson et al. (1987) e Floyd (1980) f Triggs (1996) g Mayze and Moore (1990) h Myers (1983) i pers. comm. G. Coulson 2005 (Tidbinbilla, ACT, n = 333) j Edwards (1969) k Barboza et al. (1993)

Chapter 3 Broad scale habitat use 95

Table 3.2. Results of G-tests comparing the selection of vegetation types by each of five herbivore species (n = 5 vegetation types, except for the rabbit n = 4 and for the kangaroo n = 3) and results of pairwise chi-squared comparisons of standardised ratio values for selection indices (Bi) for five vegetation types: Coastal Grassy Woodland (CGW), Coastal Shrubs and Grasslands (CSG), Heath (H), Heathy Woodland (HW), and Moist Foothill Forest (MFF). Samples (n = 110) were collected from GYI, Wilsons Promontory National Park autumn (March – May) and winter (June – August) 2004. P values in bold are significant effects.

Comparison Hog deer European rabbit Eastern grey kangaroo Swamp wallaby Common wombat 2 P 2 SE P 2 SE P 2 SE P 2 SE P All vegetation types 394.4 < 0.001 5029.5 < 0.001 1317.8 < 0.001 487.7 < 0.001 1420.2 < 0.001 CGW vs HW 50639.4 0.01 < 0.001 677 0.03 < 0.001 2155.2 0.02 < 0.001 CGW vs MFF 1054.8 0.03 < 0.001 CGW vs CSG 2845.6 0.01 < 0.001 4655.4 0.07 < 0.001 CGW vs H 426.2 0.05 < 0.001 CSG vs HW 3950.2 0.02 < 0.001 5122.1 0.06 < 0.001 H vs MFF 416.5 0.06 < 0.001 HW vs MFF 686 0.04 < 0.001

Chapter 3 Broad scale habitat use 96

Table 3.3. Standardised ratio selection index values (Bi) (with 95% confidence intervals, corrected for multiple comparisons using the Bonferroni correction) for herbivore selection of vegetation types: Coastal Grassy Woodland (CGW), Coastal Shrubs and Grasslands (CSG), Heath (H), Heathy Woodland (HW), and Moist Foothill Forest (MFF). Samples (n = 110) were collected from GYI, Wilsons Promontory National Park, in autumn and winter 2004.

Hog deer European rabbit Eastern grey kangaroo Swamp wallaby Common wombat Lower 95% Upper 95% Lower 95% Upper 95% Lower 95% Upper 95% Lower 95% Upper 95% Lower 95% Upper 95% Bi CI CI Bi CI CI Bi CI CI Bi CI CI Bi CI CI CGW 0.06 0.00 0.12 0.64 0.58 0.70 0.02 0.02 0.02 0.08 0.08 0.08 0.10 0.10 0.10 CSG 0.29 0.04 0.54 0.24 0.08 0.40 0.95 0.95 0.95 0.11 0.11 0.11 0.20 0.20 0.20 H 0.51 0.03 0.99 0.00 -0.40 0.40 0.00 0.00 0.00 0.39 0.39 0.39 0.35 0.35 0.35 HW 0.00 -0.02 0.03 0.01 -0.14 0.16 0.03 0.03 0.03 0.28 0.28 0.28 0.30 0.30 0.30 MFF 0.13 -0.37 0.64 0.11 -0.32 0.54 0.00 0.00 0.00 0.13 0.13 0.13 0.05 0.05 0.05

Chapter 3 Broad scale habitat use 97

Table 3.4. Values for Hulberts’ index of niche overlap (L) for proportional abundance of each of five herbivore species using five vegetation type resource states: Coastal Grassy Woodland, Coastal Shrubs and Grasslands, Heath, Heathy Woodland, and Moist Foothill Forest. Samples (n = 110) were collected from GYI, Wilsons Promontory National Park autumn and winter 2004.

Species Hog deer European Eastern Swamp rabbit grey wallaby kangaroo European rabbit 0.74 Eastern grey kangaroo 2.33 0.68 Swamp wallaby 1.22 0.55 0.76 Common wombat 1.19 0.63 1.10 1.33

Table 3.5. Values for Smith’s measure of niche breadth (FT) (on a scale of 0-1) with upper and lower 95% confidence limits for the proportion of individuals of each of five herbivore species using five vegetation type resource states: Coastal Grassy Woodland, Coastal Shrubs and Grasslands, Heath, Heathy Woodland, and Moist Foothill Forest. Samples (n = 110) were collected from GYI, Wilsons Promontory National Park during autumn and winter 2004.

Species FT Lower 95% CI Upper 95% CI Hog deer 0.83 0.80 0.86 European rabbit 0.85 0.85 0.86 Eastern grey kangaroo 0.64 0.60 0.67 Swamp wallaby 0.95 0.94 0.96 Common Wombat 0.97 0.96 0.97

Chapter 3 Broad scale habitat use 98

Table 3.6. Results of a generalised linear model estimating the relationship between European rabbit faecal pellet group density and: (1) the vegetation type in which counts were conducted, (2) the season in which counts were conducted, (3) the distance of faecal pellet count plots to water, and (4) vegetation cover parameters measured within plots. P values in bold are significant effects.

Compared Rate Overall Variable Level to Ratio 95% CI P P Lower Upper Vegetation type 0.268 CGW CSG 0.75 0.28 2.01 0.6 CGW Heath ∞ * * * CGW HW 2.48 0.69 8.89 0.16 CGW MFF 0.88 0.13 5.98 0.9 CSG Heath ∞ * * * CSG HW 3.29 0.77 14.07 0.11 CSG MFF 1.17 0.17 7.82 0.9 Heath HW 0.00 ** ** ** Heath MFF 0.00 ** ** ** HW MFF 0.35 0.05 2.76 0.3 Season 0.867 Distance to water x+500 x 1.04 0.77 1.41 0.8 % ground cover x+10 x 0.96 0.82 1.13 0.6 % shrub cover x+10 x 0.73 0.54 0.98 0.04 % canopy cover x+10 x 0.85 0.66 1.09 0.19 Ground cover: Leptospermum laevigatum x+1 x 0.73 0.01 47.90 0.9 Shrub cover: L. < laevigatum x+1 x 1.21 1.09 1.34 0.001 Canopy cover: L. laevigatum x+1 x 0.96 0.86 1.07 0.5 Canopy cover: Acacia longifolia x+1 x 1.14 0.97 1.33 0.11 Canopy cover: Allocasuarina verticillata x+1 x 1.05 0.89 1.24 0.6 Canopy cover: Banksia integrifolia x+1 x 0.93 0.72 1.20 0.6

* In this case the estimate and standard error are meaningless, because the value the algorithm is attempting to estimate is plus infinity on the log scale. ** In this case the estimate and standard error are meaningless, because the value the algorithm is attempting to estimate is zero.

Chapter 3 Broad scale habitat use 99

Table 3.7. Results of a generalised linear model estimating the relationship between hog deer faecal pellet group density and: (1) the vegetation type in which counts were conducted, (2) the season in which counts were conducted, (3) the distance of faecal pellet count plots to water, and (4) vegetation cover parameters measured within plots. P values in bold are significant effects.

Compared Rate Overall Variable Level to Ratio 95% CI P P Lower Upper Vegetation type < 0.001 CGW CSG 0.90 0.52 1.56 0.7 CGW Heath 1.08 0.42 2.73 0.9 CGW HW 18.38 4.49 75.25 < 0.001 CGW MFF 1.23 0.51 2.97 0.6 CSG Heath 1.19 0.53 2.66 0.7 CSG HW 20.37 5.24 79.15 < 0.001 CSG MFF 1.37 0.63 2.98 0.4 Heath HW 17.08 4.14 70.48 < 0.001 Heath MFF 1.15 0.51 2.58 0.7 HW MFF 0.07 0.02 0.28 < 0.001 Season < 0.001 Distance to water x+500 x 0.68 0.49 0.94 0.02 % ground cover x+10 x 1.16 1.04 1.28 0.008 % shrub cover x+10 x 0.91 0.81 1.02 0.10 % canopy cover x+10 x 0.91 0.79 1.05 0.2 Ground cover: Leptospermum laevigatum x+1 x 3.40 0.77 14.94 0.10 Shrub cover: L. laevigatum x+1 x 0.89 0.81 0.98 0.02 Canopy cover: L. laevigatum x+1 x 1.04 0.97 1.12 0.3 Canopy cover: Acacia longifolia x+1 x 1.18 1.04 1.33 0.009 Canopy cover: Allocasuarina verticillata x+1 x 1.21 0.91 1.61 0.18 Canopy cover: Banksia integrifolia x+1 x 1.21 0.99 1.47 0.06

Chapter 3 Broad scale habitat use 100

Table 3.8. Results of a generalised linear model estimating the relationship between eastern grey kangaroo faecal pellet density and: (1) the vegetation type in which counts were conducted, (2) the season in which counts were conducted, (3) the distance of faecal pellet count plots to water, and (4) vegetation cover parameters measured within plots. P values in bold are significant effects.

Compared Rate Overall Variable Level to Ratio 95% CI P P Lower Upper Vegetation type < 0.001 CGW CSG 0.02 0.00 0.11 < 0.001 CGW Heath ∞ * * * CGW HW 0.32 0.04 2.41 0.3 CGW MFF ∞ * * * CSG Heath ∞ * * * CSG HW 19.13 5.51 66.41 < 0.001 CSG MFF ∞ * * * Heath HW 0.00 ** ** ** Heath MFF ∞ * * * HW MFF ∞ * * * Season < 0.001 Distance to water x+500 x 1.98 1.09 3.61 0.03 % ground cover x+10 x 0.98 0.80 1.20 0.8 % shrub cover x+10 x 0.81 0.64 1.04 0.10 % canopy cover x+10 x 0.94 0.74 1.19 0.6 Ground cover: Leptospermum 2379.1 laevigatum x+1 x 10.80 0.05 9 0.4 Shrub cover: L. laevigatum x+1 x 1.07 0.78 1.47 0.7 Canopy cover: L. laevigatum x+1 x 0.73 0.55 0.98 0.03 Canopy cover: Acacia longifolia x+1 x 1.04 0.91 1.18 0.6 Canopy cover: Allocasuarina verticillata x+1 x 0.86 0.35 2.10 0.7 Canopy cover: Banksia integrifolia x+1 x 1.23 0.94 1.63 0.13

* In this case the estimate and standard error are meaningless, because the value the algorithm is attempting to estimate is plus infinity on the log scale. ** In this case the estimate and standard error are meaningless, because the value the algorithm is attempting to estimate is zero.

Chapter 3 Broad scale habitat use 101

Table 3.9. Results of a generalised linear model estimating the relationship between swamp wallaby faecal pellet density and: (1) the vegetation type in which counts were conducted, (2) the season in which counts were conducted, (3) the distance of faecal pellet count plots to water, and (4) vegetation cover parameters measured within plots. P values in bold are significant effects.

Compared Rate Overall Variable Level to Ratio 95% CI P P Lower Upper Vegetation type < 0.001 CGW CSG 1.68 0.81 3.47 0.16 CGW Heath 1.09 0.37 3.20 0.9 CGW HW 0.59 0.33 1.06 0.08 CGW MFF 1.71 0.76 3.87 0.19 CSG Heath 0.65 0.22 1.92 0.4 CSG HW 0.35 0.19 0.65 < 0.001 CSG MFF 1.02 0.46 2.26 > 0.9 Heath HW 0.54 0.21 1.39 0.2 Heath MFF 1.57 0.53 4.63 0.4 HW MFF 2.91 1.55 5.48 0.001 Season < 0.001 Distance to water x+500 x 0.96 0.81 1.15 0.7 % ground cover x+10 x 1.05 0.97 1.13 0.2 % shrub cover x+10 x 1.03 0.93 1.15 0.5 % canopy cover x+10 x 0.97 0.88 1.08 0.6 Ground cover: Leptospermum laevigatum x+1 x 1.75 0.02 146.39 0.8 Shrub cover: L. laevigatum x+1 x 1.02 0.95 1.11 0.5 Canopy cover: L. laevigatum x+1 x 0.96 0.89 1.03 0.2 Canopy cover: Acacia longifolia x+1 x 0.85 0.74 0.98 0.02 Canopy cover: Allocasuarina verticillata x+1 x 1.02 0.89 1.17 0.8 Canopy cover: Banksia integrifolia x+1 x 0.96 0.83 1.11 0.6

Chapter 3 Broad scale habitat use 102

Table 3.10. Results of a generalised linear model estimating the relationship between common wombat faecal pellet density and: (1) the vegetation type in which counts were conducted, (2) the season in which counts were conducted, (3) the distance of faecal pellet count plots to water, and (4) vegetation cover parameters measured within plots. P values in bold are significant effects.

Compared Rate Overall Variable Level to Ratio 95% CI P P Lower Upper Vegetation type < 0.001 CGW CSG 1.93 0.78 4.80 0.15 CGW Heath 0.58 0.17 1.95 0.4 CGW HW 0.95 0.43 2.10 0.9 CGW MFF 5.06 1.32 19.45 0.02 CSG Heath 0.30 0.09 0.97 0.04 CSG HW 0.49 0.24 1.01 0.05 CSG MFF 2.62 0.73 9.33 0.14 Heath HW 1.63 0.58 4.59 0.4 Heath MFF 8.72 1.98 38.44 0.004 HW MFF 5.35 1.65 17.31 0.005 Season < 0.001 Distance to water x+500 x 1.12 0.89 1.41 0.3 % ground cover x+10 x 1.02 0.92 1.13 0.7 % shrub cover x+10 x 1.16 1.00 1.35 0.05 % canopy cover x+10 x 1.12 0.97 1.31 0.13 Ground cover: Leptospermum laevigatum x+1 x 0.00 ** ** ** Shrub cover: L. laevigatum x+1 x 1.10 1.01 1.21 0.03 Canopy cover: L. laevigatum x+1 x 0.95 0.87 1.04 0.3 Canopy cover: Acacia longifolia x+1 X 0.98 0.86 1.11 0.7 Canopy cover: Allocasuarina verticillata x+1 X 1.07 0.92 1.24 0.4 Canopy cover: Banksia integrifolia x+1 X 1.07 0.91 1.26 0.4

** In this case the estimate and standard error are meaningless, because the value the algorithm is attempting to estimate is zero.

Chapter 3 Broad scale habitat use 103

1

wombat

wallaby rabbit

0.8 deer

0.6

kangaroo FT 0.4

0.2

0 1 10 100 M (kg)

Figure 3.1. Uniformity of habitat use (Smith’s measure of niche breadth: FT) versus body mass (M) for five herbivore species (hog deer, European rabbit, swamp wallaby, eastern grey kangaroo and common wombat) using five vegetation type resource states: Coastal Grassy Woodland, Coastal Shrubs and Grasslands, Heath, Heathy Woodland, and Moist Foothill Forest. Samples (n = 110) were collected from GYI, Wilsons Promontory National Park during autumn and winter 2004. The x-axis is logarithmic.

Chapter 3 Broad scale habitat use 104

) -2 25

20

15

10

5

0

Coastal Coastal Heath Heathy Moist Foothill Herbivore density (number km density Herbivore Grassy Scrubs & Woodland Forest Woodland Grasslands Vegetation type

Figure 3.2. Mean ( standard error) density of large mammalian herbivores per km2 calculated from faecal pellet surveys conducted in summer, autumn and winter 2004 in five vegetation types on Greater Yanakie Isthmus I, Wilsons Promontory National Park: Coastal Grassy Woodland (CGW, n = 80), Coastal Scrubs and Grasslands (CSG, n = 26), Heath (H, n = 6), Heathy Woodland (HW, n = 42), and Moist Foothill Forest (MFF, n = 11).

Chapter 3 Broad scale habitat use 105

(i) Greater Yanakie Isthmus

biomass basal metabolic demand abundance

4500 1600

-3 4000 1400 3500

1200 ; ; -1 3000 1000 2500 800 2000

600 abundance

1500 biomass10 kgx 1000 400

500 200 basal metabolic rate kJ/day x 10 x metabolic kJ/day basal rate

0 0

rabbit

swamp

wallaby

wombat

common

hog deer hog European

Species kangaroo easterngrey (ii) Coastal Grassy Woodland

biomass basal metabolic rate abundance

1800 1000 -3 1600 900 800

; 1400 -1 700 1200 600 1000 500 800 400 600 abundance

300 biomass10 kgx 400 200 200 100

basal metabolic rate kJ/day x 10 x metabolic kJ/day basal rate 0 0

rabbit

swamp

wallaby

wombat

common

hog deer hog

European

kangaroo easterngrey Species

Figure 3.3. Population abundance, biomass and basal metabolic demand at the population level for five herbivore species (i) over the area of Greater Yanakie Isthmus, and (ii) within Coastal Grassy Woodland on Greater Yanakie Isthmus. Population estimates are based on faecal pellet counts conducted between 2003-

2004 and body mass values from the literature.

Chapter 3 Broad scale habitat use 106

Chapter 3 Broad scale habitat use 107

Chapter 4

The influence of fire on fine scale habitat use by native and introduced mammalian herbivores in Coastal Grassy Woodland on Yanakie Isthmus, south-eastern Australia ______

Chapter 4 Fine scale habitat use 108

Chapter 4 Fine scale habitat use 109

Chapter 4 The influence of fire on fine scale habitat use by native and introduced mammalian herbivores in Coastal Grassy Woodland on Yanakie Isthmus, south-eastern Australia

Abstract Sympatric species should reduce competition by partitioning resources. Habitat is the most common resource partitioned, and fine-scale microhabitat selection can be a major dimension of habitat partitioning. On Yanakie Isthmus, Wilsons Promontory National Park, Victoria, native and introduced mammalian herbivores occur in sympatry. To examine the effect of habitat modification by an ecological process (fire) on fine-scale partitioning of habitat resources by sympatric herbivore species on Yanakie Isthmus, I monitored changes in plant and herbivore communities following an ecological burn. The burn changed vegetation composition and structure, resulting in changes in habitat use and composition of the herbivore community. Faecal pellet counts demonstrated an overall decrease in herbivore densities, and a decrease in the density of grazers, namely rabbits following the burn, probably associated with reduced ground layer vegetation cover, particularly native graminoid species. In contrast, there was an increase in the density of browsers, particularly swamp wallabies, probably associated with increased shrub cover. This study suggests that ecological processes that modify food and shelter resource availability can alter herbivore habitat use and community structure. Fine-scale partitioning of habitat resources was evident through inter-specific differences in abundance, population metabolism and use of vegetation strata within Coastal Grassy Woodland prior to and after the burn. Inter-specific differences in feeding strategies and thus resource requirements appear to facilitate coexistence within this assemblage, although overlap in fine-scale habitat use appears to be greater between native and introduced species than between native species. This suggests that mechanisms of resource partitioning have evolved in species with long coevolutionary histories, whereas species with independent evolutionary histories have inherently less resource partitioning.

Chapter 4 Fine scale habitat use 110

Chapter 4 Fine scale habitat use 111

Introduction Investigation of competitive interactions, and the ways in which species coexist, is essential to improve our understanding of the ecological principles underlying ecosystem functioning (Duncan et al. 1998). Gause (1934) first suggested that sympatric species should reduce competition by selecting different resources. Specialisation on resources along habitat, diet and temporal gradients results in niche differentiation, facilitating coexistence (Schoener 1974b, Whitfield 2002). Habitat is the most common resource partitioned (Schoener 1983). Habitat selection by herbivores is rarely at random (Duncan 1983): sympatric herbivores tend to exploit their environments in different ways (Schwartz and Ellis 1981, Forsyth 2000) based on differences in feeding strategies (Gwynne and Bell 1968, Schwartz and Ellis 1981) which lead to differences in habitat preference (Batcheler 1960, Taylor 1983, Fox 1989). As highlighted by Schoener (1986), many studies suggest that foragers respond to inter-specific competition by reducing niche overlap and narrowing patch use (Svarsden 1949). By decreasing spatial overlap between populations, habitat partitioning provides a major means of reducing inter-specific competition (Schoener 1974b). Mediation of competitive interactions through habitat selection can play an important role in structuring herbivore assemblages (Mishra et al. 2002). Understanding resource use overlap in terms of habitat is critical to understanding competitive interactions (Krebs 1998, Wegge et al. 2006), and thus sympatry (Telfer et al. 2008) and community structure (Bell 1971, Hofmann and Stewart 1972, Jarman 1974). Resource utilisation can occur over multiple spatial hierarchies, each of which relate to the requirements of individuals (Johnson 1980). Thus habitat selection at different scales can facilitate coexistence. Johnson (1980) describes three levels of habitat use: first order, which encompasses the broad distribution of species; second order, which consists of the home range of individuals; and third order, which refers to the use of habitat components within an individuals home range. Mapping habitat types at a broad scale provides a general indication of differences in vegetation communities. While important insights into habitat partitioning can be gained through studies at this broad scale (e.g., Forsyth 2000), Schmidt et al. (2010) point out that comparing the occurrence of species using

Chapter 4 Fine scale habitat use 112 broad vegetation types provides only a coarse assessment of habitat selection. It is essential to also examine habitat use on a fine scale (Southwell et al. 1999). Broad habitat types are generally comprised of fine-scale mosaics within which individuals and species make diverse fine-scale choices between microhabitats over time (e.g., Clarke et al. 1989, Fa et al. 1999). As demonstrated in studies such as that by Schmidt et al. (2010), fine-scale variation in habitat structure can be more important in influencing choices about habitat use than broad habitat features. Thus microhabitat, aspects of habitat such as lateral and overhead cover operating on a small scale, can be a major dimension of habitat partitioning and species that do not separate at the home range scale may show distinct separation at the microhabitat scale (e.g., Scognamillo et al. 2003). Ecological processes can have a strong influence on microhabitat structure, and therefore on habitat use by herbivores. Fire is an important ecological process in many regions of the world, with profound effects on the physical environment in which animals live (e.g., Whelan 1995, Tozer and Bradstock 2002, Whitehead et al. 2005, Van Dyke and Darragh 2006, Sankaran et al. 2008). Fire is particularly important in Australian vegetation communities (Braithwaite 1990, Cary et al. 2003). Burning of shrub and grassland communities often leads to increases in plant production and nutritional quality that benefit mammalian herbivores (Van Dyke and Darragh 2006) and preferential use of burnt areas by herbivores has been widely demonstrated (e.g., Cohn and Bradstock 2000, Briggs et al. 2002, Meers and Adams 2003). In Australia, studies have demonstrated grazing of post-fire regrowth by both native (e.g., Meers and Adams 2003) and introduced herbivores (e.g., Leigh and Holgate 1979). However, few studies (but see Catling et al. 2001) have examined fine-scale partitioning of habitat resources among multi-species assemblages comprised of native and introduced herbivore species in Australian ecosystems affected by fire. On Yanakie Isthmus (Wilsons Promontory National Park, Victoria), introduced European rabbits Oryctolagus cuniculus and hog deer Axis porcinus occur in sympatry with three native herbivores, the eastern grey kangaroo Macropus giganteus, swamp wallaby Wallabia bicolor and common wombat Vombatus ursinus. Broad and fine-scale habitat use by these species has been studied in other parts of their Australian range (e.g., Taylor 1971, Coulson 1993, de Munk 1999, Moseby et al. 2005, Roger et al. 2007), as well as in the native

Chapter 4 Fine scale habitat use 113 range of the two introduced species (e.g., Fa et al. 1999, Calvete et al. 2004, Odden et al. 2005, Odden and Wegge 2007). However, inter-specific overlap in habitat use has not been studied at a fine scale within the herbivore assemblage on Yanakie Isthmus, providing an opportunity to examine community niche dynamics in a complex herbivore assemblage comprised of introduced and native species. In their native range, hog deer prefer floodplain grassland associations, which provide both food and shelter (Odden et al. 2005). In Victoria, hog deer are essentially limited to flat, swampy coastal plains (Taylor 1971), where wetland adjacent to scrubland provides the mix of food and cover required (Mayze and Moore 1990). In their native range European rabbits favour open scrubland (Fa et al. 1999). Similarly, in sub-alpine regions of Australia rabbits occur in woodland associated with open grassy valleys (Williams and Myers 2008). Habitat selection by European rabbits is a compromise between food availability and shelter for protection from predators (Moreno et al. 1996). The native swamp wallaby generally lives in thick undergrowth of forests, woodlands and heath (Merchant 2008), although it is capable of utilising a variety of environments (Hollis et al. 1986, Troy et al. 1992) and will commonly forage along ecotones between forest and grassland (Hume 1999), moving into more open areas to feed during the night (Edwards and Ealey 1975). Swamp wallaby habitat selection is heavily influenced by the availability of dense vegetation cover (Di Stefano et al. 2007, Merchant 2008, Schmidt et al. 2010) which provides both food and shelter (e.g., Floyd 1980, Lunney and O'Connell 1988). Eastern grey kangaroos occupy a variety of habitats including sclerophyll forest, shrubland and heathland (Coulson 2008), but are largely absent from wet forests (Hill 1981b). Grass availability and quality are strong predictors of habitat use by kangaroos (Hill 1982, Taylor 1984), although they require habitats with a mix of open grassy areas for foraging and more densely vegetated areas offering lateral cover (McCullough and McCullough 2000, Moore et al. 2002, Schmidt et al. 2010). The common wombat utilises several vegetation types including sclerophyll forest, woodland, coastal scrub, heathland and temperate forest (McIlroy 2008), but preferentially forage in habitats with abundant high quality grass (Evans et al. 2006) and their habitat preferences may reflect suitability for burrowing (McIlroy 1995).

Chapter 4 Fine scale habitat use 114

Understanding resource partitioning among herbivore species requires knowledge of spatial patterns of habitat use and the environmental factors influencing these patterns. Previous research suggests that habitat use by each the hog deer, European rabbit, swamp wallaby, eastern grey kangaroo and common wombat is influenced by the availability of food (e.g., Taylor 1984, Lunney and O'Connell 1988, Evans et al. 2006, Odden and Wegge 2007) and shelter (e.g., Mayze and Moore 1990, McIlroy 1995, Carvalho and Gomes 2004, Di Stefano et al. 2007, Schmidt et al. 2010). Further, studies have shown that the food and/or habitat requirements of some of these species may overlap, although a degree of ecological partitioning has been recognised (Taylor 1971, de Munk 1999, Schmidt et al. 2010). Broad scale habitat use by the herbivore assemblage on Yanakie Isthmus demonstrated generally low overlap in habitat use; however, multiple herbivore species commonly occupied the same broad vegetation types (Chapter 3). One of these vegetation types, Coastal Grassy Woodland, offers a basis for studies of fine scale habitat use. This vegetation type originally consisted of open grassy woodland, but is now dominated by dense stands of the encroaching native shrubs Leptospermum laevigatum and Acacia longifolia var. sophorae (Bennett 1994, Costello et al. 2000). Mechanical slashing is undertaken by managers to retain open grassy woodland, creating a mosaic of dense stands of encroaching shrubs (Bennett 1994) interspersed by open grassy woodland. Further, Coastal Grassy Woodland is a fire-adapted vegetation community that relies on regular fires for regeneration (Chesterfield et al. 1995). Ecological burning is conducted on GYI to manipulate vegetation structure and composition by reducing the abundance of the dominant shrub, L. laevigatum, and stimulating regeneration of fire-adapted native grasses and overstorey trees (Bennett 1994). Mechanical slashing and ecological burning create spatial heterogeneity in Coastal Grassy Woodland, which is crucial for habitat selection (Kotler and Brown 1988). I aimed to improve our understanding of large herbivore ecology and organisation in a contemporary herbivore assemblage made up of species with disparate evolutionary histories. Specifically, I aimed to quantify the effect of habitat modification by an ecological process (fire) on fine-scale habitat use by (and thus resource partitioning among) sympatric native and introduced herbivore species. To do this, I conducted a large-scale experimental management trial

Chapter 4 Fine scale habitat use 115

(Walters and Holling 1990) designed to simultaneously investigate changes in the composition, structure and species richness of Coastal Grassy Woodland vegetation on Yanakie Isthmus following fire, and associated changes in fine- scale habitat use by herbivore species. Two general hypoptheses emerge:

1. Habitat modification by an ecological process (fire) should influence fine-scale habitat use by herbivores (i.e., result in differential use of burnt and unburnt habitats) and thus alter herbivore community structure. Fire can modify habitat sufficiently for mammalian herbivore communities to change in composition and abundance (Catling 1991). Altered vegetation structure following fire can provide food and shelter for some herbivores, while removing these resources for others (Catling 1991). That is, while increased quantity and quality of forage (Leigh et al. 1991) and changes in vegetation composition following fire can benefit some herbivores (Van Dyke and Darragh 2006), inter- specific differences in morphology result in differential responses to vegetation change after fire (Southwell and Jarman 1987).

2. Herbivore species should partition habitat resources at a fine scale to reduce competition and therefore use manually slashed and unslashed habitats differentially. Microhabitat can be a major dimension of habitat partitioning (Johnson 1980). In particular, herbivores can respond differentially to encroachment of habitat by woody shrubs (Riginos and Young 2007), and preferences for open and closed habitats vary between species (e.g., Fa et al. 1999, Schmidt et al. 2010).

Methods To simultaneously investigate changes in the composition, structure and species richness of Coastal Grassy Woodland vegetation on Yanakie Isthmus following fire, and associated changes in fine-scale habitat use by five herbivore species, I conducted a trial of ecological burning, whereby fire was applied to maintain biodiversity values of an area. I quantified fine-scale habitat use by herbivores, biomass accumulation and vegetation composition and structure prior to the fire and monitored changes in plant and herbivore communities in response to the burn, an approach recommended by Van Dyke and Darragh (2006).

Chapter 4 Fine scale habitat use 116

Study site I conducted this study on Yanakie Isthmus (38º 53' S; 146º 14' E), a 6874- ha area of Wilsons Promontory National Park, Victoria, Australia. My study sites were in Coastal Grassy Woodland, the dominant broad vegetation type on Yanakie Isthmus. Coastal Grassy Woodland typically has a sparse canopy of Eucalyptus pryoriana, Banksia integrifolia, Allocasuarina littoralis and Allocasuarina verticillata, occasional Acacia mearnsii shrubs in the understorey, and a ground layer of grasses, sedges, herbs and Pteridium esculentum (Davies and Oates 1999). However, Coastal Grassy Woodland on Yanakie Isthmus has undergone severe modification associated with encroachment of native shrubs (Bennett 1994). I divided Coastal Grassy Woodland into two vegetation strata: (1) scrub/dune woodland, and (2) slashed swale. Scrub/dune woodland, classified as Coastal Alkaline Scrub under the Ecological Vegetation Class (EVC) system, encompasses dune woodland (dunes retaining a woodland structure with a sparse canopy of A. verticillata, Banksia integrifolia, varying densities of L. laevigatum and an understorey of grasses and herbs) and scrub woodland (areas of inter-dune swale characterised by dense, tall stands of L. laevigatum with a sparse ground layer including bryophytes and lichens) (University of Ballarat 1999). Slashed swale, classified as Shrub-invaded Calcareous Swale Grassland, also occurs in inter-dune swales and has been mechanically slashed to maintain an open structure by preventing the development of L. laevigatum scrub. Slashed swale generally has a low shrub layer dominated by L. laevigatum and a ground layer of herbs and grasses; a canopy layer is absent (Chesterfield et al. 1995, University of Ballarat 1999). As described in Chapter 3, predation rates on Yanakie Isthmus are likely to be low. I used four sites on Yanakie Isthmus, separated by at least 800 m (Figure 4.1): Big Hummock (Geocentric Datum Australia 94: 436400, 569200), Springs Track (GDA94: 435900, 5695600), Old Burn Track (GDA94: 434200, 5693100), and Varneys (GDA94: 434900, 5693300). I selected sites at which Coastal Grassy Woodland grass and tree species were present, but L. laevigatum was the dominant species, and where park managers judged that there was adequate continuous fuel to carry a fire.

Chapter 4 Fine scale habitat use 117

Study design I aimed to use a Before-After Control-Impact (BACI) design (Smith 2002), involving sampling before and after ecological burns at two 60-ha treatment sites, each paired with an adjacent 60-ha control site. Parks Victoria conducted an ecological burn at Big Hummock, but because this fire was patchy and of low-intensity in areas, ecological burning at the second treatment site was cancelled. Furthermore, because the burn at Big Hummock resulted in unburnt (control) patches interspersed by burnt (treatment) patches, I did not conduct post- fire sampling at control sites, which allowed me to focus resources on the impact site (Stewart-Oaten and Bence 2001). This effectively reduced my design to a simple before-after design (Green 1979) at the impact site only, although I used the pre-fire data from the three control sites to assess how representative of Coastal Grassy Woodland on Yanakie Isthmus the Big Hummock site was. I conducted faecal pellet counts, biomass sampling and floristic surveys (see below for descriptions of which these parameters were selected) within a central 10-ha (200 × 500 m) area at each site to minimise edge effects associated with herbivore movement. I randomly placed plots for pre-fire pellet counts, floristic surveys and biomass sampling using intercepts of a 20-m grid as potential plot centres. In each case, I stratified sampling by two vegetation strata: (1) scrub/dune woodland, and (2) slashed swale. I conducted post-fire pellet counts at Big Hummock in the same plots used for pre-fire counts. However, because only six pre-fire floristic plots were burnt, and because biomass sampling was destructive, I selected new plots for post-fire floristic and biomass sampling. In addition, I introduced a second level of stratification for post-fire floristic and biomass sampling: (1) burnt, and (2) unburnt patches. I selected these new plots haphazardly, walking a random distance and compass bearing into burnt patches to place burnt plots, then walking a random distance and bearing from the edge of burnt patches to place unburnt plots. The study design was complex to allow simultaneous testing of several hypotheses, and because the initial study design failed due to cancellation of one of the two planned ecological burns. Therefore, the study design is summarised in table form (Table 4.1).

Chapter 4 Fine scale habitat use 118

Ecological burning Big Hummock was burnt by Parks Victoria staff on 11 December 2003 between 1320 and 1545 hours. The average air temperature during the burn was 21.1º C (range 19.9-22.7 ° C) and average humidity was 45% (range 39-50%). Average wind speed was 12.7 km h-1 (max. 31.4 km h-1) and wind direction varied from westerly to south-westerly. Where there were remnant grasses, the fire burnt well. However, the majority of slashed swale did not burn, or burnt only in patches, due to low fuel loads and breaks in fuel continuity. Areas with a heavy fuel load from manual slashing burnt hotly. The fire was influenced by a strong wind, and where it reached L. laevigatum crowns in the scrub/dune woodland it burnt thoroughly and rapidly in a narrow strip (J. Whelan, Parks Victoria, pers. comm.).

Fine scale herbivore habitat use Five mammalian herbivore species occur on Yanakie Isthmus: the native eastern grey kangaroo, swamp wallaby, and common wombat, and the introduced hog deer and European rabbit. To estimate the density of each mammalian herbivore species, as an index of activity and thus habitat use within and across sites prior to and after burning, I used faecal pellet counts (Mayle et al. 1999). Faecal pellet counts can be used to assess habitat use, providing an indicator of feeding patterns in different habitats (Bailey and Putman 1981), because increased pellet density is expected to be proportional to the time spent in a particular habitat (Hannan and Whelan 1989). There are known problems with this technique (Taylor and Williams 1956, Robinette et al. 1958, Wallmo et al. 1962, Van Etten and Bennet 1965), for example the assumption of a proportional relationship between pellet density and time spent in a habitat can be violated by non-regular defaecation (e.g., related to activity and diet), and differential abilities to locate pellet groups in different vegetation types (Bailey and Putman 1981). However, a pilot study showed that direct survey techniques are impractical on Yanakie Isthmus due to dense vegetation and low contact rates. The faecal pellet count method is considered appropriate for simultaneously surveying species whose activity patterns and behaviour differ (e.g., Lunney and O'Connell 1988) and avoids many of the detection biases that occur for direct counts in dense vegetation with variable topography (Hill 1981b). In particular, pellet counts are

Chapter 4 Fine scale habitat use 119 useful when animals are nocturnal or shy (Ellis et al. 1998, Evans and Jarman 1999), or their foraging habitats necessitate ground surveys (e.g., swamp wallabies; de Munk 1999). For these reasons, many studies have used faecal pellet counts to estimate habitat resource use (e.g., Hill 1978, Floyd 1980, Taylor 1985b, Lunney and O'Connell 1988, Ramsey and Engeman 1994, de Munk 1999), and studies such as that by Forsyth et al. (2007) have demonstrated positive correlations between herbivore densities and pellet counts. I only had time to conduct two pre-fire faecal pellet counts per site: (1) an initial count of pellet standing crop and simultaneous clearance of plots between 15-28 October 2003, and (2) a pellet accumulation count and clearance between 11-26 November 2003. I conducted four post-fire pellet accumulation counts at Big Hummock only, at bi-monthly intervals from 16 July 2004 to 16 February 2005. In each 3-m radius plot (n = 40 per site), I systematically searched for pellets, pushing the vegetation aside (Hickling 1986). I counted individual pellets for eastern grey kangaroos, swamp wallabies and common wombats, and pellet groups for hog deer and European rabbits, defined following Hickling (1986) as  6 pellets of the same defaecation, with at least one visible above the ground litter. I identified pellets based on size, shape and colour according to Triggs (2003).

Biomass accumulation To quantify the availability of forage for grazing, and off-take of this food resource by herbivores, and thus obtain a secondary index of herbivore habitat use, I measured fine plant biomass, that is, the total above-ground biomass of diameter < 1 cm ( 0.01 g). I then separated pre- and post-fire samples into live and dead plant material (e.g., leaf litter and fine woody materials). To provide an estimate of pre-fire fine plant biomass, I harvested two sub-samples of biomass from 5 plots in each vegetation stratum per site on 6-11 November 2003. Sub- samples were 0.25-m2 quadrats that I randomly selected from a 0.5-m2 sampling frame consisting of four quadrats (Morgan 1999). I harvested biomass to ground level (under Research permit 10002678 of the Flora and Fauna Guarantee Act 1988 and National Parks Act 1975). I used 20 total herbivore exclusion plots to estimates post-fire biomass at Big Hummock. Plots (1-m diameter) consisted of 1-m high wire netting (4-cm

Chapter 4 Fine scale habitat use 120 diameter hexagonal) fences pegged to the ground, and 20 unfenced control plots. I placed 10 exclusion plots and 10 control plots in each vegetation stratum, half in burnt patches and half in unburnt patches, three weeks after the fire before any regeneration had occurred. Because sampling was destructive, I harvested post- fire biomass plots once, 12 months after the fire, on 15 December 2004. To minimise edge effects, I took post-fire samples from one 0.25 m2 quadrat at the centre of plots. Before weighing samples I air-dried and then oven dried them at 80° C for 48 h (Morgan 1999).

Floristic surveys To examine the relationship between herbivore habitat use and changes in food and shelter resources following fire, I conducted a floristic survey before the fire (11-26 November 2003) at the 40 pellet count plots at each site, then repeated this survey after the fire (12-18 December 2004) at 20 new plots (i.e., 10 plots per vegetation stratum, half within burnt and half within unburnt patches) at Big Hummock. During each survey I estimated the composition, structure and species richness of plant species, with particular reference to overstorey trees that may provide shelter and browse (i.e., Banksia integrifolia, Allocasuarina verticillate), native graminoid species that may be important food sources for grazers (i.e., Themeda triandra, Imperata cylindrica, Austrodanthonia setacea and native Poa spp.) and dominant native shrubs that may provide shelter and browse (i.e., L. laevigatum and A. longifolia). To quantify vegetation composition I estimated the percent cover (to the nearest 10%, or as 1% if cover was < 5%) of all angiosperms rooted within the plot. To describe vegetation structure, I estimated the percent cover of the shrub and canopy layer (to the nearest 10%) and for the ground layer, I estimated the percent cover of litter, logs (coarse fallen litter with a diameter > 5 cm), bryophytes and lichens, other vegetation, and bare ground. During the pre-fire survey, I counted stems (seedlings, saplings, trees and shrubs) at a random sub-sample of the floristic plots per vegetation stratum at Big Hummock and Old Burn Track. I measured (or estimated for trees) plant height ( 0.1 cm) (Benwell 1998) and recorded the length ( 0.1 cm) and the number of live, fully-developed leaves (Mower et al. 1997) for one haphazardly selected

Chapter 4 Fine scale habitat use 121 branch per plant. Where there were no branches, I took these measurements for the main stem. I aimed to monitor plants over time, remeasuring these same parameters, to provide information on growth, regeneration and mortality (Benwell 1998). However, due to changes in the study design I did not resample these plots after the burn and instead used these data to compare the form of shrub and tree species between Big Hummock and Old Burn Track.

Statistical analyses I used analysis of variance (ANOVA) (Sokal and Rohlf 1995, Underwood 1997) to analyse most data, using a three-step approach: (1) inter-site comparisons of pre-fire data, (2) pre- vs. post-fire comparisons at Big Hummock (because there were inter-site differences in some variables, and because post-fire sampling was conducted at Big Hummock only), and (3) comparisons of burnt and unburnt patches during the post-fire period at Big Hummock. I plotted residuals for these linear models against the corresponding fitted values to check for distributional problems. If these plots were wedge shaped rather than data points being evenly spread across plots, I applied square-root-transformations to improve residual distributions. For all analyses I used an  = 0.05% level of significance. To evaluate how representative Big Hummock was of Coastal Grassy Woodland on Yanakie Isthmus, I tested for differences among sites in herbivore faecal pellet counts, biomass and vegetation composition, structure and richness (see below). In addition, I used two-tailed t-tests to compare mean sapling and tree height, mean branch length, mean number of leaves and mean number of leaders (all square-root-transformed) between Big Hummock and Old Burnt Track. To estimate herbivore habitat use, I converted faecal pellet accumulation ˆ counts for each species to estimates of herbivore density Da during the pre-fire period at each site, and during the post-fire period at Big Hummock, following Mayle et al. (1999):

Dˆ s Dˆ a  pˆ tˆ

ˆ ˆ where Ds is the estimated density of pellets or pellet groups, t is the estimated time between visits (days) and pˆ is the estimated daily rate of production of pellets or pellet groups per animal. Based on the densities calculated for each

Chapter 4 Fine scale habitat use 122 species, I then calculated the density of herbivores with predominantly grazer diets on Yanakie Isthmus (eastern grey kangaroos, common wombats and European rabbits, Chapter 5; Davis et al. 2008) and the density of herbivores with predominantly browser diets (swamp wallabies and hog deer, Chapter 5; Davis et al. 2008) prior to and after the burn, to allow comparison of relative use of the area by (and thus potential competition between) herbivores using these two feeding strategies. I used species-specific mean daily defaecation rates from published literature: 20.3 hog deer pellet groups (Dhungel 1985a); 18.5 European rabbit pellet groups, converted from 325 pellets (Wood 1988) based on an average of 17.5 ± 0.04 pellets per group at WPNP (Chapter 2; Davis et al. in review); 493 eastern grey kangaroo pellets (Johnson et al. 1987); 150 swamp wallaby pellets (Floyd 1980); and 90 wombat pellets, mid-point of the range provided by Triggs (1996). Several sources of error can occur when measuring defecation rates (Southwell 1989), as defecation rates are influenced by diet, activity, herbivore age and whether herbivores are captive or free-ranging. Where possible I used defecation rates estimated for free-ranging animals, however, Dhungel’s (1985a) estimate of hog deer defecetation rate was taken from captive animals, and several studies did not provide information on this parameter. Where estimates of the error associated with defaecation rates were not provided I assumed standard errors of ± 20% to approximate the error observed in other studies (e.g., Hill 1978). I estimated the precision of density estimates following Laing et al. (2003):

ˆ 2 ˆ 2 2 [cv(Da )]  [cv(Ds )] [cv( pˆ)] where cvtˆis the coefficient of variation of pˆ , and similarly for the other terms. I then used coefficient of variation values to calculate 95% confidence intervals for density estimates. The fastest mean faecal pellet decay time recorded by Davis et al. (Chapter 2; in review) in Coastal Grassy Woodland on Yanakie Isthmus was 134 (± 7) days. Therefore, given maximum intervals of two months between pellet counts, I assumed pellets persisted for the duration of the accumulation period, so did not correct for pellet loss. Because all sites were in Coastal Grassy Woodland within the same geographic region, I also assumed that environmental conditions and thus pellet decay rates were constant across sites.

Chapter 4 Fine scale habitat use 123

Dry matter intakes vary according to body size, digestive strategy and metabolic requirements (Dawson 1989). Metabolic rate can be used to predict energy and nutrient requirements (Hume 1999). Following Schaller (1967), I converted herbivore density estimates to estimates of metabolic demand at the population level, to provide a more meaningful measure than density for comparing resource use by species with varying body size. Du Toit and Owen- Smith (1989) define population metabolism as the product of population density and energy use per individual. I calculated population metabolism for the entire GYI as the product of mean population abundance for GYI and basal metabolic rate per individual. There are several measures of metabolic rate (Hume 1999). Field metabolic rate is the energy cost of free existence (Hume and Stevens 1995) and provides the most realistic estimate of the actual energetic needs of an animal (Hume 1999). However, field metabolic rates include energetic costs associated with factors such as growth and reproduction and can be highly variable between sexes and seasons (Hume 1999). Therefore, inter-specific comparisons are commonly made using Basal Metabolic Rate (BMR) (Hume 1999), which is the minimum rate of metabolism compatible with endothermy (McNab 1988) and correlates strongly with the energy required for longer term energy balance (Hume and Stevens 1995). I calculated BMR following Hume (1999): BMR = aMb where a is a proportionality coefficient that differs among species, and b is an empirically determined exponent that expresses the rate of change of R with changes in body mass (M). The estimates of a and b that I used were from Dawson and Hulbert’s (1970) ‘marsupial line’ and Kleiber’s (1961) ‘Kleiber line’, which yield BMRs for the average marsupial of 201 kJ kg-0.75 day-1 and for the average eutherian of 293 kJ kg-0.75 day-1, respectively. These traditional values are based on small sample sizes, but are widely used for comparative purposes (Hume 1999). Dawson and Hulbert’s (1970) marsupial line may overestimate metabolic demand for wombats, as species such as the southern hairy-nosed wombat Lasiorhinus latifrons have low metabolic rates relative to other marsupials (Wells and Green 1998). However, work by Gowland (1973) suggests that the standard metabolic rate of common wombats is near the marsupial average. I obtained weights for

Chapter 4 Fine scale habitat use 124 each species from the literature (Table 3.1). For the rabbit, I took the mid-point between the weights given for males and females by Myers (1983). I compared pre-fire pellet (or pellet group) standing crop (summed over the two pre-fire surveys and logx+1-transformed) for each herbivore species among sites, and between vegetation strata (slashed swale vs. scrub/dune woodland) using two-factor ANOVA. I did not record hog deer pellets at Springs Track and therefore excluded this site from inter-site comparisons for hog deer. Except for hog deer pellet groups, which were recorded in low numbers at Big Hummock, I compared daily pellet (or pellet group) accumulation rates (logx+1-transformed) for each species between pre- and post-fire surveys and between vegetation strata at Big Hummock using two-factor repeated measures ANOVA. To assess differences in pre-fire biomass among sites, between vegetation strata and between vegetation states (dead vs. live), I compared mean weights

(log10-transformed) of live and dead fine biomass, averaged over the two sub- samples per plot, using three-factor ANOVA. To determine the effect of grazing and fire on post-fire biomass, I used three-factor ANOVA to compare mean biomass (log10-transformed) between exclosure and control plots in burnt and unburnt patches within each vegetation strata. To compare live and dead biomass

(log10-transformed) in each vegetation strata between the pre- and post-fire sampling periods, I used two-factor repeated measures ANOVA. For the pre- vs. post-fire comparison, I used pre-fire data from all sites, because biomass did not vary significantly among sites. I compared vegetation structure variables (% vegetation cover in each the ground, shrub and canopy layers, and % cover of litter, logs, bryophytes and lichens, other vegetation in the ground layer, and bare ground) among sites and between vegetation strata using two-factor ANOVA. To examine changes in vegetation structure after fire, I used two-tailed t-tests to compare vegetation structure variables between the pre- and post-fire sampling periods. I used two- factor ANOVA to compare burnt and unburnt plots in each vegetation strata during the post-fire survey, for all parameters except canopy cover (recorded in one burnt plot only). I examined differences in the cover of native graminoid species, native shrubs (L. laevigatum and A. longifolia), and overstorey species (A. verticillata) at Big Hummock between the pre- and post-fire surveys, and between burnt and

Chapter 4 Fine scale habitat use 125 unburnt plots informally. I calculated the density of L. laevigatum, A. longifolia and A. verticillata using stem counts. To examine changes in species richness, I summed species in each functional group during the pre- and post-fire surveys, and in burnt and unburnt patches. I used non-metric multi-dimensional scaling (NMDS) to compare the species composition of plants within the ground, shrub and canopy layers among sites, between pre- and post-fire surveys, and for ground layer vegetation, also between burnt and unburnt patches. I based these 2-D and 3-D ordination spaces on Bray-Curtis dissimilarity matrices (Clarke 1993) for the percent cover of each plant species recorded at each site or in each survey or treatment. I applied the following standardisation within species to reduce the influence of abundant taxa on the NMDS:

1 xi  xi,min th xi  for the i variable, xi,max  xi,min where x is the percent cover of each plant species recorded per site, survey or treatment. Prior to creation of the dissimilarity matrix I excluded plant species that occurred at only one site, or in only one survey or treatment, as they did not contribute to systematic compositional differences between samples. I also excluded plots in which no cover was recorded.

Results Site comparability Except for hog deer, pre-fire herbivore densities were highest at Varneys (Table 4.2). The standing crop of deer, kangaroo and wombat faecal pellets varied significantly among sites (Table 4.3; Figure 4.2). Pre-fire deer pellet group counts were higher at Varneys and Old Burn Track than at Springs Track and Big Hummock (where only one deer pellet group was recorded). Kangaroo and wombat pellet counts were greater at Varneys than at the other three sites. Pre-fire pellet counts for rabbits and wallabies did not vary among sites. The cover of litter, bryophytes and lichen, other ground layer vegetation, bare ground, and vegetation in both the shrub and canopy layers varied significantly among sites during the pre-fire survey (Table 4.4). Litter cover was greater at Varneys and Old Burn Track than at Big Hummock and Springs Track,

Chapter 4 Fine scale habitat use 126 while the cover of bryophytes and lichen, and bare ground, were greatest at Big Hummock and Springs Track (Table 4.5). Both shrub cover and the cover of other ground layer vegetation were highest at Big Hummock, lowest at Varneys, and intermediate at Springs Track and Old Burn Track, whereas canopy cover was lowest at Springs Track, highest at Varneys and intermediate at Big Hummock and Old Burn Track. The pre-fire cover of logs did not vary significantly among sites (Table 4.4), nor did the fine biomass (Table 4.6). During the pre-fire period there was considerable overlap among sites in the species composition of ground layer plants (Figure 4.3) and shrub/canopy layer plants (Figure 4.4). The form of shrubs and trees varied between Big Hummock and Old Burn Track (Table 4.7; Figure 4.5). Seedlings, saplings, trees and shrubs were significantly taller, branch length significantly longer and the number of leaves significantly greater at Big Hummock than at Old Burn Track, while the number of leaders was significantly greater at Old Burn Track than at Big Hummock (Figure 4.5).

Fine scale herbivore habitat use During the pre-fire period, the density of rabbits and wombats was relatively high at all sites, compared to densities of the other three herbivore species; kangaroos occurred in intermediate densities, and in general the lowest densities were recorded for wallabies and deer, which occurred at similar densities (Table 4.2). Pre-fire wombat densities were particularly low at Springs Track (7- 25 times lower) compared to other sites, and hog deer were absent from Springs Track and Big Hummock. The post-fire density of rabbits at Big Hummock was greater than that of any other species, although unlike the pre-fire period, post-fire wallaby densities were also relatively high (Table 4.2). Post-fire densities of kangaroos and wombats at Big Hummock were intermediate and the densities of deer were relatively low. Population metabolism was relatively high for rabbits before and after the burn, relatively low for deer before and after the burn, relatively low for swamp wallabies before the burn but intermediate after the burn, and intermediate for wombats and kangaroos before and after the burn (Figure 4.6). Prior to the burn at Big Hummock, the density of herbivores that display predominantly grazer diets on Yanakie Isthmus (eastern grey kangaroos, common

Chapter 4 Fine scale habitat use 127 wombats and European rabbits, Chapter 5; Davis et al. 2008) was almost 400 times greater than the density of herbivores that display predominantly browser diets (swamp wallabies and hog deer, Chapter 5; Davis et al. 2008): this trend remained after the ecological burn, although it was much less pronounced (the density of grazers was 13 times greater than the density of browsers), following an increase in the density of browsers of over 800%, and a decrease in the density of grazers to only 26% of the pre-fire density. Prior to the burn at Big Hummock, the metabolic demand at the population level from herbivores that display predominantly grazer diets on Yanakie Isthmus was over 130 times greater than the density of herbivores that display predominantly browser diets. This trend remained after the ecological burn, although it was much less pronounced (the density of grazers was c. four times greater than the density of browsers), following a nine-fold increase in metabolic demand from browsers and a three- fold reduction in metabolic demand from grazers after the fire (Figure 4.6). Pre-fire deer pellet group counts were significantly higher in scrub/dune woodland than in slashed swale (Table 4.3; Figure 4.2). Post-fire deer pellet group counts at Big Hummock increased marginally, but were too low to compare formally between pre- and post-fire periods or between vegetation strata (Figure 4.7). Both pre-fire rabbit pellet group counts (Table 4.3; Figure 4.2) and post- fire rabbit pellet group accumulation rates at Big Hummock (Table 4.8; Figure 4.7) were greater in slashed swale than in scrub/dune woodland. European rabbit pellet group accumulation rates at Big Hummock were significantly lower during the post-fire period than during the pre-fire period (Table 4.8; Figure 4.7). Further, there was a significant interaction between the influence of survey period and vegetation strata on rabbit pellet group accumulation: the difference in accumulation rates between slashed swale and scrub/dune woodland was more pronounced during the pre-fire period than during the post-fire period. Pre-fire kangaroo pellet counts were significantly greater in slashed swale than in scrub/dune woodland, and this effect of strata interacted with site, with differences in pellet counts between strata most pronounced at Varneys (Table 4.3; Figure 4.2). Similarly, post-fire kangaroo pellet accumulation rates at Big Hummock were higher in slashed swale than in scrub/dune woodland (Table 4.8; Figure 4.7). Kangaroo pellet accumulation rates at Big Hummock did not vary

Chapter 4 Fine scale habitat use 128 significantly between the pre- and post-fire periods. Pre-fire wallaby pellet counts were significantly greater in scrub/dune woodland than in slashed swale (Table 4.3; Figure 4.2), yet there was no effect of strata on pre- or post-fire swamp wallaby pellet accumulation rates (Table 4.8; Figure 4.7). Wallaby pellet accumulation rates were significantly faster during the post-fire survey period than during the pre-fire survey. There was no significant difference between vegetation strata in pre-fire wombat pellet counts (Table 4.3; Figure 4.2), nor in post-fire pellet accumulation rates (Table 4.8; Figure 4.7). However, there was an interaction between the influence of site and strata on pre-fire wombat pellet counts: counts were higher in slashed swale at Varneys and Big Hummock, but higher in scrub/dune woodland at Springs Track and Old Burn Track (Table 4.3; Figure 4.2). Wombat pellet accumulation rates at Big Hummock did not vary significantly between the pre- and post-fire periods (Table 4.8; Figure 4.7).

Biomass accumulation Fine plant biomass at Big Hummock did not differ significantly between the pre- and post-fire sampling periods, nor between vegetation strata in either of these sampling periods (Tables 4.6 and 4.9). However, sampling period, vegetation strata and vegetation state interacted to influence biomass (Table 4.9; Figure 4.8). During the pre-fire period there was significantly greater dead than live biomass in both vegetation strata, particularly in scrub/dune woodland (Table 4.6; Figure 4.8). Conversely, during the post-fire sampling period there was more live than dead biomass in both strata (Table 4.9; Figure 4.8). During the post-fire period, fine biomass was significantly greater in exclosure plots (mean 90.9 g m-2 ± s.e. 12.3) than in control plots (mean 41.5 g m-2 ± s.e. 7.3), but did not vary significantly between burnt and unburnt patches (Table 4.8).

Floristic surveys The species composition of ground layer plants was clearly different between the pre- and post-fire surveys at Big Hummock (Figure 4.3). In particular, an outlier indicated that at least one species occurred in very different proportions between pre- and post-fire periods (Figure 4.3). Species richness at Big Hummock was marginally higher during the pre-fire survey than during the

Chapter 4 Fine scale habitat use 129 post-fire survey, although the proportions of species in each functional group were similar during the pre- and post-fire surveys (Figure 4.9). There was some separation in species composition of ground layer plants between burnt and unburnt patches during the post-fire survey at Big Hummock, but this difference was not marked apart from for one species, indicated by an outlier on the plot (Figure 4.3). Notably, species richness was greater in unburnt patches, particularly for grasses (Figure 4.9). There was no clear separation in the species composition of plants in the shrub/canopy layer between the pre- and post-fire surveys at Big Hummock (Figure 4.10). There was a reduction by almost five-fold in the cover of native graminoid species (Themeda triandra, Imperata cylindrica, Austrodanthonia setacea and native Poa spp.) between pre- and post-fire surveys at Big Hummock, and cover of these graminoid species was almost seven times greater in unburnt than in burnt patches (Figure 4.10). At Big Hummock, the cover of Allocasuarina verticillata in the post-fire period increased by almost one and a half times relative to the pre- fire period, and was over four times greater in unburnt than burnt patches. Banksia integrifolia was not recorded at Big Hummock. There was almost a 30-fold reduction in the cover of Acacia longifolia between the pre- and post-fire surveys at Big Hummock, with no cover recorded in unburnt patches and little cover recorded in burnt patches. Overall, the cover of Leptospermum laevigatum during the post-fire survey at Big Hummock was almost half that recorded during the pre-fire survey, although mean post-fire cover of L. laevigatum in unburnt plots was almost nine times greater than in burnt patches. The density of L. laevigatum during pre-fire surveys at Big Hummock and Old Burn Track was ~2 m-2, ~50 times greater than the densities of A. verticillata and A. longifolia (0.03 and 0.04 m-2 respectively), and over four times greater than the density of any other tree or shrub species. Vegetation structure during the pre-fire survey, measured as cover of litter, logs, shrub layer vegetation and canopy layer vegetation, was significantly higher in scrub/dune woodland than in slashed swale, but in all cases except log cover, the influence of vegetation strata varied between sites (Tables 4.4-4.5). The difference in litter cover between vegetation strata was more pronounced at Varneys and Old Burn Track than at Big Hummock or Springs Track. The difference in shrub cover between vegetation strata was most pronounced at

Chapter 4 Fine scale habitat use 130

Springs Track (where cover in scrub/dune woodland was 23 times greater than in slashed swale) and least at Old Burn Track (where cover in scrub/dune woodland was nine times greater than in slashed swale). The difference in canopy cover between vegetation strata was most pronounced at Old Burn Track (where cover in scrub/dune woodland was almost 100 times greater than in slashed swale) and least at Big Hummock (where cover in scrub/dune woodland was ~30 times greater than in slashed swale). The cover of bryophytes and lichen was significantly higher in slashed swale than in scrub/dune woodland, as was the cover of other ground layer vegetation, the difference between strata being more pronounced at Varneys and Old Burn Track than at Big Hummock or Springs Track. Although there was no overall effect of vegetation strata on bare ground, site and vegetation strata interacted to affect bare ground: bare ground was higher in scrub/dune woodland than in slashed swale at Big Hummock and Springs Track, but higher in slashed swale than in scrub/dune woodland at Old Burn Track and Varneys. Changes in vegetation structure variables between the pre- and post-fire surveys at Big Hummock were mixed (Table 4.11; Figure 4.11). The cover of bryophytes and lichen, other ground layer vegetation and shrub layer vegetation were significantly lower during the post-fire survey than they had been during the pre-fire survey. In contrast, bare ground increased significantly during the post- fire survey, while the cover of litter, logs and canopy layer vegetation did not vary between pre- and post-fire surveys. During the post-fire period, there was significant variation between burnt and unburnt patches in most of the vegetation structure variables (Table 4.12). Bare ground was significantly higher in burnt than in unburnt patches (Table 4.12; Figure 4.11). In contrast, the cover of litter, bryophytes and lichen, other ground layer vegetation and shrub layer vegetation were all significantly lower in burnt than in unburnt patches (Table 4.12; Figure 4.11). Informal comparisons suggested that canopy cover was also substantially lower in burnt patches than unburnt patches, being recorded in only one burnt patch, and being much greater in scrub/dune woodland than in slashed swale (Figure 4.11). Log cover was the only vegetation structure variable to show no response to burning, and to vary significantly between vegetation strata during the post-fire period: in contrast to

Chapter 4 Fine scale habitat use 131 the pre-fire period, log cover during the post-fire period was greater in slashed swale than in scrub/dune woodland.

Discussion Fire can modify habitat substantially for mammalian herbivore communities (Catling 1991). Some plant species decrease in abundance in response to fire, while others increase (Hoffman 1999), resulting in overall changes to the structure and composition of vegetation communities (Whelan 1995, Tozer and Bradstock 2002). Altered vegetation structure following fire can provide habitat for some herbivores, while removing it for others (Catling 1991). While increased quantity and quality of forage (Leigh et al. 1991) and changes in vegetation composition following fire can benefit herbivores (Van Dyke and Darragh 2006), different species experience these benefits differently, resulting in differential responses to vegetation change after fire (hypothesis 1), for example, depending on differences in species-specific morphology (Southwell and Jarman 1987). One year after ecological burning at Big Hummock on Yanakie Isthmus, shifts in Coastal Grassy Woodland vegetation composition, and reductions in plant species richness were evident, indicating changes in the availability of food and shelter resources to herbivores. Cover of the overstorey tree Allocasuarina verticillata increased following the burn, although canopy cover in general was low in burnt patches. Cover of encroaching native shrubs decreased following the burn, as did grass cover and richness. Overall, the density of herbivores, particularly grazers, decreased within the two years following the burn, although the density of browsers increased. These changes probably occurred in response to floristic changes in response to the ecological burn, although in part, floristic changes are likely to be a result of changes in the grazing regime associated with shifts in herbivore densities and community composition (Bridle and Kirkpatrick 2001, Allcock and Hik 2004, Parsons et al. 2006, Prober et al. 2007).

Herbivore responses to ecological burning Within the two years after ecological burning at Big Hummock, European rabbit pellet accumulation rates decreased, although rabbit densities remained high relative to the densities of other herbivore species. In contrast, swamp

Chapter 4 Fine scale habitat use 132 wallaby pellet accumulation rates increased within the two years after the burn, resulting in a shift from relatively low swamp wallaby densities prior to the burn, to relatively high densities after the burn. Pre- and post-fire common wombat and eastern grey kangaroo pellet accumulation rates did not vary, although relative to the densities of other herbivores, wombat densities shifted from being relatively high prior to the burn to intermediate after the burn, while eastern grey kangaroos remained at intermediate densities. Hog deer remained at low densities following the fire. However, variations in these density trends are apparent when inter- specific comparisons are based on population metabolism, a measure which provides a more meaningful picture of how resources are shared by species of different size (duToit and Owen-Smith 1989). Population metabolism was relatively high for rabbits before and after the burn, relatively low for deer before and after the burn, relatively low for swamp wallabies before the burn but intermediate after the burn, and intermediate for wombats and kangaroos before and after the burn. Thus it is clear that habitat modification by fire affected habitat use by herbivore species differently. When species-specific responses to the burn at Big Hummock are considered in combination, it appears that differential responses to fire were probably based on differences in diet. Prior to the burn, the density and population metabolism of herbivores that display predominantly grazer diets on Yanakie Isthmus (eastern grey kangaroos, common wombats and European rabbits, Chapter 5; Davis et al. 2008) was far greater than the density and population metabolism of herbivores that display predominantly browser diets (swamp wallabies and hog deer, Chapter 5; Davis et al. 2008). This trend remained after the ecological burn, although it was much less pronounced, following an increase in the density and population metabolism of browsers and a decrease in the density and population metabolism of grazers. Despite the overall reduction in herbivore density following the burn, and particularly the reduction in the density and metabolic demand of grazers, similar fine biomass levels during pre- and post-fire sampling periods indicate that herbivore pressure was not reduced significantly after the burn. Further, herbivory after the burn was great enough that herbivore exclusion had a significant effect on biomass accumulation. Although browsers can reduce fine biomass (Riggs et al. 2000), fine biomass reduction is more likely to reflect off-take by grazing

Chapter 4 Fine scale habitat use 133 species (e.g., Liedloff et al. 2001, Roques et al. 2001, Briggs et al. 2005, Prober et al. 2007), particularly as estimates of population metabolism for herbivore species at Big Hummock suggest that grazing wombats and rabbits consumed the greatest vegetation biomass. High pre- and post-fire metabolic demand from the wombat population at Big Hummock, relative to all species except rabbits, supports suggestions that wombats are competitively superior to most other herbivore species on Yanakie Isthmus, based on relatively high abundance and population metabolism in this system as a whole (Chapter 3). The wombat has adaptations that allow it to save energy and reach high population densities in low productivity habitats (Johnson 1998), such as an ability to burrow and utilise low quality forage (Hume 1999). These adaptations may give the wombat a competitive advantage over other herbivore species (Hume 1999). The production of open grassland communities, and increased nutritional value of grasses after fire (Leigh et al. 1987) should favour grazers (Calaby 1966, Leigh et al. 1987, Leigh et al. 1991), and preferential use of burnt areas has been demonstrated for eastern grey kangaroos (Southwell and Jarman 1987, Meers and Adams 2003), rabbits (Cohn and Bradstock 2000), and wombats (Leigh and Holgate 1979). Therefore, the lack of a positive response by kangaroos and wombats to ecological burning, and in particular, the negative response of rabbits, was surprising. Increased use of burned areas is usually a temporary phenomenon (e.g., Biondini et al. 1999), and the ability of this study to detect heavy use of Big Hummock immediately after the burn (e.g., Southwell and Jarman 1987) may have been limited by commencement of post-fire pellet counts seven months after the burn. However, except for rabbits, herbivore pellet groups generally persist on Yanakie Isthmus for at least seven months (Chapter 2; Davis et al. in review), and herbivore responses are likely to be measurable for at least a year following fire (Newsome et al. 1975, Southwell and Jarman 1987). Therefore, the post-fire changes in herbivore densities detected by this study are likely to reflect real patterns in herbivore responses to the burn at Big Hummock. In studies where herbivores have responded positively to burning (e.g., Southwell and Jarman 1987), food increased after fire, and cover was not removed, providing two important habitat elements for large ground-dwelling herbivores such as common wombats and eastern grey kangaroos (Lunney and O'Connell 1988). While the proportion of plant species present from each

Chapter 4 Fine scale habitat use 134 functional group was similar before and after the burn at Big Hummock, burning changed the species composition of ground layer plants. Grasses and forbs did not increase, as is typically expected after fire (Van Dyke and Darragh 2006); rather there was an overall reduction in ground layer vegetation after the burn, which often occurs due to vegetation removal by fire (Cheney 1981, Noy-Meir 1995). Specifically, there was a large reduction in the cover of native graminoid species following the burn, as well as reduced cover and species richness of native graminoid species in burnt patches than in unburnt patches, and forb cover did not change. The negative response by grasses to fire was unexpected, as fire typically stimulates grass production (Leigh et al. 1987), germination and growth (Antos et al. 1983, Prober et al. 2007), thus increasing grass dominance (Hodgkinson and Harrington 1985, Van Dyke and Darragh 2006). However, it is possible that post-fire recovery of grasses was limited by low abundance of fire- resilient species following a recent history of infrequent burning (Prober et al. 2007). Similarly, forbs commonly increase after fire (Drewa and Havstad 2001, Van Dyke and Darragh 2006), but their recovery after the burn at Big Hummock may have been influenced by the intensity (Hodgkinson and Harrington 1985, Leigh et al. 1987) or season of the fire (Christensen et al. 1981), and the loss of fire-adapted species due to infrequent fire (Scarlett and Parsons 1990). The reduced cover of native graminoid species, and lack of increase in forbs after the fire, indicate that food availability for grazers was reduced by the burn. This suggestion is supported by the increase in bare ground cover after the burn, particularly in burnt patches, a common result of removal of vegetation biomass by fire (e.g., Noy-Meir 1995, Wahren et al. 2001). Further, in contrast to previous findings that fire can stimulate the production of fine biomass (Liedloff et al. 2001), fine biomass at Big Hummock was no greater after the burn than prior to the burn, nor did it differ between burnt and unburnt patches, possibly due to low intensity of the fire. The lack of response of fine biomass to burning provides further support for the suggestion that fire did not increase food resources for grazers. However, prior to the burn fine biomass consisted mainly of dead material, while after the burn it was mainly live plant matter, suggesting that post-fire fine biomass was largely new growth, which would provide preferred forage for grazers. Further, shrub layer vegetation was reduced after ecological burning. Work on Yanakie Isthmus suggests that wombats may decline with

Chapter 4 Fine scale habitat use 135 reductions in shrub cover and rabbits may decline with reductions in the cover of L. laevigatum (Chapter 3). Hence, a decrease in refuge from predators near feeding grounds may have reduced habitat suitability for these grazers, as has been observed during other studies (e.g., Thompson 1994). The increase in browsers, particularly swamp wallabies, following the burn contrasts the findings of Newsome et al. (1975), who observed a decline in swamp wallaby abundance after a fire. Moreover, the increase in swamp wallaby density after the burn at Big Hummock does not support the suggestion by Hobbs and Swift (1985) that burned sites provide low quality habitat for large browsers due to reduced dry matter, metabolizable energy, and nitrogen following reductions in shrub biomass. The lack of negative response of swamp wallabies to fire at Big Hummock may have been due to the relatively small size of this fire, which was unlikely to have killed such large mobile herbivores, as can occur in large fires such as that studied by Newsome et al. (1975). Although the composition of the shrub/canopy layer was similar before and after the burn at Big Hummock, cover of the dominant shrub species, Leptospermum laevigatum and Acacia longifolia, species killed by fire (Hazard and Parsons 1977, Judd 1990), was reduced. Reduction of woody plant abundance by fire (Briggs et al. 2002, Van Dyke and Darragh 2006) may stimulate vegetative regrowth (Gill and Bradstock 1995) and increase the palatability of woody plants (Scholes and Archer 1997), providing additional forage for browsers. Further, shelter availability may have influenced changes in habitat use. Catling et al. (2001) suggest that the abundance of large wallabies such as the swamp wallaby may peak after fire if structural complexity is adequate. Although vegetation complexity at Big Hummock was reduced after the fire, which may reduce habitat quality for swamp wallabies (Di Stefano 2005), there was an increase in cover of the canopy species, Allocasuarina verticillata, possibly due to stimulation of regeneration (Gent and Morgan 2007), which may have favoured wallabies, and work on Yanakie Isthmus suggests that decreases in shrub cover do not influence wallaby density (Chapter 3), possibly due to low predation levels. Although a number of factors may be responsible for the positive response of wallabies to ecological burning at Big Hummock, the lack of response for kangaroos and wombats, and the negative response of rabbits (and grasses), suggest that forage became less suitable for grazers than for browsers following

Chapter 4 Fine scale habitat use 136 ecological burning. This suggestion is supported by the small increase in density of hog deer, another herbivore that can consume browse (Chapter 5; Davis et al. 2008) and utilise fresh regrowth after fire (Mayze and Moore 1990). Contrasting responses of herbivore species to ecological burning, particularly for herbivores with grazer versus browser diets, indicate differential resource requirements within this herbivore assemblage, and hence mechanisms for fine-scale partitioning of habitat resources.

Inter-specific differences in the use of vegetation strata As predicted (hypothesis 2), there were inter-specific differences in the use of mechanically slashed and unslashed habitats at Big Hummock, in line with differential responses of herbivores elsewhere to woody shrub encroachment (e.g., Riginos and Young 2007), and differential preferences among species for open and closed habitats (e.g., Fa et al. 1999, Schmidt et al. 2010). These results indicate partitioning of habitat resources at a fine scale. Pellet counts indicated greater use of slashed swale than scrub/dune woodland by rabbits and kangaroos at Big Hummock prior to, as well as after the burn. This pattern reflects the favoured use of open areas for grazing by eastern grey kangaroos (Taylor 1971, Poole 1995) and the favoured use of open scrubland by European rabbits (Fa et al. 1999). Although fine biomass did not differ between vegetation strata at Big Hummock, the cover of ground layer vegetation was greater in slashed swale than in scrub/dune woodland, while the cover of shrub layer vegetation was lower. The greater use of slashed swale by rabbits is therefore not surprising, given habitat selection by European rabbits is a compromise between food availability and protection from predators (Moreno et al. 1996, Lopez-Darias and Lobo 2009), but when predation pressure is low (Palomares and Delibes 1997) or when dense shrub cover hinders movement (Fa et al. 1999) or reduces ground layer cover, as it does on Yanakie Isthmus (Bennett 1994), rabbits prefer intermediate levels of cover (Carvalho and Gomes 2004). Similarly, the greater use of slashed swale than scrub/dune woodland by eastern grey kangaroos is unsurprising, given that this species favours foraging habitats with a ground layer of herbs and abundant, high-quality grasses (Hill 1982, Taylor 1984), and few shrubs (Emison et al. 1978, Short and Grigg 1982). The eastern grey kangaroo does require these open grassy areas for foraging to be interspersed

Chapter 4 Fine scale habitat use 137 with more densely vegetated areas offering lateral cover (McCullough and McCullough 2000, Moore et al. 2002, Schmidt et al. 2010), which would be provided in some slashed swale patches, or by the surrounding fringe of scrub/dune woodland. In contrast to patterns of habitat use by kangaroos and rabbits, pellet counts prior to the burn indicated greater use of scrub/dune woodland than slashed swale by hog deer and swamp wallabies. Scrub/dune woodland had greater shrub and canopy layer cover than did slashed swale. The swamp wallaby is more of a browser than larger macropods such as the eastern grey kangaroo (Hollis et al. 1986, Chapter 5; Davis et al. 2008). Greater use of scrub/dune woodland by swamp wallabies reflects their preference for dense vegetation (Di Stefano et al. 2007, Merchant 2008, Schmidt et al. 2010), particularly shrub and tree cover (Swan et al. 2008), which provides food and cover (e.g., Floyd 1980, Lunney and O'Connell 1988), factors important in swamp wallaby microhabitat selection (Swan et al. 2008). Prior to the burn, scrub/dune woodland also had greater cover of logs than did slashed swale, which swamp wallabies may use for refuge (Schmidt et al. 2010). Similarly, greater use of scrub/dune woodland by hog deer matches their preference for habitats which provide both food and shelter (Mayze and Moore 1990, Odden et al. 2005) and their ability to utilise dense vegetation (Odden et al. 2005), including Leptospermum scrub (Mayze and Moore 1990). Wombat pellet counts did not vary between slashed swale and scrub/dune woodland, probably reflecting their ability to utilise both open and closed habitats (McIlroy 2008). Further, wombats may use a mix of vegetation strata to meet their requirements, for example, they may use slashed swale to obtain preferred grasses (Evans et al. 2006), and use of dune areas for burrowing (McIlroy 1995). Further, in contrast to previous work suggesting ground layer biomass is lower under canopy cover than in open patches (Tozer and Bradstock 2002), there was no difference in fine biomass between vegetation strata at Big Hummock even though L. laevigatum is known to suppress other native plants (Bennett 1994). Thus although the cover of ground layer vegetation was greater in slashed swale, scrub/dune woodland may provide comparable levels of grazing forage. This may explain use of this stratum by wombats and hog deer, species that are generally considered grazers (Hume 1999, Wegge et al. 2006), but which can navigate within dense vegetation (Mayze and Moore 1990, Taylor 1993). In contrast, the

Chapter 4 Fine scale habitat use 138 larger-bodied kangaroos may not be able to utilise scrub/dune woodland as their movement is impeded by dense shrub cover (Taylor 1980, Bennett 1994). This explanation does not hold for small-bodied rabbits, although it may be that their small size allows them to shelter in the low shrub layer of slashed swale, which would not provide adequate shelter for larger species such as hog deer. After the burn, differences in the relative use of the two vegetation strata by rabbits became less pronounced, and for wallabies, there was no longer a difference in use of the two strata, indicating that slashed swale became less attractive to rabbits (or scrub/dune woodland became more attractive), and vice versa for wallabies. High overlap in microhabitat use generally implies a high tendency to compete (Schoener 1983). Patterns of use of the two vegetation strata at Big Hummock indicate inter-specific partitioning of habitat resources at a fine scale, possibly in response to resource competition (Pianka 1974). If this is the case, the overlap that was observed in habitat use suggests the potential for competition, particularly between rabbits, kangaroos and possibly wombats, and also between deer, wallabies and wombats, if resources become more limited and ecological separation does not occur on other dimensions (Hutchinson 1957). In particular, species with independent evolutionary histories are thought to have inherently less resource partitioning than species with common evolutionary histories (Kirchhoff and Larsen 1998, Kelley et al. 2002, Madhusudan 2004). There was clear overlap between native and introduced species with respect to the fine-scale use of vegetation strata, and although this overlap was not quantified, it appears to have been greater than overlap between native species: introduced rabbits and native kangaroos both used slashed swale to a greater extent than they used scrub/dune woodland, while the opposite was true of native swamp wallabies and introduced hog deer. Based on known habitat preferences of these species (e.g., Fa et al. 1999, Odden and Wegge 2007, Di Stefano et al. 2009, Schmidt et al. 2010), it seems that each species was using the stratum most suited to its requirements, indicating that competitive exclusion (e.g., Prins 2000, Kelley et al. 2002) has not occurred, although it is possible that competition suppresses population densities of some species (e.g., Madhusudan 2004).

Chapter 4 Fine scale habitat use 139

Conclusion Overlap in fine-scale habitat use among the five herbivore species should be low because sympatric species should reduce competition by selecting different resources (Gause 1934). Five species occur in sympatry in Coastal Grassy Woodland at Big Hummock, Yanakie Isthmus, and habitat resource partitioning was evident at a fine-scale within this vegetation type. Inter-specific variation in relative abundance and population metabolism was apparent in Coastal Grassy Woodland at Big Hummock, indicating that this habitat is preferred by some species, and avoided by others, or that subordinate species are competitively suppressed by dominant species. Overall, the density of herbivores decreased following ecological burning and associated changes in the vegetation, however, fine biomass levels remained constant before and after the fire. This suggests that herbivores remaining at Big Hummock shifted their foraging from browse towards graze. The burn at Big Hummock, and associated modification of habitat structure and composition, had differential effects on the use of Coastal Grassy Woodland by the five sympatric native and introduced herbivore species. The density of introduced European rabbits decreased after the burn, while the density of native swamp wallabies increased and densities of the other three species remained relatively constant. Although the density and population metabolism for grazers was greater than that of browsers before and after the burn, species- specific responses to fire resulted in an overall decrease in the density and population metabolism of grazers, but an increase in the density and population metabolism of browsers. The reduction in grazers after the burn was probably due to changes in the vegetation, such as decreased cover of ground layer vegetation, particularly grasses, while the increase in browsers was probably associated with vegetative regrowth and thus increased palatability of woody plants. Differential responses to burning from herbivores with differing feeding strategies suggest that the available forage became less suitable for grazers and more suitable for browsers after the burn, thus differential resource requirements appear to provide a mechanism for fine-scale habitat resource partitioning in the assemblage, facilitating coexistence. Similarly, although inter-specific overlap in the use of fine-scale vegetation strata indicates the potential for resource competition between some species, there were differences in inter-specific use of strata, suggesting resource partitioning at a fine scale. Differential use of burnt and

Chapter 4 Fine scale habitat use 140 unburnt habitat by sympatric herbivore species suggests that ecological processes which affect vegetation structure and composition, and thus food and shelter resource availability can alter herbivore community structure by favouring some species, while suppressing others. Big Hummock generally resembled other areas of Coastal Grassy Woodland with respect to herbivore densities and vegetation structure and composition; however, there were inter-site differences in a number of vegetation parameters. Therefore, further research is required to determine the applicability of these findings to the wider Yanakie Isthmus and beyond.

Chapter 4 Fine scale habitat use 141

Table 4.1. Study design for herbivore faecal pellet counts, biomass sampling and floristic surveys at four sites on Greater Yanakie Isthmus (2003 – 2004). Units are the number of replicates per site per treatment.

Site Big Hummock Springs Track Old Burn Track Varneys Treatment Burn treatment Control Control Control Scrub/dune Scrub/dune Slashed Scrub/dune Slashed Scrub/dune Slashed Vegetation strata woodland Slashed swale woodland swale woodland swale woodland swale Fire strata Burnt Unburnt Burnt Unburnt Pre- Faecal pellet Count 1: Oct fire counts 2003 0 20 0 20 20 20 20 20 20 20 Count 2: Nov 2003 0 20 0 20 20 20 20 20 20 20 Biomass Sampling 1: sampling Nov 2003 0 5 0 5 5 5 5 5 5 5 Floristic Survey 1: Nov surveys 2003 0 20 0 20 20 20 20 20 20 20 Post- Faecal pellet Count 1: July fire counts 2004 4 16 0 20 0 0 0 0 0 0 Count 2: Sep 2004 4 16 0 20 0 0 0 0 0 0 Count 3: Nov 2004 4 16 0 20 0 0 0 0 0 0 Count 4: Feb 2005 4 16 0 20 0 0 0 0 0 0 Biomass Sampling 1: sampling Dec 2004 5 5 5 5 0 0 0 0 0 0 Floristic Survey 1: Dec surveys 2004 5 5 5 5 0 0 0 0 0 0

Chapter 4 Fine scale habitat use 142

Table 4.2. Density of five herbivore species (with 95% confidence intervals) calculated from one pre-fire faecal pellet accumulation count (November 2003) at each of four sites and three post-fire counts (December 2004) at Big Hummock on Yanakie Isthmus, Wilsons Promontory National Park.

Site Hog deer European rabbit Eastern grey Common wombat Swamp wallaby kangaroo Density 95% CI Density 95% CI Density 95% CI Density 95% CI Density 95% CI ha-1 ha-1 ha-1 ha-1 ha-1 Pre-fire Old Burn Track 0.10 0.09 2.30 1.05 0.06 0.04 0.23 0.16 0.02 0.02 Varneys 0.04 0.08 4.51 2.20 0.23 0.14 0.71 0.54 0.04 0.05 Big Hummock 0.00 0.00 3.99 1.40 0.14 0.08 0.20 0.18 0.01 0.01 Springs Track 0.00 0.00 3.10 1.45 0.01 0.01 0.03 0.04 0.01 0.01 Post-fire Big Hummock 0.01 0.01 0.99 0.38 0.06 0.03 0.07 0.05 0.08 0.05

Chapter 4 Fine scale habitat use 143

Table 4.3. Results of two-factor ANOVA comparing pre-fire faecal pellet (or pellet group) standing crop for five herbivore species, summed over two surveys (October and November 2003), between two vegetation strata (slashed swale vs. scrub/dune woodland) and among four sites on Yanakie Isthmus, Wilsons Promontory National Park. P values in bold are significant effects.

Species Comparison m.s. d.f. F P Hog deer Site 14.233 2 6.173 0.003 Strata 10.208 1 4.427 0.038 Site*Strata 3.033 2 1.316 0.272 Error 2.306 114 European rabbit Site 1.621 3 1.899 0.132 Strata 186.081 1 217.991 <0.001 Site*Strata 1.628 3 1.907 0.131 Error 0.854 152 Eastern grey kangaroo Site 58.414 3 36.717 <0.001 Strata 63.395 1 39.849 <0.001 Site*Strata 8.107 3 5.096 0.002 Error 1.591 152 Common wombat Site 50.895 3 35.028 <0.001 Strata 0.315 1 0.217 0.642 Site*Strata 8.799 3 6.056 0.001 Error 1.453 152 Swamp wallaby Site 1.134 3 2.017 0.114 Strata 2.208 1 3.926 0.049 Site*Strata 0.046 3 0.082 0.970 Error 0.562 152

Chapter 4 Fine scale habitat use 144

Table 4.4. Results of two-factor ANOVA comparing the mean % cover for vegetation structure variables among sites and between vegetation strata (slashed swale vs. scrub/dune woodland) during the pre-fire survey in the Yanakie Isthmus, Wilsons Promontory National Park, November 2003. P values in bold are significant effects.

Comparison m.s. d.f. F P % cover litter Site 4227.2928 3 11.741 <0.001 Strata 42575.625 1 118.255 <0.001 Site*Strata 2978.958 3 8.274 <0.001 Error 360.033 152 % cover logs Site 62.500 3 0.963 0.412 Strata 1562.500 1 24.087 <0.001 Site*Strata 62.500 3 0.963 0.412 Error 64.868 152 % cover bryophytes & lichen Site 430.625 3 6.231 0.001 Strata 2175.625 1 31.480 <0.001 Site*Strata 350.625 3 5.073 0.002 Error 69.112 152 % cover bare ground Site 203.958 3 4.956 0.003 Strata 75.625 1 1.838 0.177 Site*Strata 110.625 3 2.688 0.048 Error 41.151 152 % cover other ground vegetation Site 1101.667 3 3.513 0.017 Strata 35402.500 1 112.908 <0.001 Site*Strata 3314.167 3 10.570 <0.001 Error 313.553 152 % cover shrub layer Site 832.292 3 4.525 0.005 Strata 22325.625 1 121.391 <0.001 Site*Strata 522.292 3 2.840 0.040 Error 183.914 152 % cover canopy layer Site 764.167 3 3.555 0.016 Strata 81000.000 1 376.860 <0.001 Site*Strata 825.000 3 3.838 0.011 Error 214.934 152

Chapter 4 Fine scale habitat use 145

Table 4.5. Mean % cover and standard error for vegetation structure variables in two vegetation strata (slashed swale and scrub/dune woodland) at four sites (n = 40 per site) during the pre-fire survey on Yanakie Isthmus, Wilsons Promontory National Park, November 2003.

Site Strata Litter Log Bryophyte & Bare ground Other ground Shrub layer Canopy layer lichen layer vegetation Mean SE Mean SE Mean SE Mean SE Mean SE Mean SE Mean SE Big Hummock Scrub/dune woodland 25.00 3.20 8.00 1.38 9.00 1.43 9.50 1.85 48.50 4.12 35.00 2.46 45.50 3.87 Slashed swale 8.00 0.92 4.50 2.11 12.50 1.60 8.00 1.17 67.00 2.06 2.00 0.92 1.50 1.09 Springs Track Scrub/dune woodland 30.00 5.18 6.00 1.52 5.50 1.70 9.50 1.98 49.00 5.18 23.00 3.00 33.50 4.12 Slashed swale 10.00 1.26 0.50 0.50 21.00 2.70 8.00 0.92 60.00 2.71 1.00 0.69 0.50 0.50 Old Burn Track Scrub/dune woodland 55.50 7.52 8.50 1.50 5.00 1.54 2.50 0.99 28.50 5.54 27.00 5.03 49.00 4.52 Slashed swale 15.50 2.85 2.00 0.92 7.50 1.23 6.50 1.31 67.50 3.15 3.00 3.00 0.50 0.50 Varneys Scrub/dune woodland 63.00 6.20 9.50 3.73 3.00 1.05 3.00 1.28 21.50 3.86 15.50 4.78 54.50 5.64 Slashed swale 9.50 1.14 0.00 0.00 11.00 2.80 7.50 1.60 72.00 3.81 0.00 0.00 0.00 0.00

Chapter 4 Fine scale habitat use 146

Table 4.6. Results of three-factor ANOVA comparing pre-fire load of two states (live vs. dead) of fine biomass between vegetation strata (slashed swale vs. scrub/dune woodland) and among four sites on Yanakie Isthmus, Wilsons Promontory National Park, November 2003. P values in bold are significant effects.

Comparison m.s. d.f. F P Site 0.371 3 0.999 0.399 Vegetation strata 0.044 1 0.119 0.731 Vegetation state 9.080 1 24.468 <0.000 Site*Vegetation strata 0.856 3 2.306 0.085 Site*Vegetation state 0.218 3 0.588 0.625 Vegetation strata*Vegetation state 6.419 1 17.298 <0.000 Site*Vegetation strata*Vegetation state 0.286 3 0.770 0.515 Error 0.371 63

Table 4.7. Results of two-tailed t-tests comparing mean sapling and tree height, mean branch length, mean number of leaves and mean number of leaders (all square-root transformed) between two sites on Yanakie Isthmus, Wilsons Promontory National Park, November 2003. P values in bold are significant effects.

Comparison t d.f. P Height 2.80 751 0.005 Branch length 4.49 736 <0.001 Number of leaves 3.43 695 0.001 Number of leaders 15.01 870 <0.001

Chapter 4 Fine scale habitat use 147

Table 4.8. Results of two-factor repeated measures ANOVA comparing daily pellet (or pellet group) accumulation rate for five herbivore species between pre- fire (October and November 2003) and post-fire (July, September, November 2003, and February 2004) surveys in two vegetation strata (slashed swale vs. scrub/dune woodland) at Big Hummock on Yanakie Isthmus, Wilsons Promontory National Park. P values in bold are significant effects.

Species Comparison m.s. d.f. F P European rabbit Between subjects Strata 0.102 1 75.897 <0.001 Error 0.001 38 Within subjects Survey 0.056 1 73.038 <0.001 Survey*Strata 0.060 1 78.277 <0.001 Error 0.029 38 Eastern grey kangaroo Between subjects Strata 0.071 1 14.650 <0.001 Error 0.005 38 Within subjects Survey 0.006 1 1.524 0.225 Survey*Strata 0.005 1 1.259 0.269 Error 0.004 38 Common wombat Between subjects Strata 0.000 1 0.080 0.779 Error 0.001 38 Within subjects Survey 0.003 1 3.276 0.078 Survey*Strata 0.001 1 0.812 0.373 Error 0.001 38 Swamp wallaby Between subjects Strata 0.000 1 1.515 0.226 Error 0.000 38 Within subjects Survey 0.001 1 11.441 0.002 Survey*Strata 0.000 1 3.787 0.059 Error 0.000 38

Chapter 4 Fine scale habitat use 148

Table 4.9. Results of two-factor repeated measures ANOVA comparing fine biomass (log10-transformed) of two states (dead vs. live) in two vegetation strata (slashed swale vs. scrub/dune woodland) between the pre-fire period at four sites and the post-fire period at Big Hummock on Yanakie Isthmus, Wilsons Promontory National Park, November 2003 and December 2004. P values in bold are significant effects.

Comparison m.s. d.f. F P Between subjects Vegetation strata 0.056 1 0.169 0.682 Vegetation state 5.474 1 16.490 <0.000 Vegetation strata*Vegetation state 3.080 1 9.280 0.003 Error 0.332 76 Within subjects Sampling period 0.000 1 0.000 0.998 Sampling period*Vegetation strata 0.237 1 0.762 0.385 Sampling period* Vegetation state 3.630 1 11.676 0.001 Sampling period*Vegetation 3.263 1 10.493 0.002 strata*Vegetation state Error 0.311 76

Table 4.10. Results of three-factor ANOVA examining the effect of grazing (exclosure vs. control plots) and fire (burnt vs. unburnt patches) on fine biomass in two vegetation strata (slashed swale vs. scrub/dune woodland) during the post- fire sampling period at Big Hummock on Yanakie Isthmus, Wilsons Promontory National Park, December 2004. P values in bold are significant effects.

Comparison m.s. d.f. F P Grazing treatment 4.221 1 20.640 <0.001 Vegetation strata 2.262 1 1.280 0.262 Burn treatment 0.003 1 0.012 0.912 Grazing treatment*Vegetation strata 0.016 1 0.077 0.782 Grazing treatment*Burn treatment 0.141 1 0.691 0.409 Vegetation strata*Burn treatment 0.467 1 2.286 0.135 Grazing treatment*Vegetation 0.586 1 2.864 0.095 strata*Burn treatment Error 0.204 72

Chapter 4 Fine scale habitat use 149

Table 4.11. Results of two-tailed t-tests comparing the mean % cover for vegetation structure variables between pre-fire (November 2003) and post-fire (December 2004) surveys at Big Hummock on Yanakie Isthmus, Wilsons Promontory National Park. P values in bold are significant effects.

Comparison t d.f. P % cover litter 0.52 36 0.606 % cover logs 1.55 52 0.128 % cover bryophytes & lichen 2.29 39 0.028 % cover bare ground 3.10 20 0.006 % cover other ground vegetation 3.17 33 0.003 % cover shrub layer 4.27 49 <0.001 % cover canopy layer 1.58 40 0.123

Chapter 4 Fine scale habitat use 150

Table 4.12. Results of two-factor ANOVA comparing the mean % cover of vegetation structure variables between burnt and unburnt patches in two vegetation strata (slashed swale vs. scrub/dune woodland) during the post-fire survey at Big Hummock on Yanakie Isthmus, Wilsons Promontory National Park, November 2003. P values in bold are significant effects.

Comparison m.s. d.f. F P % cover litter Strata 125.0 1 0.794 0.386 Treatment 845.0 1 5.365 0.034 Strata*Treatment 405.0 1 2.571 0.128 Error 157.5 16 % cover logs Strata 180.0 1 8.000 0.012 Treatment 20.0 1 0.889 0.360 Strata*Treatment 20.0 1 0.889 0.360 Error 22.5 16 % cover bryophytes & lichen Strata 45.0 1 1.385 0.257 Treatment 245.0 1 7.538 0.014 Strata*Treatment 45.0 1 1.385 0.257 Error 32.5 16 % cover bare ground Strata 80.0 1 0.640 0.435 Treatment 13520.0 1 108.160 <0.001 Strata*Treatment 180.0 1 1.440 0.248 Error 125.0 16 % cover other ground vegetation Strata 20.0 1 0.116 0.738 Treatment 4500.0 1 26.087 <0.001 Strata*Treatment 500.0 1 2.899 0.108 Error 172.5 16 % cover shrub layer Strata 80.0 1 4.000 0.063 Treatment 80.0 1 4.000 0.063 Strata*Treatment 20.0 1 1.000 0.332 Error 20.0 16

Chapter 4 Fine scale habitat use 151

Legend * ecological burn control

Big Hummock Springs Track Big Hummock

Varneys

Old Burn Track

Big Hummock

Figure 4.1. Treatment (burnt) and control (unburnt) sites used for a trial of ecological burning at Big Hummock, Yanakie Isthmus, Wilsons Promontory National Park.

Chapter 4 Fine scale habitat use 152

Chapter 4 Fine scale habitat use 153

(i) Hog deer 800 Slashed sw ale Scrub/dune w oodland 600

400

200

0 (ii) European rabbit 18000 15000 12000 9000 6000

3000

0

1 - (iii) Eastern grey kangaroo 80000

60000

40000

20000

0

Pellets (orPellets groups)ha pellet (iv) Swamp wallaby 1500 1200 900 600 300 0 (v) Common wombat 40000

30000

20000

10000

0 Old Burn Track Varneys Springs Track Big Hummock Site Figure 4.2. Mean (± standard error) pellet standing crop for five herbivore species over two pre-fire surveys (October and November 2003) at four sites (n = 40 per site) on Yanakie Isthmus, Wilsons Promontory National Park: (i) hog deer (pellet groups), (ii) European rabbit (pellet groups), (iii) eastern grey kangaroo (pellets), (iv) swamp wallaby (pellets), and (v) common wombat (pellets).

Chapter 4 Fine scale habitat use 154

(i) Pre-fire Axis 1*Axis 2 Axis 1*Axis 3 Axis 2*Axis 3 Site Big Hummock Old Burn Track Springs Track Varneys

(ii) Pre- vs. post-fire Survey Post-fire

Pre-fire

1

s

i

x A

Axis 2 (iii) Burnt vs. unburnt Treatment Burnt

Unburnt

1

s

i

x A

Axis 2 Figure 4.3. Non-metric multi-dimensional scaling two and three-dimensional configurations of ground layer plant species composition on Yanakie Isthmus, Wilsons Promontory National Park, based on Bray-Curtis matrix of dissimilarities between the % cover of ground layer plant species in plots at (i) four sites (n = 40 per site, November 2003) during the pre-fire survey (stress = 0.16, each axis displayed in two dimensions), (ii) at Big Hummock during pre-fire (n = 40, November 2003) and post-fire (n = 20, December 2004) surveys (stress = 0.11), and (iii) in burnt (n = 10) and unburnt (n = 10) patches at Big Hummock during the post-fire survey (stress = 0.08). Standardisation (1) was applied within species and plots at which no ground cover was recorded were excluded.

Chapter 4 Fine scale habitat use 155

(i) Pre-fire A xis 1*A xis 2 A xis 1*A xis 3 A xis 2*A xis 3 Site Big Hummock Old Burn Track Springs Track Varneys

(i) Pre- vs. post-fire A xis 1*A xis 2 A xis 1*A xis 3 A xis 2*A xis 3 Survey Post-fire Pre-fire

Figure 4.4. Non-metric multi-dimensional scaling three-dimensional configuration (each axis displayed in two dimensions) of shrub and canopy layer plant species composition on Yanakie Isthmus, Wilsons Promontory National Park, based on Bray-Curtis matrix of dissimilarities between the % cover of shrub and canopy layer plant species in plots at (i) at four sites (n = 40 per site) during the pre-fire survey, November 2003 (stress = 0.09), and (ii) at Big Hummock Treatment during pre-fire (n = 40, November 2003) and post-fire (n = 20, December 2004) surveys (stress = 0.11). Standardisation (1) was applied within species and plots at which no shrub or canopy cover was recorded were excluded.

Chapter 4 Fine scale habitat use 156

90 Big Hummock 80 Old Burn Track 70 60 50 40 30 20 10 0 Height (cm) Branch Number of Number of length (cm) leaves leaders Vegetation growth variable

Figure 4.5. Mean (± standard error) sapling and tree height, mean branch length, mean number of leaves and mean number of leaders at two sites (n = 10 per site) on Yanakie Isthmus, WPNP, November 2003.

Chapter 4 Fine scale habitat use 157

(i) Pre-fire

abundance biomass basal metabolic demand

; -4 70

; ; 60 -2 50 40 30

abundance 20

biomass kg x 10x kg biomass 10 0

metabolic demand kJ/day10x demand metabolic hog deer common swamp eastern European wombat wallaby grey rabbit kangaroo Species

(ii) Post-fire

20

15

10 abundance

biomass kg x 10-2;x kg biomass 5

metabolic demand kJ/day10-4;x demand metabolic 0 hog deer common swamp eastern grey European wombat wallaby kangaroo rabbit Species

Figure 4.6. Population abundance, biomass and basal metabolic demand at the population level for five herbivore species (i) pre-fire at Big Hummock, and (ii) post-fire at Big Hummock on Greater Yanakie Isthmus. Population estimates are based on faecal pellet counts conducted between 2003-2005 and body mass values from the literature.

Chapter 4 Fine scale habitat use 158

-1 -1 160 Slashed sw ale 140 Scrub/dune w oodland 120 100 80

-1 60

40 day 20

0

hog deer hog deer hog

European rabbit European rabbit European

swamp wallaby swamp wallaby swamp

Pellet pellet (or group) accumulation ha

common wombat common wombat common

easterngrey kangaroo easterngrey kangaroo Pre-fire period Post-fire period Species

Figure 4.7. Mean (± standard error) pellet (or pellet group) accumulation rate for five herbivore species over four post-fire surveys (July, September, November 2003 and February 2004) at Big Hummock (n = 40 per survey) on Yanakie Isthmus, Wilsons Promontory National Park.

Chapter 4 Fine scale habitat use 159

300

) Pre-fire -2 Post-fire 200

100 Mean fine Mean fuel m (g

0

swale

(dead)

(live)

Slashed

(dead)

Slashed

woodland woodland

swale swale (live)

Scrub/dune Scrub/dune Vegetation strata and state

Figure 4.8. Mean (± standard error) fine biomass (g m-2) of two vegetation states (live and dead) in two vegetation strata (slashed swale vs. scrub/dune woodland) during the pre-fire period (November 2003) at four sites (n = 40), and during the post-fire period (December 2004) at Big Hummock (n = 20) on Yanakie Isthmus, Wilsons Promontory National Park.

Chapter 4 Fine scale habitat use 160

(i) Pre- vs. post-fire

70 60 50 40 30 20

Number Number of species 10 0 Pre-fire Post-fire Survey

(ii) Burnt vs. unburnt

60

50

40

30

20

Number Number of species 10

0 Burnt Unburnt Treatment

Trees Grasslike herbs Shrubs Grasses Forbs Mosses

Figure 4.9. Number of species recorded at Big Hummock on Yanakie Isthmus, Wilsons Promontory National Park during (i) the pre-fire (November 2003, n = 40) and post-fire (December 2004, n = 20) surveys, and (ii) in burnt (n = 10) and unburnt (n = 10) patches during the post-fire survey.

Chapter 4 Fine scale habitat use 161

Graminoid spp. Allocasuarina verticillata Acacia longifolia Leptospermum laevigatum 35

30

25

20

15 %cover 10

5

0 Pre-fire Post-fire (burnt) Post-fire (unburnt) Survey

Figure 4.10. Mean (± standard error) % cover during pre-fire (November 2003) and post-fire (December 2004) surveys at Big Hummock on Yanakie Isthmus, Wilsons Promontory National Park.

Chapter 4 Fine scale habitat use 162

(i) Pre- vs. post-fire

70 Pre-fire Post-fire 60 50 40 30 % cover 20 10 0

(ii) Burnt vs. unburnt

Slashed swale Unburnt Slashed swale Burnt Scrub/dune woodland Unburnt Scrub/dune woodland Burnt 70 60 50 40 30

% cover % 20 10

0

Log Log

Litter

Shrub Shrub

Canopy

Bare ground Bare

vegetation

Bryophytelichen& Other ground layerOtherground Vegetation structure variable

Figure 4.11. Mean (± standard error) % cover of vegetation structure variables at Big Hummock on Yanakie Isthmus, Wilsons Promontory National Park (i) during pre-fire (n = 40, November 2003) and post-fire (n = 20, December 2004) surveys, and (ii) in burnt and unburnt patches in two strata (slashed swale vs scrub/dune woodland) (n = 5 per treatment per strata) during the post-fire survey.

Chapter 4 Fine scale habitat use 163

Chapter 5

Dietary niche relationships among sympatric native and introduced mammalian herbivores on Yanakie Isthmus, south-eastern Australia

______

Chapter 5 Diets of native and introduced herbivores 164

Chapter 5 Diets of native and introduced herbivores 165

Chapter 5 Dietary niche relationships among sympatric native and introduced mammalian herbivores on Yanakie Isthmus, south-eastern Australia

Abstract Sympatric species should reduce competition by partitioning resources. Food resource use is one of the most important components of the niche and ecological separation may be achieved by feeding in different ways. I used microhistological diet analysis to test theoretical predictions regarding inter- specific diet overlap, niche breadth and resource partitioning among five species of native and introduced mammalian herbivores in sympatry on Yanakie Isthmus at Wilsons Promontory National Park, Victoria, Australia. The diets of the relatively large-bodied introduced hog deer Axis porcinus and relatively small native swamp wallabies Wallabia bicolor consisted mainly of dicots. The diet of the small, introduced European rabbit Oryctolagus cuniculus contained similar proportions of monocots and dicots. The diets of native eastern grey kangaroos Macropus giganteus intermediate in size, and large native common wombats Vombatus ursinus consisted mainly of monocots but kangaroos also consumed moderate amounts of dicots. Overlap in food use by the five species was high, particularly between native and introduced species, but also between some native species. Despite a high potential for food resource competition, it appears that coexistence of herbivores on Yanakie Isthmus is facilitated by ecological separation. However, patterns of ecological separation, niche breadth and diet overlap in this guild did not conform well to body-size related predictions: the species with the narrowest and the broadest diet niches were intermediate in size, and the largest species consumed a greater proportion of dicots than did several smaller species. Interactions between intrinsic and extrinsic constraints on diet choice are likely to influence the diet of herbivores on Yanakie Isthmus.

Chapter 5 Diets of native and introduced herbivores 166

Chapter 5 Diets of native and introduced herbivores 167

Introduction Investigation of competitive interactions, and the ways in which species coexist, is essential to improve our understanding of the ecological principles underlying ecosystem functioning (Duncan et al. 1998). Gause (1934) first suggested that sympatric species should reduce competition by selecting different resources (i.e., reducing niche overlap; 1986). Sympatric herbivores tend to exploit their environments in different ways (Schwartz and Ellis 1981, Forsyth 2000) based on differences in feeding strategies (Gwynne and Bell 1968, Schwartz and Ellis 1981) and/or differences in habitat preference (Batcheler 1960, Taylor 1983, Fox 1989). Specialisation on resources along habitat, diet and temporal gradients results in niche differentiation, facilitating coexistence (Schoener 1974b, Whitfield 2002, Schmidt et al. 2009). Food resource use is one of the most important components of the niche of any species (Krebs 1998). Ecological separation may be achieved by feeding in different ways (Schwartz and Ellis 1981), but as noted by Child and Von Richter (1969) it is also commonly achieved by utilising different feeding habitats (e.g., ungulates in Tanzania; Voeten and Prins 1999), and may be achieved through a combination of these strategies. Indeed, divergence in diet can result from competition-driven divergence in habitat use (niche compression hypothesis: MacArthur and Wilson 1967), thus it is the sum of habitat patch and food choices that determines coexistence (Pyke et al. 1977). Mammalian herbivore diet selection, feeding ecology and foraging behaviour are influenced by body size, morphology and physiology (Schwartz and Ellis 1981). For example, differences in mouth and gut morphology result in differences in foraging behaviour (Schwartz and Ellis 1981). Body size has a particularly strong influence on diet and feeding ecology (Illius and Gordon 1993). The Jarman-Bell principle (Bell 1971, Jarman 1974) proposes that because metabolic rate decreases with increasing body weight, but gut capacity remains a constant fraction of body weight, animals with different body size will have different nutritional requirements: larger herbivores should tolerate lower quality forage than small herbivores, whereas smaller herbivores should select higher quality forage (May and MacArthur 1972, Jarman 1974, Schoener 1974b, Demment and Van Soest 1985). The Jarman-Bell principle has been demonstrated

Chapter 5 Diets of native and introduced herbivores 168 in foraging guilds of African ungulates (Bell 1971) and Australian macropods (Jarman and Phillips 1989). This allometric relationship between body size and metabolic requirements, and isometric relationship between body size and capacity to ingest and process food, strongly influences inter-specific differences in foraging behaviour (Jarman 1974), and has commonly been used to explain ecological phenomena such as grazing succession, facilitation and segregation of habitat use (Illius and Gordon 1993). Body size is often strongly related to the feeding classification of herbivore species, according to dietary grades from grazer to browser (Hofmann and Stewart 1972, Sanson 1978). Following the simplified classification system proposed by Hansen et al. (1985), browsers can be considered those species that consume >50% dicots, while grazers consume >50% monocots. Hofmann and Stewart (1972) suggest an intermediate grade, ‘intermediate mixed feeders’, which includes species that can adapt, in different seasons or in different areas, towards a bulk and roughage feeder diet (i.e., grass-eating grazers) or selectors of juicy, concentrated herbage (i.e., browsers). The occurrence of herbivore species classified as intermediate mixed feeders (Hofmann and Stewart 1972) permits temporal variation in diet, which often allows or complements niche differentiation based on food resource partitioning. Species should be active during times when they are the most efficient foragers; predation and competition are major influences on their foraging efficiency (Jones et al. 2001). Temporal partitioning of food resources through time, over the diel period (e.g., Jones et al. 2001, Borgnia et al. 2007, Rouag et al. 2007) or seasonally (e.g., Jones et al. 2001, Zapata et al. 2005, Schleuter and Eckmann 2007), may allow competitors to avoid interference and exploitative competition within and between species. Resource partitioning and niche differentiation are seen as evolutionary outcomes of competition, and have been well described among assemblages of co- evolved native herbivores (e.g., Bell 1970, Jarman and Sinclair 1979, McNaughton and Georgiadis 1986, Green 1987, Bodmer 1991). For example, native herbivore species in Nepal exhibit resource partitioning associated with seasonal changes in food availability: a high degree of spatial and food resource overlap occurs during the monsoonal growing season, but in the resource-limited dry season, the majority of species move into sub-optimal habitats, associated

Chapter 5 Diets of native and introduced herbivores 169 with their relative abilities to utilise poor quality food resources (Wegge et al. 2006). However, herbivore communities globally are increasingly composed of native and introduced species (e.g., Voeten and Prins 1999, Baldi et al. 2004, Madhusudan 2004) and there has been speculation on the effect of introducing exotic herbivore species into native herbivore assemblages (Bagchi et al. 2004, Madhusudan 2004, Mishra et al. 2004). Studies such as that by Schwartz and Ellis (1981) have highlighted relationships between recent evolutionary history and the degree of resource use overlap between native and introduced species. Subsequently, it has been asserted that species with independent evolutionary histories have inherently less resource partitioning to facilitate coexistence than species with common evolutionary histories, as they have not had the opportunity to coevolve mechanisms of resource partitioning (Kirchhoff and Larsen 1998, Kelley et al. 2002, Madhusudan 2004). Therefore, it is unsurprising that competition for diet resources between native and introduced herbivores has been suggested, and in some cases demonstrated, in many herbivore communities globally (e.g., Madhusudan 2004, Bingwan and Zhigang 2004, Wegge et al. 2006, Bonino et al. 1997, Odadi et al. 2007, Beck and Peek 2005). As in communities of sympatric herbivore species globally, studies of Australian herbivore communities have demonstrated ecological separation based on diet partitioning, for example, between northern hairy-nosed wombats Lasiorhinus krefftii and eastern grey kangaroos Macropus giganteus (Woolnough and Johnson 2000). Research has also demonstrated resource partitioning among native herbivores with respect to both habitat and diet, for example, between eastern grey kangaroos and wallaroos Macropus robustus (Taylor 1983), between eastern grey kangaroos and swamp wallabies Wallabia bicolor (de Munk 1999), and within other macropod communities (Jarman and Coulson 1989). Over the last two centuries, a wide variety of introduced mammalian herbivores have established populations in Australia (Forsyth et al. 2004) and all are now sympatric with native herbivores (e.g., Dawson and Ellis 1979, Dierenfeld 1984, Duncan 1992, Dawson and Ellis 1996). However, few studies have examined partitioning of diet resources among multi-species assemblages comprised of native and introduced herbivore species. On Yanakie Isthmus (Wilsons Promontory National Park, Victoria), introduced European rabbits Oryctolagus cuniculus and hog deer Axis porcinus

Chapter 5 Diets of native and introduced herbivores 170 occur in sympatry with three native herbivores, the eastern grey kangaroo, swamp wallaby and common wombat Vombatus ursinus. The diets of these species have been studied in other parts of their Australian range (e.g., Taylor 1971, Robley et al. 2001, Dawson et al. 2004, Evans et al. 2006, Di Stefano and Newell 2008), as well as in the native range of the two introduced species (e.g., Wegge et al. 2006, Ferreira and Alves 2009). Previous research suggests that the food and/or habitat requirements of at least some of the herbivore species that occur on Yanakie Isthmus may overlap, although a degree of ecological partitioning has been recognised (Taylor 1971, de Munk 1999, Schmidt et al. 2010). Studies based on herbivore faecal pellet counts (Chapters 3 and 4) have demonstrated some partitioning of habitat resources within this herbivore assemblage, but also a degree of inter-specific overlap in habitat use. Niche differentiation within a community is generally complementary: when species are similar on one niche dimension, they differ on another (Pianka 1976, Dunbar 1978, Fox 1989, Bagchi et al. 2003), although competing species often differ in multiple niche dimensions (e.g., Pianka 1974, le Mar and McArthur 2005). Therefore, it is likely that herbivores on Yanakie Isthmus maximise ecological separation by partitioning diet resources. Understanding resource use overlap in terms of diet is critical to understanding competitive interactions (Krebs 1998, Wegge et al. 2006), and ultimately, the divergent exploitation patterns which determine community structure (Bell 1971, Hofmann and Stewart 1972, Jarman 1974). Investigation of inter-specific overlap in diet within the herbivore assemblage on Yanakie Isthmus will allow us to attain a more complete understanding of community niche dynamics in complex herbivore assemblages comprised of introduced and native species. The largest species, hog deer (c. 40 kg; Mayze and Moore 1990), are primarily grazers in their native range (Wegge et al. 2006) and in Victoria (Taylor 1971). The smallest species, the introduced European rabbit (c. 1.6 kg; Strahan 1995) grazes selectively on forbs and grasses (Leigh et al. 1991, Martin et al. 2007). The second smallest species, the Swamp wallaby (c. 18 kg; Edwards 1969), has been classified as a browser on the basis of diet (Claridge 2001, Davis et al. 2008, Di Stefano and Newell 2008) and dentition (Sanson 1980), but includes a range of forage in its diet including grass (Norbury et al. 1989), herbs, woody perennials, ferns , tree seedlings, saplings and fungi (Edwards and Ealey

Chapter 5 Diets of native and introduced herbivores 171

1975, Waters 1985, Hollis et al. 1986). The eastern grey kangaroo, which is intermediate in size (c. 26 kg: ACT, n = 333, G. Coulson, The University of Melbourne, pers. comm.) was classified by Sanson (1978) as a grazer on the basis of dentition and accordingly, their diet is dominated by high-protein, low-fibre grasses (Taylor 1985a, McCullough and McCullough 2000). The second largest species, the common wombat (c. 28 kg; Barboza et al. 1993), is classified as a grazer (Rishworth et al. 1995, Hume 1999). I aimed to improve our understanding of large herbivore ecology and organisation by testing ecological predictions in a contemporary assemblage made up of species with disparate evolutionary histories. Specifically, I aimed to estimate the diets of sympatric native and introduced herbivore species on Yanakie Isthmus, and assess the potential for inter-specific competition (Krebs 1998) within this assemblage by quantifying inter-specific diet overlap (and thus resource partitioning) and variation among species in diet niche breadth. Five general predictions emerge.

1. Diet overlap among the five herbivore species should be low: diets of the five species should differ significantly Sympatric species should reduce competition by selecting different resources (Gause 1934). Therefore, diet overlap among the five herbivore species should be low and their diets should differ significantly (Schoener 1974a).

2. Diet overlap should be greater between native and introduced herbivore species than between native herbivore species To facilitate coexistence, mechanisms of resource partitioning have evolved in species with common evolutionary histories, while species with disparate evolutionary histories have inherently less resource partitioning (Kirchhoff and Larsen 1998, Madhusudan 2004). Therefore, diet overlap should be greatest between native and introduced herbivore species.

3. The proportion of monocot material in the diet should be greatest in the largest species and inversely, the proportion of dicot material should be greatest in the smallest species

Chapter 5 Diets of native and introduced herbivores 172

Illius and Gordon (1993) argue that herbivore body sizes and feeding niches have evolved to occupy subsets of the vegetation quantity-quality matrix, according to the Jarman-Bell principle (after Jarman (1974) and Bell (1970)) whereby body size has implications for the minimum quality of food necessary for survival, and hence for the feeding niche selected. The proportion of low quality monocot material should be highest in the diets of the largest species, and should decrease (and be progressively replaced by higher quality dicot material) with decreasing body size.

4. Diet overlap should be greatest among similarly sized animals Following prediction number three, diet overlap should be greater among similar- sized species than between species that differ greatly in body size (Schwartz and Ellis 1981).

5. Niche breadth should decrease with decreasing body size According to the Jarman-Bell principle, smaller species require more energy per unit body weight and therefore must feed selectively on high quality foods that they can digest quickly (Van Soest 1994). In contrast, larger animals with relatively lower metabolic rates and greater gut capacity relative to body size can utilise more abundant, low quality foods (Bell 1971, Jarman 1974, Illius and Gordon 1993). Therefore, diet selectivity should be negatively related to body size (Bagchi et al. 2003).

Methods Study area I conducted this study on Yanakie Isthmus (38º 53' S; 146º 14' E), a 6874- ha area of Wilsons Promontory National Park, Victoria, Australia (Figure 1.1). Five Broad Vegetation Types occur on Yanakie Isthmus, Coastal Grassy Woodland, Coastal Scrubs and Grasslands, Heath, Heathy Woodland and Moist Foothill Forest, each of which is described in Chapter 1. As described in Chapter 3, predation rates on Yanakie Isthmus are likely to be low.

Chapter 5 Diets of native and introduced herbivores 173

Collection and preparation of stomach samples I estimated the diet of the hog deer, eastern grey kangaroo, swamp wallaby, common wombat and European rabbit using microhistological identification of cuticle fragments of plant species in stomach samples (Norbury 1988). I obtained samples from animals shot by Parks Victoria staff for the purposes of this study (authorised under Section 37 of the National Parks Act; register number 04004 of the University of Melbourne Animal Experimentation Ethics Committee) between 7 June 2004 and 11 February 2005. Animals were shot at night from a vehicle with the aid of a spotlight. All species were shot in accordance with the Code of Practice for the Humane Shooting of Kangaroos (Australian National Parks and Wildlife Service 1995). I aimed to collect stomach samples from 20 mature individuals of each species. All (n = 93) of my samples were collected from Coastal Grassy Woodland and Coastal Scrubs and Grasslands except for five samples that were collected from Heath and Heathy Woodland due to difficulties in locating 20 of each species in Coastal Grassy Woodland and Coastal Scrubs and Grasslands. The number of samples collected during each season and of each sex from each vegetation type is detailed in Table 5.1. Importantly, most samples were collected during winter and no samples were collected during autumn (Table 5.1). In order to minimise bias associated with differential digestion rates of plant species, and different feeding strategies of the herbivore species studied, I sampled freshly-ingested plant material from the oesophageal region of the stomach (G. Norbury, Landcare Research New Zealand Ltd., pers. comm.). I took only a fist-full of ingesta to limit and standardise the time period being sampled from (G. Norbury, pers. comm.) and to increase the likelihood that samples contained vegetation foraged from the area in which the animals were shot. I preserved samples in 70% ethanol.

Microhistological analysis I prepared stomach samples and reference slides for over 200 plant species from Yanakie Isthmus using standard microhistological techniques (Norbury 1988). To determine the relative area of categories of plant epidermal fragments, I used point quadrat analysis (Norbury 1988), identifying 400 fragments per sample as: (i) monocotyledons (monocots), (ii) eudicotyledons (dicots), or (iii) other

Chapter 5 Diets of native and introduced herbivores 174

(Bryophytes and Pteridophytes). Where possible, I then identified fragments to the higher taxonomic levels of family, and species, and also as stem or leaf. Reference herbarium photographs coupled with descriptions of major diagnostic features (i.e., cell and stomata size, shape and arrangement, and unique features such as oil glands and trichomes; Norbury 1988), provided the basis for identifying the plant fragments in stomach samples. were not considered during this study as none of the study species were expected to consume substantial quantities of (Hume 1999, Leigh 1989, Wegge et al. 2006), although hog deer will consume fruit at times (Dhungel and O’Gara 1991). Stomach sampling was conducted under Research permit 10002857, National Parks Act 1975 and Wildlife Act 1975, and plants were collected under Research Permit 10002450 of the Wildlife Act 1975, Flora and Fauna Guarantee Act 1988 and National Parks Act 1975.

Statistical analyses To analyse the diet data I constructed seven levels of classification, each of which contained two or more categories into which plant fragments were assigned: 1, Broad taxonomic: (i) monocots, (ii) dicots and (iii) other (Bryophytes and Pteridophytes). 2, Functional group: (i) forbs, (ii) shrubs, (iii) trees, (iv) ferns, (v) grasses, (vi) grasslike plants (i.e., non-grass graminoid species such as sedges, rushes, lilies and some herbs) and (vii) mosses. 3, Structural group: (i) non-woody ground layer and (ii) woody shrub and tree layer. 4, Plant part: (i) leaf and (ii) stem. 5, Family. 6, species. I pooled data from each sub-sample to give a total of 400 fragments identified per sample (Norbury 1988) and based analyses on the proportion of plant epidermal fragments identified within categories of interest per stomach sample for the five species. To examine diet composition I calculated the mean proportion of plant epidermal fragments identified within each category of levels 1 - 5 for each of the five herbivore species. I calculated bootstrapped 95% confidence intervals for these means with Microsoft ExcelTM using the Poptools add-in (Hood 2005). To examine inter-specific variation in diet I compared the proportion of fragments identified in each category of levels 1-7 between species using one- factor ANOVA with post-hoc Tukey’s tests, because these have greater power

Chapter 5 Diets of native and introduced herbivores 175 than other post-hoc tests under most circumstances (Quinn and Keough 2003). I plotted residuals for these linear models against the corresponding fitted values to check for distributional problems. If these plots were wedge shaped rather than data points being evenly spread across plots, I applied arcsine transformations to improve residual distributions. For all analyses I used an  = 0.05% level of significance. I estimated niche breadth using Levins’ measure (Levins 1968): Y 2 B  2  N j where B = Levin’s measure of niche breadth

Pj = Proportion of individuals using resource state j, or fraction of items in the diet that are of food category j

Nj = Number of individuals found in or using resource state j

Y = ∑ Nj = Total number of individuals sampled. I then applied a standardisation procedure developed by Hurlbert (1978): B 1 B  A n 1 where BA = Levin’s standardised niche breadth B = Levin’s measure of niche breadth n = number of possible resource states. To measure overlap in resource use between herbivore species I used Horn’s index of niche overlap, where a value of zero indicates no overlap and a value of one indicates complete overlap (Horn 1966): p  p logp  p  p log p  p log p R   ij ik ij ik  ij ij  ik ik o 2log 2 where Ro = Horn’s index of overlap for species j and k

Pij = Proportion resource i is of the total resources utilised by species j

Pik = Proportion resource i is of the total resources utilised by species k I based calculations of niche breadth and diet overlap on the proportion of fragments identified in samples of each species from each plant functional group. To represent diet overlap between animal species graphically, I used non- metric multi-dimensional scaling (NMDS). I based these 3-D ordination spaces on Bray-Curtis dissimilarity matrices (Clarke 1993) for the number of fragments of

Chapter 5 Diets of native and introduced herbivores 176 each functional group, plant species or family identified per individual for the five animal species. NMDS scales objects based on a reduced set of variables derived from the original variables and these new variables are used as the axes: the actual values of the object scores are arbitrary and only the relative distances (dissimilarities) between objects are important (Quinn and Keough 2003). I applied the following standardisation within functional groups, families and species to reduce the influence of abundant groups on the NMDS:

1 xi  xi,min th xi  for the i variable, xi,max  xi,min where x is the number of fragments of each functional group, plant species or family identified per individual. Prior to creation of the dissimilarity matrix I excluded plant species or families that occurred in only one sample, as they did not contribute to systematic compositional differences between samples.

Results Diets of the five herbivores There were 182 identifiable plant species (52 families) in the diets of the five herbivore species (Table 5.2). On average, almost 90% of fragments were identifiable to at least the broad taxonomic level. The proportion of unidentifiable fragments ranged from 10% for the European rabbit to 19% for the hog deer. The diet of the hog deer consisted predominantly of dicot material, which was five times greater than the proportion of monocot material (Table 5.3; Figure 5.1). Deer samples included fragments from all seven functional groups in both the non-woody ground layer and the woody shrub and tree layer, however, forb and shrub materials were dominant, each comprising greater than three times the number of fragments identified in any other category. The diet of the European rabbit contained similar proportions of monocot and dicot material (Table 5.3; Figure 5.1). The diet of the rabbit consisted predominantly of non-woody ground layer vegetation, mainly forbs (c. 40%), but also included considerable amounts of grasses and grasslike plants. The diet of the swamp wallaby was dominated by dicots (> 90%) (Table 5.3; Figure 5.1) from both the non-woody ground layer and the woody shrub and tree layer. Although the diet of the wallaby included several functional groups, it consisted predominantly of shrub and forb material (c. 50%

Chapter 5 Diets of native and introduced herbivores 177 and 30% respectively). The diet of the eastern grey kangaroo consisted mainly of monocots (almost twice the proportion of dicots) and was dominated by grasses (almost 50%) (Table 5.3; Figure 5.1). However, moderate amounts of forb and shrub material were also present, accounting for the occurrence of almost 30% dicots in the diet of the kangaroo. Despite this mix of functional groups consumed by the kangaroo, the diet consisted mainly of material from the non-woody ground layer. The diet of the common wombat was composed predominantly of monocot material (more than eleven times the proportion of dicot material) (Table 5.3; Figure 5.1). The diet of the wombat consisted largely of material taken from the non-woody ground layer vegetation, particularly grasses (> 50%) and grasslike plants (> 20%).

Inter-specific variation in diet At the broadest level, there were significant differences between the diets of the five species in the proportions of monocot and dicot material in stomach samples (Tables 5.4-5.5; Figure 5.1). The proportion of monocot material in the diet, in ascending order (and conversely, the proportion of dicot material in descending order) was: wallaby, deer, rabbit, kangaroo and wombat (Fig. 1). In particular, deer and wallabies consumed greater proportions of dicot than did other species. Conversely, wombats consumed a greater proportion of monocot than any other species. Kangaroos and rabbits consumed significantly greater proportions of monocot than did deer or wallabies, and the amount of monocot was significantly greater in kangaroos than rabbits. There were also inter-specific differences in diet at finer scales. Deer and wallabies consumed significantly more shrub material than did the other three species, and similarly consumed significantly more material from the woody shrub and tree layer (Tables 5.4-5.5; Figure 5.1). Furthermore, wallabies consumed a greater proportion of fern fragments than did other species. In contrast, the diets of kangaroos, wombats and rabbits contained significantly more non-woody ground layer species. Wombats and kangaroos consumed significantly more grass than did other species. Wombats also consumed significantly greater proportions of grasslike plants than all species but the rabbit. Consistent with these patterns, wombats consumed the least forb material and kangaroos consumed less forb than did deer or rabbits. The largest proportion of mosses (<

Chapter 5 Diets of native and introduced herbivores 178

0.01%) occurred in deer, and the largest proportions of tree material (c. 0.01%) occurred in wallabies (Figure 5.1). Overall, the assemblage of herbivores mainly consumed material from the non-woody ground layer: grasses, forbs and shrubs were consumed in the greatest proportions, grasslike plants were consumed in intermediate proportions, and trees, ferns and mosses were consumed in the smallest proportions. There was a clear overall trend for all five species to contain much greater proportions of leaf (> 80%) than stem material (Figure 5.1). However, the proportions of stem and leaf material varied among the five species (Tables 5.4- 5.5; Figure 5.1). Deer and wallabies consumed significantly more leaf material than did rabbits. Similarly, rabbits consumed the greatest proportion of stem material (c. 15%), significantly more than either kangaroos or wombats.

Niche breadth and inter-specific overlap in diet

Levins’ measure of niche breadth (BA), in ascending order, was: kangaroo 0.57, wombat 0.61, rabbit 0.65, deer 0.82 and wallaby 0.83. Horn’s index indicated that overlap in food use among the species in this community was generally high (Table 5.6). In particular, there was extensive overlap between deer and wallabies (Ro = 0.93). Kangaroos also overlapped extensively with rabbits (Ro

= 0.90) and wombats (Ro = 0.89). Overlap was moderate between rabbits and wombats, hog deer and kangaroos, and hog deer and rabbits. The only pair for which little overlap occurred was wombats and wallabies. NMDS revealed that, although there was diet overlap among the five species, partitioning occurred at each of the three levels examined, as evidenced by slight separation in the clumps of data points for each species (Figure 5.2). At the levels of functional group, species and family, deer and wallabies were similar, and kangaroos, wombats and rabbits were similar. These ordinations also indicate variation in diets within species, particularly for deer, one of which varied considerably from its conspecifics with respect to the plant families consumed (Figure 5.2, panel iii).

Chapter 5 Diets of native and introduced herbivores 179

Discussion Herbivore diets The diets of hog deer, eastern grey kangaroos, swamp wallabies, common wombats and European rabbits on Yanakie Isthmus were broadly similar to diets described for these species elsewhere in south-eastern Australia (e.g., Taylor 1983, Osawa 1990, Rishworth et al. 1995, Martin et al. 2007). However, some important differences were evident. In particular, the diet of hog deer consisted predominantly of dicots, including both forbs and shrubs. This contrasts with previous work concluding that hog deer are primarily grazers in their native range in south and south-east Asia (Wegge et al. 2006) and in Victoria (Taylor 1971). Consistent with their classification as grazers (Taylor 1983, Sanson 1989), eastern grey kangaroos on Yanakie Isthmus consumed predominantly grasses, but ate higher proportions of dicot material than recorded in previous studies (Taylor 1983, Jarman and Phillips 1989). The diet of swamp wallabies on Yanakie Isthmus was similar to elsewhere in south-east Australia (Jarman and Phillips 1989, Osawa 1990): they consumed plants from all seven functional groups, consistent with their classification as browsers. However, as for hog deer and kangaroos, the ratio of monocots to dicots consumed by swamp wallabies (c. 1:30) was higher than observed elsewhere (1:3 - 1:11; Edwards 1969, Taylor 1971), possibly reflecting altered foraging strategies in response to food quality and availability (Osawa and Woodall 1990). The diets of common wombats and European rabbits on Yanakie Isthmus mirrored the results of previous studies. The diet of wombats consisted mainly of monocots from the non-woody ground layer, particularly grasses, sedges, rushes and lilies, supporting their classification as grazers (Rishworth et al. 1995, Hume 1999). Rabbits consumed relatively even proportions of dicots and monocots. Although rabbits predominantly ate forbs, they also consumed considerable quantities of grasses and grasslike plants; these results conform with other studies showing that rabbits graze selectively on forbs and grasses (Leigh et al. 1991, Martin et al. 2007).

Competition, ecological separation and resource limitation Patterns of food resource use by the five herbivore species on Yanakie Isthmus suggest that a number of ecological predictions based on competition

Chapter 5 Diets of native and introduced herbivores 180 theory and the Jarman-Bell principle (Bell 1971, Jarman 1974) do not hold true in contemporary herbivore guilds made up of species with disparate evolutionary histories. In contrast to my prediction (hypothesis 1), diet overlap among sympatric herbivore species on Yanakie Isthmus was high. However, there were inter-specific differences in diet as expected, providing some evidence for resource partitioning within this community, although in contrast to my predictions (hypotheses 3 and 5), body size was not a strong predictor of differences in diet quality or niche breadth. Further, diet overlap, was better predicted by species origin (hypothesis 2) than by body size (hypothesis 4). To reduce competition, sympatric species should reduce niche overlap by partitioning resources (Gause 1934, Schoener 1974a). However, on Yanakie Isthmus there was high overlap in food resource use among herbivore species, indicating a high potential for inter-specific competition if shared food resources are limiting (Gause 1934, Schoener 1983). Resource availability and thus diet selection (Evans and Jarman 1999) were not quantified, as this was considered too difficult to estimate due to the complexities involved for an assemblage of free- ranging species with differing morphology and physiology (Feinsinger et al. 1981, Norbury and Sanson 1992). However, marked differences in the diets of several herbivore species on Yanakie Isthmus, compared to diet descriptions for these species from other locations, provide insights into the potential for food resource limitation in this system. The diet of hog deer on Yanakie Isthmus consisted predominantly of dicots, strongly contrasting with previous work concluding that hog deer are primarily grazers in their native range (Wegge et al. 2006) and in Victoria (Taylor 1971). In contrast, hog deer in their native range are the only species to efficiently use grasses throughout the year, while larger bulk-feeding species switch to browse when senescent grasses are of poor quality (Wegge et al. 2006). Eastern grey kangaroos on Yanakie Isthmus consumed predominantly grasses; however they also ate greater proportions of shrubs and forbs than recorded in previous studies (Taylor 1983, Jarman and Phillips 1989). Similarly, the ratio of dicots to monocots consumed for swamp wallabies on Yanakie Isthmus (c. 30:1) was higher than observed elsewhere (3:1 – 11:1; Edwards 1969, Taylor 1971). Similarly, rabbits consumed more forbs than grasses, despite their preference for palatable grasses (Leigh 1989). Thus four of the five species on Yanakie Isthmus

Chapter 5 Diets of native and introduced herbivores 181 displayed diet shifts from grazing towards browsing relative to diets observed in other parts of their ranges. According to optimal foraging theory, food types should be added to the diet in rank order as the abundance of preferred items decreases (Pyke et al. 1977). Competition theory suggests that if competition for food resources reduces the availability of some food items more than others, this should directly alter diets (Sih 1993). Therefore, species that have similar diets when they occur separately may use different food items when they occur together (Werner 1977) as they alter their foraging strategies in response to food quality and availability (Osawa and Woodall 1990). For example, impala Aepyceros melampus modify their diet when in competition for food resources with cattle (Jarman and Jarman 1974, Dunham 1981), which are more efficient at utilizing short grasses (Bell 1970, Illius and Gordon 1987). In sympatric herbivores, diet switching can indicate food resource limitation, either inherent to the habitat or associated with inter-specific competition (Sih 1993, Fritz et al. 1996). Therefore, trends in diet switching from grazing towards browsing suggest that grass resources on Yanakie Isthmus are limited. High diet overlap coupled with resource limitation is likely to result in inter-specific competition (Gause 1934). However, despite indications of strong competition on Yanakie Isthmus, populations of these sympatric species persist. Coexistence of sympatric species with similar resource requirements is likely to be facilitated by resource partitioning (Schoener 1974a). Resource partitioning among sympatric herbivores is often based on differences in feeding strategies (Schwartz and Ellis 1981) and can occur at fine scales in species whose diets appear broadly similar (Bagchi et al. 2003). Despite the high diet overlap observed between the five herbivore species at the level of plant functional group, resource partitioning was apparent: there were broad inter-specific differences in the consumption of monocots and dicots, as well as finer scale differences in the consumption of plant structural and functional groups, plant parts, species and families. Thus it appears that like in many other communities, resource partitioning on Yanakie Isthmus facilitates coexistence (e.g., Zapata et al. 2005, Azevedo et al. 2006, Prigioni 2008, Namgail et al. 2009). However, unlike most sympatric mammalian herbivore assemblages (Demment and Van Soest 1985),

Chapter 5 Diets of native and introduced herbivores 182 patterns of diet separation on Yanakie Isthmus were not well predicted by body size.

Body size Body size influences niche differentiation due to its influence on the efficiency with which foods can be consumed and utilised (Illius and Gordon 1993). According to the Jarman-Bell principle (Bell 1970, Jarman 1974), smaller species require more energy per unit body weight and therefore must feed selectively on higher quality foods (Van Soest 1994). In contrast, larger animals with relatively lower metabolic rates and greater gut capacity can utilise more abundant foods of lower quality (Bell 1971, Jarman 1974). In contrast to my predictions (hypothesis 3), consumption of monocots by two herbivores of intermediate to large size on Yanakie Isthmus, eastern grey kangaroos and common wombats, and by the smallest herbivore, the European rabbit, was greater than consumption of monocots by the largest herbivore, the hog deer. Further, swamp wallabies consumed the greatest proportion of dicots even though they are much larger than rabbits. Similarly, hog deer consumed a greater proportion of dicots than the smaller wombat, kangaroo and rabbit. Divergence from diet predictions based on the Jarman-Bell principle has also been observed in other communities of sympatric herbivores (Schwartz and Ellis 1981, Bodmer 1991, Wegge et al. 2006). For Amazonian ungulates, Bodmer (1991) suggested that digestive morphology was a more important mechanism of food resource partitioning than body size. Moreover, niche selection is a function of the interaction between body size and vegetation abundance and quality (Illius and Gordon 1993) and the constraints imposed on diet by vegetation abundance and quality may supersede physiological digestive processing constraints (Penry 1993). On Yanakie Isthmus, divergence in herbivore diets from body size-related predictions may be explained by species-specific responses to forage quality and availability, controlled by interactions between body size, morphology and physiology, factors which vary depending on evolutionary history (Illius and Gordon 1987). A fundamental difference among herbivore species on Yanakie Isthmus is the lower basal metabolic rates of marsupials relative to eutherians, resulting in lower energy requirements for marsupials (Hume 1999). Such differences are likely to explain high diet overlap between pairs of species that

Chapter 5 Diets of native and introduced herbivores 183 differ widely in body size, and thus lack of support for predicted patterns of overlap in food resource use based on body size (Schwartz and Ellis 1981). Diet overlap was thus better predicted by species origin.

Inter-specific interactions: native versus introduced herbivore species On Yanakie Isthmus, classification according to feeder types was not related to whether herbivore species were native or introduced. There was high diet overlap between introduced hog deer and native swamp wallabies, and between introduced rabbits and native kangaroos. In addition, there was moderate diet overlap between rabbits and native wombats and between hog deer and kangaroos. These results support other work that has demonstrated the potential for high overlap in diet between sympatric native and introduced herbivores (e.g., Dawson and Ellis 1996), suggesting that diet overlap between native and introduced herbivores is influenced by recent evolutionary history (Schwartz and Ellis 1981). In contrast, diet overlap between native herbivores was generally low. These observations support my prediction (hypothesis 2) that diet overlap should be greater between native and introduced herbivore species than between native herbivore species. However, there was also high also overlap between two native species, kangaroos and common wombats. Similarly, Woolnough and Johnson (2000) found high diet overlap between the eastern grey kangaroo and the northern hairy-nosed wombat, and de Munk (1999) found considerable overlap between diets of the eastern grey kangaroo and swamp wallaby, although in both cases it was suggested that population densities were low enough that competition for food resources was negligible. Population densities estimates on Yanakie Isthmus suggest that kangaroos occur in relatively low densities (Chapter 3), which could explain the coexistence of wombats and kangaroos, despite high diet overlap. There was also moderate overlap in the diets of the two introduced species on Yanakie Isthmus, hog deer and rabbits, which superficially does not appear to match the prediction of greater overlap between native and introduced species than between native species; however, these species lack a shared evolutionary history and in this context, could therefore be considered closer to a comparison between introduced and native species. Pianka’s (1974) theory of maximal tolerable niche overlap predicts that there should be less niche overlap in resource limited situations compared to those

Chapter 5 Diets of native and introduced herbivores 184 with higher resource abundance (Pianka 1974), due to the evolution of mechanisms of resource partitioning (Schwartz and Ellis 1981). Thus the diets of sympatric herbivores should overlap least (i.e., niches should be highly segregated) when food supply is short (Schwartz and Ellis 1981). On the other hand, high resource use overlap is commonly interpreted as indicating resource abundance and reduced competition (Schoener 1982), as species can share resources when they are not limiting (Pianka 1974). However, Voeten and Prins (1999) argue that, while overlap in resource use is not expected among native species under food-limited conditions, when exotic species are introduced into native assemblages, overlap in resource use can occur under food-limited conditions because evolutionary segregation has not occurred (Kirchhoff and Larsen 1998, Madhusudan 2004). Therefore, high overlap in resource use between native and introduced species on Yanakie Isthmus is likely to indicate competition for limited resources. The suggestion of competition between native and introduced species is not surprising: for an introduced species to become established, it must find a suitable niche that is unoccupied or compete with native species (Ovington 1978), yet the existence of empty niches is rare (Williamson 1996). Marsupials are usually competitively inferior to eutherians (McNab 2005). However, despite high diet overlap on Yanakie Isthmus, introduced eutherian species have not out-competed native marsupials, and native herbivore species with high diet overlap also coexist. Such coexistence has also been observed in other sympatric populations of native and introduced herbivores (Dawson and Ellis 1979, Duncan 1992, Dawson and Ellis 1996) and in populations of sympatric native herbivores (Woolnough and Johnson 2000). We may be observing a system in flux, with species exclusion yet to occur, as Yanakie Isthmus has undergone considerable disturbance in recent history (Bennett 1994) and this herbivore assemblage has not coexisted long (Menkhorst 1995b). Alternatively, resource partitioning (possibly in dimensions other than food, for example habitat; Chapters 3 and 4) may allow coexistence (Schoener 1974a) within this assemblage. Examining species-specific adaptations, with reference to the diet shifts previously mentioned, provides some insight into the potential competitive interactions between native and introduced herbivores on Yanakie Isthmus.

Chapter 5 Diets of native and introduced herbivores 185

Taylor (1971) suggested that hog deer occupy a previously vacant ecological niche in Victoria. However, on Yanakie Isthmus hog deer consumed a greater proportion of dicots than observed by Taylor (1971), but still consumed monocots, and their diet overlapped with that of both swamp wallabies and kangaroos. The diet of hog deer reflects food availability more strongly than dietary preferences (Miller 1975), and Taylor (1971) recognised that under conditions of resource limitation, hog deer may shift their diet towards browse rather than compete for grasses, a phenomenon that has been observed in other systems where macropods compete with ruminants (Ealey 1967, Edwards et al. 1995). Despite experimental demonstration of competitive displacement (Werner 1977), whereby the diet of a forager is displaced away from that of its competitor (Sih 1993), the mechanisms by which ecological interactions alter diets are poorly understood (Sih 1993). I hypothesise that diet overlap coupled with food resource limitation results in competition among grazers on Yanakie Isthmus and that a greater capacity for digestion of high-fibre grasses by kangaroos and wombats (Hume 1999), compounded by higher energy requirements for ruminants than marsupials (Freudenberger et al. 1988), has resulted in a more pronounced shift in the diet of the ruminant hog deer from graze to browse than exhibited by native grazers. In particular, low diet overlap between hog deer and wombats, and the lack of diet switching in wombats on Yanakie Isthmus relative to diet observations elsewhere (Rishworth et al. 1995, Hume 1999) (the only species for which diet switching was not observed), may reflect a competitive advantage of this species in low productivity habitats (Johnson 1998, Hume 1999). Their efficient masticatory system and great capacity for digesta retention and fermentation (Hume 1999) enables them to process lower-quality forage than similarly sized ruminants (Illius and Gordon 1993) or foregut fermenters (Demment and Van Soest 1985), and their low metabolic rate and use of burrows allows them to conserve water and energy (Hume 1999). In contrast to inter- specific relationships between hog deer and native herbivores, evolutionary segregation between the native kangaroo and wombat is likely to facilitate coexistence, for example, based on the preference of kangaroos (foregut fermenters) for low-fibre grasses (Forbes and Tribe 1970), whereas wombats (hindgut fermenters) can utilise high-fibre monocots (Leigh et al. 1991).

Chapter 5 Diets of native and introduced herbivores 186

Alternatively, diet switching in hog deer could reflect a competitive advantage in this species. Hog deer are considered to prefer grasses (Miller 1975), and according to optimal foraging theory, an animal should never specialise on a less preferred food type regardless of its abundance (Pyke et al. 1977). However, the relative nutritional value of browse versus grasses varies (Norbury et al. 1989). Moreover, the relative value of foods varies for different animals according to their digestive capacity (Illius and Gordon 1993). Illius and Gordon (1993) suggest that forbs and browse could provide ample energy for larger animals if not for the low biomass and the small size of food items. Hog deer are opportunistic and capable of exploiting nutritious browse when available (Taylor 1971, Roberts 1977). If grass resources on Yanakie Isthmus are scarce or of poor quality, but forbs and shrubs are abundant enough that hog deer can selectively use high- quality plant parts, hog deer may have an advantage over native grazers such as the eastern grey kangaroo, which possess a digestive system that is inefficient at utilising high-quality forages (Dawson 1989, Hume 1999). European rabbits can compete with native herbivores for food (Dawson and Ellis 1979) and have been implicated in the decline of common wombats in South Australia (Cooke 1998). On Yanakie Isthmus, the diet of rabbits overlapped with that of both wombats and kangaroos. Although small selective herbivores are better able to meet their energy requirements on shorter swards than large bodied animals with grazing dentition (Gordon and Illius 1988, 1989), rabbits have high nitrogen and energy requirements (Cooke 1974), digest fibre poorly (Voris et al. 1940) and have a limited ability to increase food intake to compensate for low- quality food (Martin et al. 2007). Therefore, rabbits are restricted to higher quality foods than wombats or kangaroos (Leigh et al. 1991). If forage on Yanakie Isthmus is of poor quality, these digestive constraints, combined with control of rabbit populations (Parks Victoria 2003a) and concentration of predation by red foxes on rabbits (Lunney et al. 1990), may minimise competition between rabbits and native herbivores.

Niche breadth: body size, resource limitation and competition If two populations have access to the same resource base, the population whose members discriminate less among resources states has a broad niche relative to a population whose members concentrate on particular resource states

Chapter 5 Diets of native and introduced herbivores 187

(Levins 1968). The Jarman-Bell principle predicts an inverse relationship between body size and foraging selectively (hypothesis 5) (Bell 1970, Jarman 1974). On Yanakie Isthmus however, body size was not a good predictor of niche breadth. The largest herbivore, the hog deer, did display a broad niche, although the feeding niche of the swamp wallaby, which is considerably smaller, was the broadest. In contrast, the eastern grey kangaroo, which is larger than the wallaby but intermediate in size relative to the other species in this assemblage, had the narrowest niche. Divergence in patterns of diet selectivity from body size-related predictions has also been observed in other herbivore assemblages (Schwartz and Ellis 1981, Shrestha et al. 2005). The thresholds at which species switch their diets from selective to non-selective tactics as food quality and availability change depend on energetic demands (Schwartz and Ellis 1981). Divergence in niche breadth patterns from body size-related predictions are likely to be explained by interactions between body size and factors such as morphology and physiology, which vary according to evolutionary history, and should be interpreted with respect to resource quality and availability (Schwartz and Ellis 1981). Moreover, herbivore assemblages in which relationships between body size and food selectivity have been established generally include species with a greater range of body sizes than occurs on Yanakie Isthmus (e.g., Bell 1970). According to optimal foraging theory, broad dietary niches indicate low resource availability (inherent to the environment, or due to competition): as food becomes scarce, individuals respond by widening their choice of foods to include less nutritious plants due to resource constraints (Pyke et al. 1977). In contrast, surplus resources allow selectivity: as the abundance of a preferred food type increases, the number of less preferred food types consumed will be reduced as the species includes fewer but more nutritious plant species in its diet (Pyke et al. 1977). Therefore, the broad feeding niches observed on Yanakie Isthmus provide support for the assertion that resource availability is limited. Furthermore, the broader feeding niche of hog deer relative to sympatric herbivores on Yanakie Isthmus contrasts with inter-specific comparisons in the native range of hog deer, where they have a relatively narrow feeding niche (Wegge et al. 2006), suggesting resource availability for hog deer is relatively low on Yanakie Isthmus. Competition for limited food resources can reduce niche breadth (Hume 1999) as

Chapter 5 Diets of native and introduced herbivores 188 species adapt to use food resources selectively (Duncan 1992). However, if sympatric species broaden their diets as food becomes scarce (which often occurs if patch use is not altered in response to competition; Sih 1993), their diets will converge, resulting in competition (Pyke et al. 1977). This is a likely scenario when exotic species are introduced into an established native assemblage, where evolutionary segregation has not occurred (Voeten and Prins 1999). Therefore, the generally broad diets of herbivores on Yanakie Isthmus indicate a high potential for competition, despite demonstrating that each is capable of consuming a relatively unspecialised diet. The broad niche of the introduced hog deer relative to most native herbivores was expected, following Schwartz and Ellis (1981) and Duncan (1992). This trend may reflect an ability of introduced ruminants to broaden their diets to include less preferred food types as the abundance of preferred grasses declines; an attribute that Duncan (1992) suggested may facilitate the establishment of introduced ruminants. The native swamp wallaby also displayed a broad feeding niche, supporting the assertion by de Munk (1999) that swamp wallabies are generalists, able to utilise a heterogeneous diet. In contrast, an inability to use alternative food resources, or a greater capacity to utilise low- quality grasses, may explain the more selective feeding observed in kangaroos, which was consistent with de Munk’s findings (1999). Although specialised species may perform well in productive environments, when food is limiting, generalist species tend to do better due to their ability to switch food items (Schleuter and Eckmann 2007).

Conclusion Conclusions drawn from the patterns of food resource use by the five sympatric herbivore species on Yanakie Isthmus are limited because there was a strong seasonal bias in sample collection, and thus diets may vary at other times of the year. However, the results of this study, conducted primarily during winter, suggest that a number of ecological predictions based on competition theory and the Jarman-Bell principle do not hold true in contemporary herbivore guilds made up of species with disparate evolutionary histories. Foragers commonly respond to competition by reducing niche overlap and increasing diet specialisation (Schoener 1986). Therefore, the classical interpretation of high diet overlap and

Chapter 5 Diets of native and introduced herbivores 189 broad feeding niches, as was observed for herbivores on Yanakie Isthmus, would be that competition is not occurring. However, Voeten and Prins (1999) highlight key differences in the interpretation of diet overlap and niche breadth in sympatric populations of native and introduced species compared to populations of sympatric native species. These differences suggest that competition is likely on Yanakie Isthmus, particularly given indications of resource limitation through dietary shifts, broad niches and high diet overlap between native and introduced herbivores. However, it appears that coexistence of sympatric herbivores on Yanakie Isthmus is facilitated by food resource partitioning. In contrast to our predictions though, patterns of food resource use, niche breadth and diet overlap on Yanakie Isthmus were not well predicted by body size. This lack of conformity to the Jarman-Bell principle is probably due to interactions between intrinsic and extrinsic constraints on diet choice (Stephens and Krebs 1986): body size is likely to interact with digestive morphology and physiology, as well as food resource quality and availability (Schwartz and Ellis 1981). In particular, the relatively high overlap in food use between native and introduced species suggests that intrinsic constraints on digestion, which vary between species according to evolutionary history, are more important than body size in predicting diet on Yanakie Isthmus.

Chapter 5 Diets of native and introduced herbivores 190

Chapter 5 Diets of native and introduced herbivores 191

Table 5.1. Summary of gut samples (n = 93) collected from male and female herbivores on Yanakie Isthmus in five vegetation types over three seasons between 7 June 2004 and 11 February 2005. Coastal Scrubs Coastal Grassy and Heathy Species Season Woodland Grasslands Heath Woodland Total Female Male Female Male Female Male Female Male common wombat Winter 3 4 4 6 1 18 Spring 1 1 Summer 1 1 eastern grey kangaroo Winter 6 3 5 4 1 1 20 Spring 0 Summer 0 hog deer Winter 3 2 3 2 10 Spring 1 1 2 Summer 1 2 2 3 8 swamp wallaby Winter 1 5 6 Spring 1 1 Summer 1 1 1 2 1 6 European rabbit Winter 1 8 6 15 Spring 3 1 4 Summer 1 1 Total 20 15 24 29 0 2 3 0 93

Chapter 5 Diets of native and introduced herbivores 192

Table 5.2. Frequency of occurrence of plant species in gut samples from five herbivores. Dietary analysis was based on microhistological identification of epidermal fragments. *indicates introduced plant species.

Plant species Identified in gut samples of: Hog deer Common Swamp European Eastern grey (n = 20) wombat wallaby rabbit kangaroo (n = 20) (n = 13) (n = 20) (n = 20)

Woody plants Trees Allocasuarina littoralis 6 2 6 3 2 Banksia integrifolia 1 0 0 0 0 Eucalyptus obliqua 1 0 2 0 0 Eucalyptus viminalis 0 0 1 0 0 Shrubs ericifolia 4 0 3 0 2 Acacia longifolia 11 1 12 2 10 Acacia verticillata 4 3 9 0 3 Acrotriche serrulata 12 4 12 7 4 quadripartita 3 0 0 0 2 Alyxia buxifolia 6 2 6 1 5 Astroloma humifisium 3 1 3 8 6 Banksia marginata 1 0 0 0 0 Bossiaea cinera 2 1 0 0 0 Bursaria spinosa 5 0 5 0 0 Cassinia aculeata 2 1 1 2 2 Correa alba 0 1 2 1 0 Correa reflexa 1 0 2 0 1 Dillwynia glaberrima 1 0 1 0 0 impressa 2 0 2 0 0 Epacris lanuginosa 1 0 1 1 0 Epacris obtusifolia 0 0 2 0 0 Exocarpus syrticola 0 0 1 0 0 Hakea nodosa 0 0 1 0 0 Hibbertia sericea 6 5 4 2 7 Indigofera australis 2 0 3 0 0 Leptospermum continentale 1 0 2 0 0 Leptospermum laevigatum 8 0 9 0 1 Leptospermum lanigerum 1 0 3 0 0 Leptospermum myrsinoides 0 0 4 0 1 Leucopogon australis 1 0 0 0 0 Leucopogon ericoides 0 0 2 0 1 Leucopogon parviflorus 9 1 11 2 9 Leucopogon virgatus 3 0 2 0 2 Melaleuca ericifolia 2 0 3 0 0 Monotoca elliptica 0 0 1 0 0 Monotoca scoparia 2 0 2 0 0 Persoonia juniperina 2 0 0 0 1 Pimelia humilis 1 0 0 2 0 Pimelia phylicoides 1 2 1 0 0

Chapter 5 Diets of native and introduced herbivores 193

Table 5.2 (cont.) Pomaderris oraria 1 0 3 0 0 Pultanaea stricta 0 0 3 0 0 Rhagodia candolleana 19 2 7 4 7 Solanum aviculare* 0 0 0 0 1 Sprengelia incarnata 1 0 4 2 0 Tetragonia implexicoma 5 2 1 1 5 Tetratheca pilosa 2 0 0 0 1 Thomasia petalocalyx 2 0 4 0 0 Unidentified sp. 403 0 0 1 0 0

Ferns and allies Blechnum sp. 0 1 1 1 0 Gleichenia circinnata 1 0 5 1 0 Pteridium esculentum 7 0 9 0 2 Pteris comans 0 0 1 0 0

Grasses and ‘grass-like herbs’ Grasses Agrostis billardierei 4 12 0 12 10 Aira caryophyllea 6 16 1 16 16 Aira praecox 11 19 2 18 17 Austrodanthonia setacea 2 9 0 0 5 Austrodanthonia sp. 0 5 0 0 0 Briza minor* 3 14 0 1 7 Bromus diandrus* 1 6 0 1 4 Critesion murinum 5 12 0 5 8 Ehrharta calycina* 2 15 0 11 13 Imperata cylindrica 2 2 0 0 1 Lagurus ovatus* 13 10 1 9 13 Luzula campestus* 7 9 0 4 5 Phragmites australis 4 4 0 0 3 Poa poiformis 7 16 1 17 16 Poa sp. 10 5 9 0 8 5 Poa sp. 103 1 3 0 0 1 Poa sp. 106 6 16 0 11 16 Poa sp. 128 1 4 0 0 1 Poa sp. 131 2 13 0 13 9 Poa sp. 136 3 16 0 8 8 Poa sp. 137 2 10 0 0 3 Poa sp. 139 0 3 0 1 0 Poa sp. 154 6 16 0 8 15 Poa sp. 158 0 4 0 0 0 Poa sp. 167 6 14 0 8 17 Poa sp. 175 7 14 2 10 16 Sporobolus virginicus 0 3 0 0 0 Stipa compacta 9 19 2 16 17 Themeda triandra 6 12 0 3 7 Vulpia bromoides* 2 19 0 11 7 Grasslike plants Baumea articulata 0 10 2 1 11 Burchardia umbellata 0 7 2 3 2 Caesia parviflora 1 3 1 0 0 Dianella revoluta 0 10 1 1 6

Chapter 5 Diets of native and introduced herbivores 194

Table 5.2 (cont.) Ficinia nodosa 0 15 0 3 8 Isolepis inundata 6 16 3 7 11 Juncus kraussii ssp. 3 17 11 12 australiensis 1 Lepidosperma concavum 7 11 1 8 9 Lepidosperma filiforme 0 2 0 0 2 Lepidosperma sp. 2 14 0 17 10 Leptocarpus tenax 0 1 4 0 0 Patersonia fragili 4 15 0 2 10 Schoenus carsei 0 2 1 0 0 Stylidium graminifolium 0 1 0 2 0 Wurmbea dioica 2 7 1 1 1 Wurmbea sp. 1 10 0 8 8 Xanthorrhoea resinifera 0 7 0 1 8

Herbaceous plants Acaena sp. 1 0 1 1 0 Ajuga australis 2 6 0 6 2 Australina pusilla 2 0 0 0 0 Brachyscome diversifolia 4 7 7 10 4 Brachyscome graminea 4 5 6 14 8 Brachyscome parvula 3 1 0 0 1 Carpobrotus aequilaterus* 0 0 1 0 1 Centaurium erythraea* 0 1 1 0 1 Centella cordifolia 1 1 0 3 2 Cirsium vulgare* 7 5 5 6 6 Clematis microphylla 6 0 7 13 3 Convolvulus erubescens 4 0 0 2 1 Conyza alba* 8 6 3 18 9 Cotula reptans 8 3 5 6 6 Craspedia glauca 1 1 0 4 0 Crassula sieberiana 0 1 0 10 1 Cynoglossum australe 0 0 2 0 0 Dichondria repens 14 3 9 9 13 Galium australe 1 1 0 2 1 Galium tricornatum 1 0 2 0 0 Geranium solanderi 11 1 3 2 1 Glycine candestina 2 0 5 0 5 Hypochoeris glabra* 6 2 3 12 6 Hypolaena fastigata 0 3 1 1 1 Ipheion uniflorum* 2 1 10 1 0 Kennedia prostrata 15 2 1 4 12 Lagenifera stipitata 2 2 3 2 2 Leontodon taraxacoides* 4 2 2 4 2 Leucanthemum vulgare* 4 0 2 8 3 Lobelia alata 1 0 0 1 1 Mentha dienicia 2 0 1 0 1 Mentha gracilis* 3 1 1 1 2 Oxalis corniculata 5 0 5 10 1 Ozothamnus turbinatum 1 2 5 2 2 Plantago australis 2 0 0 7 1 Plantago lanceolata* 1 0 2 2 0 Plantago sp. 18 2 0 3 2 0

Chapter 5 Diets of native and introduced herbivores 195

Table 5.2 (cont.) Prunella vulgaris* 1 0 7 7 2 Ranunculus rivularis 7 2 10 6 2 Rubus parviflorus 13 2 0 3 6 Sambucus gaudichaudiana 6 0 1 0 1 Sebaea ovata 0 0 8 0 1 Senecio elegans* 9 2 0 6 2 Senecio hispidulus 1 0 1 3 2 Senecio jacobaea* 2 0 1 0 0 Senecio linearifolius 2 3 4 0 0 Senecio odoratus 8 1 2 2 8 Senecio sp. 119 0 0 0 0 0 Senecio sp. 148 1 0 2 0 0 Senecio variabilis* 1 2 0 2 3 Silene anglica* 0 0 1 1 1 Solanum nigrum* 5 1 0 5 4 Somolus repens 2 0 1 0 0 Sonchus olearaceus* 0 1 0 2 2 Taraxaum ruderalia* 1 0 1 0 0 Trifolium glomeratum* 1 1 1 2 1 Villarsia exaltata 2 0 0 1 0 Asteraceae sp. 2 0 1 0 0 0 Asteraceae sp. 100 0 2 0 0 0 Asteraceae sp. 109 7 1 2 1 9 Asteraceae sp. 110 1 0 0 0 1 Asteraceae sp. 112 0 1 1 0 0 Asteraceae sp. 118 2 0 0 0 0 Asteraceae sp. 120 1 1 3 1 2 Asteraceae sp. 141 0 0 0 1 0 Asteraceae sp. 142 0 0 1 0 1 Asteraceae sp. 143 0 0 1 0 0 Asteraceae sp. 149 0 1 0 1 0 Asteraceae sp. 150 1 0 0 0 0 Asteraceae sp. 151 1 0 0 0 0 Asteraceae sp. 162 0 1 0 0 1 Asteraceae sp. 1000 0 0 1 1 0 Orchidaceae sp. 147 4 0 0 4 3 Primulaceae sp. 123 1 0 0 0 0 Unidentified sp. 163 1 0 0 0 0 Unidentified sp. 652 1 1 0 0 0 Unidentified sp. 660 1 1 2 0 0 Unidentified sp. 114 1 0 0 3 0 Unidentified sp. 115 1 0 1 1 0

Mosses Hypnum cupressiforme var. 0 3 0 2 0 lacunosum Racopilum cuspidigerum var. 6 0 2 2 0 convolutaceum Thuidiopsis sparsa 0 1 1 0 0

Chapter 5 Diets of native and introduced herbivores 196

Table 5.3. Bootstrapped mean and upper and lower 95% confidence intervals for proportion of plant epidermal fragments identified within Structural, Broad Taxonomic, Functional Group and Plant Origin categories per gut sample for five herbivore species.

Category Hog deer Common womba t Swamp wallaby European rabbit Eastern grey kangaroo (n = 20) (n = 20) (n = 13) (n = 20) (n = 20) Proportion of fragments Mean Lower Upper Mean Lower Upper Mean Lower Upper Mean Lower Upper Mean Lower Upper CI CI CI CI CI CI CI CI CI CI Broad taxonomic Monocotyledon 0.166 0.114 0.224 0.918 0.882 0.948 0.030 0.016 0.045 0.469 0.393 0.546 0.648 0.570 0.722 Eudicotyledon 0.82 0.763 0.868 0.080 0.051 0.115 0.910 0.879 0.941 0.529 0.451 0.605 0.351 0.277 0.428 Other (Pteridophyte & 0.017 0.005 0.035 0.002 0.000 0.005 0.060 0.028 0.097 0.002 0.000 0.004 0.000 0 0.002 Bryophyte) Functional group Forb 0.369 0.284 0.459 0.039 0.022 0.060 0.273 0.203 0.344 0.408 0.347 0.467 0.178 0.126 0.232 Shrub 0.388 0.294 0.482 0.025 0.009 0.049 0.516 0.457 0.577 0.036 0.021 0.053 0.141 0.089 0.198 Tree 0.005 0.002 0.009 0.005 0.000 0.015 0.016 0.004 0.032 0.006 0.000 0.016 0.001 0.000 0.002 Fern 0.012 0.002 0.030 0.000 0.000 0.001 0.058 0.026 0.093 0.001 0.000 0.002 0.001 0.000 0.002 Grass 0.113 0.071 0.162 0.525 0.421 0.621 0.008 0.002 0.015 0.237 0.187 0.288 0.446 0.370 0.522 Grasslike plant 0.015 0.008 0.022 0.239 0.165 0.321 0.020 0.009 0.032 0.145 0.070 0.236 0.090 0.061 0.122 Moss 0.005 0.001 0.010 0.002 0.000 0.004 0.002 0.000 0.004 0.001 0.000 0.002 0.000 0.000 0.000 Plant part Leaf 0.807 0.776 0.838 0.715 0.715 0.715 0.781 0.781 0.781 0.652 0.574 0.725 0.741 0.701 0.777 Stem 0.083 0.061 0.105 0.059 0.037 0.086 0.105 0.075 0.137 0.152 0.088 0.221 0.058 0.037 0.082 Structural Shrub/tree layer 0.404 0.306 0.506 0.041 0.019 0.068 0.529 0.460 0.598 0.047 0.027 0.071 0.146 0.094 0.203 Ground layer 0.501 0.410 0.594 0.793 0.746 0.838 0.363 0.289 0.439 0.788 0.757 0.817 0.710 0.655 0.763 Unidentified 0.014 0.002 0.060 0.002 0.001

Chapter 5 Diets of native and introduced herbivores 197

Table 5.4. Results of one-factor analysis of variance comparing the diets of five herbivore species. Analyses are based on the proportions of plant epidermal fragments identified per stomach sample (n = 93). Plant epidermal fragments are categorised at the levels: broad taxonomic, functional, structural, plant part and plant origin. *P = < 0.05; **P = < 0.01; ***P < 0.001

Inter-specific test m.s. d.f. F P Broad taxonomic Monocot 3.88 4 114.44 *** Error 0.03 88 Dicot 3.44 4 103.85 *** Error 0.03 88 Functional group Forbs 0.50 4 20.60 *** Error 0.02 88 Shrubs 0.92 4 44.99 *** Error 0.02 88 Grasses 0.10 4 30.97 *** Error 0.03 88 Grasslike plants 0.18 4 9.21 *** Error 0.02 88 Structural group Non-woody ground layer 0.99 4 27.62 *** Error 0.04 88 Woody shrub/tree layer 0.95 4 37.97 *** Error 0.03 88 Plant part Leaf 0.07 4 5.59 *** Error 0.01 88 Stem 0.03 4 3.95 ** Error 0.01 88

Chapter 5 Diets of native and introduced herbivores 198

Table 5.5. Results of pair-wise comparisons using Tukey’s post-hoc analysis for one-factor analysis of variance (Table 2.1) comparing the diets of five herbivore species. Analyses are based on the proportions of plant epidermal fragments identified within categories of interest per stomach sample. n.s. = not significant; *P = < 0.05; **P = < 0.01; ***P < 0.001

Pairwise comparison Common Swamp European Eastern wombat wallaby rabbit grey (n = 20) (n = 13) (n = 20) kangaroo (n = 20) Hog deer Monocot *** n.s. *** *** (n = 20) Dicot *** * *** *** Forbs *** n.s. n.s. *** Shrubs *** n.s. *** *** Grasses *** n.s. n.s. *** Grasslike plants *** n.s. * n.s. Non-woody ground layer *** n.s. *** *** Woody shrub/tree layer *** n.s. *** *** Leaf n.s. n.s. *** n.s. Stem n.s. n.s. n.s. n.s. Common Monocot - *** *** *** wombat (n = 20) Dicot - *** *** *** Forbs - *** *** * Shrubs - *** n.s. n.s. Grasses - ** *** n.s. Grasslike plants - *** n.s. ** Non-woody ground layer - *** n.s. n.s. Woody shrub/tree layer - *** n.s. n.s. Leaf - n.s. n.s. n.s. Stem - n.s. * n.s. Swamp Monocot *** - *** *** wallaby (n = 13) Dicot *** - *** *** Forbs *** - n.s. n.s. Shrubs *** - *** *** Grasses *** - ** *** Grasslike plants *** - n.s. n.s. Non-woody ground layer *** - *** *** Woody shrub/tree layer *** - *** *** Leaf n.s. - * n.s. Stem n.s. - n.s. n.s. European Monocot *** *** - ** rabbit (n = 20) Dicot *** *** - ** Forbs *** n.s. - *** Shrubs n.s. *** - n.s.

Chapter 5 Diets of native and introduced herbivores 199

Table 5.5 (cont.) Grasses *** ** - *** Grasslike plants n.s. n.s. - n.s. Non-woody ground layer n.s. *** - n.s. Woody shrub/tree layer n.s. *** - n.s. Leaf n.s. * - n.s. Stem * n.s. - **

Chapter 5 Diets of native and introduced herbivores 200

Table 5.6. Values for Horn’s index of niche overlap (Ro) between five herbivore species. Calculations are based on the proportion of individuals of each of five herbivore species using the resource states: forbs, shrubs, trees, ferns, grasses, grasslike herbs, and mosses.

Hog deer Common Swamp European wombat wallaby rabbit Common wombat (n = 20) 0.49 Swamp wallaby (n = 13) 0.93 0.29 European rabbit (n = 20) 0.78 0.79 0.59 Eastern grey kangaroo (n = 20) 0.80 0.89 0.61 0.90

Chapter 5 Diets of native and introduced herbivores 201

(i) Broad taxonomic group 1

0.8

0.6 Monocot Dicot 0.4 Other 0.2

0 (ii) Functional group 0.7 Forbs 0.6 Shrubs 0.5 Trees 0.4 Ferns 0.3 Grasses 0.2 Grasslike plants 0.1 Mosses

0

1 (iii) Structural group

0.8

0.6 Non-w oody ground layer 0.4 Woody shrub/tree layer 0.2

0 (iv) Plant part

1 Mean proportion of fragmentsMeanof proportion

0.8

0.6 Stem Leaf 0.4

0.2

0

1 (v) Plant origin

0.8

0.6 Exotic Native 0.4

0.2

0 hog deer common sw amp European eastern grey w ombat w allaby rabbit kangaroo

Figure 5.1. Mean (with 95% confidence intervals) proportion of plant fragments identified in stomachs of individuals of five herbivore species (n = 20 for all species, except swamp wallaby n = 13): (i) broad taxonomic group; (ii) functional group; (iii) structural group; (iv) plant part; and (v) plant origin.

Chapter 5 Diets of native and introduced herbivores 202

(i) Functional group Axis 1*Axis 2 Axis 1*Axis 3 Axis 2*Axis 3 Species common wombat eastern grey kangaroo European rabbit hog deer swamp wallaby

(ii) Species

(iii) Family

Figure 5.2. Non-metric multi-dimensional scaling three-dimensional configuration (each axis displayed in two dimensions) of individuals of five herbivore species (n = 20 for all species, except swamp wallaby n = 13) based on a Bray-Curtis matrix of dissimilarities between the number of fragments identified in stomach samples belonging to each plant (i) functional group (stress = 0.09); (ii) species (stress = 0.19); and (iii) family (stress = 0.12). Standardisation (1) was applied within species, families and functional groups, and species and families occurring in one stomach only were excluded.

Chapter 5 Diets of native and introduced herbivores 203

Chapter 6

Conclusion ______

Chapter 6 Conclusion 204

Chapter 6 Conclusion 205

Chapter 6 Conclusion

Inter-specific interactions among sympatric species are much debated issues in theoretical ecology and practical wildlife management (Wegge et al. 2006). Competitive interactions, in particular, are pervasive in ecological systems (Schoener 1983) and are believed to be a central biotic factor structuring herbivore communities (Sinclair and Norton-Griffiths 1982, Schoener 1989). Investigation of competitive interactions, and the ways in which species coexist, is essential to improve our understanding of the ecological principles underlying ecosystem functioning (Duncan et al. 1998). Extensive work has been done to further our knowledge of inter-specific interactions in communities of native mammalian herbivores (e.g., Bell 1971). However, communities globally are increasingly composed of native and introduced mammalian herbivores (e.g., Voeten and Prins 1999, Baldi et al. 2004, Madhusudan 2004), adding a new dimension to this well-established area of study. The comparative study of multiple species in a community can give insight into the ecology of individual species and ecological partitioning between them (Telfer et al. 2008). In this study I combined multiple approaches to improve our understanding of large herbivore ecology and organisation in a contemporary assemblage made up of species with disparate evolutionary histories on Yanakie Isthmus (Figure 1.1), Wilsons Promontory National Park, Victoria, Australia (Figure 1.2). My aim was to compare niche parameters among populations of five sympatric native and introduced herbivore species by simultaneously assessing resource use along two dimensions (spatial and trophic) at multiple scales, thereby providing insight into resource overlap and partitioning within this herbivore assemblage (Krebs 1998). In this concluding chapter, I provide a synthesis of food and habitat resource use, and thus the potential for resource competition within the assemblage of native and introduced herbivores on Yanakie Isthmus, based on key findings regarding inter-specific niche overlap and niche partitioning, and trends in niche breadth and niche adjustment.

Chapter 6 Conclusion 206

Food and habitat resource use Faecal pellet counts on Yanakie Isthmus suggest that hog deer Axis porcinus, eastern grey kangaroos Macropus giganteus, swamp wallabies Wallabia bicolor, common wombats Vombatus ursinus and European rabbits Oryctolagus cuniculus used habitats selectively (Senft et al. 1985, Prins and Olff 1998), in ways which broadly reflect patterns of habitat use and food and shelter requirements described for these species in other parts of their ranges (e.g., Coulson 1990, Mayze and Moore 1990, Evans et al. 2006, Merchant 2008, Williams and Myers 2008). Similarly, microhistological estimates of herbivore diets suggest that the diets of each of the five herbivore species on Yanakie Isthmus were broadly similar to diets described for these species elsewhere in south-eastern Australia (e.g., Taylor 1983, Osawa 1990, Rishworth et al. 1995, Martin et al. 2007). However, patterns of inter-specific overlap in spatial and trophic resource use (Figure 6.1), niche breadth, and niche adjustment by some species provide important insights into niche dynamics within this herbivore assemblage, particularly inter-specific interactions involving resource partitioning and competition (Schoener 1986, Fritz et al. 1996). These are discussed in the following two sections.

Niche adjustment In multi-species assemblages, species may use only a subset of available resources due to the presence of sympatric species sharing resources (Hutchinson 1957). Species can be packed into assemblages as a result of either increasing the resource range, or narrowing the niche width, of constituent species (MacArthur 1972). When resources are scarce, niche adjustment is expected to be the predominant way of accommodating additional species in an assemblage (Namgail et al. 2009). Thus species may use similar habitats and diet when they occur separately, but feed on different food resources in different habitats when they occur in sympatry (Lawlor and Maynard Smith 1976, Werner 1977), following alterations in their habitat use and foraging strategies in response to changes in resource availability (Osawa and Woodall 1990). For example, impala Aepyceros melampus in competition for habitat and food resources in their preferred habitat (Jarman and Jarman 1974, Dunham 1981) adapt their foraging

Chapter 6 Conclusion 207 behaviour by switching to 'refuge' habitat, or by modifying their diet (Bell 1970, Illius and Gordon 1987). No clear shifts in habitat use from preferred to suboptimal habitats were apparent on Yanakie Isthmus, suggesting that inter-specific competition is not strong enough to cause competitive exclusion (Wegge et al. 2006). However, despite general similarities between habitat use on Yanakie Isthmus and habitat use described in other parts of the range of these herbivore species, there was some divergence in relationships between their habitat use and the availability of water, food and shelter resources from relationships previously presented in the literature. This divergence suggests that inter-specific interactions affect herbivore habitat use on Yanakie Isthmus. Despite broad parallels between the diets of the five herbivore species on Yanakie Isthmus and diets described for these species elsewhere in south-eastern Australia (e.g., Taylor 1983, Osawa 1990, Rishworth et al. 1995, Martin et al. 2007), important differences were observed. In particular, four of the five species on Yanakie Isthmus displayed diet shifts from grazing towards browsing. According to optimal foraging theory, food types should be added to the diet in rank order as the abundance of preferred items decreases (Pyke et al. 1977). Therefore, if competition for food resources reduces the availability of some food items more than others, this should directly alter diets (Sih 1993). In sympatric herbivores, diet switching can indicate food resource limitation, either inherent to the habitat or associated with inter-specific competition (Sih 1993, Fritz et al. 1996). Therefore, trends in diet switching from grazing towards browsing suggest that grass resources on Yanakie Isthmus are limited, and competition occurs within this herbivore assemblage (Namgail et al. 2009). This suggestion is supported by observations of niche adjustment on Yanakie Isthmus with respect to habitat use. However, inter-specific resource competition may be inferred only when all of the following necessary conditions are met: (i) habitats overlap, (ii) diets overlap; and (iii) resources are limited (De Boer and Prins 1990).

Resource use overlap For inter-specific resource competition to occur, there must be niche overlap between species in their need for resources (Hutchinson 1957). Greater ecological overlap generally implies a greater tendency to compete when niche

Chapter 6 Conclusion 208 dimensions are food or microhabitat, while the opposite may be true for macrohabitat, as spatial overlap is commonly associated with niche differentiation on other dimensions (Schoener 1983). Inter-specific overlap in broad scale habitat use among herbivores on Yanakie Isthmus was generally low (Figure 6.1). If overlap in habitat use is low, current competition is unlikely (Schoener 1983), due to differences in resource requirements. However, in some cases low inter-specific overlap in habitat use indicates competitive exclusion (MacArthur and Levins 1967, Abrams 1983), possibly associated with present-day or past competition (the ghost of competition past; Connell 1980). Despite generally low levels of overlap in habitat use among the herbivore assemblage on Yanakie Isthmus, high overlap in broad and fine-scale habitat use was apparent between some species. High overlap in microhabitat in particular can imply a high tendency to compete (Schoener 1983) if resources are limited and ecological separation does not occur on other dimensions (Hutchinson 1957). This suggestion is further supported by the similar relationships to habitat features (and thus resource requirements) that were displayed by some species. However, even if overlap in habitat use is high, if species do not exclude one another from common habitats, competition between them may be minimal due to resource abundance, or due to the evolution of mechanisms of coexistence along other dimensions in response to high spatial overlap (Schoener 1983). In contrast to habitat use, high diet overlap was observed among herbivore species on Yanakie Isthmus (Figure 6.1). Foragers commonly respond to competition by increasing diet specialisation to reduce niche overlap (Schoener 1986). Therefore, the classical interpretation of high diet overlap would be that competition is not occurring. However, Voeten and Prins (1999) highlight key differences in the interpretation of diet overlap in sympatric populations of native and introduced species compared to populations of sympatric native species (detailed bellow). If resources are limited, high diet overlap and overlap in habitat use at some scales between some species is likely to result in inter-specific competition (Gause 1934). Combining data on inter-specific differences in densities, population metabolism and resource use provides some insight into the potential implications of competition for herbivore species within this system.

Chapter 6 Conclusion 209

Species-specific implications of competition The balance between population density and energy use per individual determines how evenly community resources are shared by species of different size (duToit and Owen-Smith 1989). Distribution of resources can be particularly important when patterns of resource use by sympatric herbivore species are similar, as resource limitation due to competition can suppress population densities of one or both competing species (Madhusudan 2004). Overall, the abundance and population metabolism of herbivores with grazing diets on Yanakie Isthmus (i.e., rabbits, wombats and kangaroos) was far greater than that of herbivores with browsing diets (i.e., wallabies and deer), and this trend was even more pronounced if hog deer were considered grazers. High diet overlap among grazers and their high population densities suggest the potential for competition among grazers if food resources are limited, as is suggested by niche adjustment from grazing towards browsing in four of the five herbivore species. The species that occurred in the lowest density on Yanakie Isthmus was the eastern grey kangaroo. Eastern grey kangaroos also had a relatively low metabolic demand at the population level, closely followed by swamp wallabies. This could indicate competitive suppression (Madhusudan 2004) of macropod populations, particular grazing kangaroos, although these results may in part be an artefact of the low metabolic rate of marsupials (Hume 1982). The narrow niche breadth of kangaroos with respect to food and habitat may reflect an inability to use as great a variety of the resources available of Yanakie Isthmus as other species can. Specialised species may do well in productive environments, however, when food is limiting, generalist species tend to do better due to their ability to switch food items (Schleuter and Eckmann 2007). For example, the generalist swamp wallaby, which displayed a broad feeding niche, may be more capable of persisting on Yanakie Isthmus, as it can utilise the food niche used by kangaroos, while the specialist kangaroo is less able to utilise the diverse food species used by the swamp wallaby (de Munk 1999). The most abundant herbivore species on Yanakie Isthmus were the native wombat and the introduced rabbit. In particular, wombats had a relatively high metabolic demand at the population level over Yanakie Isthmus as a whole, as well as pre- and post-fire at Big Hummock, suggesting wombats utilise a greater

Chapter 6 Conclusion 210 proportion of resources than sympatric herbivore species, and therefore may be competitively superior in this environment relative to the other herbivore species present. In contrast, metabolic demand at the population level for small-bodied rabbits was relatively low on GYI despite their high density, indicating relatively low use of resources compared to the other species. Rather, hog deer ranked second with respect to metabolic demand at the population level despite its relatively low density, suggesting that wombats and deer are the most efficient competitors in this assemblage, dominating food resource use. In particular, the wombat may have a competitive advantage over other large herbivores in low productivity habitats due to adaptations that allow it to reach high population densities (Johnson 1998, Hume 1999): their efficient masticatory system and greater capacity for digesta retention and fermentation (Hume 1999) enables them to better process lower-quality forage than similarly sized ruminants (Illius and Gordon 1993) or foregut fermenters (Demment and Van Soest 1985), and their low metabolic rate and use of burrows allows them to conserve water and energy (Hume 1999). The suggestion of a competitive advantage of wombats is supported by their relatively broad habitat niche and, unlike the other four species, their lack of diet switching on Yanakie Isthmus relative to diet observations elsewhere (Rishworth et al. 1995, Hume 1999). Kangaroos and wombats have a greater capacity for digestion of high-fibre grasses than ruminants (Hume 1999), and ruminants have higher energy requirements than marsupials (Freudenberger et al. 1988). However, despite the possibility that competition on Yanakie Isthmus has forced hog deer to shift their diet from graze towards browse, a relatively high population metabolism for this species suggests it has a competitive advantage over native grazers. Hog deer may be more capable of exploiting nutritious browse (Taylor 1971, Roberts 1977) than species such as the eastern grey kangaroo, which possess a digestive system that is inefficient at utilising high-quality forages (Dawson 1989, Hume 1999). Hog deer may therefore have a competitive advantage if grass resources on Yanakie Isthmus are scarce, but forbs and shrubs are abundant. Taylor (1971) suggested that in Victoria, hog deer occupy a previously vacant ecological niche. However, on Yanakie Isthmus hog deer consumed a greater proportion of dicots than observed by Taylor (1971) and their diet overlapped with those of swamp wallabies, kangaroos and rabbits. Hog deer also

Chapter 6 Conclusion 211 exhibited high spatial overlap with eastern grey kangaroos at a broad scale, and with swamp wallabies at a fine scale, and displayed similar responses to changes in the cover of the dominant shrub species, Leptospermum laevigatum, as did kangaroos, indicating similarities in food and/or shelter requirements with other species and thus a high potential for inter-specific competition. Taylor (1971) recognised that under conditions of resource limitation, hog deer may shift their diet towards browse rather than compete for grasses, a phenomenon that has been observed in other systems where macropods compete with ruminants (Ealey 1967, Edwards et al. 1995). The broad diet niche of the introduced hog deer relative to most native herbivores on Yanakie Isthmus may reflect an ability of introduced ruminants to broaden their diets to include less preferred food types as the abundance of preferred grasses decline, an attribute that Duncan (1992) suggested may facilitate the establishment of introduced ruminants. European rabbits can compete with native herbivores for food (Dawson and Ellis 1979). On Yanakie Isthmus, the diet of rabbits overlapped with the diets of both wombats and kangaroos. High diet overlap and low overlap in broad habitat use between the European rabbit and sympatric grazing species may indicate competitive exclusion of other grazers by rabbits, particularly from Coastal Grassy Woodland, which was used preferentially by rabbits, but avoided by all other herbivore species. However, while small selective herbivores are better able to meet their energy requirements on shorter swards than large bodied animals with grazing dentition (Gordon and Illius 1988, 1989), rabbits have high nitrogen and energy requirements (Cooke 1974), digest fibre poorly (Voris et al. 1940) and have a limited ability to increase food intake to compensate for low- quality food (Martin et al. 2007). Therefore, rabbits are restricted to higher quality foods than wombats or kangaroos (Leigh et al. 1991). If forage on Yanakie Isthmus is of poor quality, these digestive constraints, combined with control of rabbit populations by managers (Parks Victoria 2003a) and concentration of predation by red foxes Vulpes vulpes on this species (Lunney et al. 1990), may minimise competition between rabbits and native herbivores.

Resource partitioning Despite indications of competition on Yanakie Isthmus, populations of sympatric species persist. Coexistence of sympatric species with similar resource

Chapter 6 Conclusion 212 requirements is likely to be facilitated by resource partitioning: each herbivore species partitioned at least one type of resource from every other species, and except for hog deer and swamp wallabies, there was never high overlap in use of more than one type of resource (Figure 6.1). Resource partitioning reduces niche overlap and therefore reduces competition (Gause 1934, Schoener 1974a, b, Whitfield 2002). Specialisation on resources along habitat, diet and temporal gradients results in niche differentiation, which is seen as the evolutionary outcome of competition: each species in a community becomes adapted to exploit a unique niche (Schoener 1974b, Whitfield 2002, Schmidt et al. 2010). Niche differentiation can occur along several dimensions (Pianka 1974, Schoener 1983, le Mar and McArthur 2005) but is generally complementary; when species are similar on one niche dimension, they differ on another (Pianka 1976, Dunbar 1978, Fox 1989, Bagchi et al. 2003). Habitat is the most common dimension partitioned, followed by food resources (Schoener 1983). Ecological separation can involve differences in both diet and habitat use (Jarman and Phillips 1989, de Munk 1999), but separation along these two dimensions is often complementary (Bagchi et al. 2003). For example, a number of studies have demonstrated that species with similar diets partition habitats, whereas species with different diets often used the same habitat (Pianka 1976, Dunbar 1978, Taylor 1983, Bodmer 1991). Temporal partitioning becomes important in environments where resources are rapidly renewed (Kotler et al. 1993). Overlap in food resource use among sympatric herbivore species on Yanakie Isthmus was high, suggesting that these herbivores should partition habitat resources to minimise inter-specific competition. In support of this prediction, inter-specific overlap in broad scale habitat use was generally low and broad scale habitat preferences were generally varied (Figure 6.1). These results suggest that differential habitat selection within this community facilitates coexistence, as has been observed in other communities of ecologically similar sympatric species (MacArthur and Levins 1964, Rosenzweig 1981). Fine-scale habitat resource partitioning was evidenced by differential associations of species with dominant habitat features, suggesting differential food and/or shelter requirements. In addition, fine-scale habitat resource partitioning was seen in inter-specific differences in the use of mechanically slashed and unslashed vegetation strata within Coastal Grassy Woodland at Big Hummock

Chapter 6 Conclusion 213

(e.g., greater use of slashed swale by rabbits and kangaroos, and greater use of unslashed vegetation by wallabies; Figure 6.1) which reflected differential responses of herbivores to woody shrub encroachment (e.g., Riginos and Young 2007) and open and closed habitats elsewhere (e.g., Fa et al. 1999, Schmidt et al. 2010). Fine-scale differences in habitat use suggest inter-specific partitioning of habitat resources, which suggests a low likelihood of competition, possibly aided by temporal partitioning. However, fine-scale resource partitioning could be occurring in response to resource competition (Pianka 1974). These results reflect patterns of habitat use observed in many other herbivore assemblages globally, where multiple species occupy essentially the same environment, but each species is adapted to certain habitat conditions which ecologically separate it to some extent from other species (Lamprey 1963, Schaller 1967, Jarman 1972). The results of this study suggest that ecological processes that modify habitat structure and composition can also alter patterns of habitat partitioning. Fire in Coastal Grassy Woodland appeared to modify food and shelter resources sufficiently for the mammalian herbivore community to change in composition and abundance, as has been observed in other herbivore communities (Catling 1991, Leigh et al. 1991, Van Dyke and Darragh 2006). Overall, there was a decrease in the density and population metabolism of grazers after ecological burning, probably due to changes in the vegetation such as decreased cover of ground layer vegetation, particularly grasses, and an increase in the density and population metabolism of browsers, probably associated with increased palatability of regenerating woody plants. Differential responses to burning from herbivores with differing feeding strategies suggest that differential resource requirements provide a mechanism for fine-scale habitat resource partitioning in the herbivore assemblage on Yanakie Isthmus, facilitating coexistence. Resource partitioning among sympatric herbivores is commonly based on differences in feeding strategies (Schwartz and Ellis 1981) and can occur at fine scales in species whose diets appear broadly similar (Bagchi et al. 2003). In addition to habitat partitioning, it appears that coexistence of sympatric herbivores on Yanakie Isthmus is also facilitated by food resource partitioning at a fine scale, for example, at the level of plant parts and species, despite high inter-specific diet overlap at a broad level. Indeed, differences in diet selection may drive differences in habitat selection (Swan et al. 2008). Thus on Yanakie Isthmus,

Chapter 6 Conclusion 214 niche complementarity of food and habitat dimensions appears to result in total niche overlap below the competition threshold (Bosakowski et al. 1992). In particular, partitioning of habitat resources appears to be important in organising the herbivore community of Yanakie Isthmus, while further partitioning occurs in food resource use. These results mirror the results of studies in macropod communities, where habitat selection is important in organising communities, while further partitioning may occur with respect to diet preference, but is of less importance except where it is complementary for species with high overlap in habitat use (Fox 1989). Resource availability constrains species richness in ecological communities (MacArthur 1972, Tonn et al. 1990) and only those species that show trade-offs in niche utilisation in response to competition can coexist (Chase and Leibold 2003). The results of this study indicate that, as in many other herbivore communities, herbivore species on Yanakie Isthmus make trade-offs with respect to niche utilisation and partition resources to facilitate coexistence (e.g., Zapata et al. 2005, Azevedo et al. 2006, Prigioni 2008, Namgail et al. 2009). Further, the results of this study suggest that the resource base on Yanakie Isthmus is heterogeneous enough to allow coexistence of a community comprised of five ecologically similar species. The basic tenet of competitive exclusion is that n number of species cannot coexist on fewer than n resources (Gause 1934, Hutchinson 1959). Body size and evolutionary history are two factors which have been shown to have a strong influence on niche dynamics within other herbivore communities (e.g., Jarman 1974, Schwartz and Ellis 1981). The next step in this study was to decipher whether these processes govern patterns of resource partitioning within this herbivore community.

The influence of body size on niche breadth and resource use overlap If two populations have access to the same resource base, the population whose members discriminate less among resource states has a broad niche relative to a population whose members concentrate on particular resource states (Levins 1968). Body size influences niche differentiation due to its influence on the efficiency with which foods can be consumed and utilised (Illius and Gordon 1993). According to the Jarman-Bell principle (Bell 1970, Jarman 1974), smaller species require more energy per unit body weight and therefore must feed

Chapter 6 Conclusion 215 selectively on higher quality foods (Van Soest 1994). In contrast, larger animals with relatively lower metabolic rates and greater gut capacity can utilise more abundant foods of lower quality (Bell 1971, Jarman 1974). Increased dietary tolerance among larger species, i.e., tolerance of a wider range of food items in terms of nutritional quality or food-item size (May and MacArthur 1972, Schoener 1974b), also affects habitat use, generally allowing larger species to use of a wider range of habitat patches, hence a more even use of environmental resources (duToit and Owen-Smith 1989). Therefore, it was expected that both diet and habitat use on Yanakie Isthmus would be influenced by body size, with smaller herbivore species being more selective (i.e., having narrower niches) than larger herbivore species (duToit and Owen-Smith 1989). However, in contrast to many sympatric mammalian herbivore assemblages (e.g., Demment and Van Soest 1985, duToit and Owen-Smith 1989), body size was not a good predictor of herbivore habitat or diet niche breadths, nor of diet quality on Yanakie Isthmus. Diversity in habitat use by large species is limited by the diversity of available habitats (duToit and Owen-Smith 1989). Being a temperate ecosystem, Yanakie Isthmus may not support great enough spatial heterogeneity to detect a relationship between body mass and habitat diversity compared with tropical ecosystems where relationships have been detected (Peters and Raelson 1984, duToit and Owen-Smith 1989). The lack of conformity of patterns of food resource use, niche breadth and diet overlap on Yanakie Isthmus to body size-related predictions based on the Jarman-Bell principle is probably due to interactions between intrinsic and extrinsic constraints on diet choice (Stephens and Krebs 1986): body size is likely to interact with factors such as digestive morphology and physiology, which vary according to evolutionary history, as well as food resource quality and availability (Schwartz and Ellis 1981). These results suggest that ecological predictions based on the Jarman-Bell principle do not hold true in contemporary herbivore guilds made up of species with disparate evolutionary histories.

The influence of evolutionary history on niche breadth and resource use overlap Patterns of niche differentiation can reflect phylogenetic aspects of species, for example, the environment to which species are adapted (Bagchi et al. 2003), and mechanisms of resource partitioning are likely to have evolved to

Chapter 6 Conclusion 216 facilitate coexistence among species with common evolutionary histories. However, species with disparate evolutionary histories have inherently less resource partitioning (Kirchhoff and Larsen 1998, Kelley et al. 2002, Madhusudan 2004). Therefore, overlap in resource use is generally greater between introduced and native herbivore species, than between native herbivore species (Kelley et al. 2002). This has implications for relationships between resource use overlap and resource availability (Pianka 1974). Pianka’s (1974) theory of maximal tolerable niche overlap predicts that there should be less niche overlap in resource limited situations compared to ones with higher resource abundance (Pianka 1974) due to the evolution of mechanisms of resource partitioning (Schwartz and Ellis 1981). Therefore, high resource use overlap is commonly interpreted as indicating resource abundance and reduced competition (Schoener 1982), as species can share resources when they are not limiting (Pianka 1974). However, Voeten and Prins (1999) argue that while overlap in resource use is not expected among native species under food-limited conditions, when exotic species are introduced into native assemblages, overlap in resource use can occur under food-limited conditions because evolutionary segregation has not occurred (Kirchhoff and Larsen 1998, Madhusudan 2004). Spatial segregation of native and introduced herbivore species has been observed in many herbivore communities (e.g., Hanley and Hanley 1982, Hanley 1984, Madhusudan 2004). Overlap in broad scale habitat use by herbivores on Yanakie Isthmus was low, suggesting the possibility of resource-mediated segregation due to diet overlap and resource limitation (Madhusudan 2004). However, broad scale patterns of spatial segregation were not related to evolutionary history: there was high overlap in habitat use between some native species, as well as between some introduced and native species (Figure 6.1). It is possible that adaptation resulting in niche adjustment by native and/or introduced species has occurred in response to competition within the evolutionary timeframe for which this assemblage has coexisted (de Munk 1999), as invaders can evolve rapidly in response to novel abiotic and biotic conditions, and native species can evolve in response to invasions (Sakai et al. 2001). For example, rapid adaptation may have occurred in response to high dietary overlap between native and introduced herbivores and food resource limitation (Madhusudan 2004). However, there did appear to be more overlap in fine-scale habitat use between native and

Chapter 6 Conclusion 217 introduced herbivore species than between native herbivore species (Figure 6.1), indicating that to some extent, mechanisms of resource partitioning have evolved in species with common evolutionary histories, whereas species with disparate evolutionary histories have inherently less resource partitioning (Kirchhoff and Larsen 1998, Kelley et al. 2002, Madhusudan 2004). Overlap in food resource use among sympatric herbivores on Yanakie Isthmus was high, particularly between native and introduced species (Figure 6.1). These results support other work that has demonstrated the potential for high overlap in diet between sympatric native and introduced herbivores (e.g., Dawson and Ellis 1996), and suggest that intrinsic constraints on digestion, which vary between species according to evolutionary history, are more important than body size in predicting diet of herbivores on Yanakie Isthmus. Further, based on the ideas put forward by Voeten and Prins (1999), high overlap in food resource use between native and introduced species on Yanakie Isthmus is likely to indicate competition for limited resources. Closely related to the concept of niche overlap is the concept of niche breadth. According to optimal foraging theory, broad dietary niches indicate low resource availability: as food becomes scarce, individuals respond by widening their choice of foods to include less nutritious plants due to resource constraints (Pyke et al. 1977). In contrast, surplus resources allow selectivity: as the abundance of a preferred food type increases, the number of less preferred food types consumed will be reduced as the species includes fewer but more nutritious plant species in its diet (Pyke et al. 1977). Therefore, the generally broad feeding niches observed on Yanakie Isthmus provide support for the assertion that resource availability is limited. Competition for limited food resources can reduce niche breadth (Hume 1999), as species adapt to use food resources selectively (Duncan 1992). However, if sympatric species broaden their diets as food becomes scarce, due to reduced prey encounter rates or prey depletion, which often occurs if patch use is not altered in response to competition (Sih 1993), their diets will converge, resulting in competition (Pyke et al. 1977). This is a likely scenario when exotic species are introduced into an established native assemblage, where evolutionary segregation has not occurred (Voeten and Prins 1999). Therefore, the generally broad diet niches of herbivores on Yanakie Isthmus provide further evidence for competition within this community. However, it is

Chapter 6 Conclusion 218 the sum of habitat patch and food choices that determines coexistence (Pyke et al. 1977). If a variety of habitat patches with different foods are available, patch utilisation by two species may converge or diverge depending on the similarity in responses to declining food (Pyke et al. 1977). In contrast to the relationship between diet niche width and inter-specific competition, inter-specific competition tends to limit habitat use, narrowing habitat niche breadth (Svarsden 1949, Schoener 1986). Trends in habitat niche breadth on Yanakie Isthmus were not clear: breadth varied among species, being broad for some species and narrow for others. The broad habitat niches observed for some species may reflect intra- specific competition, which tends to broaden habitat use rather than narrow it, as inter-specific competition does (Svarsden 1949, Rosenzweig 1991). Resource competition is an important process governing the influence of introduced herbivore populations on native herbivore populations (Madhusudan 2004). If habitat and diet requirements of native and introduced herbivores are similar on Yanakie Isthmus, leading to resource limitation, as has been observed during other studies (e.g., Madhusudan 2004), the long term viability of populations of native species may be threatened. Similar outcomes of resource competition have been suggested in Australia for native swamp wallabies in sympatry with introduced rusa deer Cervus timorensis (Ramp and Ben-Ami 2006), as well as for populations of sympatric native and introduced herbivores in other parts of the globe (e.g., Runyoro et al. 1995, Madhusudan 2004, Mishra et al. 2004). However, the overall metabolic demand from native herbivores on Yanakie Isthmus was considerably greater than that of introduced herbivores, suggesting that introduced herbivore species are not superior competitors to native herbivore species.

Future directions This study provides important preliminary insights into herbivore community niche dynamics on Yanakie Isthmus. High diet overlap and overlap in habitat use at some scales between some species, coupled with resource limitation is likely to result in inter-specific competition (Gause 1934), particularly given indications of resource limitation through diet niche adjustments, broad niches and high diet overlap between native and introduced herbivores on Yanakie Isthmus. However, without experimental manipulation (Sih 1993), the effect of competitive

Chapter 6 Conclusion 219 interactions on resource selection cannot be determined (Manor and Saltz 2008), nor can competition be conclusively demonstrated (Park 1962). Moreover, although competition is believed to be a central biotic factor structuring herbivore communities (Sinclair and Norton-Griffiths 1982, Schoener 1989), factors such as food quality and availability (Fritz et al. 1996), weather conditions, and predator and parasite avoidance are also potentially important in influencing resource use and thus community interactions (Duncan 1983, Schmitz 1998, Kotler and Brown 2007). Inter-specific interactions such as facilitation can also have an important, often indirect, influence on community dynamics (Hobbs et al. 1996, Schmitz 1998, Kuiters et al. 2005, Bakkera et al. 2009) and these interactions and associated resource use will vary over time and across space (Taylor 1983, Evans and Jarman 1999). Large-scale adaptive experimental management (Walters and Holling 1990), involving the measurement of species’ responses to altered abundance of potential competitors (Schoener 1983, Sih 1993), is required to obtain a mechanistic understanding of species interactions (Sih 1993). Manipulation of the density of one or more of the herbivore species on Yanakie Isthmus would provide an opportunity to test the ideas presented in this thesis. For example, population control involving eradication of hog deer by exclusion or culling, reduction of wombat densities, or cessation of rabbit control, could be followed by measurement of potential response variables such as changes in the population densities, biomass or health of other species, and shifts in their diets and habitat use. This would provide a mechanistic understanding of species interactions, and their influence on community structure, within mammalian herbivore communities comprising a range of body sizes and diverse origins.

Chapter 6 Conclusion 220

Chapter 6 Conclusion 221

Eastern Hog Common grey Swamp European Niche dimension Species deer * wombat kangaroo wallaby rabbit

Ascending body-size Legend High overlap Medium overlap Broad-scale habitat use Common wombat Low overlap (Hulberts’ index of niche overlap) Eastern grey kangaroo Swamp wallaby European rabbit *

Fine-scale habitat use Common wombat (relative use of slashed swale vs. Eastern grey kangaroo scrub/dune woodland) Swamp wallaby European rabbit *

Diet Common wombat (Horn’s index of niche overlap) Eastern grey kangaroo Swamp wallaby European rabbit *

Figure 6.1. Spatial and trophic resource use overlap among five herbivore species on GYI. High, medium and low overlap in broad-scale habitat use indicate Hulberts’ index of niche overlap values > 1.3, > 0.8 to < 1.3, and < 0.8, respectively (based on faecal pellet counts). High and low overlap in fine-scale habitat use indicate pairs of species for which faecal pellet counts are greater in the same vegetation strata, and pairs of species for which pellet counts are greater in contrasting vegetation strata, respectively, and medium overlap indicates pairs of species for which neither the high nor low overlap definition applied (i.e., faecal pellet counts were not greater in the same strata, nor were they greater in opposite strata). High, medium and low overlap in diet indicate Hulberts’ index of niche overlap values of > 0.8, > 0.3 to < 0.8, and < 0.3, respectively (based on microhistological diet analysis). * indicates introduced species.

Chapter 6 Conclusion 222

Chapter 6 Conclusion 223

References ______

References 224

References 225

References

Abrams, P. A. 1983. The theory of limiting similarity. Annual Review of Ecology and Systematics 14:359-376. Abramsky, Z., M. Rosenzweig, and B. Pinshow. 1991. The shape of a gerbil isocline: an experimental field study using principles of optimal habitat selection. Ecology 72:328-340. Allcock, K. G., and D. S. Hik. 2004. Survival, growth, and escape from herbivory are determined by habitat and herbivore species for three Australian woodland plants. Oecologia 138:231-241. Anderson, T. M., J. G. C. Hopcraft, S. Eby, M. Ritchie, J. B. Grace, and H. Olff. 2010. Landscape-scale analyses suggest both nutrient and antipredator advantages to Serengeti herbivore hotspots. Ecology 91:1519-1529. Antos, J. A., B. McCune, and C. Bara. 1983. The effect of fire on an ungrazed western Montana grassland. The American Midland Naturalist 110:354– 364. Australian National Parks and Wildlife Service. 1995. Code of practise for the humane shooting of kangaroos. Australian National Parks and Wildlife Service, Canberra. Azevedo, F., V. Lester, W. Gorsuch, S. Lariviere, A. Wirsing, and D. Murray. 2006. Dietary breadth and overlap among five sympatric prairie carnivores. Journal of Zoology 269:127-135. Bagchi, S., S. P. Goyal, and K. Sankar. 2003. Niche relationships of an ungulate assemblage in a dry tropical forest. Journal of Mammalogy 84:981-988. Bagchi, S., C. Mishra, and Y. V. Bhatnagar. 2004. Conflicts between traditional pastoralism and conservation of Himalayan ibex (Capra sibirica) in the Trans-Himalayan mountains. Animal Conservation 7:121-128. Bailey, R. E., and R. J. Putman. 1981. Estimation of fallow deer (Dama dama) populations from faecal accumulation. Journal of Applied Ecology 18:697- 702. Bakkera, E. S., H. Olffb, and J. M. Gleichmana. 2009. Contrasting effects of large herbivore grazing on smaller herbivores. Basic and Applied Ecology 10:141-150.

References 226

Baldi, R., A. Pelliza-Sbriller, D. Elston, and S. Albon. 2004. High potential for competition between guanacos and in Patagonia. Journal of Wildlife Management 68:924-938. Barboza, P. S., I. D. Hume, and J. V. Nolan. 1993. Nitrogen metabolism and requirements of nitrogen and energy in wombats (Marsupalia: Vombatidae). Physiological Zoology 66:807-828. Barnes, R. F. W., and A. Dunn. 2002. Estimating forest elephant density in Sapo National Park (Liberia) with a rainfall model. African Journal of Ecology 40:159-163. Batcheler, C. L. 1960. A study of the relations between roe, red and fallow deer, with special reference to Drummond Hill forest, Scotland. Journal of Animal Ecology 29:375-384. Beale, I. F. 1973. Tree density effects on yields of herbage and tree components in south-west Queensland mulga (Acacia aneura F. Muell.) scrub. Tropical Grasslands 7:135-142. Bell, R. H. V. 1970. The use of the herb layer by grazing ungulates in the serengeti. Pages 111-123 in Animal populations in relation to their food resources. Blackwell Scientific Publications, Aberdeen. Bell, R. H. V. 1971. A grazing ecosystem in the Serengeti. Scientific American 225:86-93. Bennett, L. T. 1994. The expansion of Leptospermum laevigatum on the Yanakie Isthmus, Wilsons Promontory, under changes in the burning and grazing regimes. Australian Journal of Botany 42:555-564. Benwell, A. S. 1998. Post-fire seedling recruitment in coastal heathland in relation to regeneration strategy and habitat. Australian Journal of Botany 46:75- 101. Bingwan, L., and J. Zhigang. 2004. Dietary overlap between Przewalski's gazelle and domestic sheep in the Qinghai Lake region and implications for rangeland management. Journal of Wildlife Management 68:241-246. Biondini, M. E., A. A. Steuter, and R. G. Hamilton. 1999. Bison use of fire- managed remnant prairies. Journal of Range Management 52:454–461. Biwas, T. a. M., V. B. 2000. A review of the present conservation scenario of hog deer (Axis porcinus) in its native range. Indian Forester 126:1068-1084.

References 227

Bodmer, R. E. 1991. Influence of digestive morphology on resource partitioning in Amazonian ungulates. Oecologia 85:361-365. Bonino, N., A. Sbriller, M. M. Manacorda, and F. Larosa. 1997. Food partitioning between the mara (Dolichotis patagonum) and the introduced hare (Lepus europaeus) in the Monte Desert, Argentina. Studies of Neotropical Fauna and Environment 32:129-134. Borgnia, M., B. Vila, and M. Cassini. 2007. Interaction between wild camelids and livestock in an Andeansemi-desert. Journal of Arid Environments 72:2150-2158. Bosakowski, T., D. G. Smith, and R. Speiser. 1992. Niche overlap of two sympatric-nesting hawks Accipiter spp. in the New Jersey-New York highlands. Ecography 15:358-372. Bradley, L. R. 1978. A comparison of methods for censusing macropods in the Styx River State Forest, New South Wales. Honours thesis. University of New England. Braithwaite, R. W. 1990. Australia's unique biota: implications for ecological processes. Journal of Biogeography 17:347-354. Bridle, K., and J. Kirkpatrick. 2001. Impacts of grazing by vertebrate herbivores on the stem production of tall alpine herbs, Eastern Central Plateau, Tasmania. Australian Journal of Botany 49:459-470. Briggs, J. M., A. K. Knapp, J. M. Blair, J. L. Heisler, G. A. Hoch, M. S. Lett, and J. K. McCarron. 2005. An ecosystem in transition: Causes and consequences of the conversion of mesic grassland to shrubland. BioScience 55:243-254. Briggs, J. M., A. K. Knapp, and B. L. Brock. 2002. Expansion of woody plants in tallgrass prairie: A fifteen-year study of fire and fire-grazing interactions. The American Midland Naturalist 147:287. Brodie, J. 2006. An experimentally determined persistence-rate correction factor for scat-based abundance indices. Wildlife Society Bulletin 34:1216-1219. Brown, J. H., and B. A. Maurer. 1986. Body size, ecological dominance and Cope's rule. Nature (Lond.) 324:248-250. Cabrera-Rodriguez, F. 2006. Microhabitat selection of the European rabbit on La Palma, Canary Islands, Spain. Acta Theriologica 51:435-442.

References 228

Calaby, J. H. 1966. Mammals of the upper Richmond and Clarence Rivers, New South Wales. Technical Paper, Division of Wildlife Survey, CSIRO, Australia. Calvete, C., R. Estrada, E. Angulo, and S. Cabezas-Ruiz. 2004. Habitat factors related to wild rabbit conservation in an agricultural landscape. Landscape Ecology 19:531-542. Carvalho, J. C., and P. Gomes. 2004. Influence of herbaceous cover, shelter and land cover structure on wild rabbit abundance in NW Portugal. Acta Theriologica 49:63-74. Cary, G., D. Lindenmayer, and S. Dovers. 2003. Australian vegetation is fire- adapted CSIRO Publishing, Collingwood, Victoria, Australia. Catling, P. C. 1991. Ecological effects of prescribed burning practices on the mammals of south-eastern Australia. Pages 353–363 in D. Lunney, editor. Conservation of Australia’s Forest Fauna. Royal Zoological Society of New South Wales, . Catling, P. C., and R. J. Burt. 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 22:271-288. Catling, P. C., R. J. Burt, and R. I. Forrester. 2000. Models of the distribution and abundance of ground-dwelling mammals in the eucalypt forests of north- eastern New South Wales in relation to habitat variables. Wildlife Research 27:639-654. Catling, P. C., N. C. Coops, and R. J. Burt. 2001. The distribution and abundance of ground-dwelling mammals in relation to time since wildfire and vegetation structure in south-eastern Australia. Wildlife Research 28:555- 564. Caughley, G. J. 1964. Density and dispersion of two species of kangaroo in relation to habitat. Australian Journal of Zoology 12:238-249. Caughley, G. J., J. Short, G. C. Grigg, and H. Nix. 1987. Kangaroos and climate: an analysis of distribution. Journal of Animal Ecology 56:751-761. Chamaille-Jammes, S., M. Valeix, and H. Fritz. 2007. Managing heterogeneity in elephant distribution: interactions between elephant population density and surface-water availability. Journal of Applied Ecology 44:625-633.

References 229

Chapman, N. G. 2004. Faecal pellets of Reeves' muntjac, Muntiacus reevesi: defecation rate, decomposition period, size and weight. European Journal of Wildlife Research 50:141-145. Chase, J. M., and M. A. Leibold. 2003. Ecological niches: linking classical and contemporary approaches. University of Chicago Press, Chicago. Cheney, N. P. 1981. Fire behaviour. Pages 151-156 in A. M. Gill, R. H. Groves, and I. R. Noble, editors. Fire and the Australian Biota. Australian Academy of Science, Canberra. Chesterfield, E., A. Trumbull-Ward, P. Hopmans, and J. Whelan. 1995. Early Changes in Vegetation from a Grazing Trial on Yanakie Isthmus, Wilsons Promontory National Park. Flora and Fauna Technical Report 139, Department of Conservation and Natural Resources, Melbourne. Child, G., and W. Von Richter. 1969. Observations on the ecology and behaviour of lechwe, puku and waterbuck along the Chobe River, Botswana. Zeitschrift fur Saugetierkunde 34:275-295. Christensen, P., H. Recher, and J. Hoare. 1981. Responses of open forest to fire regimes. Pages 367-394 in A. M. Gill, R. H. Groves, and I. R. Noble, editors. Fire and the Australian Biota. Australian Academy of Science, Canberra. Claridge, A. W., Trappe, J. M. and Claridge, D. L. 2001. Mycophagy by the swamp wallaby. Wildlife Research 28:643-645. Clarke, J. L., M. E. Jones, and P. J. Jarman. 1989. A day in the life of a kangaroo: activities and movements of Eastern Grey kangaroos Macropus giganticus at Wallaby Creek. Pages 611-618 in Kangaroos, Wallabies and Rat- kangaroos. Surrey Beatty and Sons Pty Limited, Sydney. Clarke, K. R. 1993. Non-parametric multivariate analyses of changes in community structure. Australian Journal of Ecology 18:117-143. Clarke, P. J. 2002. Experiments on tree and shrub establishment in temperate grassy woodlands: Seedling survival. Austral Ecology 27:606-615. Cohn, J. S., and R. A. Bradstock. 2000. Factors affecting post-fire seedling establishment of selected mallee understorey species. Australian Journal of Botany 48:59-70. Connell, J. 1980. Diversity and the coevolution of competitors, or the ghost of competition past. Oikos 35:131-138.

References 230

Connell, J. 1983. On the prevalence and relative importance of interspecific competition: evidence from field experiments. The American Naturalist 122:661-696. Cooke, B. D. 1974. Food and other resources of the wild rabbit Oryctolagus cuniculus (L.) PhD thesis. University of Adelaide. Cooke, B. D. 1998. Did introduced rabbits, Oryctolagus cuniculus (L.) displace Common Wombats, Vombatus ursinus (Shaw) from part of their range of South Australia? Pages 262-270 in R. T. Wells and P. A. Pridmore, editors. Wombats in Australia. Surrey Beatty and Sons, Sydney. Costello, D. A., I. D. Lunt, and J. E. Williams. 2000. Effects of invasion by the indigenous shrub Acacia sophorae on plant composition of coastal grasslands in south-eastern Australia. Biological Conservation 96:113-121. Coulson, G. 1990. Habitat separation in the grey kangaroos, Macropus giganteus (Shaw) and M. fuliginosus (Desmarest) (Marsupialia: Macropodidae), in Grampians National Park, Western Victoria. Aust. Mammal. 13:33–40. Coulson, G. 1993. Use of heterogeneous habitat by the western grey kangaroo, Macropus fuliginosus. Wildlife Research 20:137-149. Coulson, G. 2008. Eastern Grey Kangaroo Macropus giganteus. Pages 335-338 in S. Van Dyck and R. Strahan, editors. The Mammals of Australia. Reed New Holland, Sydney. Cowan, D. P. 1987. Aspects of the Social Organisation of the European Wild Rabbit (Oryctolagus cuniculus). Ethology 75:197 - 210. Croft, D. B. 1989. Social organisation of the Macropodoidea. Pages 505-525 in Kangaroos, Wallabies and Rat-kangaroos. Surrey Beatty and Sons Pty Limited, Chipping Norton, Sydney. Damuth, J. 1981. Population density and body size in mammals. Nature (Lond.) 290:699-700. Damuth, J. 1987. Interspecific allometry in population density in mammals and other animals: the independence of body mass and energy-use. Biological Journal of the Linnean Society 31:193-246. Davies, J. B., and A. M. Oates. 1999. Ecological Vegetation Class (EVC) mapping of Wilsons Promontory and adjacent islands. Department of Natural Resources and Environment, Gippsland Region, Melbourne.

References 231

Davis, N. E., G. Coulson, and D. M. Forsyth. 2008. Diets of native and introduced mammalian herbivores in shrub-encroached grassy woodland, south- eastern Australia. Wildlife Research 35:684-694. Davis, N. E., G. Coulson, and D. M. Forsyth. in review. Habitat-specific faecal pellet decay rates for five mammalian herbivores in south-eastern Australia. Australian Mammalogy. Davison, A. C., and D. Hinkley. 2006. Bootstrap Methods and their Applications. 8th edition. Cambridge Series in Statistical and Probabilistic Mathematics, Cambridge. Dawson, T. J. 1989. Diets of macropodoid marsupials: general patterns and environmental influences. Pages 129-142 in Kangaroos, Wallabies and Rat-kangaroos. Surrey Beatty and Sons Pty Limited, Sydney. Dawson, T. J., and B. A. Ellis. 1979. Comparison of the diets of yellow-footed rock wallabies Petrogale xanthopus and sympatric herbivores in western New South Wales, Australia. Australian Wildlife Research 6:245-254. Dawson, T. J., and B. A. Ellis. 1996. Diets of mammalian herbivores in Australian arid shrublands: seasonal effects on overlap between euros, sheep and feral goats, and on dietary niche breadths and electivities. Journal of Arid Environments 34:491-506. Dawson, T. J., and A. J. Hulbert. 1970. Standard metabolism, body temperature, and surface areas of Australian marsupials. American Journal of Physiology 218:1233-1238. De Boer, W. F., and H. H. T. Prins. 1990. Large herbivores that strive mightily but eat and drink as friends. Oecologia 82:264–274. de Munk, F. D. 1999. Resource Use by the Eastern Grey Kangaroo and the Black Wallaby. PhD thesis. Deakin University, Melbourne. Demment, M. W., and P. J. Van Soest. 1985. A nutritional explanation for body- size patterns of ruminant and non-ruminant herbivores. American Naturalist 152:641-672. Denny, M. J. S., and T. J. Dawson. 1975. Comparative metabolism of tritiated water by macropodid marsupials. American Journal of Physiology 228:1794-1795.

References 232

Department of Natural Resources and Environment. 1998. Victoria's biodiversity: directions in management. Department of Natural Resources and Environment, Melbourne. Dhungel, S. K. 1985a. Defecation rate of hog deer in captivity. Journal of Wildlife Management 49:925-926. Dhungel, S. K. 1985b. Ecology of the hog deer in Royal Chitwan National Park, Nepal. PhD thesis. University of Montana, Montana. Dhungel, S. K., and B. W. O'Gara. 1991. Ecology of the hog deer in Royal Chitwan National Park, Nepal. Wildlife Monographs 119:1-40. Di Stefano, J. 2005. Mammalian browsing damage in the Mt. Cole State forest, southeastern Australia: analysis of browsing patterns, spatial relationships and browse selection. New Forests 29:43-61. Di Stefano, J., J. A. Anson, A. York, A. Greenfield, G. Coulson, A. Berman, and M. Bladen. 2007. Interactions between timber harvesting and swamp wallabies (Wallabia bicolor): Space use, density and browsing impact. Forest Ecology and Management 253:128–137. Di Stefano, J., and G. R. Newell. 2008. Diet selection by the swamp wallaby (Wallabia bicolor): feeding strategies under conditions of changed food availability. Journal of Mammalogy 89:1540-1549. Di Stefano, J., A. York, M. Swan, A. Greenfield, and G. Coulson. 2009. Habitat selection by the swamp wallaby (Wallabia bicolor) in relation to diel period, food and shelter. Austral Ecology 34:143-155. Diamond, J. M. 1975. Assembly of species communities. Pages 342–444 in M. L. Cody and J. M. Diamond, editors. Ecology and evolution of communities. Harvard University Press, Cambridge, MA, USA. Dierenfeld, E. S. 1984. Diet quality of sympatric wombats, kangaroos and rabbits during severe drought. PhD thesis. Cornell University, Ithaca, NY. Dinerstein, E. 1979. An ecological survey of the Royal Karnali-Bardia Wildlife Reserve, Nepal. Part II: Habitat/animal interactions. Biological Conservation 16:265-300. Drewa, P. B., and K. M. Havstad. 2001. Effects of fire, grazing, and the presence of shrubs on Chihuahuan desert grasslands. Journal of Arid Environments 48:429-443.

References 233

Dudzinski, M. L., W. A. Low, W. J. Müller, and B. S. Low. 1982. Joint use by red kangaroos and shorthorn cattle in arid central Australia. Australian Journal of Ecology 7:69-74. Dunbar, R. I. M. 1978. Competition and niche separation in a high altitude herbivore community in Ethiopia. East African Wildlife Journal 16:183- 199. Duncan, A. 1992. The winter diets of fallow deer (Dama dama) and forester kangaroos (Macropus giganteus tasmaniensis) in the midlands of Tasmania. The Tasmanian Naturalist 110:1-6. Duncan, A., H. Tixier, R. R. Hoffman, and M. Lechner-Doll. 1998. Feeding strategies and the physiology of digestion in roe deer. Pages 91-116 in R. Andersen, P. Duncan, and D. C. Linnell, editors. The European Roe Deer: The Biology of Success. Scandinavian University Press, Oslo. Duncan, P. 1983. Determinants of the use of habitat by horses in a Mediterranean wetland. Journal of Animal Ecology 52:93-109. Dunham, K. M. 1981. The foraging behaviour of impala Aepyceros melampus. South African Journal of Wildlife Research 12:36-40. duToit, J. T., and N. Owen-Smith. 1989. Body size, population British metabolism, and habitat specialization among large African herbivores. American Naturalist 133:736-740. Ealey, E. H. M. 1967. Ecology of the euro, Macropus robustus (Gould) in north- : The environment and changes in euro and sheep populations. Wildlife Research 12:9-25. Edwards, G. 1969. Some aspects of the ecology of the Swamp Wallaby, Wallabia bicolor. Honours thesis. Monash University, Melbourne. Edwards, G. P., T. J. Dawson, and D. B. Croft. 1995. The dietary overlap between red kangaroos (Macropus rufus) and sheep (Ovis aries) in the arid rangelands of Australia. Australian Journal of Ecology 20:324-334. Edwards, G. P., and E. H. M. Ealey. 1975. Aspects of the ecology of the swamp wallaby Wallabia bicolor (Marsupialia: Macropodidae). Australian Mammalogy 1:307-317. Ellis, W. A. H., B. J. Sullivan, A. T. Lisle, and F. N. Carrick. 1998. The spatial and temporal distribution of faecal pellets. Wildlife Research 25:663-668.

References 234

Emison, W. B., J. W. Porter, K. C. Norris, and G. J. Apps. 1978. Survey of the vertebrate fauna in the Grampians-Edenhope area of southwestern Victoria. Memoirs of the National Museum of Victoria 39:281–363. Evans, M. C., and P. J. Jarman. 1999. Diets and feeding selectivities of bridled nailtail wallabies and black-striped wallabies. Wildlife Research 26:1-19. Evans, M. C., C. Macgregor, and P. J. Jarman. 2006. Diet and feeding selectivity of common wombats. Wildlife Research 33:321–330. Fa, J. E., C. M. Sharples, and D. J. Bell. 1999. Habitat correlates of European rabbit (Oryctolagus cuniculus) distribution after the spread of RVHD in Cadiz Province, Spain. Journal of the Zoological Society of 249:83-96. Feinsinger, P., E. E. Spears, and R. W. Poole. 1981. A simple measure of niche breadth. Ecology 62:27-32. Fernández, N. 2005. Spatial patterns in European rabbit abundance after a population collapse. Landscape Ecology 20:897-910. Ferreira, C., and P. C. Alves. 2009. Influence of habitat management on the abundance and diet of wild rabbit (Oryctolagus cuniculus algirus) populations in Mediterranean ecosystems. European Journal of Wildlife Research 55:487–496. Floyd, R. B. 1980. Density of Wallabia bicolor (Desmarest) (Marsupialia: Macropodidae) in eucalypt plantations of different ages. Australian Wildlife Research 7:333-337. Forbes, D. K., and D. E. Tribe. 1970. The utilization of roughages by sheep and kangaroos. Australian Journal of Zoology 18:247-256. Forsyth, D. M. 2000. Habitat selection and coexistence of the Alpine chamois (Rupicapra rupicapra) and Himalayan tahr (Hemitragus jemlahicus) in the eastern Southern Alps, New Zealand. Journal of the Zoological Society of London 252:215-225. Forsyth, D. M. 2005. Protocol for estimating changes in the relative abundance of deer in New Zealand forests using the Faecal Pellet Index (FPI). Report prepared for Landcare Research, Wellington, by the Arthur Rylah Institute for Environmental Research, Melbourne.

References 235

Forsyth, D. M., R. J. Barker, G. Morriss, and M. P. Scroggie. 2007. Modeling the relationship between fecal pellet indices and deer density. Journal of Wildlife Management 71:964-970. Forsyth, D. M., R. P. Duncan, M. Bomford, and G. Moore. 2004. Climatic suitability, life-history traits, introduction effort, and the establishment and spread of introduced mammals in Australia. Conservation Biology 18:557- 569. Fox, B. J. 1989. Community ecology of macropodoids. Pages 89-104 in Kangaroos, Wallabies and Rat-Kangaroos. Surrey Beatty and Sons Pty Limited, Chipping Norton, Sydney. Freudenberger, D. O., I. R. Wallis, and I. D. Hume. 1988. Digestive adaptations in kangaroos, wallabies and rat-kangaroos. Pages 179-187 in Kangaroos, Wallabies and Rat-kangaroos. Surrey Beatty and Sons Pty Limited, Sydney. Frith, H. J. 1979. Wildlife Conservation. Angus and Robertson, Sydney. Fritz, H., M. de Garine-Wichatitsky, and G. Letessier. 1996. Habitat use by sympatric wild and domestic herbivores in an African savanna woodland: The influence of cattle spatial behaviour. Journal of Applied Ecology 33:589-598. Futuyma, D., and G. Moreno. 1988. The evolution of ecological specialisation. Annual Review of Ecology and Systematics 19:207-233. Garshelis, D. L. 2000. Delusions in habitat evaluation: Measuring use, selection, and importance. Pages 111-164 in L. Boitani and T. K. Fuller, editors. Research Techniques in Animal Ecology; Controversies and Consequences. Columbia University Press, New York. Gause, G. F. 1934. The struggle for existence. Williams and Wilkins, Baltimore. Gent, M. L., and J. W. Morgan. 2007. Changes in the stand structure (1975-2000) of coastal Banksia forest in the long absence of fire. Austral Ecology 32:239-244. Gill, A. M., and R. A. Bradstock. 1995. Extinction of biota by fire. Pages 309-322 in R. A. Bradstock, T. D. Auld, D. A. Keith, R. T. Kingsford, D. Lunney, and D. P. Silversen, editors. Conserving Biodiversity: Threats and Solutions. Surrey Beaty and Sons, Sydney.

References 236

Gordon, I. J., and A. W. Illius. 1988. Incisor arcade structure and diet selection in ruminants. Functional Ecology 2:15-22. Gordon, I. J., and A. W. Illius. 1989. Resource partitioning by ungulates on the Isle of Rhum Oecologia 79:383-389. Gowland, P. N. 1973. Aspects of the digestive physiology of the common wombat Vombatus ursinus (Shaw 1800). Australian National University, Canberra. Green, M. J. B. 1987. Ecological separation in Himalayan ungulates. Journal of Zoology 1:693-719. Green, R. H. 1979. Sampling Design and Statistical Methods for Environmental Biologists. Wiley, Chichester. Gwynne, M. D., and R. H. Bell. 1968. Selection of vegetation components by grazing ungulates in the Serengeti National Park. Nature, London 220:390- 393. Hairston, N. G., F. E. Smith, and L. B. Slobodkin. 1960. Community structures, population control, and competition. American Naturalist 94:421-425. Hanley, T., . A., and K. A. Hanley. 1982. Food resource partitioning by sympatric ungulates on Great Basin rangeland. Journal of Range Management 35:152-158. Hanley, T. A. 1984. Habitat patches and their selection by wapiti and black-tailed deer in coastal montane coniferous forest. Journal of Applied Ecology 21:423-436. Hannan, M. J., and J. Whelan. 1989. Deer and habitat relations in managed forests. Pages 116-127 in R. J. Putman, editor. Mammals as pests. Chapman and Hall, New York. Hansen, R. M., W. M. Mugambi, and A. M. Bauni. 1985. Diets and trophic ranking of ungulates in the Northern Serengeti. Journal of Wildlife Management 49:823-829. Hardin, G. 1960. The competitive exclusion principle. Science 131:1292-1296. Harestad, A. S., and F. L. Bunnell. 1987. Persistence of Black-tailed Deer fecal pellets in coastal habitats. Journal of Wildlife Management 51:33-37. Hazard, J., and R. F. Parsons. 1977. Size-class analysis of coastal scrub and woodland, , southern Australia. Australian Journal of Botany 2:187-197.

References 237

Hemami, M. R., and P. M. Dolman. 2005. The disappearance of muntjac (Muntiacus reevesi) and roe deer (Capreolus capreolus) pellet groups in a pine forest of lowland England. European Journal of Wildlife Research 51:19-24. Hickling, G. J. 1986. Red deer population surveys in the Harper-Avoca catchment (1956-1983). Forest Research Institute Bulletin 107:1-10. Hilborn, R., and M. Mangel. 1997. The Ecological Detective: Confronting Models with Data. Princeton University Press, Princeton, New Jersey. Hill, G. J. E. 1978. Preliminary assessment of defecation patterns for the eastern grey kangaroo (Macropus giganteus). Australian Zoologist 19:291-300. Hill, G. J. E. 1981a. Distribution of Grey Kangaroos in south, inland Queensland. Australian Rangeland Journal 3:58-66. Hill, G. J. E. 1981b. A study of grey kangaroo density using pellet counts. Australian Wildlife Research 8:237-243. Hill, G. J. E. 1981c. A study of the habitat preferences in the Grey Kangaroo. Australian Wildlife Research 8:245-254. Hill, G. J. E. 1982. Seasonal movement patterns of the Eastern Grey Kangaroo in southern Queensland. Australian Wildlife Research 9:373-387. Hobbs, N. T., D. L. Baker, G. D. Bear, and D. C. Bowden. 1996. Ungulate grazing in sagebrush grassland: mechanisms of resource competition. Ecological Applications 6:200-217. Hobbs, N. T., and D. M. Swift. 1985. Estimates of habitat carrying capacity incorporating explicit nutritional constraints. Journal of Wildlife Management 49:814–822. Hodgkinson, K. C., and G. N. Harrington. 1985. The case for prescribed burning to control shrubs in eastern semi-arid woodlands. Australian Rangelands Journal 7:64-74. Hoffman, W. A. 1999. Fire and population dynamics of woody plants in a Neotropical savanna: Matrix model projections. Ecology 80:1354-1369. Hofmann, R. R., and D. R. M. Stewart. 1972. Grazer or browser: A classification based on the stomach-structure and feeding habits of East African ruminants. Mammalia 36:226-240.

References 238

Hollis, C. J., J. D. Robertshaw, and R. H. Harden. 1986. Ecology of the swamp wallaby (Wallabia bicolor) in north-eastern New South Wales. I. Diet. Australian Wildlife Research 13:335-365. Hone, J., and W. Martin. 1998. A study of dung decay and plot size for surveying feral pigs using dung counts. Wildlife Research 25:255-260. Hood, G. M. 2005. PopTools. CSIRO. Hopcraft, J. G. C., H. Olff, and A. R. E. Sinclair. 2010. Herbivores, resources and risks: alternating regulation along primary environmental gradients in savannas. Trends in Ecology and Evolution 25:119-128. Horn, H. S. 1966. Measurement of "overlap" in comparative ecological studies. American Naturalist 100:419-424. Hume, I., D. 1982. Digestive Physiology and Nutrition of Marsupials. Cambridge University Press, Cambridge. Hume, I. D. 1999. Marsupial Nutrition. Cambridge University Press, Cambridge. Hume, I. D., and C. E. Stevens. 1995. Comparative physiology of the vertebrate digestive system. Second edition. Cambridge University Press, Cambridge. Hurlbert, S. H. 1978. The measurement of niche overlap and some relatives. Ecology 59:67-77. Hutchinson, G. 1959. Homage to Santa Rosalia or why are there so many kinds of animals? American Naturalist 93:145-159. Hutchinson, G. E. 1957. Concluding remarks. Cold Springs Harbour symposia on quantitative biology 22:415-427. Illius, A. W., and I. J. Gordon. 1987. The allometry of food intake in grazing ruminants. Journal of Animal Ecology 56:969-1000. Illius, A. W., and I. J. Gordon. 1993. Diet selection in mammalian herbivores: constraints and tactics. Pages 157-181 in R. H. Hughes, editor. Diet Selection: An Interdisciplinary Approach to Foraging Behaviour. Blackwell Scientific Publications, Oxford. Jarman, P., and G. Coulson. 1989. Dynamics and adaptiveness of grouping in macropods. Pages 527-547 in G. Grigg, P. Jarman, and I. Hume, editors. Kangaroos, Wallabies and Rat-Kangaroos. Chipping Norton: Surrey Beatty. Jarman, P. J. 1972. Seasonal distribution of the large mammal populations in the unflooded middle Zambezi valley. Journal of Applied Ecology 9:283-299.

References 239

Jarman, P. J. 1974. The social organisation of antelope in relation to their ecology. Behaviour 48:215-267. Jarman, P. J., and M. V. Jarman. 1974. Impala behaviour and its relevance to management. Pages 871-881 in V. Geist and F. Walther, editors. The Behaviour of Ungulates and its Relation to Management. IUCN, Morges, Switzerland. Jarman, P. J., and C. M. Phillips. 1989. Diets in a community of macropod species. Pages 139-149 in Kangaroos, Wallabies and Rat-kangaroos. Surrey Beatty and Sons Pty Limited, Sydney. Jarman, P. J., and A. R. E. Sinclair. 1979. Feeding strategy and the pattern of resource partitioning in ungulates. Pages 130-163 in A. R. E. Sinclair and N. Norton-Griffiths, editors. Serengeti: Dynamics of an Ecosystem. University of Chicago Press, Chicago, IL. Johnson, C. N. 1998. The evolutionary ecology of wombats.in R. T. Wells and P. A. Pridmore, editors. Wombats. Surrey Beatty, Sydney. Johnson, C. N., and P. J. Jarman. 1987. Macropod studies at Wallaby Creek VI. A validation of the use of dung-pellet counts for measuring absolute densities of populations of macropodids. Australian Wildlife Research 14:139-145. Johnson, C. N., P. J. Jarman, and C. J. Southwell. 1987. Macropod studies at Wallaby Creek V. Patterns of defaecation by eastern grey kangaroos and red-necked wallabies. Australian Wildlife Research 14:133-138. Johnson, D. H. 1980. The comparison of usage and availability measurements for evaluating resource preference. Ecology 61:65-71. Jones, M., Y. Mandelik, and T. Dayan. 2001. Coexistence of temporally partitioned spiny mice: roles of habitat structure and foraging behaviour. Ecology 82:2164-2176. Judd, T. S. 1990. The ecology and water relations of the invasive shrubs, Kunzea ambigua (sm.) druce, Kunzea ericoides (A. Rich) J. Thompson and Leptospermum laevigatum (J. Gaertn) F. Muell. PhD thesis. The University of Melbourne. Kelley, M., R. Stewart, T. Bowyer, J. G. Kie, N. J. Cimon, and B. K. Johnsone. 2002. Temporospatial distributions of elk, mule deer, and cattle: Resource

References 240

partitioning and competitive displacement. Journal of Mammalogy 83:229-244. Kirchhoff, M. D., and D. N. Larsen. 1998. Dietary overlap between native sitka black-tailed deer and introduced elk in southeast Alaska. The Journal of Wildlife Management 62:236-242. Kleiber, M. 1961. The Fire of Life. John Wiley and Sons, New York. Kotler, B., and J. Brown. 1988. Environmental heterogeneity and the coexistence of desert rodents. Annual Review of Ecology and Systematics 19:281-307. Kotler, B., and J. Brown. 2007. Community Ecology. Pages 397-438 in D. Stephens, J. Brown, and R. Ydenberg, editors. Foraging: Behavior and Ecology. University of Chicago Press, Chicago. Kotler, B. P., J. S. Brown, and A. Subach. 1993. Mechanisms of species coexistence of optimal foragers: temporal partitioning of two species of sand dune gerbils. Oikos 67:548-556. Krebs, C. J. 1998. Niche measures and resource preferences. Pages 455-495 in E. Fogarty, V. McDougal, and N. Murray, editors. Ecological Methodology. Addison-Welsey Educational Publishers Inc., California. Kuehl, H. S., A. Todd, C. Boesch, and P. D. Walsh. 2007. Manipulating decay time for efficient large-mammal density estimation: gorillas and dung height. Ecological Applications 17:2403-2414. Kuiters, A. T., G. W. T. A. G. Bruinderink, and D. R. Lammertsma. 2005. Facilitative and competitive interactions between sympatric cattle, red deer and wild boar in Dutch woodland pastures. Acta Theriologica 50:241-252. Laing, S. E., S. T. Buckland, R. W. Burn, D. Lambie, and A. Amphlett. 2003. Dung and nest surveys: estimating decay rates. Journal of Applied Ecology 40:1102-1111. Lamprey, H. F. 1963. Ecological separation of the large species in the Tarangire Game Reserve, Tanganyika. East African Wildlife Journal 1:63-92. Lawlor, R., and J. Maynard Smith. 1976. The coevolution and stability of competing species. Amer. Naturalist 110:7999. le Mar, K., and C. McArthur. 2005. Comparison of habitat selection by two sympatric macropods, Thyogale billardierii and Macropus rufogriseus rufogriseus, in a patchy eucalypt-forestry environment. Austral Ecology 30:674-683.

References 241

Lehmkuhl, J. F., C. A. Hansen, and K. Sloan. 1994. Elk pellet-group decomposition and detectability in coastal forests of Washington. Journal of Wildlife Management 58:664-669. Leigh, J. 1989. Effects of rabbit and kangaroo grazing on two semi-arid grassland communities in central-western New South Wales. Australian Journal of Botany 37:375-396. Leigh, J. H., and M. D. Holgate. 1979. The response of the understorey of forests and woodlands of the Southern Tablelands to grazing and burning. Australian Journal of Ecology 4:25-45. Leigh, J. H., D. J. Wimbush, D. H. Wood, M. D. Holgate, A. V. Slee, M. G. Stanger, and R. I. Forrester. 1987. Effects of rabbit grazing and fire on a subalpine environment. I. Herbaceous and shrubby vegetation. Australian Journal of Botany 35:433-464. Leigh, J. H., D. H. Wood, A. V. Slee, and M. D. Holgate. 1991. The effects of burning and simulated grazing on productivity, forage quality, mortality and flowering of eight subalpine herbs in Kosciusko National Park. Australian Journal of Botany 39:97-118. Levins, R. 1968. Evolution in Changing Environments: Some Theoretical Explorations. Princeton University Press, Princeton, New Jersey. Liedloff, A. C., M. B. Coughenour, J. A. Ludwig, and R. Dyer. 2001. Modelling the trade-off between fire and grazing in a tropical savanna landscape, northern Australia. Environment International 27:173-180. Linklater, W. L., E. Z. Cameron, K. J. Stafford, and C. J. Veltman. 2000. Social and spatial structure and range use by Kaimanawa wild horses (Equus caballus: Equidae). New Zealand Journal of Ecology 24:139-152. Lopez-Darias, M., and J. M. Lobo. 2009. Micro-scale distribution of rabbits on Fuerteventura Island Biodiversity and Conservation 8:3687-3704. Lunney, D., and M. O'Connell. 1988. Habitat selection by the swamp wallaby Wallabia bicolor the red-necked wallaby Macropus rufogriseus and the common wombat Vombatus ursinus in logged burnt forest near Bega New South Wales Australia. Australian Wildlife Research 15:695-706. Lunney, D., B. Triggs, P. Eby, and E. Ashby. 1990. Analysis of scats of dogs Canis familiaris and foxes Vulpes vulpes (Canidae: Carnivora) in coastal

References 242

forests near Bega, New South Wales. Australian Wildlife Research 17:61– 68. MacArthur, R., and R. Levins. 1964. Competition, habitat selection, and character displacement in a patchy environment. Proceedings of the National Academy of Science USA 51:1207–1210. MacArthur, R. H. 1972. Geographical ecology: patterns in the distribution of species. Harper and Row, New York. MacArthur, R. H., and R. Levins. 1967. The limiting similarity, convergence, and divergence of coexisting species. American Naturalist 101:377-385. MacArthur, R. H., and E. O. Wilson. 1967. The Theory of Island Biogeography. Princeton University Press, Princeton, N. J., USA. MacNally, R. C. 1995. Ecological versatility and community ecology. Cambridge University Press. Madhusudan, M. D. 2004. Recovery of wild large herbivores following livestock decline in a tropical Indian wildlife reserve Journal of Applied Ecology 41:858-869. Mallett, K. J., and B. D. Cooke. 1986. The Ecology of the Common Wombat in South Australia. Nature Conservation Society of South Australia, Adelaide. Manly, B. F. J., L. L. McDonald, and D. L. Thomas. 2002. Resource selection by animals: statistical design and analysis for field studies. Second edition. Chapman & Hall, London. Manor, R., and D. Saltz. 2008. Conservation implications of competition between generalist and specialist rodents in Mediterranean afforested landscape. Biodiversity Conservation 17:2513-2523. Martin, G. R., L. E. Twigg, and L. Zampichelli. 2007. Seasonal changes in the diet of the European rabbit (Oryctolagus cuniculus) from three different Mediterranean habitats in south-western Australia. Wildlife Research 34:25-42. Martin, W., and W. Sobey. 1983. Improvement of seabird nesting habitat on Bowen Island, New South Wales by eradication of rabbits. Corella 7:40. Massei, G., P. Bacon, and P. V. Genov. 1998. Fallow deer and wild boar pellet group disappearance in a Mediterranean area. Journal of Wildlife Management 62:1086-1094.

References 243

May, R. M. 1973. Stability and complexity in model ecosystems. Princeton University Press, Princeton, New Jersey. May, R. M., and R. H. MacArthur. 1972. Niche overlap as a function of environmental variability. Proceedings of the National Academy of Science USA 69:1109-1113. May, S. 1997. The threat of feral predators to native forest fauna–implications for forest management. 3, Centre for Resource and Environmental Studies, Australian National University, Canberra, ACT. Mayle, B. A., A. J. Peace, and R. M. A. Gill. 1999. How Many Deer? A Field Guide to Estimating Deer Population Size. Forestry Commission, Edinburgh. Mayze, R. J., and G. I. Moore. 1990. The Hog Deer. The Australian Deer Research Foundation Ltd., Croydon, Victoria. McCullough, D., and Y. McCullough. 2000. Kangaroos in Outback Australia. Columbia University Press, New York. McDonald, R. A. 2002. Resource partitioning among British and Irish mustelids. Journal of Animal Ecology 71:185–200. McIlroy, J. C. 1973. Aspects of the ecology of the Common Wombat, Vombatus ursinus (Shaw, 1800). PhD thesis. Australian National University, Canberra. McIlroy, J. C. 1995. Common Wombat. Pages 204–205 in R. Strahan, editor. The Mammals of Australia. Reed Books, New South Wales. McIlroy, J. C. 2008. Common Wombat Vombatus ursinus. Pages 206-208 in S. Van Dyck and R. Strahan, editors. The Mammals of Australia. Reed New Holland, Sydney. McKenzie, D. C. 1976. The distribution and abundance of macropod species in rainforest, eucalypt forest and exotic pine plantation at Clouds Creek, New South Wales. Honours thesis. University of New England. McNab, B. K. 1988. Complications inherent in scaling the basal rate of metabolism in mammals. Quarterly Review of Biology 63:25-54. McNab, B. K. 2005. Uniformity in the basal metabolic rate of marsupials: its causes and consequences. Revista Chilena de Historia Natural 78:183-198.

References 244

McNaughton, S. J., and N. J. Georgiadis. 1986. Ecology of African grazing and browsing mammals. Annual Review of Ecology and Systematics 17:39- 65. Meers, T., and R. Adams. 2003. The impact of grazing by Eastern Grey Kangaroos (Macropus giganteus) on vegetation recovery after fire at Reef Hills Regional Park, Victoria. Ecological Management and Restoration 4:126-132. Menkhorst, P., and J. Seebeck. 1998. A list of native mammals of Wilsons Promontory National Park. The Victorian Naturalist 115:264-265. Menkhorst, P. W. 1995a. Hog deer. Pages 268-270 in P. W. Menkhorst, editor. Mammals of Victoria: Distribution, Ecology and Conservation. Department of Conservation and Natural Resources and Oxford University Press, Melbourne. Menkhorst, P. W. 1995b. Mammals of Victoria: distribution, ecology and conservation. Oxford University Press, Melbourne. Merchant, J. C. 2008. Swamp Wallaby, Wallabia bicolor. Pages 404-405 in S. Van Dyck and R. Strahan, editors. The Mammals of Australia. Reed New Holland, Sydney. Miller, R. 1975. Notes on the behaviour of hog deer in an enclosure. The Natural History Bulletin of the Siam Society 26:105-131. Mishra, C., S. van Wieren, P. Ketner, I. M. A. Heitkonig, and H. H. T. Prins. 2004. Competition between domestic livestock and wild bharal Pseudois nayaur in the Indian trans-Himalaya. Journal of Applied Ecology 41:344- 354. Mishra, C., S. E. van Wieren, I. M. A. Heitkonig, and H. H. T. Prins. 2002. A theoretical analysis of competitive exclusion in a trans-Himalayan large- herbivore assemblage. Animal Conservation 5:251-258. Mishra, H. R., and C. Wemmer. 1987. The comparative breeding ecology of four Cervids in Royal Chitwan National Park.in C. Wemmer, editor. Biology and Management of the Cervidae. Smithsonian Institute Press, Washington. Moore, B., G. Coulson, and S. Way. 2002. Habitat selection by adult female eastern grey kangaroos. Wildlife Research 29:439-445.

References 245

Moreno, S., R. Villafuerte, and M. Delibes. 1996. Cover is safe during the day but dangerous at night: the use of vegetation by European wild rabbits (Oryctolagus cuniculus L.) in Portugal. Canadian Journal of Zoology 74:1656-1660. Morgan, J. W. 1999. Defining grassland fire events and the response of perennial plants to annual fire in temperate grasslands of south-eastern Australia. Plant Ecology 144:127-144. Moriarty, A. 2004. The ecology and environmental impact of rusa deer (Cervus timorensis) in the Royal National Park. Ph D. University of Western Sydney, Sydney, Australia. Moseby, K. E., S. De Jong, N. Munro, and A. Pieck. 2005. Home range, activity and habitat use of European rabbits (Oryctolagus cuniculus) in arid Australia: implications for control. Wildlife Research 32:305–311. Mower, K. J., T. W. Townsend, and W. J. Tyznik. 1997. Sample sizes to measure young apple trees browsed by deer in Ohio. Wildlife Society Bulletin 25:344-347. Murden, S. B., and K. L. Risenhoover. 1993. Effects of habitat enrichment on patterns of diet selection. Ecological Application 3:497-505. Murray, D., E. Ellsworth, and A. Zack. 2005. Assessment of potential bias with snowshoe hare fecal-pellet plot counts. Journal of Wildlife Management 69:385-395. Myers, K. 1983. The Australian Museum Complete Book of Australian Mammals. Angus and Robertson Publishers, London, Sydney, Melbourne. Namgail, T., C. Mishra, C. B. de Jong, S. E. van Wieren, and H. H. T. Prins. 2009. Effects of herbivore species richness on the niche dynamics and distribution of blue sheep in the Trans-Himalaya. Diversity and Distributions 15:940–947. Natural Resources and Environment 2000. Wilsons Promontory National Park Outdoor Leisure Map. Vicmap published. Newsome, A. E., J. C. McIlroy, and P. C. Catling. 1975. The effects of an extensive wildfire on populations of twenty ground vertebrates in south- east Australia. Proceedings of the Ecological Society of Australia 9:107- 123.

References 246

Norbury, G. L. 1988. Microscopic analysis of herbivore diets - a problem and a solution. Australian Wildlife Research 15:51-57. Norbury, G. L., and G. D. Sanson. 1992. Problems with measuring diet selection of terrestrial mammalian herbivores. Australian Journal of Ecology 17:1-7. Norbury, G. L., G. D. Sanson, and A. K. Lee. 1989. Feeding ecology of the Macropodoidea. Pages 169-178 in Kangaroos, Wallabies and Rat- kangaroos. Surrey Beatty and Sons Pty Limited, Sydney. Noy-Meir, I. 1995. Interactive effects of fire and grazing on structure and diversity of Mediterranean grasslands. Journal of Vegetation Science 6:701-710. NPWS. 2001. Pest animal management programs 2001. National Parks and Wildlife Service, New South Wales. Odden, M., and P. Wegge. 2007. Predicting spacing behaviour and mating systems of solitary cervids: A study of hog deer and Indian muntjac. Zoology 110:261-270. Odadi, W. O., T. P. Young, and J. B. Okeyo-Owuor. 2007. Effects of wildlife on cattle diets in Laikipia rangeland, Kenya. Rangeland Ecology and Management 60:179-185. Odden, M., P. Wegge, and T. Storaas. 2005. Hog deer Axis porcinus need threatened tallgrass floodplains: a study of habitat selection in lowland Nepal. Animal Conservation 8:99-104. Osawa, R. 1990. Feeding strategies of the swamp wallaby, Wallabia bicolor, on North Stradbroke Island, Queensland. I: Composition of diets. Australian Wildlife Research 17:615-621. Osawa, R., and P. F. Woodall. 1990. Feeding strategies of the swamp wallaby, Wallabia bicolor, on North Stradbroke Island Queensland Australia II. Effects of seasonal changes in diet quality on intestinal morphology. Australian Wildlife Research 17:623-632. Ovadia, O., and Z. Abramsky. 1995. Density-dependent habitat selection: evaluation of the isodar method. Oikos 73:86-94 Ovington, D. 1978. Australian Endangered Species: Mammals, Birds and Reptiles. Cassell Australia Limited, Melbourne. Paddle, R. 2000. The Last Tasmanian Tiger: The History and Extinction of the Thylacine. Cambridge University Press, Cambridge.

References 247

Palomares, F., and M. Delibes. 1997. Predation upon European rabbits and their use of open and closed patches in Mediterranean habitats. Oikos 80:407- 410. Park, T. 1962. Beetles, competition, and populations. Science 138:1369-1375. Parks Victoria. 2003a. Environmental Risks Report. Parks Victoria, Melbourne. Parks Victoria. 2003b. Environmental Values Report. Parks Victoria, Melbourne. Parks Victoria 2011. Wilsons Promontory National Park Timeline 1700’s to present. Parks Victoria Resource Kit. . Accessed 31 January 2011. Parrish, J. D. 1995. Effects of needle architecture on warbler habitat selection in a coastal spruce forest. Ecology 76:1813-1820. Parsons, M. H., J. Koch, B. B. Lamont, S. Vlahos, and M. M. Fairbanks. 2006. Planting density effects and selective herbivory by kangaroos on species used in restoring forest communities. Forest Ecology and Management 229:39-49. Penry, D. L. 1993. Digestive constraints on diet selection. Pages 32-55 in R. N. Hughes, editor. Diet Selection: An Interdisciplinary Approach to Foraging Behaviour. Blackwell Scientific Publications, Oxford. Perry, R. J., and M. L. Braysher. 1986. A technique for estimating the numbers of eastern grey kangaroos, Macropus giganteus, grazing a given area of pasture. Australian Journal of Wildlife Research 13:335-338. Peters, R. H. 1983. The ecological implications of body size. Cambridge University Press, Cambridge. Peters, R. H., and J. V. Raelson. 1984. Relations between individual size and mammalian population density. American Naturalist 124:498-517. Petren, K., and T. Case. 1996. An experimental demonstration of exploitation competition in an ongoing invasion. Ecology 77:118-132. Pianka, E. 1974. Niche overlap and diffuse competition. Proceedings of the National Academy of Science, USA 71:2141-2145. Pianka, E. R. 1976. Competition and niche theory. Pages 114–141 in R. M. May, editor. Theoretical Ecology: Principles and Applications. Blackwell Scientific Publications, Oxford.

References 248

Plumptre, A. J., and S. Harris. 1995. Estimating the biomass of large mammalian herbivores in a tropical montane forest: A method of faecal counting that avoids assuming a 'steady state' system. Journal of Applied Ecology 32:111-120. Poole, W. E. 1995. Eastern Grey Kangaroo Pages 335–338 in R. Strahan, editor. The mammals of Australia. Reed Books, New South Wales. Prigioni, C. 2008. Differential use of food and habitat by sympatric carnivores in the eastern Italian Alps. Italian Journal of Zoology 75:173-184. Prins, H. H. T. 2000. Competition between wildlife and livestock in Africa. Pages 51–80 in H. H. T. Prins, J. G. Grootenhuis, and T. T. Dolan, editors. Wildlife Conservation by Sustainable Use Kluwer Academic Publishers, Boston, MA. Prins, H. H. T., and H. Olff. 1998. Species richness of African grazer assemblages: towards a functional explanation. Pages 449-490 in D. M. Newberry, H. H. T. Prins, and N. D. Brown, editors. Dynamics of Tropical Communities. Blackwell Science, Oxford, UK. Prober, S. M., K. Thiele, and I. D. Lunt. 2007. Fire frequency regulates tussock grass composition, structure and resilience in endangered temperate woodlands. Austral Ecology 32:808–824. Prugh, L., and C. J. Krebs. 2004. Snowshoe hare pellet-decay rates and aging in different habitats. Wildlife Society Bulletin 32:386-393. Pyke, G. H., H. R. Pulliam, and E. L. Charnov. 1977. Optimal foraging: A selective review of theory and tests. Quarterly Review of Biology 52:137- 154. Quinn, G. P., and M. J. Keough. 2003. Experimental design and data analysis for biologists. Cambridge University Press, Cambridge. Ramp, D., and D. Ben-Ami. 2006. The effect of road-based fatalities on the viability of a peri-urban swamp wallaby population. Journal of Wildlife Management 70:1615–1624. Ramsey, D. S. L., and R. M. Engeman. 1994. Patterns of grazing on coastal dune systems by insular populations of two species of macropod. Wildlife Research 21:107-114. Ricklefs, R. E., and D. Schluter. 1993. Species diversity in ecological communities. University of Chicago Press, Chicago.

References 249

Riggs, R. A., A. R. Tiedemann, J. G. Cook, T. M. Ballard, P. J. Edgerton, M. Vavra, W. C. Krueger, F. C. Hall, L. D. Bryant, L. L. Irwin, and T. Delcurto. 2000. Modification of mixed-conifer forests by ruminant herbivores in the Blue Mountains Ecological Province. Riginos, C., and T. P. Young. 2007. Positive and negative effects of grass, cattle, and wild herbivores on Acacia saplings in an East African savanna. Oecologia 153:985-995. Rishworth, C., J. C. McIlroy, and M. T. Tanton. 1995. Diet of the common wombat, Vombatus ursinus, in plantations of Pinus radiata. Wildlife Research 22:333-339. Roberts, T. J. 1977. The Mammals of Pakistan. Ernest Benn Ltd, London. Robinette, W. L., R. B. Ferguson, and J. S. Gashwiler. 1958. Problems involved in the use of deer pellet group counts. Transactions of the North American Wildlife Conference 23:411-425. Roger, E., S. W. Laffan, and D. Ramp. 2007. Habitat selection by the common wombat (Vombatus ursinus) in disturbed environments: Implications for the conservation of a 'common' species. Biological Conservation 137:437- 449. Roger, E., and D. Ramp. 2009. Incorporating habitat use in models of fauna fatalities on roads. Diversity and Distributions 15:222-231. Roques, K. G., T. G. O'Connor, and A. R. Watkinson. 2001. Dynamics of shrub encroachment in an African savanna: relative influences of fire, herbivory, rainfall and density dependence. Journal of Applied Ecology 38:268-280. Rosenzweig, M. 1991. Habitat selection and population interactions: the search for mechanism. The American Naturalist 137:S5-S28. Rosenzweig, M. L. 1981. A theory of habitat selection. Ecology 62:327-335. Rouag, R., H. Djilali, H. Gueraiche, and L. Luiselli. 2007. Resource partitioning patterns between two sympatric lizard species from Algeria. Journal of Arid Environments 69:158-168. Rowntree, J. B. 1935. Habits of the Hog Deer (Hyelaphus porcinus Zimm.). Journal of Bombay Natural History Society 38:396. Runyoro, V. A., H. Hofer, E. B. Chausi, and P. D. Moehlman. 1995. Long-term trends in the herbivore populations of the Ngorongoro Crater, Tanzania. Pages 146–168 in A. R. E. Sinclair and P. Arcese, editors. Serengeti II:

References 250

Dynamics, Management, and Conservation of an Ecosystem. University of Chicago Press, Chicago, IL. Sakai, A. K., F. W. Allendorf, J. S. Holt, D. M. Lodge, J. Molofsky, K. A. With, S. Baughman, R. J. Cabin, J. E. Cohen, N. C. Ellstrand, D. E. McCauley, P. O’Neil, I. M. Parker, J. N. Thompson, and S. G. Weller. 2001. The population biology of invasive species. Annual Review of Ecology and Systematics 32:305-332. Sankaran, M., J. Ratnam, and N. Hanan. 2008. Woody cover in African savannas: the role of resources, fire and herbivory. Global Ecology and Biogeography 17:236-245. Sanson, G. D. 1978. The evolution and significance of mastication in the Macropodidae. Australian Mammalogy 2:23-28. Sanson, G. D. 1980. The morphology and occlusion of the molariform cheek teeth in some Macropodinae (Marsupialia: Macropodidae). Australian Journal of Zoology 28:341-365. Sanson, G. D. 1989. Morphological adaptations of teeth to diets and feeding in the Macropodidae. Pages 151-168 in G. Grigg, P. Jarman, and I. Hume, editors. Kangaroos, Wallabies and Rat-kangaroos. Surrey Beatty, Sydney. Savage, R. E. 1931. The relation between the feeding of the herring off the east coast of England and the plankton of the surrounding waters. Fishery Investigations, Ministry of Agriculture, Food and Fisheries, Series 2 12:1- 88. Sawyer, H., R. M. Nielson, F. Lindzey, and L. McDonald. 2006. Winter habitat selection of mule deer before and during development of a natural gas field. Journal of Wildlife Management 70:395–402. Scarlett, N. H., and R. F. Parsons. 1990. Conservation biology of the southern Australian daisy Rutidosis leptorrhynchoides Pages 195-205 in T. W. Clark and J. H. Seebeck, editors. Management and conservation of small populations. Chicago Zoological Society, Illinois. Schaller, G. B. 1967. The Deer and the Tiger: a study of wildlife in India. The University of Chicago Press, Chicago and London. Schleuter, D., and R. Eckmann. 2007. Generalist versus specialist: the performances of perch and ruffe in a lake of low productivity. Ecology of Freshwater Fish 17:86-99.

References 251

Schmidt, B., G. Coulson, and J. Di Stefano. 2010. Habitat partitioning among sympatric grey kangaroos and swamp wallabies in box-ironbark remnants. Pages 219-230 in G. Coulson and M. Eldridge, editors. Macropods: The biology of kangaroos, wallabies and rat-kangaroos. CSIRO Publishing, Collingwood, Victoria. Schmitz, O. 1998. Direct and indirect effects of predation and predation risk in old-field interaction webs. American Naturalist 151:327-342. Schoener, T. S. 1989. Food webs from the small to the large. Ecology 70:1559- 1589. Schoener, T. W. 1974a. Competition and the form of habitat shift. Theoretical Population Biology 6:265-307. Schoener, T. W. 1974b. Resource partitioning in ecological communities. Science 185:27-39. Schoener, T. W. 1982. The controversy over interspecific competition. American Scientist 70:586-595. Schoener, T. W. 1983. Field experiments on interspecific competition. The American Naturalist 122:240-285. Schoener, T. W. 1986. Resource partitioning. Pages 91-126 in J. Kikkawa and D. J. Anderson, editors. Community Ecology. Patterns and Process. Blackwell Scientific Publications, Palo Alto, CA. Scholes, R. J., and S. R. Archer. 1997. Tree-grass interactions in savannas. Annual Review of Ecology and Systematics 28:517-544. Schwartz, C. C., and J. E. Ellis. 1981. Feeding ecology and niche separation in some native and domestic ungulates on the shortgrass pairie. Journal of Applied Ecology 18:343-353. Scognamillo, D., I. Maxit, M. Sunquist, and J. Polisar. 2003. Coexistence of jaguar (Puma onca) and puma (Puma concolor) in a mosaic landscape in the Venezuelan llanos. Journal of Zoology, London 259:269-279. Senft, R. L., L. R. Rittenhouse, and R. G. Woodmansee. 1985. Factors influencing patterns of cattle grazing behaviour on shortgrass steppe. Journal of Range Management 38:82-87. Short, J., and G. C. Grigg. 1982. The abundance of kangaroos in suboptimal habitats: wheat, intensive pastoral, and mallee. Australian Wildlife Research 9:221–227.

References 252

Shrestha, R., P. Wegge, and R. A. Koirala. 2005. Summer diets of wild and domestic ungulates in Nepal Himalaya. Journal of the Zoological Society of London 266:111-119. Sih, A. 1993. Effects of ecological interactions on forager diets: competition, predation risk, parasitism and prey behaviour. Pages 182-211 in R. N. Hughes, editor. Diet Selection: An Interdisciplinary Approach to Foraging Behaviour. Blackwell Scientific Publications, Oxford. Sinclair, A. R. E., and M. Norton-Griffiths. 1982. Does competition or facilitation regulate migrant ungulate populations in the Serengeti? A test of hypotheses. Oecologia 53:364-369. Skerratt, L. F., J. H. L. Skerratt, S. Banks, R. Martin, and K. Handasyde. 2004. Aspects of the ecology of common wombats (Vombatus ursinus) at high density on pastoral land in Victoria. Australian Journal of Zoology 52:303- 330. Smith, A. D. 1964. Defecation rates of mule deer. Journal of Wildlife Management 28:435-444. Smith, E. P. 1982. Niche breadth, resource availability, and inference. Ecology 63:1675-1681. Smith, E. P. 2002. BACI design. Pages 141-148 in A. H. El-Shaarawi and W. W. Piegorsch, editors. Encyclopedia of Environmetrics. John Wiley & Sons Ltd, Chichester. Sokal, R. R., and F. J. Rohlf. 1995. Biometry: the principles and practice of statistics in biological research. Third edition. W. H. Freeman and Co., New York. Southwell, C. 1987. Activity pattern of the Eastern Grey Kangaroo. Mammalia 51:211-223. Southwell, C. 1989. Techniques for monitoring the abundance of kangaroo and wallaby populations. Pages 659-693 in Kangaroos, Wallabies and Rat- kangaroos. Surrey Beatty and Sons Pty Limited, Sydney. Southwell, C. J. 1984. Variability in grouping in the Eastern Grey Kangaroo, Macropus giganteus. I. Group density and group size. Australian Wildlife Research 11:424-436.

References 253

Southwell, C. J., S. C. Cairns, A. R. Pople, and R. Delaney. 1999. Gradient analysis of macropod distribution in open forest and woodland of eastern Australia. Australian Journal of Ecology 24:132–143. Southwell, C. J., and P. Jarman. 1987. Macropod studies at Wallaby Creek. III The effect of fire on pasture utilization by macropods and cattle. Australian Wildlife Research 14:117-124. Stephens, D. W., and J. R. Krebs. 1986. Foraging Theory. Princeton University Press, Princeton, New Jersey. Stewart-Oaten, A., and J. R. Bence. 2001. Temporal and spatial variation in environmental assessment. Ecological Monographs 71:305–339. Stewart, A. 1982. Evaluation of kangaroo pellet counts. Bulletin of the Australian Mammal Society 7:25. Stott, P. 2003. Use of space by sympatric European hares (Lepus europaeus) and European rabbits (Oryctolagus cuniculus) in Australia. Mammalian Biology 68:317-327. Strahan, R. e. 1995. The mammals of Australia. Reed Books, Australia. Svarsden, G. 1949. Competition and habitat selection in birds. Oikos 1:157-174. Swan, M., J. Di Stefano, A. Greenfield, and G. Coulson. 2008. Fine-scale habitat selection by adult female swamp wallabies (Wallabia bicolor). Australian Journal of Zoology 56:305-309. Taylor, P. G. 1971. Aspects of the biology of the hog deer (Axis porcinus Zimmerman 1780). PhD thesis. Monash University, Melbourne. Taylor, R. 1984. Foraging in the eastern grey kangaroo and the wallaroo. Journal of Animal Ecology 53:65-74. Taylor, R. 1985a. Effects of pasture improvement on the nutrition of eastern grey kangaroos and wallaroos. Journal of Applied Ecology 22:717-725. Taylor, R. H., and R. M. Williams. 1956. The use of pellet counts for estimating the density of populations of the wild rabbit, Oryctolagus cuniculus (L.). New Zealand Journal of Science and Technology 38B:236-256. Taylor, R. J. 1980. Distribution of feeding activity of the eastern grey kangaroo, Macropus giganteus, in coastal lowland of south-east Queensland. Australian Wildlife Research 7:317-325.

References 254

Taylor, R. J. 1983. The diet of the eastern grey kangaroo and wallaroo in areas of improved and native pasture in the New England Tablelands. Australian Wildlife Research 10:203-211. Taylor, R. J. 1985b. Habitat use by the eastern grey kangaroo and wallaroo in an area of sympatry. Mammalia 49:173-186. Taylor, R. J. 1993. Observations on the behavior and ecology of the common wombat Vombatus ursinus in northeast Tasmania. Australian Mammalogy 16:1-7. Telfer, W. R., A. D. Griffiths, and D. M. J. S. Bowman. 2008. The habitat requirements of four sympatric rock-dwelling macropods of the Australian monsoon tropics. Austral Ecology 33:1-33-1044. Thompson, H. V. 1994. The rabbit in Britain. Pages 64-98 in H. V. Thompson and C. M. King, editors. The European Rabbit: The history and biology of a successful colonizer. Oxford Scientific Publications, New York. Toda, M. J., M. T. Kimura, and N. Tuno. 1999. Coexistence mechanisms of mycophagus drosophilids on multispecies fungal hosts: aggregation and resource partitioning. Journal of Animal Ecology 68:794–803. Tonn, W. M., J. J. Magnuson, M. Rask, and J. Toivonen. 1990. Intercontinental comparison of small-lake fish assemblages: the balance between regional and local processes. The American Naturalist 136:345–375. Tozer, M. G., and R. A. Bradstock. 2002. Fire-mediated effects of overstorey on plant species diversity and abundance in an eastern Australian heath. Plant Ecology 164:213-223. Triggs, B. 1996. The Wombat: Common Wombats in Australia. Second edition. NSW University Press, Sydney. Triggs, B. 2003. Tracks, Scats and Other Traces; a field guide to Australian mammals. Oxford University Press, Melbourne. Troy, S., and G. Coulson. 1993. Home range of the swamp wallaby, Wallabia bicolor. Wildlife Research 20:571-577. Troy, S., G. Coulson, and D. Middleton. 1992. A comparison of radiotracking and line transect techniques to determine habitat preferences in the swamp wallaby, Wallabia bicolor, in south-eastern Australia. Pages 651-660 in I. G. Priede and S. M. Swift, editors. Wildlife Telemetry. Remote Monitoring and Tracking of Animals. Ellis Horwood, Chichester, U.K.

References 255

Tyndale-Biscoe, M. 1994. During burial by native and introduced dung beetles (Scarabaeidae). Australian Journal of Agricultural Research 45:1799-1808. Underwood, A. J. 1997. Experiments in Ecology. Their Logical Design and Interpretation Using Analysis of Variance. Cambridge University Press, Cambridge. University of Ballarat. 1999. Assessment of the condition of vegetation on the Yanakie Isthmus, Wilsons Promontory National Park. Report prepared for Parks Victoria by the University of Ballarat, Centre for Environmental Management. Van Dyke, F., and J. A. Darragh. 2006. Short- and long-term changes in elk use and forage production in sagebrush communities following prescribed burning. Biodiversity and Conservation 15:4375–4398. Van Etten, R. C., and C. L. Bennet. 1965. Some sources of error in using pellet- group counts for censusing deer. Journal of Wildlife Management 29:723- 729. Van Soest, P. J. 1994. The nutritional ecology of the ruminant. Second edition. O&B Books Inc., Corvallis, Oregon. Vernes, K. 1999. Pellet counts to estimate density of a rainforest kangaroo. Wildlife Society Bulletin 27:991-996. Villafuerte, R., M. B. Kufner, M. Delibes, and S. Moreno. 1993. Environmental factors influencing the seasonal daily activity of the European rabbit (Orycolagus cuniculus) in a Mediterranean area. Mammalia 57:341-347. Voeten, M. M., and H. H. T. Prins. 1999. Resource partitioning between sympatric wild and domestic herbivores in the Tarangire region of Tanzania. Oecologia 120:287-294. Voris, L., L. F. Marcy, E. J. Thacker, and W. W. Wainio. 1940. Digestible nutrients of feeding stuffs for the domestic rabbit. Journal of Agricultural Research 61: 673-684. Wahren, H. A., W. A. Papst, and R. J. Williams. 2001. Early post-fire regeneration in subalpine heathland and grassland in the Victorian Alpine National Park, south-eastern Australia. Austral Ecology 26:670-679. Wallmo, O., C. Jackson, T. L. Hartley, and R. L. Carlisle. 1962. Influence of rain on the count of deer pellet groups. Journal of Wildlife Management 26:50- 55.

References 256

Walters, C. J., and C. S. Holling. 1990. Large-scale management experiments and learning by doing. Ecology 71:2060-2068. Waters, R. 1985. Aspects of the ecology of the swamp wallaby, Wallabia biocolor (Desmarest), in E. regnans (F. Muell.) and P. radiata (D. Don) plantations. MSc thesis. Monash University, Melbourne. Wegge, P., A. K. Shrestha, and S. R. Moe. 2006. Dry season diets of sympatric ungulates in lowland Nepal: competition and facilitation in alluvial tall grasslands. Ecological Research 21:698-706. Wells, R. T. 1989. The Vombatidae. Pages 755-768 in D. W. Walton and B. J. Richardson, editors. Fauna of Australia. Australian Government Publishing Service, Canberra. Wells, R. T., and B. Green. 1998. Aspects of water metabolism in the southern hairy-nosed wombat Lasiorhinus latifrons Pages 61-66 in R. T. Wells and P. A. Pridmore, editors. Wombats. Surrey Beatty, Chipping Norton. Werner, E. E. 1977. Species packing and niche complementarity in three sunfishes. American Naturalist 111:553-578. Wheeler, A. G., and M. C. Calver. 1996. Resource partitioning in an island community of insectivorous birds during winter. EMU 96:23–31. Whelan, J. 2008. Are kangaroos indigenous to Wilsons Promontory National Park? The Victorian Naturalist 125:172-177. Whelan, R. J. 1995. The ecology of fire. Cambridge University Press, Cambridge. Whitehead, P. J., J. Russell-Smith, and J. C. Z. Woinarski. 2005. Fire, landscape heterogeneity and wildlife management in Australia’s tropical savannas: introduction and overview. Wildlife Research 32:369–375. Whitfield, J. 2002. Neutrality versus the niche. Nature 417:480-481. Williams, C. K., and K. Myers. 2008. Family Laeporidae rabbit and hare. Pages 745-747 in S. Van Duck and R. Strahan, editors. The Mammals of Australia. Reed New Holland, Sydney. Williams, K., I. Parer, B. Coman, J. Burley, and M. Braysher. 1995. Managing Vertebrate Pests: Rabbits. Bureau of Rural Sciences, Canberra. Williamson, M. 1996. Biological Invasions. Chapman & Hall, London. Wood, D. H. 1988. Estimating rabbit density by counting dung pellets. Australian Journal of Wildlife Research 15:665-671.

References 257

Woolnough, A. P., and C. N. Johnson. 2000. Assessment of the potential for competition between two sympatric herbivores - the northern hairy-nosed wombat, Lasiorhinus krefftii, and the eastern grey kangaroo, Macropus giganteus. Wildlife Research 27:301-308. Zapata, S., A. Travaini, M. Delibes, and R. Martinez-Peck. 2005. Food habits and resource partitioning between grey and culpeo foxes in southeastern Argentine Patagonia. Studies on Neotropical Fauna and Environment 40:97-103.

References

Minerva Access is the Institutional Repository of The University of Melbourne

Author/s: Davis, Naomi Ezra

Title: Resource partitioning among five sympatric mammalian herbivores on Yanakie Isthmus, south-eastern Australia

Date: 2010

Citation: Davis, N. E. (2010). Resource partitioning among five sympatric mammalian herbivores on Yanakie Isthmus, south-eastern Australia. PhD thesis, Science - Zoology, The University of Melbourne.

Publication Status: Unpublished

Persistent Link: http://hdl.handle.net/11343/35801

File Description: Resource partitioning among five sympatric mammalian herbivores on Yanakie Isthmus, south-eastern Australia

Terms and Conditions: Terms and Conditions: Copyright in works deposited in Minerva Access is retained by the copyright owner. The work may not be altered without permission from the copyright owner. Readers may only download, print and save electronic copies of whole works for their own personal non-commercial use. Any use that exceeds these limits requires permission from the copyright owner. Attribution is essential when quoting or paraphrasing from these works.